Compare commits

...
294 Commits
Author SHA1 Message Date
zemion fa2d5d40dd Release Core v0.1.33 with redirect-sensitive headers
Module Package Release / publish-packages (push) Successful in 13s
2026-08-22 16:11:02 +02:00
zemion 6ccef162f6 Release Core v0.1.32 with bounded HTTP request bodies
Module Package Release / publish-packages (push) Successful in 14s
2026-08-22 14:32:24 +02:00
zemion 48dac139a5 feat(wiki): compose governed Wiki WebUI
Module Package Release / publish-packages (push) Successful in 12s
2026-08-22 13:32:20 +02:00
zemion a090e5af20 feat(tickets): add optional integration contracts and WebUI composition
Module Package Release / publish-packages (push) Successful in 13s
2026-08-22 12:06:04 +02:00
zemion 0c1358b862 feat(policy): add delegation and escalation contracts
Module Package Release / publish-packages (push) Failing after 6s
2026-08-22 03:12:30 +02:00
zemion a9035c4c3b feat: add Campaign work orchestration contract
Module Package Release / publish-packages (push) Failing after 6s
2026-08-22 02:15:32 +02:00
zemion 137c7c005f test(core): recognize Campaign work integrations
Module Package Release / publish-packages (push) Successful in 12s
2026-08-22 01:06:12 +02:00
zemion 8eeea968f2 chore(release): load Campaign 0.1.22
Module Package Release / publish-packages (push) Successful in 12s
2026-08-22 01:01:35 +02:00
zemion 0ca6568005 chore(release): load Campaign 0.1.21
Module Package Release / publish-packages (push) Successful in 12s
2026-08-22 00:54:00 +02:00
zemion af90db44c9 chore(webui): load Campaign collaboration release
Module Package Release / publish-packages (push) Successful in 12s
2026-08-22 00:18:06 +02:00
zemion 5de46e9c0e chore(webui): load Files folder sync release
Module Package Release / publish-packages (push) Successful in 12s
2026-08-21 22:50:48 +02:00
zemion 1d9b677c1b chore(webui): load IDM relationship administration release
Module Package Release / publish-packages (push) Successful in 12s
2026-08-21 22:06:29 +02:00
zemion 54178ee56c chore(webui): load Access credential help release
Module Package Release / publish-packages (push) Successful in 12s
2026-08-21 21:15:31 +02:00
zemion 10e7597612 feat(core): define datasource visibility policy contract
Module Package Release / publish-packages (push) Failing after 6s
2026-08-21 20:31:51 +02:00
zemion 142ccbc587 feat(core): define managed tabular file contract
Module Package Release / publish-packages (push) Successful in 12s
2026-08-21 19:29:48 +02:00
zemion f75ad48d78 test(core): isolate entitlement navigation fixture 2026-08-21 19:26:00 +02:00
zemion 5a9e8f79f9 test(webui): cover contextual help in the browser 2026-08-21 17:36:06 +02:00
zemion fbea74a74b feat(core): define durable datasource artifact contract 2026-08-21 17:36:05 +02:00
zemion 925dc33696 feat(docs): define semantic subject contract 2026-08-21 14:55:58 +02:00
zemion 4b0737e1cd feat: support autonomous campaign schedule results 2026-08-20 21:10:04 +02:00
zemion 4f4007aff1 feat: define policy impact subject contract 2026-08-20 20:27:17 +02:00
zemion 8e687c4420 feat: define bulk governance projection contract 2026-08-20 19:46:53 +02:00
zemion 604f20eed7 feat: classify audit evidence export scope 2026-08-20 18:00:18 +02:00
zemion 6a2da94e47 feat: load connector governance webui 2026-08-20 12:46:01 +02:00
zemion e121ca900e feat: load identity administration webui 2026-08-20 12:27:37 +02:00
zemion 79629c5a2c feat: define campaign archive encryption policy contract 2026-08-20 12:12:27 +02:00
zemion 026e451aa4 feat: configure scheduling self-enrollment limits 2026-08-20 11:45:07 +02:00
zemion f11c675d11 feat: govern accessible appearance overrides 2026-08-20 10:50:55 +02:00
zemion 0fae09ba3c feat: extend calendar scheduling reconciliation contract 2026-08-20 07:44:16 +02:00
zemion 8f642bd618 feat: govern effective appearance defaults 2026-08-20 07:03:28 +02:00
zemion 6643c8fc1e feat: add validated appearance palettes 2026-08-20 06:26:52 +02:00
zemion fd90b60430 feat: centralize access explanation subject selection 2026-08-20 06:16:43 +02:00
zemion 0aae6f0539 feat(postbox): schedule lifecycle notification reconciliation 2026-08-20 04:58:35 +02:00
zemion be7b79612c feat(postbox): expose governed grouping policy 2026-08-20 04:22:29 +02:00
zemion 557c77670b feat(postbox): expose configurable protection contracts 2026-08-20 03:42:58 +02:00
zemion d277218784 feat(webui): preserve state in shared widget launches 2026-08-20 02:39:41 +02:00
zemion cf16a7b27a feat(core): add layered side rail preferences 2026-08-20 01:47:04 +02:00
zemion 51bf14f376 docs(core): define collaboration connector strategy 2026-08-19 23:15:35 +02:00
zemion 8d9bcfd8b5 test(core): prove Views cannot grant access 2026-08-19 22:55:19 +02:00
zemion 3c7a593f63 feat(templates): carry content field requirements 2026-08-19 21:54:51 +02:00
zemion 9d1352ba30 feat(mail): define standard IMAP folder mappings 2026-08-19 21:27:42 +02:00
zemion 4cf2bfeb3e feat(operations): add runtime work status contract 2026-08-19 20:52:44 +02:00
zemion 94c94fefb4 test(webui): cover View-focused Quick Access catalogues 2026-08-19 20:14:02 +02:00
zemion d600bca374 test: make full discovery optional-state independent 2026-08-19 19:20:05 +02:00
zemion ffaab543d2 feat(webui): govern actions, quick access, and metrics
Refs #264, #285, #289
2026-08-19 18:47:46 +02:00
zemion 41db78c201 Define semantic page archetypes 2026-08-19 14:26:25 +02:00
zemion c042244da8 Define semantic page action layouts 2026-08-19 13:33:51 +02:00
zemion 5a2e99f496 feat: add status and payment capability contracts 2026-08-19 12:33:34 +02:00
zemion 8a925782ab feat: add UI conformance and launch context 2026-08-18 21:32:25 +02:00
zemion 7685a103e8 Centralize shared WebUI structural primitives 2026-08-18 13:17:31 +02:00
zemion ee5c881df9 Centralize metric and description layouts 2026-08-18 11:30:39 +02:00
zemion 6814a41ae4 Add shared WebUI layout primitives 2026-08-18 10:42:55 +02:00
zemion dd7ad4d9c7 feat: add shared workspace layouts 2026-08-18 02:17:13 +02:00
zemion 934db6d44b feat: centralize shared page layouts 2026-08-18 01:03:33 +02:00
zemion d307e29145 feat: add exact credential help contexts 2026-08-17 19:51:51 +02:00
zemion ff88142471 feat: add exact retention help contexts 2026-08-17 17:41:45 +02:00
zemion 887e9beb9e docs: update meta documentation links 2026-08-17 16:52:53 +02:00
zemion e6457b3f6b feat: add governed DSAR workflow 2026-08-07 14:53:07 +02:00
zemion eb0c01c5d2 feat: expose typed deployment capability receipts 2026-08-07 11:15:49 +02:00
zemion 40cc012124 feat(modules): validate catalog permission declarations 2026-08-07 01:59:24 +02:00
zemion 44196f5620 Validate module catalog provenance and availability 2026-08-06 22:42:08 +02:00
zemion 9ceb1b8c22 Add trusted public module catalog installs 2026-08-06 21:12:54 +02:00
zemion 32c234fbdb Add product presentation extension contracts 2026-08-06 19:02:53 +02:00
zemion d65d7a8e5f Add shared work-item provider contracts 2026-08-06 16:06:17 +02:00
zemion b5f5be15f6 Add governed form evidence contract 2026-08-06 12:42:20 +02:00
zemion f5949427cc feat(records): define archive transfer provider contract 2026-08-06 05:36:11 +02:00
zemion 5d1287735e feat(core): add provider-neutral records filing contract 2026-08-06 01:43:08 +02:00
zemion 7ea0cb8655 Fix responsive DataGrid contraction 2026-08-05 22:42:21 +02:00
zemion b553513c9f Release v0.1.18
Module Package Release / publish-packages (push) Successful in 12s
2026-08-05 21:08:17 +02:00
zemion b5a4eb177a Release v0.1.17
Module Package Release / publish-packages (push) Successful in 12s
2026-08-05 20:34:47 +02:00
zemion f1a5be2a93 Release v0.1.16
Module Package Release / publish-packages (push) Successful in 12s
2026-08-05 19:53:46 +02:00
zemion add7a99f6d feat: add temporal context and contextual help 2026-08-05 00:03:31 +02:00
zemion 982ef636b8 Release v0.1.15
Module Package Release / publish-packages (push) Successful in 13s
2026-08-04 15:22:52 +02:00
zemion 9aad49f16d Make package publication retries hash-safe 2026-08-04 14:19:04 +02:00
zemion 2b5c14385d Add Voting provider assurance contract 2026-08-04 14:01:05 +02:00
zemion bca3e46293 Publish npm artifacts from explicit local paths 2026-08-04 14:00:57 +02:00
zemion f09d2bf9df fix(ci): normalize historical package refs 2026-08-04 13:34:00 +02:00
zemion 702421be48 fix(ci): rely on protected release boundary 2026-08-04 13:32:28 +02:00
zemion bb471df21c fix(ci): bind Gitea action context explicitly 2026-08-04 13:28:28 +02:00
zemion bfb0d7d7c9 Allow Dataflow datasets to pin published runs 2026-08-04 12:48:37 +02:00
zemion 1974bf1a2b Define bounded tabular source contracts 2026-08-04 12:05:09 +02:00
zemion 0c9bf6758c Add secure password generator 2026-08-04 11:07:55 +02:00
zemion 25da7d49a9 Add controlled first-admin enrollment 2026-08-04 10:07:10 +02:00
zemion 40c10089ab Enforce tenant module entitlements beyond requests 2026-08-04 09:29:36 +02:00
zemion d6e7c8b0b1 Add governed module and interface controls 2026-08-04 05:20:47 +02:00
zemion 5bc7d748f8 Add protected package release workflow 2026-08-04 04:14:03 +02:00
zemion 7117673ecc Add durable search event indexing contract 2026-08-04 03:03:15 +02:00
zemion 14351b0c94 Register trust and encryption WebUI modules 2026-08-04 01:27:58 +02:00
zemion ad57fad1ea Reject duplicate module migration revisions 2026-08-03 15:04:18 +02:00
zemion fa32cca03f Complete guided Core configuration patterns 2026-08-03 10:16:26 +02:00
zemion 2d0551a845 Explain disabled mail connection tests 2026-08-03 10:05:59 +02:00
zemion bb84122061 Expose object modification evidence 2026-08-03 09:23:15 +02:00
zemion b823a22b9b Link contextual guidance to Docs 2026-08-03 07:33:57 +02:00
zemion 70fc6da811 Localize shared blocker guidance 2026-08-03 07:22:16 +02:00
zemion 729b84d3af Fence and reconcile module lifecycle effects 2026-08-03 07:02:07 +02:00
zemion b962f6756e Extend action recovery reconciliation 2026-08-03 06:37:45 +02:00
zemion 79d00b84e3 Commit verified recovery projections atomically 2026-08-03 06:09:52 +02:00
zemion 842be5edb5 Commit atomic domain and recovery state together 2026-08-03 05:43:39 +02:00
zemion 7e59a7f2b3 Resolve unknown recovery outcomes from evidence 2026-08-03 05:00:08 +02:00
zemion 5bfbe9a887 Bridge legacy WebUI router imports 2026-08-03 03:59:08 +02:00
zemion 01f91154e0 Add durable external-effect runtime identity 2026-08-03 03:47:41 +02:00
zemion 6c2940aebc Refresh shared contract test fixtures 2026-08-03 03:07:59 +02:00
zemion 670693bde8 Support repository-root WebUI packages 2026-08-03 03:07:51 +02:00
zemion bca6a7c8aa Implement durable module recovery operations 2026-08-03 03:07:42 +02:00
zemion 21c1fa49b6 Add worker delivery acceptance probe 2026-08-03 00:20:51 +02:00
zemion 435b924fd9 Extend calendar invitation capability contract 2026-08-02 16:38:34 +02:00
zemion af5c6af0e7 Define connector runtime preview contract 2026-08-02 14:54:44 +02:00
zemion c6ef644842 Add typed IDM relationship contracts 2026-08-02 14:44:41 +02:00
zemion 5783d43547 Validate Campaign template contracts 2026-08-02 13:58:49 +02:00
zemion e4d2d10c7e Add template and generated artifact contracts 2026-08-02 12:37:58 +02:00
zemion fe62fd4644 Validate Campaign Distribution Lists contract 2026-08-02 11:53:36 +02:00
zemion 9ecdc6d713 Add contact point resolution contract 2026-08-02 06:20:18 +02:00
zemion ca35aad286 Add external calendar profile contract 2026-08-02 05:59:06 +02:00
zemion d6255f9f8f feat: harden multi-host runtime coordination 2026-08-02 05:30:11 +02:00
zemion b58c9c55cf feat: define governed report provider contract 2026-08-02 05:30:06 +02:00
zemion 3c4bcc28f1 refactor: consolidate shared catalog and API helpers 2026-08-02 05:30:02 +02:00
zemion 972c681650 feat: finalize encryption and voting provider contracts 2026-08-02 03:40:44 +02:00
zemion 2b4eb0151f Add governed institutional capability contracts 2026-08-01 20:57:25 +02:00
zemion 7192d32e65 feat: add institutional governance and recovery contracts 2026-08-01 17:46:54 +02:00
zemion b65b48832b Add cross-module operational and interaction contracts 2026-07-31 22:48:07 +02:00
zemion 4cb334c912 Add distribution audience contracts and module wiring 2026-07-31 20:59:49 +02:00
zemion 6ebb299d6c Add workflow baseline orchestration contracts 2026-07-31 19:39:59 +02:00
zemion 42b5019464 Extend Postbox message authoring contract 2026-07-31 18:40:30 +02:00
zemion 4c0e6435ab Fix IDM lifecycle capability identifier 2026-07-31 18:40:30 +02:00
zemion e37d8fee94 Extend Postbox access classification contract 2026-07-31 18:21:36 +02:00
zemion 50a8d459e7 Add IDM assignment lifecycle worker contract 2026-07-31 18:07:36 +02:00
zemion 5ee85d07d6 Define governed View projection contract 2026-07-31 17:52:57 +02:00
zemion cf01545806 Allow session-aware definition policy resolution 2026-07-31 17:34:16 +02:00
zemion 1884274f8d feat: add workflow engine contribution contracts 2026-07-31 16:58:57 +02:00
zemion 5211e07d0b fix: enforce DataGrid cover boundary 2026-07-31 04:44:24 +02:00
zemion 7b8072d049 feat: complete shared platform UI contracts 2026-07-31 04:21:34 +02:00
zemion 5b55f59a92 Add shared automation and WebUI editing primitives 2026-07-31 02:48:56 +02:00
zemion f0898fcdee feat: add governed ownership workflows and admin tree navigation 2026-07-30 17:42:05 +02:00
zemion 9b88ae388b feat: harden shared platform contracts 2026-07-30 14:26:36 +02:00
zemion 6970bf7457 Support observable grid queries and file filters 2026-07-30 05:22:09 +02:00
zemion 47e106684d perf(webui): lazily load module descriptors 2026-07-30 04:35:57 +02:00
zemion 9e219bc4d3 feat(core): dispatch bounded postbox routes 2026-07-30 03:59:38 +02:00
zemion ea436a513f feat(core): define organization hierarchy contracts 2026-07-30 03:26:19 +02:00
zemion e7c84e3227 feat(core): reconcile durable workflow instances 2026-07-30 03:09:58 +02:00
zemion cf7afe9dda feat(core): dispatch durable dataflow runs 2026-07-30 02:32:49 +02:00
zemion ca8a8c5111 feat(core): define sanctions screening gate contract 2026-07-30 01:54:58 +02:00
zemion f3b388fe7e perf(core): define bounded reference search 2026-07-30 01:29:45 +02:00
zemion af3e0a055d perf(core): batch tenant summary providers 2026-07-30 01:15:04 +02:00
zemion 51d4032b86 docs(core): define compatibility retention policy 2026-07-30 01:03:39 +02:00
zemion 0beb9ffea9 build: reject duplicate generated translations 2026-07-30 00:40:25 +02:00
zemion 9e6a6b5fdc feat(webui): center optional global search 2026-07-29 22:00:45 +02:00
zemion 48fb953b93 feat: add durable auth principal cache revisions 2026-07-29 19:23:52 +02:00
zemion 4bde0495f7 feat: expose workflow view resolution input 2026-07-29 19:09:03 +02:00
zemion a80caf7933 feat: support focused dev reload scopes 2026-07-29 18:52:55 +02:00
zemion 790790ab37 feat: expose sanctions screening integration 2026-07-29 18:46:53 +02:00
zemion 920e3c9834 feat: version search source contracts 2026-07-29 18:08:52 +02:00
zemion 13893c80cd feat: version automation principal subjects 2026-07-29 17:48:36 +02:00
zemion a192a2215f Add durable platform event delivery contract 2026-07-29 17:34:52 +02:00
zemion e8fed6d25a fix: resolve postbox inbox icon 2026-07-29 16:01:45 +02:00
zemion d9b5708df0 feat: add search and external integration contracts 2026-07-29 15:50:08 +02:00
zemion 68328f3d8e feat: strengthen module contracts and shared WebUI runtime 2026-07-29 14:16:28 +02:00
zemion 53e947935a fix: standardize direct page scroll viewports 2026-07-28 22:50:11 +02:00
zemion 324c26da78 fix: allow fallback dashboard scrolling 2026-07-28 22:13:22 +02:00
zemion 389f98e349 Keep normalized View roots acyclic 2026-07-28 21:32:20 +02:00
zemion ce9ef8d88f Add governed View surface runtime 2026-07-28 21:04:54 +02:00
zemion 13bc3d3b4e Add shared credential envelope infrastructure 2026-07-28 19:32:41 +02:00
zemion 3f5870281a Color all shared alert tones 2026-07-28 18:35:53 +02:00
zemion a46df85479 Resolve XyFlow styles for linked modules 2026-07-28 15:47:15 +02:00
zemion c31581b1b9 Prebundle XyFlow for linked module development 2026-07-28 15:39:00 +02:00
zemion 26ae034153 Add governed automation contracts 2026-07-28 15:02:42 +02:00
zemion baa2143a26 feat: add dataflow publication contracts and workflow webui 2026-07-28 13:47:50 +02:00
zemion 8b1910b5b7 feat: add datasource and definition graph contracts 2026-07-28 12:42:49 +02:00
zemion d36bb94335 Add provider-neutral tabular source contracts 2026-07-28 11:12:24 +02:00
zemion 74034947c6 Update vulnerable PostCSS dependency 2026-07-28 01:36:39 +02:00
zemion c7183fe7f1 Integrate Dataflow module into core 2026-07-28 01:33:37 +02:00
zemion 139a352c80 chore: update GovOPlaN repository references 2026-07-27 15:46:51 +02:00
zemion 336c94137f chore(release): align Mail bundle with Campaign contract 2026-07-23 00:47:34 +02:00
zemion 93225b6487 docs: move system status badges to meta repository 2026-07-22 23:45:42 +02:00
zemion e11ea81008 chore(release): prepare Core 0.1.14 2026-07-22 20:31:49 +02:00
zemion bc8afeb139 test(campaign): align synchronous send security contract 2026-07-22 20:30:00 +02:00
zemion f876345656 test(db): prove PostgreSQL retirement atomicity 2026-07-22 15:28:00 +02:00
zemion d487726f4d chore(release): bump Core to 0.1.13 2026-07-22 10:40:27 +02:00
zemion e6fc07da37 chore(release): bundle Campaign 0.1.10 2026-07-22 10:38:37 +02:00
zemion e6d589eb07 fix(release): package Core migration runtime 2026-07-22 10:34:34 +02:00
zemion 59610e21d2 chore(release): record reviewed 0.1.12 migration heads 2026-07-22 09:06:56 +02:00
zemion cece71d945 feat(webui): synchronize external DataGrid queries 2026-07-22 09:03:11 +02:00
zemion 22e8183846 fix(webui): translate MetricCard content 2026-07-22 08:48:42 +02:00
zemion aa111a5fe1 chore(release): bump Core to 0.1.12 2026-07-22 08:41:54 +02:00
zemion e6062fe9e4 fix(webui): enforce full-result DataGrid queries 2026-07-22 08:05:11 +02:00
zemion 987ca894ed chore(release): record reviewed 0.1.11 migration heads 2026-07-22 04:42:30 +02:00
zemion 4caa326878 chore(core): bump version to 0.1.11 2026-07-22 03:41:24 +02:00
zemion 8c4c4456c6 feat(core): define auditable poll response retirement 2026-07-22 03:31:05 +02:00
zemion 6abe292ac8 feat(core): define governed poll participation contract 2026-07-22 03:21:03 +02:00
zemion fea2807754 feat(core): define atomic poll option ordering 2026-07-22 03:20:28 +02:00
zemion 22646c614c feat(core): add bounded people picker foundation 2026-07-22 03:01:56 +02:00
zemion 17376332a2 feat(core): configure bounded scheduling cancellation notices 2026-07-22 02:58:09 +02:00
zemion 0946bc84a9 feat(core): support explicit public module routes 2026-07-22 02:58:03 +02:00
zemion a18499cbb5 feat(core): require scoped user workflows 2026-07-22 01:55:24 +02:00
zemion 36d7b73bb5 fix(webui): allow card content to overflow 2026-07-22 01:49:24 +02:00
zemion b89a2d15f1 chore(core): bump version to 0.1.10 2026-07-21 20:47:54 +02:00
zemion 7f923afdad docs(core): refine function-bound postbox encryption 2026-07-21 20:47:54 +02:00
zemion a7683c5d4a feat(core): add resilient fixed-window throttling 2026-07-21 20:47:54 +02:00
zemion 41ad057f7e feat(webui): standardize discard and table actions 2026-07-21 20:47:54 +02:00
zemion bf0729eb59 feat(core): add authenticated baseline role templates 2026-07-21 20:47:54 +02:00
zemion c4b90181e0 fix(webui): localize contextual Mail help 2026-07-21 19:15:56 +02:00
zemion 55ed194a99 fix(webui): respect configured documentation access 2026-07-21 19:01:00 +02:00
zemion b3b0cf0fca feat(webui): open contextual configured handbooks 2026-07-21 18:42:31 +02:00
zemion fa9119bea7 security(webui): block remote mail preview content 2026-07-21 17:52:02 +02:00
zemion 70ca772138 test: gate Mail on Campaign access interface 2026-07-21 17:51:35 +02:00
zemion 2eae5c4df6 test: align module contracts with Mail-owned delivery 2026-07-21 17:15:19 +02:00
zemion 57fe6c6006 security(connectors): reject tunneled metadata addresses 2026-07-21 16:54:10 +02:00
zemion 713afdb39b fix(installer): enlist table retirement transaction 2026-07-21 16:44:20 +02:00
zemion 77f8d15d17 docs: clarify fail-closed connector SDKs 2026-07-21 15:53:22 +02:00
zemion fda99d40eb docs: clarify institutional identity ownership 2026-07-21 15:46:55 +02:00
zemion 5ab1af803b docs: align technical roadmap to reference program 2026-07-21 15:45:39 +02:00
zemion 0845e99cf6 Block SDK-managed secondary connector peers 2026-07-21 15:43:04 +02:00
zemion 28a0a596a6 Fail closed for unpinned connector transports 2026-07-21 15:36:32 +02:00
zemion ae74189588 fix(webui): complete central admin control styles 2026-07-21 14:01:56 +02:00
zemion 09b5009187 feat(webui): extend shared explorer tree actions 2026-07-21 13:58:38 +02:00
zemion 2ca61059dc refactor(webui): describe organization function actions 2026-07-21 13:51:19 +02:00
zemion 865901f090 Centralize shared layout style contracts 2026-07-21 13:46:59 +02:00
zemion 2ac1e64daa refactor(webui): share outside-dismiss behavior 2026-07-21 13:35:09 +02:00
zemion 7526c5ebb2 fix(webui): centralize form control layout 2026-07-21 13:31:32 +02:00
zemion 8e1f64c790 feat(webui): add central selection list 2026-07-21 13:30:39 +02:00
zemion 66e4783d2e feat(webui): add central icon button 2026-07-21 13:23:59 +02:00
zemion 7af86b42eb Centralize explorer work-surface styling 2026-07-21 13:19:20 +02:00
zemion ad202f1267 Centralize shared small-note styling 2026-07-21 13:19:20 +02:00
zemion 6526f37aae feat(webui): centralize resource access explanations 2026-07-21 13:18:30 +02:00
zemion 9dabd9356d Document Core test bootstrap lint exemptions 2026-07-21 13:11:24 +02:00
zemion 6502775bf7 Block connector limited-broadcast targets 2026-07-21 13:02:29 +02:00
zemion b2492b820f Harden private connector address validation 2026-07-21 12:55:01 +02:00
zemion 78d9ae48b2 feat(webui): centralize disabled button reasons 2026-07-21 12:38:45 +02:00
zemion 4cb3e94de3 feat(webui): expose central pagination bar 2026-07-21 12:36:49 +02:00
zemion 9131838b98 refactor(webui): generalize central selection lists 2026-07-21 12:34:35 +02:00
zemion 8e9eb6e1f5 feat(webui): centralize contextual table actions 2026-07-21 12:22:35 +02:00
zemion 249bf63eb8 fix(release): remove unavailable tagged webui packages 2026-07-21 12:22:07 +02:00
zemion 248e3dc70e test(release): isolate catalog contract projection 2026-07-21 12:21:38 +02:00
zemion 230ecf42b0 Enforce deployment security boundaries 2026-07-21 12:10:05 +02:00
zemion 825791e9b0 Harden outbound connector transports 2026-07-21 12:09:44 +02:00
zemion 7184b6cdd6 Dispose replaced migration database handles 2026-07-21 03:22:59 +02:00
zemion ea8c600dce Cover Notifications in WebUI permutations 2026-07-21 03:18:08 +02:00
zemion 1839693575 Manage nested dialogs through a central stack 2026-07-21 03:18:08 +02:00
zemion 183bf7aef0 Ignore generated file drop test builds 2026-07-21 03:18:08 +02:00
zemion 37a5dfb182 Use managed Files in Campaign smoke fixtures 2026-07-21 03:18:08 +02:00
zemion 844f934379 Redact devserver database credentials 2026-07-21 03:18:08 +02:00
zemion a98475f7bc Harden installer runtime secrets 2026-07-21 03:18:08 +02:00
zemion 1153c9dd36 Clean Core security audit findings 2026-07-21 03:18:07 +02:00
zemion 7eef52776c docs: link product vision to technical roadmap 2026-07-20 20:48:47 +02:00
zemion c50ce58ad8 refactor(webui): clarify shared controls and status feedback 2026-07-20 20:03:42 +02:00
zemion 344fc0077f test(integration): harden cross-module API coverage 2026-07-20 20:03:27 +02:00
zemion 28afc01371 fix(notifications): commit worker delivery transactions 2026-07-20 20:03:21 +02:00
zemion 1a29e75db4 feat(identity): define the search provider contract 2026-07-20 20:03:15 +02:00
zemion 57ec960f40 docs(release): record the v0.1.8 composition 2026-07-20 20:03:10 +02:00
zemion 6388afdad8 chore(release): move tooling tests to meta repository 2026-07-20 20:03:10 +02:00
zemion 1a0e90b22d fix(migrations): preserve runtime logging during upgrades 2026-07-20 20:03:02 +02:00
zemion 7ad6f6328a feat(poll): add filtered response lookup contract 2026-07-20 18:28:56 +02:00
zemion c79a7124b7 chore(core): bump contract version to 0.1.9 2026-07-20 18:27:13 +02:00
zemion abbef5a10b feat(policy): add scheduling participant privacy contract 2026-07-20 18:23:59 +02:00
zemion 2f559e3f0b feat(poll): extend response editing contract 2026-07-20 18:16:42 +02:00
zemion 9b5418db78 fix(webui): keep DataGrid actions out of form submission 2026-07-20 17:58:32 +02:00
zemion 1678602fd6 docs(ui): bind interfaces to central components 2026-07-20 17:56:18 +02:00
zemion 6e373dcdd6 feat(poll): define the scheduling capability contract 2026-07-20 17:33:47 +02:00
zemion d1c033edc7 feat(calendar): document outbox retention policy 2026-07-20 17:06:19 +02:00
zemion b5cfba666c fix(files): make dropped-file handling explicit 2026-07-20 16:57:29 +02:00
zemion 78b4afdec4 feat(calendar): define picker UI capability 2026-07-20 16:57:17 +02:00
zemion 15596f0742 feat(calendar): schedule durable outbox recovery 2026-07-20 16:57:01 +02:00
zemion a2320fcb5d docs(ui): record placement and focused-view rules 2026-07-20 16:49:57 +02:00
zemion e6f7c45f0a intermittent commit 2026-07-14 13:22:10 +02:00
zemion 8aa1943581 Register Scheduling WebUI package 2026-07-12 19:00:54 +02:00
zemion b9badc9153 Harden module installation and imports 2026-07-11 18:49:35 +02:00
zemion 9a0c467d55 Release v0.1.8 2026-07-11 17:00:37 +02:00
zemion a00ef54821 Release v0.1.7
Dependency Audit / dependency-audit (push) Successful in 1m35s
2026-07-11 03:11:06 +02:00
zemion edb4687826 Cover access canonical directory migration
Dependency Audit / dependency-audit (push) Successful in 1m33s
2026-07-11 01:13:53 +02:00
zemion fcfe0b69a3 Install cloned WebUI dependencies in release integration
Dependency Audit / dependency-audit (push) Successful in 1m38s
2026-07-11 01:00:09 +02:00
zemion 715bdcbebe Treat empty unittest discovery as release skip
Dependency Audit / dependency-audit (push) Successful in 1m28s
2026-07-11 00:43:14 +02:00
zemion 060f4da751 Add resource access explanation contract
Dependency Audit / dependency-audit (push) Has been cancelled
2026-07-11 00:39:39 +02:00
zemion c32951393a Skip empty cloned backend test suites
Dependency Audit / dependency-audit (push) Successful in 1m31s
2026-07-11 00:38:07 +02:00
zemion 7b85d6deae Install release WebUI modules as real packages
Dependency Audit / dependency-audit (push) Successful in 1m34s
2026-07-11 00:32:25 +02:00
zemion 2d2d9e7bc7 Temporarily limit heavy CI workflows to manual runs
Dependency Audit / dependency-audit (push) Has been cancelled
2026-07-11 00:25:40 +02:00
zemion dc1a250797 Install release WebUI modules sequentially in CI
Dependency Audit / dependency-audit (push) Successful in 1m36s
Release Integration / release-integration (push) Failing after 1m32s
Module Matrix / module-matrix (push) Successful in 3m54s
2026-07-11 00:23:44 +02:00
zemion 7b7cc8ada7 Fail CI after exhausted WebUI install retries
Dependency Audit / dependency-audit (push) Successful in 1m38s
Release Integration / release-integration (push) Has been cancelled
Module Matrix / module-matrix (push) Has been cancelled
2026-07-11 00:19:17 +02:00
zemion 722c9e5d1c Retry WebUI release dependency installs in CI
Dependency Audit / dependency-audit (push) Successful in 1m44s
Module Matrix / module-matrix (push) Successful in 4m9s
Release Integration / release-integration (push) Failing after 2m47s
2026-07-11 00:01:44 +02:00
zemion 63e54a67be Use checkout Alembic root in release integration
Dependency Audit / dependency-audit (push) Failing after 1m36s
Release Integration / release-integration (push) Has been cancelled
Module Matrix / module-matrix (push) Has been cancelled
2026-07-10 23:58:04 +02:00
zemion 12b623bec9 Include IDM in release integration install
Dependency Audit / dependency-audit (push) Successful in 1m47s
Module Matrix / module-matrix (push) Failing after 2m37s
Release Integration / release-integration (push) Failing after 1m30s
2026-07-10 23:36:52 +02:00
zemion b788afcae1 Harden module update compatibility cleanup
Dependency Audit / dependency-audit (push) Successful in 1m46s
Module Matrix / module-matrix (push) Failing after 2m38s
Release Integration / release-integration (push) Failing after 1m41s
2026-07-10 23:27:48 +02:00
zemion 2b0cdf13f3 Add release integration workflow
Dependency Audit / dependency-audit (push) Successful in 1m44s
Module Matrix / module-matrix (push) Successful in 4m19s
Release Integration / release-integration (push) Failing after 1m40s
2026-07-10 23:19:09 +02:00
zemion 013e0b883d Refresh WebUI release install inputs in CI
Dependency Audit / dependency-audit (push) Successful in 1m45s
Module Matrix / module-matrix (push) Successful in 4m14s
2026-07-10 22:57:01 +02:00
zemion 50f60e4d43 Install test extras in module matrix workflow
Dependency Audit / dependency-audit (push) Successful in 1m41s
Module Matrix / module-matrix (push) Has been cancelled
2026-07-10 22:54:23 +02:00
zemion ff4b514475 Align release WebUI dependency metadata
Dependency Audit / dependency-audit (push) Failing after 1m21s
Module Matrix / module-matrix (push) Failing after 1m30s
2026-07-10 22:47:38 +02:00
zemion ca1206e8c9 Harden Gitea Actions release dependency SSH
Dependency Audit / dependency-audit (push) Failing after 49s
Module Matrix / module-matrix (push) Failing after 47s
2026-07-10 22:24:47 +02:00
zemion 94236a7d7e Prepare GovOPlaN self-hosted release workflow
Dependency Audit / dependency-audit (push) Failing after 50s
Module Matrix / module-matrix (push) Failing after 49s
2026-07-10 21:57:22 +02:00
zemion 8dd5123aab Cover identity and organization API cleanup
Dependency Audit / dependency-audit (push) Failing after 1m51s
Module Matrix / module-matrix (push) Failing after 13s
2026-07-10 18:57:40 +02:00
zemion c61abe154c Add module event producer coverage
Dependency Audit / dependency-audit (push) Has been cancelled
Module Matrix / module-matrix (push) Has been cancelled
2026-07-10 18:40:38 +02:00
zemion 83784ea19d Add retention policy option regression coverage
Dependency Audit / dependency-audit (push) Has been cancelled
Module Matrix / module-matrix (push) Has been cancelled
2026-07-10 17:50:04 +02:00
zemion 81f0f649b5 Define governed platform event envelope
Dependency Audit / dependency-audit (push) Has been cancelled
Module Matrix / module-matrix (push) Has been cancelled
2026-07-10 17:48:33 +02:00
zemion 0a130962d3 Mark DataGrid layout work deferred
Dependency Audit / dependency-audit (push) Has been cancelled
Module Matrix / module-matrix (push) Has been cancelled
2026-07-10 17:43:28 +02:00
zemion 79af252e88 Decouple scopes from tenancy schema
Dependency Audit / dependency-audit (push) Has been cancelled
Module Matrix / module-matrix (push) Has been cancelled
2026-07-10 17:33:43 +02:00
zemion 635d25c74c chore: consolidate platform split checks
Dependency Audit / dependency-audit (push) Has been cancelled
Module Matrix / module-matrix (push) Has been cancelled
2026-07-10 12:51:19 +02:00
zemion 150b720f12 Release v0.1.6 2026-07-07 16:00:38 +02:00
633 changed files with 128959 additions and 15673 deletions
+28
View File
@@ -0,0 +1,28 @@
.git
.venv
node_modules
webui/node_modules
audit-reports
runtime
dist
build
coverage
htmlcov
__pycache__
*.pyc
*.pyo
*.log
.mypy_cache
.pytest_cache
.ruff_cache
.cache
.component-test-build
.module-test-build
.policy-test-build
.template-preview-test-build
.import-test-build
webui/.component-test-build
webui/.module-test-build
webui/.policy-test-build
webui/.template-preview-test-build
webui/.import-test-build
+35
View File
@@ -0,0 +1,35 @@
---
name: "Bug"
about: "Report a reproducible defect, regression, or incorrect behavior"
title: "[Bug] "
labels:
- type/bug
- status/triage
- module/core
---
## Scope
- Repository:
- Area/module:
- Affected version or commit:
## Behavior
Expected:
Actual:
## Reproduction
1.
2.
3.
## Evidence
Logs, screenshots, traces, or failing test output:
## Verification Target
Command or workflow that should pass when fixed:
+1
View File
@@ -0,0 +1 @@
blank_issues_enabled: false
+27
View File
@@ -0,0 +1,27 @@
---
name: "Docs / workflow"
about: "Request documentation, process, or developer workflow changes"
title: "[Docs] "
labels:
- type/docs
- status/triage
- module/core
- area/docs
---
## Scope
- Repository:
- Document or workflow:
## Current State
What is missing, unclear, duplicated, or stale?
## Desired State
What should the docs or workflow make clear?
## Verification Target
How should this be checked?
+32
View File
@@ -0,0 +1,32 @@
---
name: "Feature"
about: "Propose new user-visible behavior or platform capability"
title: "[Feature] "
labels:
- type/feature
- status/triage
- module/core
---
## Problem
What user, operator, or developer problem should this solve?
## Proposed Capability
What should exist when this is done?
## Ownership
- Owning repository:
- Related module repositories:
- Extension point or integration boundary:
## Acceptance Criteria
- [ ]
- [ ]
## Verification Target
Command, scenario, or UI flow that should prove completion:
+28
View File
@@ -0,0 +1,28 @@
---
name: "Task"
about: "Track implementation, maintenance, or migration work"
title: "[Task] "
labels:
- type/task
- status/triage
- module/core
---
## Objective
What needs to be completed?
## Scope
- Owning repository:
- In-scope:
- Out-of-scope:
## Checklist
- [ ]
- [ ]
## Verification Target
Command or manual check:
+25
View File
@@ -0,0 +1,25 @@
---
name: "Tech debt"
about: "Track cleanup, refactoring, risk reduction, or deferred engineering work"
title: "[Debt] "
labels:
- type/debt
- status/triage
- module/core
---
## Current Cost
What does this make harder, riskier, slower, or more fragile?
## Desired Shape
What should the code, tests, or architecture look like afterwards?
## Constraints
What behavior, compatibility, or module boundary must be preserved?
## Verification Target
Focused checks that should pass:
+15
View File
@@ -0,0 +1,15 @@
## Issue
Closes #
## Summary
-
## Verification
-
## Notes
Follow-up issues:
+270
View File
@@ -0,0 +1,270 @@
name: Module Package Release
on:
push:
tags:
- "v*"
workflow_dispatch:
inputs:
release_tag:
description: Existing protected version tag to publish
required: true
type: string
jobs:
publish-packages:
runs-on: ubuntu-latest
env:
GITEA_REPOSITORY: ${{ gitea.repository }}
steps:
- uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
with:
fetch-depth: 0
- uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065
with:
python-version: "3.12"
- uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020
with:
node-version: "22"
- name: Select and validate protected release tag
shell: bash
env:
REQUESTED_TAG: ${{ inputs.release_tag }}
TRIGGER_TAG: ${{ gitea.ref_name }}
run: |
set -euo pipefail
tag="${REQUESTED_TAG:-$TRIGGER_TAG}"
case "$tag" in
v[0-9]*.[0-9]*.[0-9]*) ;;
*) echo "Release tag must start with a SemVer-shaped vX.Y.Z value" >&2; exit 1 ;;
esac
git fetch --force origin "refs/tags/$tag:refs/tags/$tag" refs/heads/main:refs/remotes/origin/main
tag_commit="$(git rev-list -n 1 "$tag")"
git merge-base --is-ancestor "$tag_commit" refs/remotes/origin/main || {
echo "Release tag is not contained in main" >&2
exit 1
}
git checkout --detach "$tag"
printf 'RELEASE_TAG=%s\n' "$tag" >> "$GITEA_ENV"
printf 'SOURCE_DATE_EPOCH=%s\n' "$(git show -s --format=%ct HEAD)" >> "$GITEA_ENV"
- name: Validate package versions
run: |
python - <<'PY'
import json
from pathlib import Path
import os
import re
import tomllib
tag = os.environ["RELEASE_TAG"]
expected = tag.removeprefix("v")
project = tomllib.loads(Path("pyproject.toml").read_text(encoding="utf-8"))["project"]
if project.get("version") != expected:
raise SystemExit(f"pyproject version {project.get('version')!r} does not match {tag}")
if re.fullmatch(r"govoplan-[a-z0-9-]+", str(project.get("name", ""))) is None:
raise SystemExit("Python distribution name must use the govoplan-* namespace")
webui = Path("webui/package.json")
if webui.is_file():
package = json.loads(webui.read_text(encoding="utf-8"))
if package.get("version") != expected:
raise SystemExit(f"WebUI version {package.get('version')!r} does not match {tag}")
if re.fullmatch(r"@govoplan/[a-z0-9-]+-webui", str(package.get("name", ""))) is None:
raise SystemExit("WebUI package name must use the @govoplan/*-webui namespace")
release = Path("webui/package.release.json")
if release.is_file():
release_package = json.loads(release.read_text(encoding="utf-8"))
if (
release_package.get("name") != package.get("name")
or release_package.get("version") != expected
):
raise SystemExit("WebUI release package identity does not match package.json and the release tag")
PY
- name: Build immutable package artifacts
shell: bash
run: |
set -euo pipefail
python -m pip install --disable-pip-version-check build==1.5.0 twine==7.0.0
rm -rf dist .package-webui
python -m build --wheel --outdir dist
python -m twine check dist/*.whl
if [[ -f webui/package.json ]]; then
mkdir .package-webui
cp -a webui/. .package-webui/
rm -rf .package-webui/node_modules .package-webui/dist
if [[ -f .package-webui/package.release.json ]]; then
cp .package-webui/package.release.json .package-webui/package.json
fi
node <<'NODE'
const fs = require("node:fs");
const path = ".package-webui/package.json";
const packageJson = JSON.parse(fs.readFileSync(path, "utf8"));
const groups = ["dependencies", "optionalDependencies", "peerDependencies"];
for (const group of groups) {
for (const [name, specifier] of Object.entries(packageJson[group] || {})) {
if (!name.startsWith("@govoplan/")) continue;
if (typeof specifier !== "string") {
throw new Error(`${group}.${name} must use a string version`);
}
const packageSlug = name.slice("@govoplan/".length);
if (!packageSlug.endsWith("-webui")) {
throw new Error(`${group}.${name} is outside the WebUI package namespace`);
}
const repository = `govoplan-${packageSlug.slice(0, -"-webui".length)}`;
const escapedRepository = repository.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
const gitTag = specifier.match(
new RegExp(
`^git\\+(?:ssh://git@|https://)git\\.add-ideas\\.de/(?:GovOPlaN|add-ideas)/${escapedRepository}\\.git#v([0-9]+\\.[0-9]+\\.[0-9]+)$`,
),
);
if (gitTag) {
packageJson[group][name] = gitTag[1];
continue;
}
if (specifier.startsWith("file:") || specifier.startsWith("git+")) {
throw new Error(
`${group}.${name} must resolve to an exact registry version for publication`,
);
}
}
}
delete packageJson.private;
fs.writeFileSync(path, `${JSON.stringify(packageJson, null, 2)}\n`);
NODE
npm pkg delete private --prefix .package-webui
(cd .package-webui && npm pack --ignore-scripts --pack-destination ../dist)
fi
python - <<'PY'
import hashlib
import json
from pathlib import Path
import os
import subprocess
artifacts = []
for path in sorted(Path("dist").iterdir()):
if path.suffix not in {".whl", ".tgz"}:
continue
digest = hashlib.sha256(path.read_bytes()).hexdigest()
artifacts.append({"filename": path.name, "sha256": digest, "size": path.stat().st_size})
payload = {
"schema_version": "1",
"repository": os.environ["GITEA_REPOSITORY"],
"tag": os.environ["RELEASE_TAG"],
"commit": subprocess.check_output(["git", "rev-parse", "HEAD"], text=True).strip(),
"artifacts": artifacts,
}
Path("dist/package-artifacts.json").write_text(
json.dumps(payload, indent=2, sort_keys=True) + "\n",
encoding="utf-8",
)
PY
- name: Retain package hash evidence
uses: actions/upload-artifact@a8a3f3ad30e3422c9c7b888a15615d19a852ae32
with:
name: module-packages-${{ gitea.ref_name }}
path: dist/package-artifacts.json
- name: Check immutable registry state
shell: bash
env:
PACKAGE_TOKEN: ${{ secrets.GOVOPLAN_PACKAGE_TOKEN }}
run: |
set -euo pipefail
test -n "$PACKAGE_TOKEN"
python - <<'PY'
import hashlib
import json
import os
from pathlib import Path
import tomllib
from urllib.error import HTTPError
from urllib.parse import quote
from urllib.request import Request, urlopen
api_root = "https://git.add-ideas.de/api/v1/packages/GovOPlaN"
token = os.environ["PACKAGE_TOKEN"]
def should_publish(kind, name, version, path):
package_url = "/".join(
(api_root, kind, quote(name, safe=""), quote(version, safe=""), "files")
)
request = Request(
package_url,
headers={"Accept": "application/json", "Authorization": f"token {token}"},
)
try:
with urlopen(request, timeout=30) as response:
files = json.load(response)
except HTTPError as exc:
if exc.code == 404:
print(f"{kind} package {name}=={version} is not published yet")
return True
raise
if not isinstance(files, list) or len(files) != 1:
raise SystemExit(
f"immutable {kind} package {name}=={version} has an unexpected file set"
)
expected_sha256 = hashlib.sha256(path.read_bytes()).hexdigest()
if files[0].get("sha256") != expected_sha256:
raise SystemExit(
f"immutable {kind} package {name}=={version} already exists with a different SHA-256"
)
print(f"verified existing {kind} package {name}=={version} ({expected_sha256})")
return False
project = tomllib.loads(Path("pyproject.toml").read_text(encoding="utf-8"))["project"]
wheels = tuple(Path("dist").glob("*.whl"))
if len(wheels) != 1:
raise SystemExit("release build must contain exactly one wheel")
publish_pypi = should_publish(
"pypi", str(project["name"]), str(project["version"]), wheels[0]
)
tarballs = tuple(Path("dist").glob("*.tgz"))
if len(tarballs) > 1:
raise SystemExit("release build must contain at most one npm package")
publish_npm = False
if tarballs:
webui = json.loads(
Path(".package-webui/package.json").read_text(encoding="utf-8")
)
publish_npm = should_publish(
"npm", str(webui["name"]), str(webui["version"]), tarballs[0]
)
with Path(os.environ["GITEA_ENV"]).open("a", encoding="utf-8") as env_file:
env_file.write(f"PUBLISH_PYPI={int(publish_pypi)}\n")
env_file.write(f"PUBLISH_NPM={int(publish_npm)}\n")
PY
- name: Publish wheel and WebUI package
shell: bash
env:
PACKAGE_USERNAME: ${{ secrets.GOVOPLAN_PACKAGE_USERNAME }}
PACKAGE_TOKEN: ${{ secrets.GOVOPLAN_PACKAGE_TOKEN }}
run: |
set -euo pipefail
test -n "$PACKAGE_USERNAME"
test -n "$PACKAGE_TOKEN"
if [[ "$PUBLISH_PYPI" == 1 ]]; then
TWINE_USERNAME="$PACKAGE_USERNAME" TWINE_PASSWORD="$PACKAGE_TOKEN" \
python -m twine upload --non-interactive \
--repository-url https://git.add-ideas.de/api/packages/GovOPlaN/pypi \
dist/*.whl
else
echo "Exact wheel is already present; skipping immutable retry."
fi
shopt -s nullglob
webui_packages=(dist/*.tgz)
if (( ${#webui_packages[@]} )) && [[ "$PUBLISH_NPM" == 1 ]]; then
npmrc="$(mktemp)"
trap 'rm -f "$npmrc"' EXIT
chmod 600 "$npmrc"
printf '%s\n' \
'@govoplan:registry=https://git.add-ideas.de/api/packages/GovOPlaN/npm/' \
"//git.add-ideas.de/api/packages/GovOPlaN/npm/:_authToken=$PACKAGE_TOKEN" \
> "$npmrc"
NPM_CONFIG_USERCONFIG="$npmrc" npm publish "./${webui_packages[0]}" \
--ignore-scripts --access public \
--registry https://git.add-ideas.de/api/packages/GovOPlaN/npm/
elif (( ${#webui_packages[@]} )); then
echo "Exact WebUI package is already present; skipping immutable retry."
fi
+19
View File
@@ -136,6 +136,25 @@ dist
.yarn/install-state.gz
.pnp.*
# Local WebUI test/build scratch directories
.component-test-build/
.file-drop-test-build/
.module-test-build/
.policy-test-build/
.template-preview-test-build/
.import-test-build/
webui/.component-test-build/
webui/.file-drop-test-build/
webui/.module-test-build/
webui/.policy-test-build/
webui/.template-preview-test-build/
webui/.import-test-build/
webui/dist-conformance/
webui/test-results/
# Security audit reports
audit-reports/
# ---> Python
# Byte-compiled / optimized / DLL files
__pycache__/
+9 -5
View File
@@ -6,6 +6,7 @@ This repository is the platform runner and shared core for GovOPlaN. It owns the
Sibling module repositories usually used with this repo:
- `/mnt/DATA/git/govoplan-access`
- `/mnt/DATA/git/govoplan-files`
- `/mnt/DATA/git/govoplan-mail`
- `/mnt/DATA/git/govoplan-campaign`
@@ -18,9 +19,9 @@ Use targeted commands first:
```bash
cd /mnt/DATA/git/govoplan-core
./.venv/bin/python -m govoplan_core.devserver --smoke --no-reload
./.venv/bin/python -m unittest tests.test_module_system
./.venv/bin/python -m unittest tests.test_api_smoke.ApiSmokeTests.test_mailbox_message_listing_reports_total_count
/mnt/DATA/git/govoplan/.venv/bin/python -m govoplan_core.devserver --smoke --no-reload
/mnt/DATA/git/govoplan/.venv/bin/python -m unittest tests.test_module_system
/mnt/DATA/git/govoplan/.venv/bin/python -m unittest tests.test_api_smoke.ApiSmokeTests.test_mailbox_message_listing_reports_total_count
```
For WebUI checks:
@@ -35,8 +36,8 @@ PATH=/mnt/DATA/git/govoplan-core/webui/node_modules/.bin:/home/zemion/.nvm/versi
Run the consolidated focused check when a change touches module discovery, optional integrations, shared mail components, or mailbox listing:
```bash
cd /mnt/DATA/git/govoplan-core
./scripts/check-focused.sh
cd /mnt/DATA/git/govoplan
tools/checks/check-focused.sh
```
## Working Rules
@@ -44,5 +45,8 @@ cd /mnt/DATA/git/govoplan-core
- Prefer `rg`, `sed`, and targeted tests over broad recursive scans or full builds.
- Avoid DataGrid changes unless explicitly requested; it is intentionally brittle and has known deferred work.
- Do not add module-to-module imports for optional integrations. Use core registry/capability/module metadata paths.
- Treat documentation as part of every behavior change. Update the owning module's manifest-driven `DocumentationTopic` contributions for each affected user and administrator workflow, setting, permission, limitation, and operational consequence. Feature modules own this content; `govoplan-docs` projects it and must not import feature internals.
- Keep a static user and administrator documentation baseline in every module manifest, even when richer configured-state topics come from `documentation_providers`. Run `/mnt/DATA/git/govoplan/tools/checks/check-manifest-shapes.py` after changing a manifest or module behavior.
- Treat Gitea issues as the canonical backlog and state log. Treat Gitea wiki pages as durable project context mirrored from repository and product docs. Use `docs/GITEA_ISSUES.md` for labels, templates, TODO import, wiki sync, and Codex issue updates.
- Do not keep generated WebUI test folders in git status; they should be ignored and removable.
- Do not start persistent dev servers unless the user asks.
+91 -17
View File
@@ -1,6 +1,10 @@
# govoplan-core
GovOPlaN core is the platform runner and shared foundation. It owns the server entry point, database/session primitives, tenant and RBAC infrastructure, governance policy, audit/auth helpers, module discovery, migration registration, and the shared WebUI shell. Feature code is supplied by installed modules.
<!-- govoplan-repository-type:start -->
**Repository type:** system (kernel).
<!-- govoplan-repository-type:end -->
GovOPlaN core is the platform runner and shared foundation. It owns the server entry point, database/session primitives, module discovery, migration orchestration, capability contracts, install/uninstall orchestration, and the shared WebUI shell. Platform and feature behavior is supplied by installed modules.
## Repository ownership
@@ -9,37 +13,48 @@ Core owns:
- `govoplan_core.server.app:app`, the FastAPI entry point used by uvicorn
- `GovoplanServerConfig`, module discovery, registry validation, and route aggregation
- SQLAlchemy base/session helpers and module migration registration
- tenant/account/session/RBAC/governance/audit models and services
- core API routes for auth, admin, platform metadata, audit, and system health
- kernel APIs for platform metadata, module lifecycle, health, and development diagnostics
- `@govoplan/core-webui`, including login, CSRF/API helpers, shell layout, generic UI components, IconRail, DataGrid, access boundaries, and module route/nav contracts
Feature modules own their backend routers, models, migrations, permissions, frontend packages, nav items, and route contributions. Core should not import feature pages directly; it imports module manifests and renders their route contributions.
The shared DataGrid sizing and resize invariants are specified in
[`docs/DATAGRID_SIZING_CONTRACT.md`](docs/DATAGRID_SIZING_CONTRACT.md).
Platform and feature modules own their backend routers, models, migrations,
permissions, frontend packages, nav items, and route contributions. Access,
tenancy, policy, audit, and admin behavior live in their owning platform
modules. Core should not import feature pages directly; it imports module
manifests and renders their route contributions.
## Governance docs
Canonical policy documents live in `docs/`:
- [RBAC_MANIFEST.md](docs/RBAC_MANIFEST.md)
- [SYSTEM_GOVERNANCE_MANIFEST.md](docs/SYSTEM_GOVERNANCE_MANIFEST.md)
- [DOCUMENTATION_MAP.md](docs/DOCUMENTATION_MAP.md)
- [ACCESS_RBAC_MODEL.md](docs/ACCESS_RBAC_MODEL.md)
- [GOVERNANCE_MODEL.md](docs/GOVERNANCE_MODEL.md)
- [MODULE_ARCHITECTURE.md](docs/MODULE_ARCHITECTURE.md)
- [DEPLOYMENT_OPERATOR_GUIDE.md](docs/DEPLOYMENT_OPERATOR_GUIDE.md)
- [CODEX_WORKFLOW.md](docs/CODEX_WORKFLOW.md)
Modules may define module-specific permissions and policy behavior, but the platform-level permission model and governance hierarchy belong here.
Modules define module-specific permissions and policy behavior. Shared DTOs and
composition rules live in core only where they are stable kernel contracts.
## Backend development
Create or activate the core virtual environment, then install core and sibling modules from this repository:
For whole-product development, create the virtualenv from the meta repository:
```bash
cd /mnt/DATA/git/govoplan-core
cd /mnt/DATA/git/govoplan
python3 -m venv .venv
./.venv/bin/python -m pip install --upgrade pip
./.venv/bin/python -m pip install -r requirements-dev.txt
```
Run the platform server from core through the module-aware development runner. The default config reads `ENABLED_MODULES` and discovers installed module entry points. Local development defaults to `access,campaigns,files,mail`; set `ENABLED_MODULES` explicitly when testing a smaller module permutation.
Run the platform server from core through the module-aware development runner. The default config reads `ENABLED_MODULES` and discovers installed module entry points. Local development defaults to the modules listed by `govoplan_core.settings.Settings.enabled_modules`, including Views when its package is installed; set `ENABLED_MODULES` explicitly when testing a smaller module permutation.
```bash
cd /mnt/DATA/git/govoplan-core
./.venv/bin/python -m govoplan_core.devserver \
/mnt/DATA/git/govoplan/.venv/bin/python -m govoplan_core.devserver \
--host 127.0.0.1 \
--port 8000
```
@@ -48,29 +63,84 @@ For example, to test campaign without files or mail:
```bash
cd /mnt/DATA/git/govoplan-core
ENABLED_MODULES=access,campaigns ./.venv/bin/python -m govoplan_core.devserver \
ENABLED_MODULES=access,campaigns /mnt/DATA/git/govoplan/.venv/bin/python -m govoplan_core.devserver \
--host 127.0.0.1 \
--port 8000
```
The runner loads the same `GovoplanServerConfig` as `govoplan_core.server.app:app`, builds the platform registry, and passes core plus enabled module source roots to uvicorn as reload directories. After reinstalling the editable package, the same command is also available as `govoplan-devserver`.
The default development SQLite database lives at `runtime/multimailer-dev.db`, alongside other local runtime state.
For focused backend work, keep the complete module graph active while watching
only the module being edited. Core/config sources and explicit `--reload-dir`
paths remain watched:
```bash
/mnt/DATA/git/govoplan/.venv/bin/python -m govoplan_core.devserver \
--reload-module calendar \
--reload-module campaign
```
Use `--reload-core-only` when no optional module source tree should trigger a
restart. Omitting both options preserves the broad default and watches every
enabled module. Startup, migration, and compatibility checks still run against
the complete enabled graph whenever the backend restarts.
The default development database is PostgreSQL at `postgresql+psycopg://govoplan_dev@127.0.0.1:5432/govoplan_dev`. Store the password in `~/.pgpass`. To force the disposable SQLite fallback, run with `GOVOPLAN_DEV_DATABASE_BACKEND=sqlite`; that database lives below `runtime/`.
Local devserver runs do not require Redis. `CELERY_ENABLED` defaults to `false`, so campaign queue actions update database state without publishing Celery tasks. Use the synchronous send flow for local send tests, or set `CELERY_ENABLED=true` only when a Redis broker and worker are running.
If the configured local SQLite database is missing or empty, `govoplan_core.devserver` enables the development bootstrap before loading settings. This creates the schema and the default development login on startup. Explicitly setting `DEV_BOOTSTRAP_ENABLED=false` disables this convenience. Production deployments should use migrations and managed database provisioning instead.
To run the production-like local profile with PostgreSQL, Redis, a Celery
worker, explicit module configuration, and persistent local file storage:
To verify the effective runtime paths and missing-SQLite bootstrap without starting uvicorn, run the smoke mode:
```bash
cd /mnt/DATA/git/govoplan
tools/launch/launch-production-like-dev.sh
```
See `/mnt/DATA/git/govoplan/dev/production-like/README.md` for ports,
environment overrides, and cleanup commands. Core keeps wrapper commands during
the migration, but whole-product profiles are owned by the meta repository.
## Security audit
The repository includes a containerized audit toolbox for SAST, secret scanning,
dependency checks, filesystem misconfiguration scans, duplication, and complexity
reports. See [SECURITY_AUDIT.md](docs/SECURITY_AUDIT.md) for the operating model.
```bash
cd /mnt/DATA/git/govoplan
tools/checks/security-audit/run.sh --mode ci --scope govoplan
tools/checks/security-audit/run.sh --mode full --scope govoplan
```
CI runs the `ci` profile in report-only mode and uploads `audit-reports/` as an
artifact. Once the baseline is clean, set `SECURITY_AUDIT_FAIL_ON_FINDINGS=1`
or pass `--strict` locally to turn findings into a failing gate.
`govoplan_core.devserver` enables the development bootstrap before loading settings. In dev, startup migrations create or upgrade the schema and the bootstrap creates the default development login if needed. Explicitly setting `DEV_BOOTSTRAP_ENABLED=false` disables this convenience. Production deployments use the separate, expiring single-use flow exposed by `python -m govoplan_core.commands.first_admin`; it cannot enable or consume development bootstrap credentials.
To verify the effective runtime paths and bootstrap behavior without starting uvicorn, run the smoke mode:
```bash
cd /mnt/DATA/git/govoplan-core
./.venv/bin/python -m govoplan_core.devserver --smoke --no-reload
/mnt/DATA/git/govoplan/.venv/bin/python -m govoplan_core.devserver --smoke --no-reload
```
The smoke mode prints the effective config, runtime root, database URL, modules, reload state, and bootstrap decision, then creates the ASGI app and runs startup once.
`requirements-dev.txt` links local `govoplan-files`, `govoplan-mail`, and `govoplan-campaign` checkouts for development. `requirements-release.txt` installs those modules from tagged git refs for release builds. See [RELEASE_DEPENDENCIES.md](docs/RELEASE_DEPENDENCIES.md).
The meta repository owns whole-product `requirements-dev.txt`,
`requirements-release.txt`, and the root `.env.example` operator template. Core
keeps package metadata and runtime commands. See
[RELEASE_DEPENDENCIES.md](docs/RELEASE_DEPENDENCIES.md).
For the install/runtime configuration contract and operator deployment flow, see [DEPLOYMENT_OPERATOR_GUIDE.md](docs/DEPLOYMENT_OPERATOR_GUIDE.md).
For self-hosted config bootstrap and validation:
```bash
cd /mnt/DATA/git/govoplan-core
/mnt/DATA/git/govoplan/.venv/bin/python -m govoplan_core.commands.config env-template --profile self-hosted --generate-secrets
/mnt/DATA/git/govoplan/.venv/bin/python -m govoplan_core.commands.config validate --profile self-hosted
```
## WebUI development
@@ -84,6 +154,10 @@ PATH=/home/zemion/.nvm/versions/node/v22.22.3/bin:$PATH /home/zemion/.nvm/versio
The local host links sibling module WebUI packages through local file dependencies and Vite filesystem allowances. Release builds should use `webui/package.release.json`, which points WebUI module packages at tagged git refs. See [RELEASE_DEPENDENCIES.md](docs/RELEASE_DEPENDENCIES.md).
Production builds lazy-load enabled module descriptors and enforce initial and
asynchronous JavaScript budgets. See
[WEBUI_BUNDLE_BUDGETS.md](docs/WEBUI_BUNDLE_BUDGETS.md).
## Module contract
Backend modules register through the `govoplan.modules` entry point and return a `ModuleManifest`. A manifest can contribute:
+2 -1
View File
@@ -1,7 +1,8 @@
[alembic]
script_location = alembic
path_separator = os
prepend_sys_path = .
sqlalchemy.url = sqlite:///./runtime/multimailer-dev.db
sqlalchemy.url = sqlite:///./runtime/govoplan-dev.db
[loggers]
keys = root,sqlalchemy,alembic
@@ -0,0 +1,52 @@
"""add audit outbox events to detailed dev track
Revision ID: 0f1e2d3c4b5a
Revises: 4f2a9c8e7b6d
Create Date: 2026-07-11 00:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "0f1e2d3c4b5a"
down_revision = "4f2a9c8e7b6d"
branch_labels = None
depends_on = None
def upgrade() -> None:
op.create_table(
"audit_outbox_events",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("event_id", sa.String(length=36), nullable=False),
sa.Column("event_type", sa.String(length=200), nullable=False),
sa.Column("module_id", sa.String(length=100), nullable=False),
sa.Column("correlation_id", sa.String(length=128), nullable=True),
sa.Column("causation_id", sa.String(length=128), nullable=True),
sa.Column("classification", sa.String(length=40), nullable=False),
sa.Column("payload", sa.JSON(), nullable=False),
sa.Column("status", sa.String(length=20), nullable=False),
sa.Column("attempts", sa.Integer(), nullable=False),
sa.Column("next_attempt_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("dispatched_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("last_error", sa.Text(), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint("id", name=op.f("pk_audit_outbox_events")),
sa.UniqueConstraint("event_id", name="uq_audit_outbox_events_event_id"),
)
op.create_index("ix_audit_outbox_events_correlation_id", "audit_outbox_events", ["correlation_id"], unique=False)
op.create_index("ix_audit_outbox_events_event_type", "audit_outbox_events", ["event_type"], unique=False)
op.create_index(op.f("ix_audit_outbox_events_status"), "audit_outbox_events", ["status"], unique=False)
op.create_index(
"ix_audit_outbox_events_status_next_attempt_at",
"audit_outbox_events",
["status", "next_attempt_at"],
unique=False,
)
def downgrade() -> None:
op.drop_table("audit_outbox_events")
@@ -94,10 +94,10 @@ def _scrub_policy_column(table_name: str) -> None:
def upgrade() -> None:
inspector = sa.inspect(op.get_bind())
tables = set(inspector.get_table_names())
for table_name in ("system_settings", "tenants"):
for table_name in ("core_system_settings", "tenancy_tenants"):
if table_name in tables and "settings" in {column["name"] for column in inspector.get_columns(table_name)}:
_scrub_settings_table(table_name)
for table_name in ("users", "groups", "campaigns"):
for table_name in ("access_users", "access_groups", "campaigns"):
if table_name in tables and "mail_profile_policy" in {column["name"] for column in inspector.get_columns(table_name)}:
_scrub_policy_column(table_name)
@@ -30,7 +30,7 @@ def upgrade() -> None:
batch_op.add_column(sa.Column("locked_by_user_id", sa.String(length=36), nullable=True))
batch_op.create_foreign_key(
op.f("fk_campaign_versions_locked_by_user_id_users"),
"users",
"access_users",
["locked_by_user_id"],
["id"],
ondelete="SET NULL",
@@ -0,0 +1,130 @@
"""auth sessions and RBAC assignments
Revision ID: 2c3d4e5f6a7b
Revises: 1f8d4c2a0b7e
Create Date: 2026-06-08 10:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "2c3d4e5f6a7b"
down_revision = "1f8d4c2a0b7e"
branch_labels = None
depends_on = None
def upgrade() -> None:
with op.batch_alter_table("access_users") as batch_op:
batch_op.add_column(sa.Column("auth_provider", sa.String(length=50), nullable=False, server_default="local"))
batch_op.add_column(sa.Column("password_hash", sa.String(length=500), nullable=True))
batch_op.add_column(sa.Column("last_login_at", sa.DateTime(timezone=True), nullable=True))
op.create_table(
"access_user_group_memberships",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=False),
sa.Column("user_id", sa.String(length=36), nullable=False),
sa.Column("group_id", sa.String(length=36), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["user_id"], ["access_users.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["group_id"], ["access_groups.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("tenant_id", "user_id", "group_id", name="uq_user_group_memberships"),
)
op.create_index(op.f("ix_access_user_group_memberships_tenant_id"), "access_user_group_memberships", ["tenant_id"])
op.create_index(op.f("ix_access_user_group_memberships_user_id"), "access_user_group_memberships", ["user_id"])
op.create_index(op.f("ix_access_user_group_memberships_group_id"), "access_user_group_memberships", ["group_id"])
op.create_table(
"access_user_role_assignments",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=False),
sa.Column("user_id", sa.String(length=36), nullable=False),
sa.Column("role_id", sa.String(length=36), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["user_id"], ["access_users.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["role_id"], ["access_roles.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("tenant_id", "user_id", "role_id", name="uq_user_role_assignments"),
)
op.create_index(op.f("ix_access_user_role_assignments_tenant_id"), "access_user_role_assignments", ["tenant_id"])
op.create_index(op.f("ix_access_user_role_assignments_user_id"), "access_user_role_assignments", ["user_id"])
op.create_index(op.f("ix_access_user_role_assignments_role_id"), "access_user_role_assignments", ["role_id"])
op.create_table(
"access_group_role_assignments",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=False),
sa.Column("group_id", sa.String(length=36), nullable=False),
sa.Column("role_id", sa.String(length=36), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["group_id"], ["access_groups.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["role_id"], ["access_roles.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("tenant_id", "group_id", "role_id", name="uq_group_role_assignments"),
)
op.create_index(op.f("ix_access_group_role_assignments_tenant_id"), "access_group_role_assignments", ["tenant_id"])
op.create_index(op.f("ix_access_group_role_assignments_group_id"), "access_group_role_assignments", ["group_id"])
op.create_index(op.f("ix_access_group_role_assignments_role_id"), "access_group_role_assignments", ["role_id"])
op.create_table(
"access_auth_sessions",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=False),
sa.Column("user_id", sa.String(length=36), nullable=False),
sa.Column("token_hash", sa.String(length=128), nullable=False),
sa.Column("expires_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("last_seen_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("revoked_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("user_agent", sa.String(length=500), nullable=True),
sa.Column("ip_address", sa.String(length=100), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["user_id"], ["access_users.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("token_hash"),
)
op.create_index(op.f("ix_access_auth_sessions_tenant_id"), "access_auth_sessions", ["tenant_id"])
op.create_index(op.f("ix_access_auth_sessions_user_id"), "access_auth_sessions", ["user_id"])
op.create_index(op.f("ix_access_auth_sessions_token_hash"), "access_auth_sessions", ["token_hash"])
op.create_index(op.f("ix_access_auth_sessions_expires_at"), "access_auth_sessions", ["expires_at"])
op.create_index(op.f("ix_access_auth_sessions_revoked_at"), "access_auth_sessions", ["revoked_at"])
def downgrade() -> None:
op.drop_index(op.f("ix_access_auth_sessions_revoked_at"), table_name="access_auth_sessions")
op.drop_index(op.f("ix_access_auth_sessions_expires_at"), table_name="access_auth_sessions")
op.drop_index(op.f("ix_access_auth_sessions_token_hash"), table_name="access_auth_sessions")
op.drop_index(op.f("ix_access_auth_sessions_user_id"), table_name="access_auth_sessions")
op.drop_index(op.f("ix_access_auth_sessions_tenant_id"), table_name="access_auth_sessions")
op.drop_table("access_auth_sessions")
op.drop_index(op.f("ix_access_group_role_assignments_role_id"), table_name="access_group_role_assignments")
op.drop_index(op.f("ix_access_group_role_assignments_group_id"), table_name="access_group_role_assignments")
op.drop_index(op.f("ix_access_group_role_assignments_tenant_id"), table_name="access_group_role_assignments")
op.drop_table("access_group_role_assignments")
op.drop_index(op.f("ix_access_user_role_assignments_role_id"), table_name="access_user_role_assignments")
op.drop_index(op.f("ix_access_user_role_assignments_user_id"), table_name="access_user_role_assignments")
op.drop_index(op.f("ix_access_user_role_assignments_tenant_id"), table_name="access_user_role_assignments")
op.drop_table("access_user_role_assignments")
op.drop_index(op.f("ix_access_user_group_memberships_group_id"), table_name="access_user_group_memberships")
op.drop_index(op.f("ix_access_user_group_memberships_user_id"), table_name="access_user_group_memberships")
op.drop_index(op.f("ix_access_user_group_memberships_tenant_id"), table_name="access_user_group_memberships")
op.drop_table("access_user_group_memberships")
with op.batch_alter_table("access_users") as batch_op:
batch_op.drop_column("last_login_at")
batch_op.drop_column("password_hash")
batch_op.drop_column("auth_provider")
@@ -0,0 +1,82 @@
"""namespace platform-owned tables
Revision ID: 2e3f4a5b6c7d
Revises: 1b2c3d4e5f70
Create Date: 2026-07-09 00:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "2e3f4a5b6c7d"
down_revision = "1b2c3d4e5f70"
branch_labels = None
depends_on = None
_TABLE_RENAMES = (
("tenants", "tenancy_tenants"),
("accounts", "access_accounts"),
("users", "access_users"),
("groups", "access_groups"),
("roles", "access_roles"),
("system_role_assignments", "access_system_role_assignments"),
("user_group_memberships", "access_user_group_memberships"),
("user_role_assignments", "access_user_role_assignments"),
("group_role_assignments", "access_group_role_assignments"),
("api_keys", "access_api_keys"),
("auth_sessions", "access_auth_sessions"),
("system_settings", "core_system_settings"),
("governance_templates", "admin_governance_templates"),
("governance_template_assignments", "admin_governance_template_assignments"),
)
_KNOWN_TABLE_NAMES = {name for pair in _TABLE_RENAMES for name in pair}
def _row_count(bind: sa.Connection, table_name: str) -> int:
if table_name not in _KNOWN_TABLE_NAMES:
raise RuntimeError(f"Unexpected table name: {table_name}")
quoted = bind.dialect.identifier_preparer.quote(table_name)
return int(bind.execute(sa.text(f"SELECT COUNT(*) FROM {quoted}")).scalar_one()) # nosemgrep: python.sqlalchemy.security.audit.avoid-sqlalchemy-text.avoid-sqlalchemy-text
def _rename_tables(renames: tuple[tuple[str, str], ...]) -> None:
bind = op.get_bind()
inspector = sa.inspect(bind)
tables = set(inspector.get_table_names())
old_tables_to_drop: set[str] = set()
new_tables_to_drop: set[str] = set()
for old_name, new_name in renames:
if old_name not in tables or new_name not in tables:
continue
if _row_count(bind, old_name) == 0:
old_tables_to_drop.add(old_name)
elif _row_count(bind, new_name) == 0:
new_tables_to_drop.add(new_name)
else:
raise RuntimeError(f"Cannot rename non-empty {old_name} over non-empty {new_name}")
for old_name, _new_name in reversed(renames):
if old_name in old_tables_to_drop:
op.drop_table(old_name)
tables.remove(old_name)
for _old_name, new_name in reversed(renames):
if new_name in new_tables_to_drop:
op.drop_table(new_name)
tables.remove(new_name)
for old_name, new_name in renames:
if old_name not in tables:
continue
op.rename_table(old_name, new_name)
tables.remove(old_name)
tables.add(new_name)
def upgrade() -> None:
_rename_tables(_TABLE_RENAMES)
def downgrade() -> None:
_rename_tables(tuple((new_name, old_name) for old_name, new_name in reversed(_TABLE_RENAMES)))
@@ -32,7 +32,7 @@ def upgrade() -> None:
sa.Column("retained_until", sa.DateTime(timezone=True), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("tenant_id", "checksum_sha256", "size_bytes", name="uq_file_blobs_tenant_checksum_size"),
)
@@ -55,10 +55,10 @@ def upgrade() -> None:
sa.Column("metadata", sa.JSON(), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["created_by_user_id"], ["users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["owner_group_id"], ["groups.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["owner_user_id"], ["users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["created_by_user_id"], ["access_users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["owner_group_id"], ["access_groups.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["owner_user_id"], ["access_users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
)
for col in ["tenant_id", "owner_type", "owner_user_id", "owner_group_id", "current_version_id", "display_path", "filename", "created_by_user_id", "deleted_at"]:
@@ -80,9 +80,9 @@ def upgrade() -> None:
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["blob_id"], ["file_blobs.id"], ondelete="RESTRICT"),
sa.ForeignKeyConstraint(["created_by_user_id"], ["users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["created_by_user_id"], ["access_users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["file_asset_id"], ["file_assets.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("file_asset_id", "version_number", name="uq_file_versions_asset_number"),
)
@@ -101,9 +101,9 @@ def upgrade() -> None:
sa.Column("revoked_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["created_by_user_id"], ["users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["created_by_user_id"], ["access_users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["file_asset_id"], ["file_assets.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("file_asset_id", "target_type", "target_id", "revoked_at", name="uq_file_shares_active_target"),
)
@@ -136,7 +136,7 @@ def upgrade() -> None:
sa.ForeignKeyConstraint(["file_asset_id"], ["file_assets.id"], ondelete="RESTRICT"),
sa.ForeignKeyConstraint(["file_blob_id"], ["file_blobs.id"], ondelete="RESTRICT"),
sa.ForeignKeyConstraint(["file_version_id"], ["file_versions.id"], ondelete="RESTRICT"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("campaign_job_id", "file_version_id", "filename_used", "use_stage", name="uq_campaign_attachment_uses_job_file_stage"),
)
@@ -0,0 +1,111 @@
"""add core change sequence
Revision ID: 3f4a5b6c7d8e
Revises: 2e3f4a5b6c7d
Create Date: 2026-07-09 00:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "3f4a5b6c7d8e"
down_revision = "2e3f4a5b6c7d"
branch_labels = None
depends_on = None
def upgrade() -> None:
inspector = sa.inspect(op.get_bind())
if "core_change_sequence" not in inspector.get_table_names():
op.create_table(
"core_change_sequence",
sa.Column("id", sa.BigInteger().with_variant(sa.Integer(), "sqlite"), autoincrement=True, nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=True),
sa.Column("module_id", sa.String(length=100), nullable=False),
sa.Column("collection", sa.String(length=150), nullable=False),
sa.Column("resource_type", sa.String(length=100), nullable=False),
sa.Column("resource_id", sa.String(length=255), nullable=False),
sa.Column("operation", sa.String(length=30), nullable=False),
sa.Column("actor_type", sa.String(length=30), nullable=True),
sa.Column("actor_id", sa.String(length=255), nullable=True),
sa.Column("payload", sa.JSON(), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint("id", name=op.f("pk_core_change_sequence")),
)
inspector = sa.inspect(op.get_bind())
indexes = {item["name"] for item in inspector.get_indexes("core_change_sequence")}
for column in (
"tenant_id",
"module_id",
"collection",
"resource_type",
"resource_id",
"operation",
"actor_type",
"actor_id",
"created_at",
):
name = op.f(f"ix_core_change_sequence_{column}")
if name not in indexes:
op.create_index(name, "core_change_sequence", [column], unique=False)
for name, columns in (
("ix_core_change_sequence_tenant_module_id", ["tenant_id", "module_id", "id"]),
("ix_core_change_sequence_collection_id", ["collection", "id"]),
("ix_core_change_sequence_resource", ["module_id", "resource_type", "resource_id"]),
):
if name not in indexes:
op.create_index(name, "core_change_sequence", columns, unique=False)
inspector = sa.inspect(op.get_bind())
if "core_change_sequence_retention_floor" not in inspector.get_table_names():
op.create_table(
"core_change_sequence_retention_floor",
sa.Column("id", sa.BigInteger().with_variant(sa.Integer(), "sqlite"), autoincrement=True, nullable=False),
sa.Column("tenant_key", sa.String(length=36), nullable=False),
sa.Column("module_id", sa.String(length=100), nullable=False),
sa.Column("collection", sa.String(length=150), nullable=False),
sa.Column("min_valid_sequence", sa.BigInteger().with_variant(sa.Integer(), "sqlite"), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint("id", name=op.f("pk_core_change_sequence_retention_floor")),
sa.UniqueConstraint("tenant_key", "module_id", "collection", name="uq_core_change_sequence_retention_scope"),
)
indexes = {item["name"] for item in inspector.get_indexes("core_change_sequence_retention_floor")}
if "ix_core_change_sequence_retention_scope" not in indexes:
op.create_index(
"ix_core_change_sequence_retention_scope",
"core_change_sequence_retention_floor",
["tenant_key", "module_id", "collection"],
unique=False,
)
def downgrade() -> None:
inspector = sa.inspect(op.get_bind())
if "core_change_sequence_retention_floor" in inspector.get_table_names():
indexes = {item["name"] for item in inspector.get_indexes("core_change_sequence_retention_floor")}
if "ix_core_change_sequence_retention_scope" in indexes:
op.drop_index("ix_core_change_sequence_retention_scope", table_name="core_change_sequence_retention_floor")
op.drop_table("core_change_sequence_retention_floor")
if "core_change_sequence" not in inspector.get_table_names():
return
indexes = {item["name"] for item in inspector.get_indexes("core_change_sequence")}
for name in (
"ix_core_change_sequence_resource",
"ix_core_change_sequence_collection_id",
"ix_core_change_sequence_tenant_module_id",
op.f("ix_core_change_sequence_created_at"),
op.f("ix_core_change_sequence_actor_id"),
op.f("ix_core_change_sequence_actor_type"),
op.f("ix_core_change_sequence_operation"),
op.f("ix_core_change_sequence_resource_id"),
op.f("ix_core_change_sequence_resource_type"),
op.f("ix_core_change_sequence_collection"),
op.f("ix_core_change_sequence_module_id"),
op.f("ix_core_change_sequence_tenant_id"),
):
if name in indexes:
op.drop_index(name, table_name="core_change_sequence")
op.drop_table("core_change_sequence")
@@ -31,10 +31,10 @@ def upgrade() -> None:
sa.Column("metadata", sa.JSON(), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["created_by_user_id"], ["users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["owner_group_id"], ["groups.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["owner_user_id"], ["users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["created_by_user_id"], ["access_users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["owner_group_id"], ["access_groups.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["owner_user_id"], ["access_users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
)
for col in ["tenant_id", "owner_type", "owner_user_id", "owner_group_id", "path", "created_by_user_id", "deleted_at"]:
@@ -0,0 +1,79 @@
"""rename tenancy scope table to core scopes
Revision ID: 4f2a9c8e7b6d
Revises: 3f4a5b6c7d8e
Create Date: 2026-07-10 00:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "4f2a9c8e7b6d"
down_revision = "3f4a5b6c7d8e"
branch_labels = None
depends_on = None
LEGACY_SCOPE_TABLE = "tenancy_tenants"
CORE_SCOPE_TABLE = "core_scopes"
LEGACY_SLUG_INDEX = "ix_tenancy_tenants_slug"
CORE_SLUG_INDEX = "ix_core_scopes_slug"
_KNOWN_SCOPE_TABLES = {LEGACY_SCOPE_TABLE, CORE_SCOPE_TABLE}
def _row_count(bind: sa.Connection, table_name: str) -> int:
if table_name not in _KNOWN_SCOPE_TABLES:
raise RuntimeError(f"Unexpected scope table name: {table_name}")
quoted = bind.dialect.identifier_preparer.quote(table_name)
return int(bind.execute(sa.text(f"SELECT COUNT(*) FROM {quoted}")).scalar_one()) # nosemgrep: python.sqlalchemy.security.audit.avoid-sqlalchemy-text.avoid-sqlalchemy-text
def _scope_tables(bind: sa.Connection) -> set[str]:
return set(sa.inspect(bind).get_table_names())
def _drop_table_if_empty(bind: sa.Connection, table_name: str) -> bool:
if _row_count(bind, table_name) != 0:
return False
op.drop_table(table_name)
return True
def _ensure_slug_index(bind: sa.Connection, table_name: str, index_name: str, old_index_name: str) -> None:
indexes = {index["name"] for index in sa.inspect(bind).get_indexes(table_name)}
if old_index_name in indexes:
op.drop_index(old_index_name, table_name=table_name)
indexes.remove(old_index_name)
if index_name not in indexes:
op.create_index(op.f(index_name), table_name, ["slug"], unique=True)
def _rename_scope_table(old_name: str, new_name: str) -> None:
bind = op.get_bind()
tables = _scope_tables(bind)
if old_name not in tables:
if new_name in tables:
_ensure_slug_index(bind, new_name, CORE_SLUG_INDEX if new_name == CORE_SCOPE_TABLE else LEGACY_SLUG_INDEX, LEGACY_SLUG_INDEX if new_name == CORE_SCOPE_TABLE else CORE_SLUG_INDEX)
return
if new_name in tables:
if _drop_table_if_empty(bind, new_name):
tables.remove(new_name)
elif _drop_table_if_empty(bind, old_name):
_ensure_slug_index(bind, new_name, CORE_SLUG_INDEX if new_name == CORE_SCOPE_TABLE else LEGACY_SLUG_INDEX, LEGACY_SLUG_INDEX if new_name == CORE_SCOPE_TABLE else CORE_SLUG_INDEX)
return
else:
raise RuntimeError(f"Cannot reconcile non-empty {old_name} over non-empty {new_name}")
if old_name in tables and new_name not in tables:
op.rename_table(old_name, new_name)
_ensure_slug_index(bind, new_name, CORE_SLUG_INDEX if new_name == CORE_SCOPE_TABLE else LEGACY_SLUG_INDEX, LEGACY_SLUG_INDEX if new_name == CORE_SCOPE_TABLE else CORE_SLUG_INDEX)
def upgrade() -> None:
_rename_scope_table(LEGACY_SCOPE_TABLE, CORE_SCOPE_TABLE)
def downgrade() -> None:
_rename_scope_table(CORE_SCOPE_TABLE, LEGACY_SCOPE_TABLE)
@@ -24,7 +24,7 @@ def upgrade() -> None:
batch_op.add_column(sa.Column("user_locked_by_user_id", sa.String(length=36), nullable=True))
batch_op.create_foreign_key(
"fk_campaign_versions_user_locked_by_user_id_users",
"users",
"access_users",
["user_locked_by_user_id"],
["id"],
ondelete="SET NULL",
@@ -62,25 +62,25 @@ def upgrade() -> None:
bind = op.get_bind()
inspector = sa.inspect(bind)
tables = set(inspector.get_table_names())
user_columns = {column["name"] for column in inspector.get_columns("users")}
user_columns = {column["name"] for column in inspector.get_columns("access_users")}
# Base.metadata.create_all() from a newer application can create brand-new
# tables while leaving existing tables unaltered. Repair that known drift by
# removing only empty, unreferenced administration tables before applying the
# real migration. A non-empty table is never guessed at or discarded.
if "account_id" not in user_columns and "system_role_assignments" in tables:
count = bind.execute(sa.text("SELECT COUNT(*) FROM system_role_assignments")).scalar_one()
if "account_id" not in user_columns and "access_system_role_assignments" in tables:
count = bind.execute(sa.text("SELECT COUNT(*) FROM access_system_role_assignments")).scalar_one()
if count:
raise RuntimeError("Cannot reconcile non-empty create_all system_role_assignments table")
op.drop_table("system_role_assignments")
tables.remove("system_role_assignments")
if "account_id" not in user_columns and "accounts" in tables:
count = bind.execute(sa.text("SELECT COUNT(*) FROM accounts")).scalar_one()
op.drop_table("access_system_role_assignments")
tables.remove("access_system_role_assignments")
if "account_id" not in user_columns and "access_accounts" in tables:
count = bind.execute(sa.text("SELECT COUNT(*) FROM access_accounts")).scalar_one()
if count:
raise RuntimeError("Cannot reconcile non-empty create_all accounts table")
op.drop_table("accounts")
op.drop_table("access_accounts")
op.create_table(
"accounts",
"access_accounts",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("email", sa.String(length=320), nullable=False),
sa.Column("normalized_email", sa.String(length=320), nullable=False),
@@ -95,29 +95,29 @@ def upgrade() -> None:
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("normalized_email", name="uq_accounts_normalized_email"),
)
op.create_index(op.f("ix_accounts_normalized_email"), "accounts", ["normalized_email"])
op.create_index(op.f("ix_access_accounts_normalized_email"), "access_accounts", ["normalized_email"])
with op.batch_alter_table("tenants") as batch_op:
with op.batch_alter_table("tenancy_tenants") as batch_op:
batch_op.add_column(sa.Column("description", sa.Text(), nullable=True))
batch_op.add_column(sa.Column("default_locale", sa.String(length=20), nullable=False, server_default="en"))
batch_op.add_column(sa.Column("settings", sa.JSON(), nullable=False, server_default="{}"))
with op.batch_alter_table("groups") as batch_op:
with op.batch_alter_table("access_groups") as batch_op:
batch_op.add_column(sa.Column("description", sa.Text(), nullable=True))
batch_op.add_column(sa.Column("is_active", sa.Boolean(), nullable=False, server_default=sa.true()))
with op.batch_alter_table("roles") as batch_op:
with op.batch_alter_table("access_roles") as batch_op:
batch_op.add_column(sa.Column("description", sa.Text(), nullable=True))
batch_op.add_column(sa.Column("is_builtin", sa.Boolean(), nullable=False, server_default=sa.false()))
batch_op.add_column(sa.Column("is_assignable", sa.Boolean(), nullable=False, server_default=sa.true()))
with op.batch_alter_table("users") as batch_op:
with op.batch_alter_table("access_users") as batch_op:
batch_op.add_column(sa.Column("account_id", sa.String(length=36), nullable=True))
with op.batch_alter_table("auth_sessions") as batch_op:
with op.batch_alter_table("access_auth_sessions") as batch_op:
batch_op.add_column(sa.Column("account_id", sa.String(length=36), nullable=True))
users = bind.execute(sa.text(
"SELECT id, tenant_id, email, display_name, is_active, is_tenant_admin, auth_provider, password_hash, last_login_at, created_at, updated_at FROM users ORDER BY created_at, id"
"SELECT id, tenant_id, email, display_name, is_active, is_tenant_admin, auth_provider, password_hash, last_login_at, created_at, updated_at FROM access_users ORDER BY created_at, id"
)).mappings().all()
accounts_by_email: dict[str, str] = {}
@@ -155,7 +155,7 @@ def upgrade() -> None:
first_system_owner_account_id = account_id
accounts_table = sa.table(
"accounts",
"access_accounts",
sa.column("id", sa.String), sa.column("email", sa.String), sa.column("normalized_email", sa.String),
sa.column("display_name", sa.String), sa.column("is_active", sa.Boolean), sa.column("auth_provider", sa.String),
sa.column("password_hash", sa.String), sa.column("password_reset_required", sa.Boolean),
@@ -166,54 +166,54 @@ def upgrade() -> None:
bind.execute(accounts_table.insert(), list(account_rows.values()))
for row in users:
bind.execute(
sa.text("UPDATE users SET account_id = :account_id WHERE id = :user_id"),
sa.text("UPDATE access_users SET account_id = :account_id WHERE id = :user_id"),
{"account_id": accounts_by_email[_normalize_email(row["email"])], "user_id": row["id"]},
)
bind.execute(sa.text(
"UPDATE auth_sessions SET account_id = (SELECT users.account_id FROM users WHERE users.id = auth_sessions.user_id)"
"UPDATE access_auth_sessions SET account_id = (SELECT access_users.account_id FROM access_users WHERE access_users.id = access_auth_sessions.user_id)"
))
with op.batch_alter_table("users") as batch_op:
with op.batch_alter_table("access_users") as batch_op:
batch_op.alter_column("account_id", existing_type=sa.String(length=36), nullable=False)
batch_op.create_foreign_key("fk_users_account_id_accounts", "accounts", ["account_id"], ["id"], ondelete="CASCADE")
batch_op.create_foreign_key("fk_users_account_id_accounts", "access_accounts", ["account_id"], ["id"], ondelete="CASCADE")
batch_op.create_unique_constraint("uq_users_tenant_account", ["tenant_id", "account_id"])
op.create_index(op.f("ix_users_account_id"), "users", ["account_id"])
op.create_index(op.f("ix_access_users_account_id"), "access_users", ["account_id"])
with op.batch_alter_table("auth_sessions") as batch_op:
with op.batch_alter_table("access_auth_sessions") as batch_op:
batch_op.alter_column("account_id", existing_type=sa.String(length=36), nullable=False)
batch_op.create_foreign_key("fk_auth_sessions_account_id_accounts", "accounts", ["account_id"], ["id"], ondelete="CASCADE")
op.create_index(op.f("ix_auth_sessions_account_id"), "auth_sessions", ["account_id"])
batch_op.create_foreign_key("fk_auth_sessions_account_id_accounts", "access_accounts", ["account_id"], ["id"], ondelete="CASCADE")
op.create_index(op.f("ix_access_auth_sessions_account_id"), "access_auth_sessions", ["account_id"])
op.create_table(
"system_role_assignments",
"access_system_role_assignments",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("account_id", sa.String(length=36), nullable=False),
sa.Column("role_id", sa.String(length=36), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["account_id"], ["accounts.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["role_id"], ["roles.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["account_id"], ["access_accounts.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["role_id"], ["access_roles.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("account_id", "role_id", name="uq_system_role_assignments"),
)
op.create_index(op.f("ix_system_role_assignments_account_id"), "system_role_assignments", ["account_id"])
op.create_index(op.f("ix_system_role_assignments_role_id"), "system_role_assignments", ["role_id"])
op.create_index(op.f("ix_access_system_role_assignments_account_id"), "access_system_role_assignments", ["account_id"])
op.create_index(op.f("ix_access_system_role_assignments_role_id"), "access_system_role_assignments", ["role_id"])
roles_table = sa.table(
"roles",
"access_roles",
sa.column("id", sa.String), sa.column("tenant_id", sa.String), sa.column("slug", sa.String),
sa.column("name", sa.String), sa.column("description", sa.Text), sa.column("permissions", sa.JSON),
sa.column("is_builtin", sa.Boolean), sa.column("is_assignable", sa.Boolean),
sa.column("created_at", sa.DateTime(timezone=True)), sa.column("updated_at", sa.DateTime(timezone=True)),
)
now = _now()
tenant_ids = [row[0] for row in bind.execute(sa.text("SELECT id FROM tenants")).all()]
tenant_ids = [row[0] for row in bind.execute(sa.text("SELECT id FROM tenancy_tenants")).all()]
definitions = _role_definitions()
tenant_role_ids: dict[tuple[str, str], str] = {}
for tenant_id in tenant_ids:
for slug, definition in definitions.items():
existing = bind.execute(
sa.text("SELECT id FROM roles WHERE tenant_id = :tenant_id AND slug = :slug"),
sa.text("SELECT id FROM access_roles WHERE tenant_id = :tenant_id AND slug = :slug"),
{"tenant_id": tenant_id, "slug": slug},
).scalar_one_or_none()
if existing:
@@ -253,7 +253,7 @@ def upgrade() -> None:
op.create_index(
"uq_roles_system_slug",
"roles",
"access_roles",
["slug"],
unique=True,
sqlite_where=sa.text("tenant_id IS NULL"),
@@ -267,11 +267,11 @@ def upgrade() -> None:
continue
owner_role_id = tenant_role_ids[(row["tenant_id"], "owner")]
exists = bind.execute(sa.text(
"SELECT 1 FROM user_role_assignments WHERE tenant_id = :tenant_id AND user_id = :user_id AND role_id = :role_id"
"SELECT 1 FROM access_user_role_assignments WHERE tenant_id = :tenant_id AND user_id = :user_id AND role_id = :role_id"
), {"tenant_id": row["tenant_id"], "user_id": row["id"], "role_id": owner_role_id}).first()
if not exists:
bind.execute(sa.text(
"INSERT INTO user_role_assignments (id, tenant_id, user_id, role_id, created_at, updated_at) VALUES (:id, :tenant_id, :user_id, :role_id, :created_at, :updated_at)"
"INSERT INTO access_user_role_assignments (id, tenant_id, user_id, role_id, created_at, updated_at) VALUES (:id, :tenant_id, :user_id, :role_id, :created_at, :updated_at)"
), {"id": str(uuid.uuid4()), "tenant_id": row["tenant_id"], "user_id": row["id"], "role_id": owner_role_id, "created_at": now, "updated_at": now})
# Bootstrap rule for existing installations: the earliest active legacy
@@ -284,40 +284,40 @@ def upgrade() -> None:
), None)
if first_system_owner_account_id:
bind.execute(sa.text(
"INSERT INTO system_role_assignments (id, account_id, role_id, created_at, updated_at) VALUES (:id, :account_id, :role_id, :created_at, :updated_at)"
"INSERT INTO access_system_role_assignments (id, account_id, role_id, created_at, updated_at) VALUES (:id, :account_id, :role_id, :created_at, :updated_at)"
), {"id": str(uuid.uuid4()), "account_id": first_system_owner_account_id, "role_id": system_owner_role_id, "created_at": now, "updated_at": now})
def downgrade() -> None:
op.drop_index("uq_roles_system_slug", table_name="roles")
op.drop_index(op.f("ix_system_role_assignments_role_id"), table_name="system_role_assignments")
op.drop_index(op.f("ix_system_role_assignments_account_id"), table_name="system_role_assignments")
op.drop_table("system_role_assignments")
op.drop_index("uq_roles_system_slug", table_name="access_roles")
op.drop_index(op.f("ix_access_system_role_assignments_role_id"), table_name="access_system_role_assignments")
op.drop_index(op.f("ix_access_system_role_assignments_account_id"), table_name="access_system_role_assignments")
op.drop_table("access_system_role_assignments")
op.drop_index(op.f("ix_auth_sessions_account_id"), table_name="auth_sessions")
with op.batch_alter_table("auth_sessions") as batch_op:
op.drop_index(op.f("ix_access_auth_sessions_account_id"), table_name="access_auth_sessions")
with op.batch_alter_table("access_auth_sessions") as batch_op:
batch_op.drop_constraint("fk_auth_sessions_account_id_accounts", type_="foreignkey")
batch_op.drop_column("account_id")
op.drop_index(op.f("ix_users_account_id"), table_name="users")
with op.batch_alter_table("users") as batch_op:
op.drop_index(op.f("ix_access_users_account_id"), table_name="access_users")
with op.batch_alter_table("access_users") as batch_op:
batch_op.drop_constraint("uq_users_tenant_account", type_="unique")
batch_op.drop_constraint("fk_users_account_id_accounts", type_="foreignkey")
batch_op.drop_column("account_id")
with op.batch_alter_table("roles") as batch_op:
with op.batch_alter_table("access_roles") as batch_op:
batch_op.drop_column("is_assignable")
batch_op.drop_column("is_builtin")
batch_op.drop_column("description")
with op.batch_alter_table("groups") as batch_op:
with op.batch_alter_table("access_groups") as batch_op:
batch_op.drop_column("is_active")
batch_op.drop_column("description")
with op.batch_alter_table("tenants") as batch_op:
with op.batch_alter_table("tenancy_tenants") as batch_op:
batch_op.drop_column("settings")
batch_op.drop_column("default_locale")
batch_op.drop_column("description")
op.drop_index(op.f("ix_accounts_normalized_email"), table_name="accounts")
op.drop_table("accounts")
op.drop_index(op.f("ix_access_accounts_normalized_email"), table_name="access_accounts")
op.drop_table("access_accounts")
@@ -16,6 +16,7 @@ revision = "9d0e1f2a3b4c"
down_revision = "8c9d0e1f2a3b"
branch_labels = None
depends_on = None
_RECONCILE_CREATE_ALL_TABLES = ("admin_governance_template_assignments", "admin_governance_templates", "core_system_settings")
def _now() -> datetime:
@@ -28,15 +29,16 @@ def upgrade() -> None:
tables = set(inspector.get_table_names())
# Reconcile only the empty create_all shape for the newly introduced tables.
for table_name in ("governance_template_assignments", "governance_templates", "system_settings"):
for table_name in _RECONCILE_CREATE_ALL_TABLES:
if table_name in tables:
count = bind.execute(sa.text(f"SELECT COUNT(*) FROM {table_name}")).scalar_one()
quoted = bind.dialect.identifier_preparer.quote(table_name)
count = bind.execute(sa.text(f"SELECT COUNT(*) FROM {quoted}")).scalar_one() # nosemgrep: python.sqlalchemy.security.audit.avoid-sqlalchemy-text.avoid-sqlalchemy-text
if count:
raise RuntimeError(f"Cannot reconcile non-empty create_all table {table_name}")
op.drop_table(table_name)
op.create_table(
"system_settings",
"core_system_settings",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("default_locale", sa.String(length=20), nullable=False, server_default="en"),
sa.Column("allow_tenant_custom_groups", sa.Boolean(), nullable=False, server_default=sa.true()),
@@ -50,16 +52,23 @@ def upgrade() -> None:
now = _now()
bind.execute(sa.text(
"""
INSERT INTO system_settings
INSERT INTO core_system_settings
(id, default_locale, allow_tenant_custom_groups, allow_tenant_custom_roles,
allow_tenant_api_keys, settings, created_at, updated_at)
VALUES
('global', 'en', 1, 1, 1, '{}', :created_at, :updated_at)
('global', 'en', :allow_tenant_custom_groups, :allow_tenant_custom_roles,
:allow_tenant_api_keys, '{}', :created_at, :updated_at)
"""
), {"created_at": now, "updated_at": now})
), {
"allow_tenant_custom_groups": True,
"allow_tenant_custom_roles": True,
"allow_tenant_api_keys": True,
"created_at": now,
"updated_at": now,
})
op.create_table(
"governance_templates",
"admin_governance_templates",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("kind", sa.String(length=20), nullable=False),
sa.Column("slug", sa.String(length=100), nullable=False),
@@ -72,51 +81,51 @@ def upgrade() -> None:
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("kind", "slug", name="uq_governance_templates_kind_slug"),
)
op.create_index(op.f("ix_governance_templates_kind"), "governance_templates", ["kind"])
op.create_index(op.f("ix_admin_governance_templates_kind"), "admin_governance_templates", ["kind"])
op.create_table(
"governance_template_assignments",
"admin_governance_template_assignments",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("template_id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=False),
sa.Column("mode", sa.String(length=20), nullable=False, server_default="available"),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["template_id"], ["governance_templates.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["template_id"], ["admin_governance_templates.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("template_id", "tenant_id", name="uq_governance_template_tenant"),
)
op.create_index(op.f("ix_governance_template_assignments_template_id"), "governance_template_assignments", ["template_id"])
op.create_index(op.f("ix_governance_template_assignments_tenant_id"), "governance_template_assignments", ["tenant_id"])
op.create_index(op.f("ix_admin_governance_template_assignments_template_id"), "admin_governance_template_assignments", ["template_id"])
op.create_index(op.f("ix_admin_governance_template_assignments_tenant_id"), "admin_governance_template_assignments", ["tenant_id"])
with op.batch_alter_table("tenants") as batch_op:
with op.batch_alter_table("tenancy_tenants") as batch_op:
batch_op.add_column(sa.Column("allow_custom_groups", sa.Boolean(), nullable=True))
batch_op.add_column(sa.Column("allow_custom_roles", sa.Boolean(), nullable=True))
batch_op.add_column(sa.Column("allow_api_keys", sa.Boolean(), nullable=True))
with op.batch_alter_table("groups") as batch_op:
with op.batch_alter_table("access_groups") as batch_op:
batch_op.add_column(sa.Column("system_template_id", sa.String(length=36), nullable=True))
batch_op.add_column(sa.Column("system_required", sa.Boolean(), nullable=False, server_default=sa.false()))
batch_op.create_foreign_key(
"fk_groups_system_template_id_governance_templates",
"governance_templates", ["system_template_id"], ["id"], ondelete="SET NULL",
"admin_governance_templates", ["system_template_id"], ["id"], ondelete="SET NULL",
)
op.create_index(op.f("ix_groups_system_template_id"), "groups", ["system_template_id"])
op.create_index(op.f("ix_access_groups_system_template_id"), "access_groups", ["system_template_id"])
with op.batch_alter_table("roles") as batch_op:
with op.batch_alter_table("access_roles") as batch_op:
batch_op.add_column(sa.Column("system_template_id", sa.String(length=36), nullable=True))
batch_op.add_column(sa.Column("system_required", sa.Boolean(), nullable=False, server_default=sa.false()))
batch_op.create_foreign_key(
"fk_roles_system_template_id_governance_templates",
"governance_templates", ["system_template_id"], ["id"], ondelete="SET NULL",
"admin_governance_templates", ["system_template_id"], ["id"], ondelete="SET NULL",
)
op.create_index(op.f("ix_roles_system_template_id"), "roles", ["system_template_id"])
op.create_index(op.f("ix_access_roles_system_template_id"), "access_roles", ["system_template_id"])
# Existing system owners use system:* and need no data change. Extend the
# read-only built-in auditor role to the newly introduced read scopes.
auditor = bind.execute(sa.text(
"SELECT id, permissions FROM roles WHERE tenant_id IS NULL AND slug = 'system_auditor'"
"SELECT id, permissions FROM access_roles WHERE tenant_id IS NULL AND slug = 'system_auditor'"
)).mappings().first()
if auditor:
raw_permissions = auditor["permissions"] or []
@@ -125,7 +134,7 @@ def upgrade() -> None:
if scope not in permissions:
permissions.append(scope)
roles_table = sa.table(
"roles",
"access_roles",
sa.column("id", sa.String),
sa.column("permissions", sa.JSON),
sa.column("updated_at", sa.DateTime(timezone=True)),
@@ -138,26 +147,26 @@ def upgrade() -> None:
def downgrade() -> None:
op.drop_index(op.f("ix_roles_system_template_id"), table_name="roles")
with op.batch_alter_table("roles") as batch_op:
op.drop_index(op.f("ix_access_roles_system_template_id"), table_name="access_roles")
with op.batch_alter_table("access_roles") as batch_op:
batch_op.drop_constraint("fk_roles_system_template_id_governance_templates", type_="foreignkey")
batch_op.drop_column("system_required")
batch_op.drop_column("system_template_id")
op.drop_index(op.f("ix_groups_system_template_id"), table_name="groups")
with op.batch_alter_table("groups") as batch_op:
op.drop_index(op.f("ix_access_groups_system_template_id"), table_name="access_groups")
with op.batch_alter_table("access_groups") as batch_op:
batch_op.drop_constraint("fk_groups_system_template_id_governance_templates", type_="foreignkey")
batch_op.drop_column("system_required")
batch_op.drop_column("system_template_id")
with op.batch_alter_table("tenants") as batch_op:
with op.batch_alter_table("tenancy_tenants") as batch_op:
batch_op.drop_column("allow_api_keys")
batch_op.drop_column("allow_custom_roles")
batch_op.drop_column("allow_custom_groups")
op.drop_index(op.f("ix_governance_template_assignments_tenant_id"), table_name="governance_template_assignments")
op.drop_index(op.f("ix_governance_template_assignments_template_id"), table_name="governance_template_assignments")
op.drop_table("governance_template_assignments")
op.drop_index(op.f("ix_governance_templates_kind"), table_name="governance_templates")
op.drop_table("governance_templates")
op.drop_table("system_settings")
op.drop_index(op.f("ix_admin_governance_template_assignments_tenant_id"), table_name="admin_governance_template_assignments")
op.drop_index(op.f("ix_admin_governance_template_assignments_template_id"), table_name="admin_governance_template_assignments")
op.drop_table("admin_governance_template_assignments")
op.drop_index(op.f("ix_admin_governance_templates_kind"), table_name="admin_governance_templates")
op.drop_table("admin_governance_templates")
op.drop_table("core_system_settings")
@@ -148,9 +148,9 @@ def upgrade() -> None:
sa.Column("revoked_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["campaign_id"], ["campaigns.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["created_by_user_id"], ["users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["created_by_user_id"], ["access_users.id"], ondelete="SET NULL"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("campaign_id", "target_type", "target_id", name="uq_campaign_share_target"),
)
@@ -160,32 +160,32 @@ def upgrade() -> None:
op.create_index("ix_campaign_shares_target_id", "campaign_shares", ["target_id"])
op.create_index("ix_campaign_shares_created_by_user_id", "campaign_shares", ["created_by_user_id"])
op.create_index("ix_campaign_shares_revoked_at", "campaign_shares", ["revoked_at"])
if "roles" not in tables:
if "access_roles" not in tables:
return
rows = bind.execute(sa.text("SELECT id, permissions FROM roles")).mappings().all()
rows = bind.execute(sa.text("SELECT id, permissions FROM access_roles")).mappings().all()
for row in rows:
bind.execute(
sa.text("UPDATE roles SET permissions = :permissions WHERE id = :id"),
sa.text("UPDATE access_roles SET permissions = :permissions WHERE id = :id"),
{"id": row["id"], "permissions": _json(_expand_legacy(_decode(row["permissions"])))},
)
for slug, permissions in TENANT_ROLE_PERMISSIONS.items():
bind.execute(
sa.text("UPDATE roles SET permissions = :permissions WHERE tenant_id IS NOT NULL AND slug = :slug AND is_builtin = TRUE"),
sa.text("UPDATE access_roles SET permissions = :permissions WHERE tenant_id IS NOT NULL AND slug = :slug AND is_builtin = TRUE"),
{"slug": slug, "permissions": _json(permissions)},
)
now = datetime.now(timezone.utc)
for slug, permissions in SYSTEM_ROLE_PERMISSIONS.items():
existing = bind.execute(
sa.text("SELECT id FROM roles WHERE tenant_id IS NULL AND slug = :slug"), {"slug": slug}
sa.text("SELECT id FROM access_roles WHERE tenant_id IS NULL AND slug = :slug"), {"slug": slug}
).first()
is_protected = slug == "system_owner"
if existing:
bind.execute(
sa.text(
"UPDATE roles SET permissions = :permissions, is_builtin = :is_builtin, is_assignable = TRUE "
"UPDATE access_roles SET permissions = :permissions, is_builtin = :is_builtin, is_assignable = TRUE "
"WHERE tenant_id IS NULL AND slug = :slug"
),
{"slug": slug, "permissions": _json(permissions), "is_builtin": is_protected},
@@ -199,7 +199,7 @@ def upgrade() -> None:
name, description = names[slug]
bind.execute(
sa.text(
"INSERT INTO roles (id, tenant_id, slug, name, description, permissions, is_builtin, is_assignable, "
"INSERT INTO access_roles (id, tenant_id, slug, name, description, permissions, is_builtin, is_assignable, "
"system_template_id, system_required, created_at, updated_at) "
"VALUES (:id, NULL, :slug, :name, :description, :permissions, :is_builtin, TRUE, NULL, FALSE, :created_at, :updated_at)"
),
@@ -0,0 +1,23 @@
from __future__ import annotations
from importlib.util import module_from_spec, spec_from_file_location
from pathlib import Path
_path = (
Path(__file__).resolve().parents[1]
/ "versions"
/ "a36d8e4f9b12_german_reference_locale.py"
)
_spec = spec_from_file_location("govoplan_german_reference_locale_migration", _path)
if _spec is None or _spec.loader is None:
raise RuntimeError(f"Unable to load migration implementation from {_path}")
_module = module_from_spec(_spec)
_spec.loader.exec_module(_module)
revision = _module.revision
down_revision = _module.down_revision
branch_labels = _module.branch_labels
depends_on = _module.depends_on
upgrade = _module.upgrade
downgrade = _module.downgrade
@@ -49,7 +49,7 @@ def upgrade() -> None:
placeholders = ", ".join(f":action_{index}" for index, _ in enumerate(SYSTEM_ACTIONS))
params = {f"action_{index}": action for index, action in enumerate(SYSTEM_ACTIONS)}
bind.execute(
sa.text(f"UPDATE audit_log SET scope = 'system' WHERE action IN ({placeholders})"),
sa.text(f"UPDATE audit_log SET scope = 'system' WHERE action IN ({placeholders})"), # nosemgrep: python.sqlalchemy.security.audit.avoid-sqlalchemy-text.avoid-sqlalchemy-text
params,
)
bind.execute(sa.text("UPDATE audit_log SET scope = 'tenant' WHERE scope IS NULL OR scope NOT IN ('tenant', 'system')"))
@@ -0,0 +1,23 @@
from __future__ import annotations
from importlib.util import module_from_spec, spec_from_file_location
from pathlib import Path
_path = (
Path(__file__).resolve().parents[1]
/ "versions"
/ "b47e6f809a13_data_subject_requests.py"
)
_spec = spec_from_file_location("govoplan_data_subject_requests_migration", _path)
if _spec is None or _spec.loader is None:
raise RuntimeError(f"Unable to load migration implementation from {_path}")
_module = module_from_spec(_spec)
_spec.loader.exec_module(_module)
revision = _module.revision
down_revision = _module.down_revision
branch_labels = _module.branch_labels
depends_on = _module.depends_on
upgrade = _module.upgrade
downgrade = _module.downgrade
@@ -18,7 +18,7 @@ depends_on = None
def upgrade() -> None:
# ### commands auto generated by Alembic - please adjust! ###
op.create_table('tenants',
op.create_table('tenancy_tenants',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('slug', sa.String(length=100), nullable=False),
sa.Column('name', sa.String(length=255), nullable=False),
@@ -27,7 +27,7 @@ def upgrade() -> None:
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint('id', name=op.f('pk_tenants'))
)
op.create_index(op.f('ix_tenants_slug'), 'tenants', ['slug'], unique=True)
op.create_index(op.f('ix_tenancy_tenants_slug'), 'tenancy_tenants', ['slug'], unique=True)
op.create_table('attachment_blobs',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
@@ -38,25 +38,25 @@ def upgrade() -> None:
sa.Column('storage_key', sa.String(length=1000), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['tenant_id'], ['tenants.id'], name=op.f('fk_attachment_blobs_tenant_id_tenants'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenancy_tenants.id'], name=op.f('fk_attachment_blobs_tenant_id_tenants'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_attachment_blobs')),
sa.UniqueConstraint('tenant_id', 'sha256', name='uq_attachment_blobs_tenant_sha256')
)
op.create_index(op.f('ix_attachment_blobs_sha256'), 'attachment_blobs', ['sha256'], unique=False)
op.create_index(op.f('ix_attachment_blobs_tenant_id'), 'attachment_blobs', ['tenant_id'], unique=False)
op.create_table('groups',
op.create_table('access_groups',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('slug', sa.String(length=100), nullable=False),
sa.Column('name', sa.String(length=255), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['tenant_id'], ['tenants.id'], name=op.f('fk_groups_tenant_id_tenants'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenancy_tenants.id'], name=op.f('fk_groups_tenant_id_tenants'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_groups')),
sa.UniqueConstraint('tenant_id', 'slug', name='uq_groups_tenant_slug')
)
op.create_index(op.f('ix_groups_tenant_id'), 'groups', ['tenant_id'], unique=False)
op.create_table('roles',
op.create_index(op.f('ix_access_groups_tenant_id'), 'access_groups', ['tenant_id'], unique=False)
op.create_table('access_roles',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=True),
sa.Column('slug', sa.String(length=100), nullable=False),
@@ -64,12 +64,12 @@ def upgrade() -> None:
sa.Column('permissions', sa.JSON(), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['tenant_id'], ['tenants.id'], name=op.f('fk_roles_tenant_id_tenants'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenancy_tenants.id'], name=op.f('fk_roles_tenant_id_tenants'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_roles')),
sa.UniqueConstraint('tenant_id', 'slug', name='uq_roles_tenant_slug')
)
op.create_index(op.f('ix_roles_tenant_id'), 'roles', ['tenant_id'], unique=False)
op.create_table('users',
op.create_index(op.f('ix_access_roles_tenant_id'), 'access_roles', ['tenant_id'], unique=False)
op.create_table('access_users',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('email', sa.String(length=320), nullable=False),
@@ -78,13 +78,13 @@ def upgrade() -> None:
sa.Column('is_tenant_admin', sa.Boolean(), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['tenant_id'], ['tenants.id'], name=op.f('fk_users_tenant_id_tenants'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenancy_tenants.id'], name=op.f('fk_users_tenant_id_tenants'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_users')),
sa.UniqueConstraint('tenant_id', 'email', name='uq_users_tenant_email')
)
op.create_index(op.f('ix_users_email'), 'users', ['email'], unique=False)
op.create_index(op.f('ix_users_tenant_id'), 'users', ['tenant_id'], unique=False)
op.create_table('api_keys',
op.create_index(op.f('ix_access_users_email'), 'access_users', ['email'], unique=False)
op.create_index(op.f('ix_access_users_tenant_id'), 'access_users', ['tenant_id'], unique=False)
op.create_table('access_api_keys',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('user_id', sa.String(length=36), nullable=False),
@@ -97,13 +97,13 @@ def upgrade() -> None:
sa.Column('revoked_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['tenant_id'], ['tenants.id'], name=op.f('fk_api_keys_tenant_id_tenants'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['user_id'], ['users.id'], name=op.f('fk_api_keys_user_id_users'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenancy_tenants.id'], name=op.f('fk_api_keys_tenant_id_tenants'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['user_id'], ['access_users.id'], name=op.f('fk_api_keys_user_id_users'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_api_keys'))
)
op.create_index(op.f('ix_api_keys_prefix'), 'api_keys', ['prefix'], unique=False)
op.create_index(op.f('ix_api_keys_tenant_id'), 'api_keys', ['tenant_id'], unique=False)
op.create_index(op.f('ix_api_keys_user_id'), 'api_keys', ['user_id'], unique=False)
op.create_index(op.f('ix_access_api_keys_prefix'), 'access_api_keys', ['prefix'], unique=False)
op.create_index(op.f('ix_access_api_keys_tenant_id'), 'access_api_keys', ['tenant_id'], unique=False)
op.create_index(op.f('ix_access_api_keys_user_id'), 'access_api_keys', ['user_id'], unique=False)
op.create_table('campaigns',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
@@ -115,8 +115,8 @@ def upgrade() -> None:
sa.Column('current_version_id', sa.String(length=36), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['created_by_user_id'], ['users.id'], name=op.f('fk_campaigns_created_by_user_id_users'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenants.id'], name=op.f('fk_campaigns_tenant_id_tenants'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['created_by_user_id'], ['access_users.id'], name=op.f('fk_campaigns_created_by_user_id_users'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenancy_tenants.id'], name=op.f('fk_campaigns_tenant_id_tenants'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_campaigns')),
sa.UniqueConstraint('tenant_id', 'external_id', name='uq_campaigns_tenant_external_id')
)
@@ -138,8 +138,8 @@ def upgrade() -> None:
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['blob_id'], ['attachment_blobs.id'], name=op.f('fk_attachment_instances_blob_id_attachment_blobs'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['campaign_id'], ['campaigns.id'], name=op.f('fk_attachment_instances_campaign_id_campaigns'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['owner_user_id'], ['users.id'], name=op.f('fk_attachment_instances_owner_user_id_users'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenants.id'], name=op.f('fk_attachment_instances_tenant_id_tenants'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['owner_user_id'], ['access_users.id'], name=op.f('fk_attachment_instances_owner_user_id_users'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenancy_tenants.id'], name=op.f('fk_attachment_instances_tenant_id_tenants'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_attachment_instances'))
)
op.create_index(op.f('ix_attachment_instances_blob_id'), 'attachment_instances', ['blob_id'], unique=False)
@@ -157,9 +157,9 @@ def upgrade() -> None:
sa.Column('details', sa.JSON(), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['api_key_id'], ['api_keys.id'], name=op.f('fk_audit_log_api_key_id_api_keys'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenants.id'], name=op.f('fk_audit_log_tenant_id_tenants'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['user_id'], ['users.id'], name=op.f('fk_audit_log_user_id_users'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['api_key_id'], ['access_api_keys.id'], name=op.f('fk_audit_log_api_key_id_api_keys'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenancy_tenants.id'], name=op.f('fk_audit_log_tenant_id_tenants'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['user_id'], ['access_users.id'], name=op.f('fk_audit_log_user_id_users'), ondelete='SET NULL'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_audit_log'))
)
op.create_index(op.f('ix_audit_log_action'), 'audit_log', ['action'], unique=False)
@@ -213,7 +213,7 @@ def upgrade() -> None:
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['campaign_id'], ['campaigns.id'], name=op.f('fk_campaign_jobs_campaign_id_campaigns'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['campaign_version_id'], ['campaign_versions.id'], name=op.f('fk_campaign_jobs_campaign_version_id_campaign_versions'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenants.id'], name=op.f('fk_campaign_jobs_tenant_id_tenants'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenancy_tenants.id'], name=op.f('fk_campaign_jobs_tenant_id_tenants'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_campaign_jobs')),
sa.UniqueConstraint('campaign_version_id', 'entry_index', name='uq_campaign_jobs_version_entry')
)
@@ -243,7 +243,7 @@ def upgrade() -> None:
sa.ForeignKeyConstraint(['campaign_id'], ['campaigns.id'], name=op.f('fk_campaign_issues_campaign_id_campaigns'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['campaign_version_id'], ['campaign_versions.id'], name=op.f('fk_campaign_issues_campaign_version_id_campaign_versions'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['job_id'], ['campaign_jobs.id'], name=op.f('fk_campaign_issues_job_id_campaign_jobs'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenants.id'], name=op.f('fk_campaign_issues_tenant_id_tenants'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['tenant_id'], ['tenancy_tenants.id'], name=op.f('fk_campaign_issues_tenant_id_tenants'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_campaign_issues'))
)
op.create_index(op.f('ix_campaign_issues_campaign_id'), 'campaign_issues', ['campaign_id'], unique=False)
@@ -327,20 +327,20 @@ def downgrade() -> None:
op.drop_index(op.f('ix_campaigns_external_id'), table_name='campaigns')
op.drop_index(op.f('ix_campaigns_created_by_user_id'), table_name='campaigns')
op.drop_table('campaigns')
op.drop_index(op.f('ix_api_keys_user_id'), table_name='api_keys')
op.drop_index(op.f('ix_api_keys_tenant_id'), table_name='api_keys')
op.drop_index(op.f('ix_api_keys_prefix'), table_name='api_keys')
op.drop_table('api_keys')
op.drop_index(op.f('ix_users_tenant_id'), table_name='users')
op.drop_index(op.f('ix_users_email'), table_name='users')
op.drop_table('users')
op.drop_index(op.f('ix_roles_tenant_id'), table_name='roles')
op.drop_table('roles')
op.drop_index(op.f('ix_groups_tenant_id'), table_name='groups')
op.drop_table('groups')
op.drop_index(op.f('ix_access_api_keys_user_id'), table_name='access_api_keys')
op.drop_index(op.f('ix_access_api_keys_tenant_id'), table_name='access_api_keys')
op.drop_index(op.f('ix_access_api_keys_prefix'), table_name='access_api_keys')
op.drop_table('access_api_keys')
op.drop_index(op.f('ix_access_users_tenant_id'), table_name='access_users')
op.drop_index(op.f('ix_access_users_email'), table_name='access_users')
op.drop_table('access_users')
op.drop_index(op.f('ix_access_roles_tenant_id'), table_name='access_roles')
op.drop_table('access_roles')
op.drop_index(op.f('ix_access_groups_tenant_id'), table_name='access_groups')
op.drop_table('access_groups')
op.drop_index(op.f('ix_attachment_blobs_tenant_id'), table_name='attachment_blobs')
op.drop_index(op.f('ix_attachment_blobs_sha256'), table_name='attachment_blobs')
op.drop_table('attachment_blobs')
op.drop_index(op.f('ix_tenants_slug'), table_name='tenants')
op.drop_table('tenants')
op.drop_index(op.f('ix_tenancy_tenants_slug'), table_name='tenancy_tenants')
op.drop_table('tenancy_tenants')
# ### end Alembic commands ###
@@ -0,0 +1,87 @@
"""add reusable core credential envelopes
Revision ID: c91f0a72be34
Revises: 0f1e2d3c4b5a
Create Date: 2026-07-23 00:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "c91f0a72be34"
down_revision = "0f1e2d3c4b5a"
branch_labels = None
depends_on = None
def upgrade() -> None:
inspector = sa.inspect(op.get_bind())
if "core_credential_envelopes" in inspector.get_table_names():
return
op.create_table(
"core_credential_envelopes",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=True),
sa.Column("scope_type", sa.String(length=20), nullable=False),
sa.Column("scope_id", sa.String(length=255), nullable=True),
sa.Column("name", sa.String(length=255), nullable=False),
sa.Column("description", sa.Text(), nullable=True),
sa.Column("credential_kind", sa.String(length=40), nullable=False),
sa.Column("public_data", sa.JSON(), nullable=False),
sa.Column("secret_data_encrypted", sa.Text(), nullable=True),
sa.Column("secret_keys", sa.JSON(), nullable=False),
sa.Column("allowed_modules", sa.JSON(), nullable=False),
sa.Column("allowed_server_refs", sa.JSON(), nullable=False),
sa.Column("inherit_to_lower_scopes", sa.Boolean(), nullable=False),
sa.Column("is_active", sa.Boolean(), nullable=False),
sa.Column("revision", sa.String(length=36), nullable=False),
sa.Column("created_by_user_id", sa.String(length=36), nullable=True),
sa.Column("updated_by_user_id", sa.String(length=36), nullable=True),
sa.Column("deleted_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("metadata", sa.JSON(), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(
["tenant_id"],
["core_scopes.id"],
name=op.f("fk_core_credential_envelopes_tenant_id_core_scopes"),
ondelete="CASCADE",
),
sa.PrimaryKeyConstraint("id", name=op.f("pk_core_credential_envelopes")),
)
op.create_index(
"ix_core_credential_envelopes_scope",
"core_credential_envelopes",
["tenant_id", "scope_type", "scope_id"],
unique=False,
)
op.create_index(
"ix_core_credential_envelopes_active",
"core_credential_envelopes",
["tenant_id", "is_active", "deleted_at"],
unique=False,
)
for column in (
"tenant_id",
"scope_type",
"scope_id",
"credential_kind",
"is_active",
"created_by_user_id",
"updated_by_user_id",
"deleted_at",
):
op.create_index(
op.f(f"ix_core_credential_envelopes_{column}"),
"core_credential_envelopes",
[column],
unique=False,
)
def downgrade() -> None:
inspector = sa.inspect(op.get_bind())
if "core_credential_envelopes" in inspector.get_table_names():
op.drop_table("core_credential_envelopes")
@@ -0,0 +1,24 @@
"""development-track wrapper for generic ownership transfers."""
from __future__ import annotations
from importlib.util import module_from_spec, spec_from_file_location
from pathlib import Path
_path = (
Path(__file__).resolve().parents[1]
/ "versions"
/ "d03a7b9c1e5f_core_ownership_transfers.py"
)
_spec = spec_from_file_location("govoplan_core_ownership_transfer_migration", _path)
if _spec is None or _spec.loader is None:
raise RuntimeError(f"Unable to load ownership migration from {_path}")
_migration = module_from_spec(_spec)
_spec.loader.exec_module(_migration)
revision = _migration.revision
down_revision = _migration.down_revision
branch_labels = _migration.branch_labels
depends_on = _migration.depends_on
upgrade = _migration.upgrade
downgrade = _migration.downgrade
@@ -19,10 +19,10 @@ def upgrade() -> None:
bind = op.get_bind()
inspector = sa.inspect(bind)
tables = set(inspector.get_table_names())
if "auth_sessions" in tables:
columns = {column["name"] for column in inspector.get_columns("auth_sessions")}
if "access_auth_sessions" in tables:
columns = {column["name"] for column in inspector.get_columns("access_auth_sessions")}
if "csrf_token_hash" not in columns:
op.add_column("auth_sessions", sa.Column("csrf_token_hash", sa.String(length=128), nullable=True))
op.add_column("access_auth_sessions", sa.Column("csrf_token_hash", sa.String(length=128), nullable=True))
if "mail_server_profiles" not in tables:
op.create_table(
"mail_server_profiles",
@@ -40,9 +40,9 @@ def upgrade() -> None:
sa.Column("updated_by_user_id", sa.String(length=36), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["created_by_user_id"], ["users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["updated_by_user_id"], ["users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["created_by_user_id"], ["access_users.id"], ondelete="SET NULL"),
sa.ForeignKeyConstraint(["tenant_id"], ["tenancy_tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["updated_by_user_id"], ["access_users.id"], ondelete="SET NULL"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("tenant_id", "slug", name="uq_mail_server_profiles_tenant_slug"),
)
@@ -67,7 +67,7 @@ def downgrade() -> None:
except Exception:
pass
op.drop_table("mail_server_profiles")
if "auth_sessions" in inspector.get_table_names():
columns = {column["name"] for column in inspector.get_columns("auth_sessions")}
if "access_auth_sessions" in inspector.get_table_names():
columns = {column["name"] for column in inspector.get_columns("access_auth_sessions")}
if "csrf_token_hash" in columns:
op.drop_column("auth_sessions", "csrf_token_hash")
op.drop_column("access_auth_sessions", "csrf_token_hash")
@@ -0,0 +1,24 @@
"""development-track wrapper for runtime coordination and recovery."""
from __future__ import annotations
from importlib.util import module_from_spec, spec_from_file_location
from pathlib import Path
_path = (
Path(__file__).resolve().parents[1]
/ "versions"
/ "e14b8c2d6f90_runtime_coordination_recovery.py"
)
_spec = spec_from_file_location("govoplan_runtime_coordination_migration", _path)
if _spec is None or _spec.loader is None:
raise RuntimeError(f"Unable to load runtime coordination migration from {_path}")
_migration = module_from_spec(_spec)
_spec.loader.exec_module(_migration)
revision = _migration.revision
down_revision = _migration.down_revision
branch_labels = _migration.branch_labels
depends_on = _migration.depends_on
upgrade = _migration.upgrade
downgrade = _migration.downgrade
@@ -31,7 +31,7 @@ def upgrade() -> None:
inspector = sa.inspect(bind)
tables = set(inspector.get_table_names())
for table_name in ("users", "groups", "campaigns"):
for table_name in ("access_users", "access_groups", "campaigns"):
if table_name not in tables:
continue
columns = _columns(inspector, table_name)
@@ -83,6 +83,6 @@ def downgrade() -> None:
batch.drop_column("scope_type")
batch.alter_column("tenant_id", existing_type=sa.String(length=36), nullable=False)
for table_name in ("campaigns", "groups", "users"):
for table_name in ("campaigns", "access_groups", "access_users"):
if table_name in tables and "mail_profile_policy" in _columns(inspector, table_name):
op.drop_column(table_name, "mail_profile_policy")
@@ -0,0 +1,23 @@
from __future__ import annotations
from importlib.util import module_from_spec, spec_from_file_location
from pathlib import Path
_path = (
Path(__file__).resolve().parents[1]
/ "versions"
/ "f25c9d3e7a01_first_admin_enrollment.py"
)
_spec = spec_from_file_location("govoplan_first_admin_enrollment_migration", _path)
if _spec is None or _spec.loader is None:
raise RuntimeError(f"Unable to load migration implementation from {_path}")
_module = module_from_spec(_spec)
_spec.loader.exec_module(_module)
revision = _module.revision
down_revision = _module.down_revision
branch_labels = _module.branch_labels
depends_on = _module.depends_on
upgrade = _module.upgrade
downgrade = _module.downgrade
@@ -25,7 +25,7 @@ def upgrade() -> None:
bind = op.get_bind()
inspector = sa.inspect(bind)
tables = set(inspector.get_table_names())
for table_name in ("users", "groups", "campaigns"):
for table_name in ("access_users", "access_groups", "campaigns"):
if table_name not in tables:
continue
if "settings" not in _columns(inspector, table_name):
@@ -36,6 +36,6 @@ def downgrade() -> None:
bind = op.get_bind()
inspector = sa.inspect(bind)
tables = set(inspector.get_table_names())
for table_name in ("campaigns", "groups", "users"):
for table_name in ("campaigns", "access_groups", "access_users"):
if table_name in tables and "settings" in _columns(inspector, table_name):
op.drop_column(table_name, "settings")
+22 -6
View File
@@ -5,29 +5,45 @@ from logging.config import fileConfig
from alembic import context
from sqlalchemy import engine_from_config, pool
from govoplan_core.access.db import models as access_models # noqa: F401 - populate access metadata
try:
from govoplan_access.backend.db import models as access_models # noqa: F401 - populate optional access metadata
except ModuleNotFoundError as exc:
if exc.name != "govoplan_access":
raise
from govoplan_core.admin import models as core_admin_models # noqa: F401 - populate core admin metadata
from govoplan_core.core import change_sequence as core_change_sequence_models # noqa: F401 - populate core metadata
from govoplan_core.core import first_admin as core_first_admin_models # noqa: F401 - populate core metadata
from govoplan_core.core import ownership as core_ownership_models # noqa: F401 - populate core metadata
from govoplan_core.core import recovery as core_recovery_models # noqa: F401 - populate core metadata
from govoplan_core.core import runtime_coordination as core_runtime_models # noqa: F401 - populate core metadata
from govoplan_core.security import credential_envelopes as core_credential_models # noqa: F401 - populate core metadata
from govoplan_core.core.migrations import migration_metadata_plan
from govoplan_core.db.base import Base
from govoplan_core.db import models # noqa: F401 - populate core metadata
from govoplan_core.server.default_config import get_server_config
from govoplan_core.server.registry import build_platform_registry
from govoplan_core.settings import settings
from govoplan_core.tenancy.scope import scope_registry
config = context.config
database_url = config.attributes.get("database_url") or settings.database_url
config.set_main_option("sqlalchemy.url", database_url)
if config.config_file_name is not None:
fileConfig(config.config_file_name)
# Migrations can run inside the long-lived application process when module
# state changes. Do not let Alembic's logging setup disable loggers that the
# server already created (for example slow-request diagnostics).
fileConfig(config.config_file_name, disable_existing_loggers=False)
def _target_metadata():
server_config = get_server_config()
enabled_modules = config.attributes.get("enabled_modules", server_config.enabled_modules)
manifest_factories = config.attributes.get("manifest_factories", server_config.manifest_factories)
registry = build_platform_registry(
server_config.enabled_modules,
manifest_factories=server_config.manifest_factories,
enabled_modules,
manifest_factories=manifest_factories,
)
plan = migration_metadata_plan(registry, extra_metadata=(Base.metadata,))
plan = migration_metadata_plan(registry, extra_metadata=(scope_registry.metadata, Base.metadata))
return tuple(dict.fromkeys(plan.metadata))
@@ -1,130 +0,0 @@
"""auth sessions and RBAC assignments
Revision ID: 2c3d4e5f6a7b
Revises: 1f8d4c2a0b7e
Create Date: 2026-06-08 10:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "2c3d4e5f6a7b"
down_revision = "1f8d4c2a0b7e"
branch_labels = None
depends_on = None
def upgrade() -> None:
with op.batch_alter_table("users") as batch_op:
batch_op.add_column(sa.Column("auth_provider", sa.String(length=50), nullable=False, server_default="local"))
batch_op.add_column(sa.Column("password_hash", sa.String(length=500), nullable=True))
batch_op.add_column(sa.Column("last_login_at", sa.DateTime(timezone=True), nullable=True))
op.create_table(
"user_group_memberships",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=False),
sa.Column("user_id", sa.String(length=36), nullable=False),
sa.Column("group_id", sa.String(length=36), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["user_id"], ["users.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["group_id"], ["groups.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("tenant_id", "user_id", "group_id", name="uq_user_group_memberships"),
)
op.create_index(op.f("ix_user_group_memberships_tenant_id"), "user_group_memberships", ["tenant_id"])
op.create_index(op.f("ix_user_group_memberships_user_id"), "user_group_memberships", ["user_id"])
op.create_index(op.f("ix_user_group_memberships_group_id"), "user_group_memberships", ["group_id"])
op.create_table(
"user_role_assignments",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=False),
sa.Column("user_id", sa.String(length=36), nullable=False),
sa.Column("role_id", sa.String(length=36), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["user_id"], ["users.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["role_id"], ["roles.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("tenant_id", "user_id", "role_id", name="uq_user_role_assignments"),
)
op.create_index(op.f("ix_user_role_assignments_tenant_id"), "user_role_assignments", ["tenant_id"])
op.create_index(op.f("ix_user_role_assignments_user_id"), "user_role_assignments", ["user_id"])
op.create_index(op.f("ix_user_role_assignments_role_id"), "user_role_assignments", ["role_id"])
op.create_table(
"group_role_assignments",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=False),
sa.Column("group_id", sa.String(length=36), nullable=False),
sa.Column("role_id", sa.String(length=36), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["group_id"], ["groups.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["role_id"], ["roles.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("tenant_id", "group_id", "role_id", name="uq_group_role_assignments"),
)
op.create_index(op.f("ix_group_role_assignments_tenant_id"), "group_role_assignments", ["tenant_id"])
op.create_index(op.f("ix_group_role_assignments_group_id"), "group_role_assignments", ["group_id"])
op.create_index(op.f("ix_group_role_assignments_role_id"), "group_role_assignments", ["role_id"])
op.create_table(
"auth_sessions",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=False),
sa.Column("user_id", sa.String(length=36), nullable=False),
sa.Column("token_hash", sa.String(length=128), nullable=False),
sa.Column("expires_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("last_seen_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("revoked_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("user_agent", sa.String(length=500), nullable=True),
sa.Column("ip_address", sa.String(length=100), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(["tenant_id"], ["tenants.id"], ondelete="CASCADE"),
sa.ForeignKeyConstraint(["user_id"], ["users.id"], ondelete="CASCADE"),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("token_hash"),
)
op.create_index(op.f("ix_auth_sessions_tenant_id"), "auth_sessions", ["tenant_id"])
op.create_index(op.f("ix_auth_sessions_user_id"), "auth_sessions", ["user_id"])
op.create_index(op.f("ix_auth_sessions_token_hash"), "auth_sessions", ["token_hash"])
op.create_index(op.f("ix_auth_sessions_expires_at"), "auth_sessions", ["expires_at"])
op.create_index(op.f("ix_auth_sessions_revoked_at"), "auth_sessions", ["revoked_at"])
def downgrade() -> None:
op.drop_index(op.f("ix_auth_sessions_revoked_at"), table_name="auth_sessions")
op.drop_index(op.f("ix_auth_sessions_expires_at"), table_name="auth_sessions")
op.drop_index(op.f("ix_auth_sessions_token_hash"), table_name="auth_sessions")
op.drop_index(op.f("ix_auth_sessions_user_id"), table_name="auth_sessions")
op.drop_index(op.f("ix_auth_sessions_tenant_id"), table_name="auth_sessions")
op.drop_table("auth_sessions")
op.drop_index(op.f("ix_group_role_assignments_role_id"), table_name="group_role_assignments")
op.drop_index(op.f("ix_group_role_assignments_group_id"), table_name="group_role_assignments")
op.drop_index(op.f("ix_group_role_assignments_tenant_id"), table_name="group_role_assignments")
op.drop_table("group_role_assignments")
op.drop_index(op.f("ix_user_role_assignments_role_id"), table_name="user_role_assignments")
op.drop_index(op.f("ix_user_role_assignments_user_id"), table_name="user_role_assignments")
op.drop_index(op.f("ix_user_role_assignments_tenant_id"), table_name="user_role_assignments")
op.drop_table("user_role_assignments")
op.drop_index(op.f("ix_user_group_memberships_group_id"), table_name="user_group_memberships")
op.drop_index(op.f("ix_user_group_memberships_user_id"), table_name="user_group_memberships")
op.drop_index(op.f("ix_user_group_memberships_tenant_id"), table_name="user_group_memberships")
op.drop_table("user_group_memberships")
with op.batch_alter_table("users") as batch_op:
batch_op.drop_column("last_login_at")
batch_op.drop_column("password_hash")
batch_op.drop_column("auth_provider")
@@ -0,0 +1,892 @@
"""v0.1.7 core baseline
Revision ID: 4f2a9c8e7b6d
Revises: None
Create Date: 2026-07-11 00:00:00.000000
"""
from __future__ import annotations
from datetime import datetime, timezone
from uuid import uuid4
from alembic import op
import sqlalchemy as sa
revision = '4f2a9c8e7b6d'
down_revision = None
branch_labels = None
depends_on = None
def _now() -> datetime:
return datetime.now(timezone.utc)
def _seed_core_defaults() -> None:
bind = op.get_bind()
now = _now()
if not bind.execute(sa.text("SELECT 1 FROM core_system_settings WHERE id = 'global'")).first():
settings_table = sa.table(
"core_system_settings",
sa.column("id", sa.String),
sa.column("default_locale", sa.String),
sa.column("allow_tenant_custom_groups", sa.Boolean),
sa.column("allow_tenant_custom_roles", sa.Boolean),
sa.column("allow_tenant_api_keys", sa.Boolean),
sa.column("settings", sa.JSON),
sa.column("created_at", sa.DateTime(timezone=True)),
sa.column("updated_at", sa.DateTime(timezone=True)),
)
bind.execute(
settings_table.insert().values({
"id": "global",
"default_locale": "en",
"allow_tenant_custom_groups": True,
"allow_tenant_custom_roles": True,
"allow_tenant_api_keys": True,
"settings": {},
"created_at": now,
"updated_at": now,
}),
)
system_roles = {
"system_owner": {
"name": "System owner",
"description": "Protected full instance-wide administration.",
"permissions": ["system:*"],
"is_builtin": True,
},
"system_admin": {
"name": "System administrator",
"description": "Manage the instance without granting the protected System owner role.",
"permissions": [
"system:tenants:read", "system:tenants:create", "system:tenants:update", "system:tenants:suspend",
"system:accounts:read", "system:accounts:create", "system:accounts:update", "system:accounts:suspend",
"system:roles:read", "system:roles:write", "system:roles:assign", "system:access:read", "system:access:assign",
"system:audit:read", "system:settings:read", "system:settings:write", "system:governance:read", "system:governance:write",
],
"is_builtin": False,
},
"system_auditor": {
"name": "System auditor",
"description": "Read-only access to system administration and audit.",
"permissions": [
"system:tenants:read", "system:accounts:read", "system:roles:read", "system:access:read",
"system:audit:read", "system:settings:read", "system:governance:read",
],
"is_builtin": False,
},
}
roles_table = sa.table(
"access_roles",
sa.column("id", sa.String),
sa.column("tenant_id", sa.String),
sa.column("slug", sa.String),
sa.column("name", sa.String),
sa.column("description", sa.Text),
sa.column("permissions", sa.JSON),
sa.column("is_builtin", sa.Boolean),
sa.column("is_assignable", sa.Boolean),
sa.column("system_template_id", sa.String),
sa.column("system_required", sa.Boolean),
sa.column("created_at", sa.DateTime(timezone=True)),
sa.column("updated_at", sa.DateTime(timezone=True)),
)
for slug, role in system_roles.items():
if bind.execute(
sa.text("SELECT 1 FROM access_roles WHERE tenant_id IS NULL AND slug = :slug"),
{"slug": slug},
).first():
continue
bind.execute(
roles_table.insert().values(
id=str(uuid4()),
tenant_id=None,
slug=slug,
name=role["name"],
description=role["description"],
permissions=role["permissions"],
is_builtin=bool(role["is_builtin"]),
is_assignable=True,
system_template_id=None,
system_required=False,
created_at=now,
updated_at=now,
),
)
def upgrade() -> None:
op.create_table('core_scopes',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('slug', sa.String(length=100), nullable=False),
sa.Column('name', sa.String(length=255), nullable=False),
sa.Column('description', sa.Text(), nullable=True),
sa.Column('default_locale', sa.String(length=20), nullable=False),
sa.Column('settings', sa.JSON(), nullable=False),
sa.Column('allow_custom_groups', sa.Boolean(), nullable=True),
sa.Column('allow_custom_roles', sa.Boolean(), nullable=True),
sa.Column('allow_api_keys', sa.Boolean(), nullable=True),
sa.Column('is_active', sa.Boolean(), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint('id', name=op.f('pk_core_scopes'))
)
op.create_index(op.f('ix_core_scopes_slug'), 'core_scopes', ['slug'], unique=True)
op.create_table('access_accounts',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('email', sa.String(length=320), nullable=False),
sa.Column('normalized_email', sa.String(length=320), nullable=False),
sa.Column('display_name', sa.String(length=255), nullable=True),
sa.Column('is_active', sa.Boolean(), nullable=False),
sa.Column('auth_provider', sa.String(length=50), nullable=False),
sa.Column('password_hash', sa.String(length=500), nullable=True),
sa.Column('password_reset_required', sa.Boolean(), nullable=False),
sa.Column('last_login_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint('id', name=op.f('pk_access_accounts'))
)
op.create_index(op.f('ix_access_accounts_normalized_email'), 'access_accounts', ['normalized_email'], unique=True)
op.create_table('access_groups',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('slug', sa.String(length=100), nullable=False),
sa.Column('name', sa.String(length=255), nullable=False),
sa.Column('description', sa.Text(), nullable=True),
sa.Column('is_active', sa.Boolean(), nullable=False),
sa.Column('system_template_id', sa.String(length=36), nullable=True),
sa.Column('system_required', sa.Boolean(), nullable=False),
sa.Column('settings', sa.JSON(), nullable=False),
sa.Column('mail_profile_policy', sa.JSON(), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint('id', name=op.f('pk_access_groups')),
sa.UniqueConstraint('tenant_id', 'slug', name='uq_groups_tenant_slug')
)
op.create_index(op.f('ix_access_groups_system_template_id'), 'access_groups', ['system_template_id'], unique=False)
op.create_index(op.f('ix_access_groups_tenant_id'), 'access_groups', ['tenant_id'], unique=False)
op.create_table('access_roles',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=True),
sa.Column('slug', sa.String(length=100), nullable=False),
sa.Column('name', sa.String(length=255), nullable=False),
sa.Column('description', sa.Text(), nullable=True),
sa.Column('permissions', sa.JSON(), nullable=False),
sa.Column('is_builtin', sa.Boolean(), nullable=False),
sa.Column('is_assignable', sa.Boolean(), nullable=False),
sa.Column('system_template_id', sa.String(length=36), nullable=True),
sa.Column('system_required', sa.Boolean(), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint('id', name=op.f('pk_access_roles')),
sa.UniqueConstraint('tenant_id', 'slug', name='uq_roles_tenant_slug')
)
op.create_index(op.f('ix_access_roles_system_template_id'), 'access_roles', ['system_template_id'], unique=False)
op.create_index(op.f('ix_access_roles_tenant_id'), 'access_roles', ['tenant_id'], unique=False)
op.create_index('uq_roles_system_slug', 'access_roles', ['slug'], unique=True, sqlite_where=sa.text('tenant_id IS NULL'), postgresql_where=sa.text('tenant_id IS NULL'))
op.create_table('admin_governance_templates',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('kind', sa.String(length=20), nullable=False),
sa.Column('slug', sa.String(length=100), nullable=False),
sa.Column('name', sa.String(length=255), nullable=False),
sa.Column('description', sa.Text(), nullable=True),
sa.Column('permissions', sa.JSON(), nullable=False),
sa.Column('is_active', sa.Boolean(), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint('id', name=op.f('pk_admin_governance_templates')),
sa.UniqueConstraint('kind', 'slug', name='uq_governance_templates_kind_slug')
)
op.create_index(op.f('ix_admin_governance_templates_kind'), 'admin_governance_templates', ['kind'], unique=False)
op.create_table('attachment_blobs',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('sha256', sa.String(length=64), nullable=False),
sa.Column('size_bytes', sa.Integer(), nullable=False),
sa.Column('mime_type', sa.String(length=255), nullable=True),
sa.Column('storage_bucket', sa.String(length=255), nullable=False),
sa.Column('storage_key', sa.String(length=1000), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint('id', name=op.f('pk_attachment_blobs')),
sa.UniqueConstraint('tenant_id', 'sha256', name='uq_attachment_blobs_tenant_sha256')
)
op.create_index(op.f('ix_attachment_blobs_sha256'), 'attachment_blobs', ['sha256'], unique=False)
op.create_index(op.f('ix_attachment_blobs_tenant_id'), 'attachment_blobs', ['tenant_id'], unique=False)
op.create_table('audit_outbox_events',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('event_id', sa.String(length=36), nullable=False),
sa.Column('event_type', sa.String(length=200), nullable=False),
sa.Column('module_id', sa.String(length=100), nullable=False),
sa.Column('correlation_id', sa.String(length=128), nullable=True),
sa.Column('causation_id', sa.String(length=128), nullable=True),
sa.Column('classification', sa.String(length=40), nullable=False),
sa.Column('payload', sa.JSON(), nullable=False),
sa.Column('status', sa.String(length=20), nullable=False),
sa.Column('attempts', sa.Integer(), nullable=False),
sa.Column('next_attempt_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('dispatched_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('last_error', sa.Text(), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint('id', name=op.f('pk_audit_outbox_events')),
sa.UniqueConstraint('event_id', name='uq_audit_outbox_events_event_id')
)
op.create_index('ix_audit_outbox_events_correlation_id', 'audit_outbox_events', ['correlation_id'], unique=False)
op.create_index('ix_audit_outbox_events_event_type', 'audit_outbox_events', ['event_type'], unique=False)
op.create_index(op.f('ix_audit_outbox_events_status'), 'audit_outbox_events', ['status'], unique=False)
op.create_index('ix_audit_outbox_events_status_next_attempt_at', 'audit_outbox_events', ['status', 'next_attempt_at'], unique=False)
op.create_table('core_change_sequence',
sa.Column('id', sa.BigInteger().with_variant(sa.Integer(), 'sqlite'), autoincrement=True, nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=True),
sa.Column('module_id', sa.String(length=100), nullable=False),
sa.Column('collection', sa.String(length=150), nullable=False),
sa.Column('resource_type', sa.String(length=100), nullable=False),
sa.Column('resource_id', sa.String(length=255), nullable=False),
sa.Column('operation', sa.String(length=30), nullable=False),
sa.Column('actor_type', sa.String(length=30), nullable=True),
sa.Column('actor_id', sa.String(length=255), nullable=True),
sa.Column('payload', sa.JSON(), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint('id', name=op.f('pk_core_change_sequence'))
)
op.create_index(op.f('ix_core_change_sequence_actor_id'), 'core_change_sequence', ['actor_id'], unique=False)
op.create_index(op.f('ix_core_change_sequence_actor_type'), 'core_change_sequence', ['actor_type'], unique=False)
op.create_index(op.f('ix_core_change_sequence_collection'), 'core_change_sequence', ['collection'], unique=False)
op.create_index('ix_core_change_sequence_collection_id', 'core_change_sequence', ['collection', 'id'], unique=False)
op.create_index(op.f('ix_core_change_sequence_created_at'), 'core_change_sequence', ['created_at'], unique=False)
op.create_index(op.f('ix_core_change_sequence_module_id'), 'core_change_sequence', ['module_id'], unique=False)
op.create_index(op.f('ix_core_change_sequence_operation'), 'core_change_sequence', ['operation'], unique=False)
op.create_index('ix_core_change_sequence_resource', 'core_change_sequence', ['module_id', 'resource_type', 'resource_id'], unique=False)
op.create_index(op.f('ix_core_change_sequence_resource_id'), 'core_change_sequence', ['resource_id'], unique=False)
op.create_index(op.f('ix_core_change_sequence_resource_type'), 'core_change_sequence', ['resource_type'], unique=False)
op.create_index(op.f('ix_core_change_sequence_tenant_id'), 'core_change_sequence', ['tenant_id'], unique=False)
op.create_index('ix_core_change_sequence_tenant_module_id', 'core_change_sequence', ['tenant_id', 'module_id', 'id'], unique=False)
op.create_table('core_change_sequence_retention_floor',
sa.Column('id', sa.BigInteger().with_variant(sa.Integer(), 'sqlite'), autoincrement=True, nullable=False),
sa.Column('tenant_key', sa.String(length=36), nullable=False),
sa.Column('module_id', sa.String(length=100), nullable=False),
sa.Column('collection', sa.String(length=150), nullable=False),
sa.Column('min_valid_sequence', sa.BigInteger().with_variant(sa.Integer(), 'sqlite'), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint('id', name=op.f('pk_core_change_sequence_retention_floor')),
sa.UniqueConstraint('tenant_key', 'module_id', 'collection', name='uq_core_change_sequence_retention_scope')
)
op.create_index('ix_core_change_sequence_retention_scope', 'core_change_sequence_retention_floor', ['tenant_key', 'module_id', 'collection'], unique=False)
op.create_table('core_system_settings',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('default_locale', sa.String(length=20), nullable=False),
sa.Column('allow_tenant_custom_groups', sa.Boolean(), nullable=False),
sa.Column('allow_tenant_custom_roles', sa.Boolean(), nullable=False),
sa.Column('allow_tenant_api_keys', sa.Boolean(), nullable=False),
sa.Column('settings', sa.JSON(), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint('id', name=op.f('pk_core_system_settings'))
)
op.create_table('file_blobs',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('storage_backend', sa.String(length=50), nullable=False),
sa.Column('storage_bucket', sa.String(length=255), nullable=True),
sa.Column('storage_key', sa.String(length=1000), nullable=False),
sa.Column('checksum_sha256', sa.String(length=64), nullable=False),
sa.Column('size_bytes', sa.Integer(), nullable=False),
sa.Column('content_type', sa.String(length=255), nullable=True),
sa.Column('ref_count', sa.Integer(), nullable=False),
sa.Column('retained_until', sa.DateTime(timezone=True), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint('id', name=op.f('pk_file_blobs')),
sa.UniqueConstraint('tenant_id', 'checksum_sha256', 'size_bytes', name='uq_file_blobs_tenant_checksum_size')
)
op.create_index(op.f('ix_file_blobs_checksum_sha256'), 'file_blobs', ['checksum_sha256'], unique=False)
op.create_index(op.f('ix_file_blobs_tenant_id'), 'file_blobs', ['tenant_id'], unique=False)
op.create_table('access_group_role_assignments',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('group_id', sa.String(length=36), nullable=False),
sa.Column('role_id', sa.String(length=36), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['group_id'], ['access_groups.id'], name=op.f('fk_access_group_role_assignments_group_id_access_groups'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['role_id'], ['access_roles.id'], name=op.f('fk_access_group_role_assignments_role_id_access_roles'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_access_group_role_assignments')),
sa.UniqueConstraint('tenant_id', 'group_id', 'role_id', name='uq_group_role_assignments')
)
op.create_index(op.f('ix_access_group_role_assignments_group_id'), 'access_group_role_assignments', ['group_id'], unique=False)
op.create_index(op.f('ix_access_group_role_assignments_role_id'), 'access_group_role_assignments', ['role_id'], unique=False)
op.create_index(op.f('ix_access_group_role_assignments_tenant_id'), 'access_group_role_assignments', ['tenant_id'], unique=False)
op.create_table('access_system_role_assignments',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('account_id', sa.String(length=36), nullable=False),
sa.Column('role_id', sa.String(length=36), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['account_id'], ['access_accounts.id'], name=op.f('fk_access_system_role_assignments_account_id_access_accounts'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['role_id'], ['access_roles.id'], name=op.f('fk_access_system_role_assignments_role_id_access_roles'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_access_system_role_assignments')),
sa.UniqueConstraint('account_id', 'role_id', name='uq_system_role_assignments')
)
op.create_index(op.f('ix_access_system_role_assignments_account_id'), 'access_system_role_assignments', ['account_id'], unique=False)
op.create_index(op.f('ix_access_system_role_assignments_role_id'), 'access_system_role_assignments', ['role_id'], unique=False)
op.create_table('access_users',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('account_id', sa.String(length=36), nullable=False),
sa.Column('email', sa.String(length=320), nullable=False),
sa.Column('display_name', sa.String(length=255), nullable=True),
sa.Column('is_active', sa.Boolean(), nullable=False),
sa.Column('is_tenant_admin', sa.Boolean(), nullable=False),
sa.Column('auth_provider', sa.String(length=50), nullable=False),
sa.Column('password_hash', sa.String(length=500), nullable=True),
sa.Column('last_login_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('settings', sa.JSON(), nullable=False),
sa.Column('mail_profile_policy', sa.JSON(), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['account_id'], ['access_accounts.id'], name=op.f('fk_access_users_account_id_access_accounts'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_access_users')),
sa.UniqueConstraint('tenant_id', 'account_id', name='uq_users_tenant_account'),
sa.UniqueConstraint('tenant_id', 'email', name='uq_users_tenant_email')
)
op.create_index(op.f('ix_access_users_account_id'), 'access_users', ['account_id'], unique=False)
op.create_index(op.f('ix_access_users_email'), 'access_users', ['email'], unique=False)
op.create_index(op.f('ix_access_users_tenant_id'), 'access_users', ['tenant_id'], unique=False)
op.create_table('admin_governance_template_assignments',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('template_id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('mode', sa.String(length=20), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['template_id'], ['admin_governance_templates.id'], name=op.f('fk_admin_governance_template_assignments_template_id_admin_governance_templates'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_admin_governance_template_assignments')),
sa.UniqueConstraint('template_id', 'tenant_id', name='uq_governance_template_tenant')
)
op.create_index(op.f('ix_admin_governance_template_assignments_template_id'), 'admin_governance_template_assignments', ['template_id'], unique=False)
op.create_index(op.f('ix_admin_governance_template_assignments_tenant_id'), 'admin_governance_template_assignments', ['tenant_id'], unique=False)
op.create_table('access_api_keys',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('user_id', sa.String(length=36), nullable=False),
sa.Column('name', sa.String(length=255), nullable=False),
sa.Column('prefix', sa.String(length=16), nullable=False),
sa.Column('key_hash', sa.String(length=128), nullable=False),
sa.Column('scopes', sa.JSON(), nullable=False),
sa.Column('expires_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('last_used_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('revoked_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['user_id'], ['access_users.id'], name=op.f('fk_access_api_keys_user_id_access_users'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_access_api_keys'))
)
op.create_index(op.f('ix_access_api_keys_prefix'), 'access_api_keys', ['prefix'], unique=False)
op.create_index(op.f('ix_access_api_keys_tenant_id'), 'access_api_keys', ['tenant_id'], unique=False)
op.create_index(op.f('ix_access_api_keys_user_id'), 'access_api_keys', ['user_id'], unique=False)
op.create_table('access_auth_sessions',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('user_id', sa.String(length=36), nullable=False),
sa.Column('account_id', sa.String(length=36), nullable=False),
sa.Column('token_hash', sa.String(length=128), nullable=False),
sa.Column('csrf_token_hash', sa.String(length=128), nullable=True),
sa.Column('expires_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('last_seen_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('revoked_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('user_agent', sa.String(length=500), nullable=True),
sa.Column('ip_address', sa.String(length=100), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['account_id'], ['access_accounts.id'], name=op.f('fk_access_auth_sessions_account_id_access_accounts'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['user_id'], ['access_users.id'], name=op.f('fk_access_auth_sessions_user_id_access_users'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_access_auth_sessions'))
)
op.create_index(op.f('ix_access_auth_sessions_account_id'), 'access_auth_sessions', ['account_id'], unique=False)
op.create_index(op.f('ix_access_auth_sessions_expires_at'), 'access_auth_sessions', ['expires_at'], unique=False)
op.create_index(op.f('ix_access_auth_sessions_revoked_at'), 'access_auth_sessions', ['revoked_at'], unique=False)
op.create_index(op.f('ix_access_auth_sessions_tenant_id'), 'access_auth_sessions', ['tenant_id'], unique=False)
op.create_index(op.f('ix_access_auth_sessions_token_hash'), 'access_auth_sessions', ['token_hash'], unique=True)
op.create_index(op.f('ix_access_auth_sessions_user_id'), 'access_auth_sessions', ['user_id'], unique=False)
op.create_table('access_user_group_memberships',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('user_id', sa.String(length=36), nullable=False),
sa.Column('group_id', sa.String(length=36), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['group_id'], ['access_groups.id'], name=op.f('fk_access_user_group_memberships_group_id_access_groups'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['user_id'], ['access_users.id'], name=op.f('fk_access_user_group_memberships_user_id_access_users'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_access_user_group_memberships')),
sa.UniqueConstraint('tenant_id', 'user_id', 'group_id', name='uq_user_group_memberships')
)
op.create_index(op.f('ix_access_user_group_memberships_group_id'), 'access_user_group_memberships', ['group_id'], unique=False)
op.create_index(op.f('ix_access_user_group_memberships_tenant_id'), 'access_user_group_memberships', ['tenant_id'], unique=False)
op.create_index(op.f('ix_access_user_group_memberships_user_id'), 'access_user_group_memberships', ['user_id'], unique=False)
op.create_table('access_user_role_assignments',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('user_id', sa.String(length=36), nullable=False),
sa.Column('role_id', sa.String(length=36), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['role_id'], ['access_roles.id'], name=op.f('fk_access_user_role_assignments_role_id_access_roles'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['user_id'], ['access_users.id'], name=op.f('fk_access_user_role_assignments_user_id_access_users'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_access_user_role_assignments')),
sa.UniqueConstraint('tenant_id', 'user_id', 'role_id', name='uq_user_role_assignments')
)
op.create_index(op.f('ix_access_user_role_assignments_role_id'), 'access_user_role_assignments', ['role_id'], unique=False)
op.create_index(op.f('ix_access_user_role_assignments_tenant_id'), 'access_user_role_assignments', ['tenant_id'], unique=False)
op.create_index(op.f('ix_access_user_role_assignments_user_id'), 'access_user_role_assignments', ['user_id'], unique=False)
op.create_table('campaigns',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('created_by_user_id', sa.String(length=36), nullable=True),
sa.Column('owner_user_id', sa.String(length=36), nullable=True),
sa.Column('owner_group_id', sa.String(length=36), nullable=True),
sa.Column('external_id', sa.String(length=255), nullable=False),
sa.Column('name', sa.String(length=255), nullable=False),
sa.Column('description', sa.Text(), nullable=True),
sa.Column('status', sa.String(length=50), nullable=False),
sa.Column('current_version_id', sa.String(length=36), nullable=True),
sa.Column('settings', sa.JSON(), nullable=False),
sa.Column('mail_profile_policy', sa.JSON(), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['created_by_user_id'], ['access_users.id'], name=op.f('fk_campaigns_created_by_user_id_access_users'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['owner_group_id'], ['access_groups.id'], name=op.f('fk_campaigns_owner_group_id_access_groups'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['owner_user_id'], ['access_users.id'], name=op.f('fk_campaigns_owner_user_id_access_users'), ondelete='SET NULL'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_campaigns')),
sa.UniqueConstraint('tenant_id', 'external_id', name='uq_campaigns_tenant_external_id')
)
op.create_index(op.f('ix_campaigns_created_by_user_id'), 'campaigns', ['created_by_user_id'], unique=False)
op.create_index(op.f('ix_campaigns_external_id'), 'campaigns', ['external_id'], unique=False)
op.create_index(op.f('ix_campaigns_owner_group_id'), 'campaigns', ['owner_group_id'], unique=False)
op.create_index(op.f('ix_campaigns_owner_user_id'), 'campaigns', ['owner_user_id'], unique=False)
op.create_index(op.f('ix_campaigns_status'), 'campaigns', ['status'], unique=False)
op.create_index(op.f('ix_campaigns_tenant_id'), 'campaigns', ['tenant_id'], unique=False)
op.create_table('file_assets',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('owner_type', sa.String(length=20), nullable=False),
sa.Column('owner_user_id', sa.String(length=36), nullable=True),
sa.Column('owner_group_id', sa.String(length=36), nullable=True),
sa.Column('current_version_id', sa.String(length=36), nullable=True),
sa.Column('display_path', sa.String(length=1000), nullable=False),
sa.Column('filename', sa.String(length=500), nullable=False),
sa.Column('description', sa.Text(), nullable=True),
sa.Column('created_by_user_id', sa.String(length=36), nullable=True),
sa.Column('deleted_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('metadata', sa.JSON(), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['created_by_user_id'], ['access_users.id'], name=op.f('fk_file_assets_created_by_user_id_access_users'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['owner_group_id'], ['access_groups.id'], name=op.f('fk_file_assets_owner_group_id_access_groups'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['owner_user_id'], ['access_users.id'], name=op.f('fk_file_assets_owner_user_id_access_users'), ondelete='SET NULL'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_file_assets'))
)
op.create_index(op.f('ix_file_assets_created_by_user_id'), 'file_assets', ['created_by_user_id'], unique=False)
op.create_index(op.f('ix_file_assets_current_version_id'), 'file_assets', ['current_version_id'], unique=False)
op.create_index(op.f('ix_file_assets_deleted_at'), 'file_assets', ['deleted_at'], unique=False)
op.create_index(op.f('ix_file_assets_display_path'), 'file_assets', ['display_path'], unique=False)
op.create_index(op.f('ix_file_assets_filename'), 'file_assets', ['filename'], unique=False)
op.create_index(op.f('ix_file_assets_owner_group_id'), 'file_assets', ['owner_group_id'], unique=False)
op.create_index(op.f('ix_file_assets_owner_type'), 'file_assets', ['owner_type'], unique=False)
op.create_index(op.f('ix_file_assets_owner_user_id'), 'file_assets', ['owner_user_id'], unique=False)
op.create_index(op.f('ix_file_assets_tenant_id'), 'file_assets', ['tenant_id'], unique=False)
op.create_table('file_folders',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('owner_type', sa.String(length=20), nullable=False),
sa.Column('owner_user_id', sa.String(length=36), nullable=True),
sa.Column('owner_group_id', sa.String(length=36), nullable=True),
sa.Column('path', sa.String(length=1000), nullable=False),
sa.Column('created_by_user_id', sa.String(length=36), nullable=True),
sa.Column('deleted_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('metadata', sa.JSON(), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['created_by_user_id'], ['access_users.id'], name=op.f('fk_file_folders_created_by_user_id_access_users'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['owner_group_id'], ['access_groups.id'], name=op.f('fk_file_folders_owner_group_id_access_groups'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['owner_user_id'], ['access_users.id'], name=op.f('fk_file_folders_owner_user_id_access_users'), ondelete='SET NULL'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_file_folders'))
)
op.create_index(op.f('ix_file_folders_created_by_user_id'), 'file_folders', ['created_by_user_id'], unique=False)
op.create_index(op.f('ix_file_folders_deleted_at'), 'file_folders', ['deleted_at'], unique=False)
op.create_index(op.f('ix_file_folders_owner_group_id'), 'file_folders', ['owner_group_id'], unique=False)
op.create_index(op.f('ix_file_folders_owner_type'), 'file_folders', ['owner_type'], unique=False)
op.create_index(op.f('ix_file_folders_owner_user_id'), 'file_folders', ['owner_user_id'], unique=False)
op.create_index(op.f('ix_file_folders_path'), 'file_folders', ['path'], unique=False)
op.create_index(op.f('ix_file_folders_tenant_id'), 'file_folders', ['tenant_id'], unique=False)
op.create_index('uq_file_folders_active_group_path', 'file_folders', ['tenant_id', 'owner_group_id', 'path'], unique=True, sqlite_where=sa.text("owner_type = 'group' AND deleted_at IS NULL"), postgresql_where=sa.text("owner_type = 'group' AND deleted_at IS NULL"))
op.create_index('uq_file_folders_active_user_path', 'file_folders', ['tenant_id', 'owner_user_id', 'path'], unique=True, sqlite_where=sa.text("owner_type = 'user' AND deleted_at IS NULL"), postgresql_where=sa.text("owner_type = 'user' AND deleted_at IS NULL"))
op.create_table('mail_server_profiles',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=True),
sa.Column('scope_type', sa.String(length=20), nullable=False),
sa.Column('scope_id', sa.String(length=36), nullable=True),
sa.Column('name', sa.String(length=255), nullable=False),
sa.Column('slug', sa.String(length=100), nullable=False),
sa.Column('description', sa.Text(), nullable=True),
sa.Column('is_active', sa.Boolean(), nullable=False),
sa.Column('smtp_config', sa.JSON(), nullable=False),
sa.Column('smtp_username', sa.String(length=320), nullable=True),
sa.Column('smtp_password_encrypted', sa.Text(), nullable=True),
sa.Column('imap_config', sa.JSON(), nullable=True),
sa.Column('imap_username', sa.String(length=320), nullable=True),
sa.Column('imap_password_encrypted', sa.Text(), nullable=True),
sa.Column('created_by_user_id', sa.String(length=36), nullable=True),
sa.Column('updated_by_user_id', sa.String(length=36), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['created_by_user_id'], ['access_users.id'], name=op.f('fk_mail_server_profiles_created_by_user_id_access_users'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['updated_by_user_id'], ['access_users.id'], name=op.f('fk_mail_server_profiles_updated_by_user_id_access_users'), ondelete='SET NULL'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_mail_server_profiles')),
sa.UniqueConstraint('tenant_id', 'slug', name='uq_mail_server_profiles_tenant_slug')
)
op.create_index(op.f('ix_mail_server_profiles_created_by_user_id'), 'mail_server_profiles', ['created_by_user_id'], unique=False)
op.create_index(op.f('ix_mail_server_profiles_is_active'), 'mail_server_profiles', ['is_active'], unique=False)
op.create_index('ix_mail_server_profiles_scope', 'mail_server_profiles', ['scope_type', 'scope_id'], unique=False)
op.create_index(op.f('ix_mail_server_profiles_scope_id'), 'mail_server_profiles', ['scope_id'], unique=False)
op.create_index(op.f('ix_mail_server_profiles_scope_type'), 'mail_server_profiles', ['scope_type'], unique=False)
op.create_index(op.f('ix_mail_server_profiles_tenant_id'), 'mail_server_profiles', ['tenant_id'], unique=False)
op.create_index(op.f('ix_mail_server_profiles_updated_by_user_id'), 'mail_server_profiles', ['updated_by_user_id'], unique=False)
op.create_table('attachment_instances',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('owner_user_id', sa.String(length=36), nullable=True),
sa.Column('campaign_id', sa.String(length=36), nullable=True),
sa.Column('blob_id', sa.String(length=36), nullable=False),
sa.Column('logical_name', sa.String(length=500), nullable=True),
sa.Column('filename', sa.String(length=500), nullable=False),
sa.Column('tags', sa.JSON(), nullable=False),
sa.Column('metadata', sa.JSON(), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['blob_id'], ['attachment_blobs.id'], name=op.f('fk_attachment_instances_blob_id_attachment_blobs'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['campaign_id'], ['campaigns.id'], name=op.f('fk_attachment_instances_campaign_id_campaigns'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['owner_user_id'], ['access_users.id'], name=op.f('fk_attachment_instances_owner_user_id_access_users'), ondelete='SET NULL'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_attachment_instances'))
)
op.create_index(op.f('ix_attachment_instances_blob_id'), 'attachment_instances', ['blob_id'], unique=False)
op.create_index(op.f('ix_attachment_instances_campaign_id'), 'attachment_instances', ['campaign_id'], unique=False)
op.create_index(op.f('ix_attachment_instances_owner_user_id'), 'attachment_instances', ['owner_user_id'], unique=False)
op.create_index(op.f('ix_attachment_instances_tenant_id'), 'attachment_instances', ['tenant_id'], unique=False)
op.create_table('audit_log',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('scope', sa.String(length=20), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=True),
sa.Column('user_id', sa.String(length=36), nullable=True),
sa.Column('api_key_id', sa.String(length=36), nullable=True),
sa.Column('action', sa.String(length=100), nullable=False),
sa.Column('object_type', sa.String(length=100), nullable=True),
sa.Column('object_id', sa.String(length=100), nullable=True),
sa.Column('details', sa.JSON(), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['api_key_id'], ['access_api_keys.id'], name=op.f('fk_audit_log_api_key_id_access_api_keys'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['user_id'], ['access_users.id'], name=op.f('fk_audit_log_user_id_access_users'), ondelete='SET NULL'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_audit_log'))
)
op.create_index(op.f('ix_audit_log_action'), 'audit_log', ['action'], unique=False)
op.create_index(op.f('ix_audit_log_api_key_id'), 'audit_log', ['api_key_id'], unique=False)
op.create_index(op.f('ix_audit_log_object_id'), 'audit_log', ['object_id'], unique=False)
op.create_index(op.f('ix_audit_log_object_type'), 'audit_log', ['object_type'], unique=False)
op.create_index(op.f('ix_audit_log_scope'), 'audit_log', ['scope'], unique=False)
op.create_index('ix_audit_log_scope_created_at', 'audit_log', ['scope', 'created_at'], unique=False)
op.create_index(op.f('ix_audit_log_tenant_id'), 'audit_log', ['tenant_id'], unique=False)
op.create_index('ix_audit_log_tenant_scope_created_at', 'audit_log', ['tenant_id', 'scope', 'created_at'], unique=False)
op.create_index(op.f('ix_audit_log_user_id'), 'audit_log', ['user_id'], unique=False)
op.create_table('campaign_shares',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('campaign_id', sa.String(length=36), nullable=False),
sa.Column('target_type', sa.String(length=20), nullable=False),
sa.Column('target_id', sa.String(length=36), nullable=False),
sa.Column('permission', sa.String(length=20), nullable=False),
sa.Column('created_by_user_id', sa.String(length=36), nullable=True),
sa.Column('revoked_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['campaign_id'], ['campaigns.id'], name=op.f('fk_campaign_shares_campaign_id_campaigns'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['created_by_user_id'], ['access_users.id'], name=op.f('fk_campaign_shares_created_by_user_id_access_users'), ondelete='SET NULL'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_campaign_shares')),
sa.UniqueConstraint('campaign_id', 'target_type', 'target_id', name='uq_campaign_share_target')
)
op.create_index(op.f('ix_campaign_shares_campaign_id'), 'campaign_shares', ['campaign_id'], unique=False)
op.create_index(op.f('ix_campaign_shares_created_by_user_id'), 'campaign_shares', ['created_by_user_id'], unique=False)
op.create_index(op.f('ix_campaign_shares_revoked_at'), 'campaign_shares', ['revoked_at'], unique=False)
op.create_index(op.f('ix_campaign_shares_target_id'), 'campaign_shares', ['target_id'], unique=False)
op.create_index(op.f('ix_campaign_shares_target_type'), 'campaign_shares', ['target_type'], unique=False)
op.create_index(op.f('ix_campaign_shares_tenant_id'), 'campaign_shares', ['tenant_id'], unique=False)
op.create_table('campaign_versions',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('campaign_id', sa.String(length=36), nullable=False),
sa.Column('version_number', sa.Integer(), nullable=False),
sa.Column('raw_json', sa.JSON(), nullable=False),
sa.Column('schema_version', sa.String(length=50), nullable=False),
sa.Column('source_filename', sa.String(length=500), nullable=True),
sa.Column('source_base_path', sa.String(length=1000), nullable=True),
sa.Column('workflow_state', sa.String(length=50), nullable=False),
sa.Column('current_flow', sa.String(length=50), nullable=False),
sa.Column('current_step', sa.String(length=100), nullable=True),
sa.Column('is_complete', sa.Boolean(), nullable=False),
sa.Column('editor_state', sa.JSON(), nullable=False),
sa.Column('autosaved_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('published_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('locked_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('locked_by_user_id', sa.String(length=36), nullable=True),
sa.Column('user_lock_state', sa.String(length=20), nullable=True),
sa.Column('user_locked_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('user_locked_by_user_id', sa.String(length=36), nullable=True),
sa.Column('validation_summary', sa.JSON(), nullable=True),
sa.Column('build_summary', sa.JSON(), nullable=True),
sa.Column('execution_snapshot', sa.JSON(), nullable=True),
sa.Column('execution_snapshot_hash', sa.String(length=64), nullable=True),
sa.Column('execution_snapshot_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['campaign_id'], ['campaigns.id'], name=op.f('fk_campaign_versions_campaign_id_campaigns'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['locked_by_user_id'], ['access_users.id'], name=op.f('fk_campaign_versions_locked_by_user_id_access_users'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['user_locked_by_user_id'], ['access_users.id'], name=op.f('fk_campaign_versions_user_locked_by_user_id_access_users'), ondelete='SET NULL'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_campaign_versions')),
sa.UniqueConstraint('campaign_id', 'version_number', name='uq_campaign_versions_campaign_number')
)
op.create_index(op.f('ix_campaign_versions_campaign_id'), 'campaign_versions', ['campaign_id'], unique=False)
op.create_index(op.f('ix_campaign_versions_current_flow'), 'campaign_versions', ['current_flow'], unique=False)
op.create_index(op.f('ix_campaign_versions_execution_snapshot_hash'), 'campaign_versions', ['execution_snapshot_hash'], unique=False)
op.create_index(op.f('ix_campaign_versions_locked_by_user_id'), 'campaign_versions', ['locked_by_user_id'], unique=False)
op.create_index(op.f('ix_campaign_versions_user_lock_state'), 'campaign_versions', ['user_lock_state'], unique=False)
op.create_index(op.f('ix_campaign_versions_user_locked_by_user_id'), 'campaign_versions', ['user_locked_by_user_id'], unique=False)
op.create_index(op.f('ix_campaign_versions_workflow_state'), 'campaign_versions', ['workflow_state'], unique=False)
op.create_table('file_shares',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('file_asset_id', sa.String(length=36), nullable=False),
sa.Column('target_type', sa.String(length=20), nullable=False),
sa.Column('target_id', sa.String(length=36), nullable=False),
sa.Column('permission', sa.String(length=20), nullable=False),
sa.Column('created_by_user_id', sa.String(length=36), nullable=True),
sa.Column('revoked_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['created_by_user_id'], ['access_users.id'], name=op.f('fk_file_shares_created_by_user_id_access_users'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['file_asset_id'], ['file_assets.id'], name=op.f('fk_file_shares_file_asset_id_file_assets'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_file_shares')),
sa.UniqueConstraint('file_asset_id', 'target_type', 'target_id', 'revoked_at', name='uq_file_shares_active_target')
)
op.create_index(op.f('ix_file_shares_created_by_user_id'), 'file_shares', ['created_by_user_id'], unique=False)
op.create_index(op.f('ix_file_shares_file_asset_id'), 'file_shares', ['file_asset_id'], unique=False)
op.create_index(op.f('ix_file_shares_revoked_at'), 'file_shares', ['revoked_at'], unique=False)
op.create_index(op.f('ix_file_shares_target_id'), 'file_shares', ['target_id'], unique=False)
op.create_index(op.f('ix_file_shares_target_type'), 'file_shares', ['target_type'], unique=False)
op.create_index(op.f('ix_file_shares_tenant_id'), 'file_shares', ['tenant_id'], unique=False)
op.create_table('file_versions',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('file_asset_id', sa.String(length=36), nullable=False),
sa.Column('blob_id', sa.String(length=36), nullable=False),
sa.Column('version_number', sa.Integer(), nullable=False),
sa.Column('filename_at_upload', sa.String(length=500), nullable=False),
sa.Column('display_path_at_upload', sa.String(length=1000), nullable=False),
sa.Column('content_type', sa.String(length=255), nullable=True),
sa.Column('size_bytes', sa.Integer(), nullable=False),
sa.Column('checksum_sha256', sa.String(length=64), nullable=False),
sa.Column('created_by_user_id', sa.String(length=36), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['blob_id'], ['file_blobs.id'], name=op.f('fk_file_versions_blob_id_file_blobs'), ondelete='RESTRICT'),
sa.ForeignKeyConstraint(['created_by_user_id'], ['access_users.id'], name=op.f('fk_file_versions_created_by_user_id_access_users'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['file_asset_id'], ['file_assets.id'], name=op.f('fk_file_versions_file_asset_id_file_assets'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_file_versions')),
sa.UniqueConstraint('file_asset_id', 'version_number', name='uq_file_versions_asset_number')
)
op.create_index(op.f('ix_file_versions_blob_id'), 'file_versions', ['blob_id'], unique=False)
op.create_index(op.f('ix_file_versions_checksum_sha256'), 'file_versions', ['checksum_sha256'], unique=False)
op.create_index(op.f('ix_file_versions_created_by_user_id'), 'file_versions', ['created_by_user_id'], unique=False)
op.create_index(op.f('ix_file_versions_file_asset_id'), 'file_versions', ['file_asset_id'], unique=False)
op.create_index(op.f('ix_file_versions_tenant_id'), 'file_versions', ['tenant_id'], unique=False)
op.create_table('campaign_jobs',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('campaign_id', sa.String(length=36), nullable=False),
sa.Column('campaign_version_id', sa.String(length=36), nullable=False),
sa.Column('entry_index', sa.Integer(), nullable=False),
sa.Column('entry_id', sa.String(length=255), nullable=True),
sa.Column('recipient_email', sa.String(length=320), nullable=True),
sa.Column('subject', sa.String(length=998), nullable=True),
sa.Column('message_id_header', sa.String(length=255), nullable=True),
sa.Column('eml_storage_key', sa.String(length=1000), nullable=True),
sa.Column('eml_local_path', sa.String(length=1000), nullable=True),
sa.Column('eml_size_bytes', sa.Integer(), nullable=True),
sa.Column('eml_sha256', sa.String(length=64), nullable=True),
sa.Column('build_status', sa.String(length=50), nullable=False),
sa.Column('validation_status', sa.String(length=50), nullable=False),
sa.Column('queue_status', sa.String(length=50), nullable=False),
sa.Column('send_status', sa.String(length=50), nullable=False),
sa.Column('imap_status', sa.String(length=50), nullable=False),
sa.Column('attempt_count', sa.Integer(), nullable=False),
sa.Column('last_error', sa.Text(), nullable=True),
sa.Column('queued_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('claimed_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('claim_token', sa.String(length=36), nullable=True),
sa.Column('smtp_started_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('outcome_unknown_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('sent_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('resolved_recipients', sa.JSON(), nullable=True),
sa.Column('resolved_attachments', sa.JSON(), nullable=False),
sa.Column('issues_snapshot', sa.JSON(), nullable=False),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['campaign_id'], ['campaigns.id'], name=op.f('fk_campaign_jobs_campaign_id_campaigns'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['campaign_version_id'], ['campaign_versions.id'], name=op.f('fk_campaign_jobs_campaign_version_id_campaign_versions'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_campaign_jobs')),
sa.UniqueConstraint('campaign_version_id', 'entry_index', name='uq_campaign_jobs_version_entry')
)
op.create_index(op.f('ix_campaign_jobs_build_status'), 'campaign_jobs', ['build_status'], unique=False)
op.create_index(op.f('ix_campaign_jobs_campaign_id'), 'campaign_jobs', ['campaign_id'], unique=False)
op.create_index(op.f('ix_campaign_jobs_campaign_version_id'), 'campaign_jobs', ['campaign_version_id'], unique=False)
op.create_index(op.f('ix_campaign_jobs_claim_token'), 'campaign_jobs', ['claim_token'], unique=False)
op.create_index(op.f('ix_campaign_jobs_eml_sha256'), 'campaign_jobs', ['eml_sha256'], unique=False)
op.create_index(op.f('ix_campaign_jobs_entry_id'), 'campaign_jobs', ['entry_id'], unique=False)
op.create_index(op.f('ix_campaign_jobs_imap_status'), 'campaign_jobs', ['imap_status'], unique=False)
op.create_index(op.f('ix_campaign_jobs_queue_status'), 'campaign_jobs', ['queue_status'], unique=False)
op.create_index(op.f('ix_campaign_jobs_recipient_email'), 'campaign_jobs', ['recipient_email'], unique=False)
op.create_index(op.f('ix_campaign_jobs_send_status'), 'campaign_jobs', ['send_status'], unique=False)
op.create_index(op.f('ix_campaign_jobs_tenant_id'), 'campaign_jobs', ['tenant_id'], unique=False)
op.create_index(op.f('ix_campaign_jobs_validation_status'), 'campaign_jobs', ['validation_status'], unique=False)
op.create_table('campaign_attachment_uses',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('campaign_id', sa.String(length=36), nullable=False),
sa.Column('campaign_version_id', sa.String(length=36), nullable=False),
sa.Column('campaign_job_id', sa.String(length=36), nullable=True),
sa.Column('entry_index', sa.Integer(), nullable=True),
sa.Column('entry_id', sa.String(length=255), nullable=True),
sa.Column('file_asset_id', sa.String(length=36), nullable=False),
sa.Column('file_version_id', sa.String(length=36), nullable=False),
sa.Column('file_blob_id', sa.String(length=36), nullable=False),
sa.Column('filename_used', sa.String(length=500), nullable=False),
sa.Column('checksum_sha256', sa.String(length=64), nullable=False),
sa.Column('size_bytes', sa.Integer(), nullable=False),
sa.Column('content_type', sa.String(length=255), nullable=True),
sa.Column('use_stage', sa.String(length=20), nullable=False),
sa.Column('used_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['campaign_id'], ['campaigns.id'], name=op.f('fk_campaign_attachment_uses_campaign_id_campaigns'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['campaign_job_id'], ['campaign_jobs.id'], name=op.f('fk_campaign_attachment_uses_campaign_job_id_campaign_jobs'), ondelete='SET NULL'),
sa.ForeignKeyConstraint(['campaign_version_id'], ['campaign_versions.id'], name=op.f('fk_campaign_attachment_uses_campaign_version_id_campaign_versions'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['file_asset_id'], ['file_assets.id'], name=op.f('fk_campaign_attachment_uses_file_asset_id_file_assets'), ondelete='RESTRICT'),
sa.ForeignKeyConstraint(['file_blob_id'], ['file_blobs.id'], name=op.f('fk_campaign_attachment_uses_file_blob_id_file_blobs'), ondelete='RESTRICT'),
sa.ForeignKeyConstraint(['file_version_id'], ['file_versions.id'], name=op.f('fk_campaign_attachment_uses_file_version_id_file_versions'), ondelete='RESTRICT'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_campaign_attachment_uses')),
sa.UniqueConstraint('campaign_job_id', 'file_version_id', 'filename_used', 'use_stage', name='uq_campaign_attachment_uses_job_file_stage')
)
op.create_index(op.f('ix_campaign_attachment_uses_campaign_id'), 'campaign_attachment_uses', ['campaign_id'], unique=False)
op.create_index(op.f('ix_campaign_attachment_uses_campaign_job_id'), 'campaign_attachment_uses', ['campaign_job_id'], unique=False)
op.create_index(op.f('ix_campaign_attachment_uses_campaign_version_id'), 'campaign_attachment_uses', ['campaign_version_id'], unique=False)
op.create_index(op.f('ix_campaign_attachment_uses_entry_id'), 'campaign_attachment_uses', ['entry_id'], unique=False)
op.create_index(op.f('ix_campaign_attachment_uses_file_asset_id'), 'campaign_attachment_uses', ['file_asset_id'], unique=False)
op.create_index(op.f('ix_campaign_attachment_uses_file_blob_id'), 'campaign_attachment_uses', ['file_blob_id'], unique=False)
op.create_index(op.f('ix_campaign_attachment_uses_file_version_id'), 'campaign_attachment_uses', ['file_version_id'], unique=False)
op.create_index(op.f('ix_campaign_attachment_uses_tenant_id'), 'campaign_attachment_uses', ['tenant_id'], unique=False)
op.create_index(op.f('ix_campaign_attachment_uses_use_stage'), 'campaign_attachment_uses', ['use_stage'], unique=False)
op.create_index(op.f('ix_campaign_attachment_uses_used_at'), 'campaign_attachment_uses', ['used_at'], unique=False)
op.create_table('campaign_issues',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('tenant_id', sa.String(length=36), nullable=False),
sa.Column('campaign_id', sa.String(length=36), nullable=False),
sa.Column('campaign_version_id', sa.String(length=36), nullable=True),
sa.Column('job_id', sa.String(length=36), nullable=True),
sa.Column('severity', sa.String(length=20), nullable=False),
sa.Column('code', sa.String(length=100), nullable=False),
sa.Column('message', sa.Text(), nullable=False),
sa.Column('source', sa.String(length=255), nullable=True),
sa.Column('behavior', sa.String(length=50), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['campaign_id'], ['campaigns.id'], name=op.f('fk_campaign_issues_campaign_id_campaigns'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['campaign_version_id'], ['campaign_versions.id'], name=op.f('fk_campaign_issues_campaign_version_id_campaign_versions'), ondelete='CASCADE'),
sa.ForeignKeyConstraint(['job_id'], ['campaign_jobs.id'], name=op.f('fk_campaign_issues_job_id_campaign_jobs'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_campaign_issues'))
)
op.create_index(op.f('ix_campaign_issues_campaign_id'), 'campaign_issues', ['campaign_id'], unique=False)
op.create_index(op.f('ix_campaign_issues_campaign_version_id'), 'campaign_issues', ['campaign_version_id'], unique=False)
op.create_index(op.f('ix_campaign_issues_code'), 'campaign_issues', ['code'], unique=False)
op.create_index(op.f('ix_campaign_issues_job_id'), 'campaign_issues', ['job_id'], unique=False)
op.create_index(op.f('ix_campaign_issues_severity'), 'campaign_issues', ['severity'], unique=False)
op.create_index(op.f('ix_campaign_issues_tenant_id'), 'campaign_issues', ['tenant_id'], unique=False)
op.create_table('imap_append_attempts',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('job_id', sa.String(length=36), nullable=False),
sa.Column('attempt_number', sa.Integer(), nullable=False),
sa.Column('folder', sa.String(length=500), nullable=True),
sa.Column('status', sa.String(length=50), nullable=False),
sa.Column('error_message', sa.Text(), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['job_id'], ['campaign_jobs.id'], name=op.f('fk_imap_append_attempts_job_id_campaign_jobs'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_imap_append_attempts'))
)
op.create_index(op.f('ix_imap_append_attempts_job_id'), 'imap_append_attempts', ['job_id'], unique=False)
op.create_table('send_attempts',
sa.Column('id', sa.String(length=36), nullable=False),
sa.Column('job_id', sa.String(length=36), nullable=False),
sa.Column('attempt_number', sa.Integer(), nullable=False),
sa.Column('status', sa.String(length=50), nullable=False),
sa.Column('claim_token', sa.String(length=36), nullable=True),
sa.Column('smtp_status_code', sa.Integer(), nullable=True),
sa.Column('smtp_response', sa.Text(), nullable=True),
sa.Column('error_type', sa.String(length=255), nullable=True),
sa.Column('error_message', sa.Text(), nullable=True),
sa.Column('started_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('finished_at', sa.DateTime(timezone=True), nullable=True),
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(['job_id'], ['campaign_jobs.id'], name=op.f('fk_send_attempts_job_id_campaign_jobs'), ondelete='CASCADE'),
sa.PrimaryKeyConstraint('id', name=op.f('pk_send_attempts'))
)
op.create_index(op.f('ix_send_attempts_claim_token'), 'send_attempts', ['claim_token'], unique=False)
op.create_index(op.f('ix_send_attempts_job_id'), 'send_attempts', ['job_id'], unique=False)
op.create_index(op.f('ix_send_attempts_status'), 'send_attempts', ['status'], unique=False)
_seed_core_defaults()
def downgrade() -> None:
op.drop_table('send_attempts')
op.drop_table('imap_append_attempts')
op.drop_table('campaign_issues')
op.drop_table('campaign_attachment_uses')
op.drop_table('campaign_jobs')
op.drop_table('file_versions')
op.drop_table('file_shares')
op.drop_table('campaign_versions')
op.drop_table('campaign_shares')
op.drop_table('audit_log')
op.drop_table('attachment_instances')
op.drop_table('mail_server_profiles')
op.drop_table('file_folders')
op.drop_table('file_assets')
op.drop_table('campaigns')
op.drop_table('access_user_role_assignments')
op.drop_table('access_user_group_memberships')
op.drop_table('access_auth_sessions')
op.drop_table('access_api_keys')
op.drop_table('admin_governance_template_assignments')
op.drop_table('access_users')
op.drop_table('access_system_role_assignments')
op.drop_table('access_group_role_assignments')
op.drop_table('file_blobs')
op.drop_table('core_system_settings')
op.drop_table('core_change_sequence_retention_floor')
op.drop_table('core_change_sequence')
op.drop_table('audit_outbox_events')
op.drop_table('attachment_blobs')
op.drop_table('admin_governance_templates')
op.drop_table('access_roles')
op.drop_table('access_groups')
op.drop_table('access_accounts')
op.drop_table('core_scopes')
@@ -0,0 +1,44 @@
"""adopt German as the untouched system reference locale
Revision ID: a36d8e4f9b12
Revises: f25c9d3e7a01
Create Date: 2026-08-05 00:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "a36d8e4f9b12"
down_revision = "f25c9d3e7a01"
branch_labels = None
depends_on = None
def upgrade() -> None:
bind = op.get_bind()
if "core_system_settings" not in set(sa.inspect(bind).get_table_names()):
return
settings = sa.table(
"core_system_settings",
sa.column("id", sa.String),
sa.column("default_locale", sa.String),
sa.column("created_at", sa.DateTime(timezone=True)),
sa.column("updated_at", sa.DateTime(timezone=True)),
)
bind.execute(
settings.update()
.where(settings.c.id == "global")
.where(settings.c.default_locale == "en")
.where(settings.c.created_at == settings.c.updated_at)
.values(default_locale="de", updated_at=sa.func.now())
)
def downgrade() -> None:
# Locale selection is user-visible state. A downgrade must not overwrite a
# German value that may have been selected explicitly after this migration.
pass
@@ -0,0 +1,77 @@
"""add governed data-subject request workflow
Revision ID: b47e6f809a13
Revises: a36d8e4f9b12
Create Date: 2026-08-07 00:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "b47e6f809a13"
down_revision = "a36d8e4f9b12"
branch_labels = None
depends_on = None
def upgrade() -> None:
inspector = sa.inspect(op.get_bind())
if "core_data_subject_requests" in inspector.get_table_names():
return
op.create_table(
"core_data_subject_requests",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=False),
sa.Column("reference", sa.String(length=120), nullable=False),
sa.Column("request_kind", sa.String(length=30), nullable=False),
sa.Column("status", sa.String(length=30), nullable=False),
sa.Column("subject", sa.JSON(), nullable=False),
sa.Column("purpose", sa.String(length=1000), nullable=False),
sa.Column("legal_basis", sa.String(length=1000), nullable=True),
sa.Column("due_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("requested_by_account_id", sa.String(length=36), nullable=False),
sa.Column("search_result", sa.JSON(), nullable=False),
sa.Column("erasure_plan", sa.JSON(), nullable=False),
sa.Column("execution_result", sa.JSON(), nullable=False),
sa.Column("coverage", sa.JSON(), nullable=False),
sa.Column("evidence_sha256", sa.String(length=64), nullable=True),
sa.Column("resource_revision", sa.Integer(), nullable=False),
sa.Column("completed_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("notes", sa.Text(), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint("id", name=op.f("pk_core_data_subject_requests")),
)
op.create_index(
op.f("ix_core_data_subject_requests_tenant_id"),
"core_data_subject_requests",
["tenant_id"],
unique=False,
)
op.create_index(
op.f("ix_core_data_subject_requests_status"),
"core_data_subject_requests",
["status"],
unique=False,
)
op.create_index(
op.f("ix_core_data_subject_requests_due_at"),
"core_data_subject_requests",
["due_at"],
unique=False,
)
op.create_index(
"ix_core_data_subject_requests_tenant_status",
"core_data_subject_requests",
["tenant_id", "status"],
unique=False,
)
def downgrade() -> None:
inspector = sa.inspect(op.get_bind())
if "core_data_subject_requests" in inspector.get_table_names():
op.drop_table("core_data_subject_requests")
@@ -0,0 +1,119 @@
"""add reusable core credential envelopes
Revision ID: c91f0a72be34
Revises: 4f2a9c8e7b6d
Create Date: 2026-07-23 00:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "c91f0a72be34"
down_revision = "4f2a9c8e7b6d"
branch_labels = None
depends_on = None
def upgrade() -> None:
inspector = sa.inspect(op.get_bind())
if "core_credential_envelopes" in inspector.get_table_names():
return
op.create_table(
"core_credential_envelopes",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=True),
sa.Column("scope_type", sa.String(length=20), nullable=False),
sa.Column("scope_id", sa.String(length=255), nullable=True),
sa.Column("name", sa.String(length=255), nullable=False),
sa.Column("description", sa.Text(), nullable=True),
sa.Column("credential_kind", sa.String(length=40), nullable=False),
sa.Column("public_data", sa.JSON(), nullable=False),
sa.Column("secret_data_encrypted", sa.Text(), nullable=True),
sa.Column("secret_keys", sa.JSON(), nullable=False),
sa.Column("allowed_modules", sa.JSON(), nullable=False),
sa.Column("allowed_server_refs", sa.JSON(), nullable=False),
sa.Column("inherit_to_lower_scopes", sa.Boolean(), nullable=False),
sa.Column("is_active", sa.Boolean(), nullable=False),
sa.Column("revision", sa.String(length=36), nullable=False),
sa.Column("created_by_user_id", sa.String(length=36), nullable=True),
sa.Column("updated_by_user_id", sa.String(length=36), nullable=True),
sa.Column("deleted_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("metadata", sa.JSON(), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(
["tenant_id"],
["core_scopes.id"],
name=op.f("fk_core_credential_envelopes_tenant_id_core_scopes"),
ondelete="CASCADE",
),
sa.PrimaryKeyConstraint("id", name=op.f("pk_core_credential_envelopes")),
)
op.create_index(
"ix_core_credential_envelopes_scope",
"core_credential_envelopes",
["tenant_id", "scope_type", "scope_id"],
unique=False,
)
op.create_index(
"ix_core_credential_envelopes_active",
"core_credential_envelopes",
["tenant_id", "is_active", "deleted_at"],
unique=False,
)
op.create_index(
op.f("ix_core_credential_envelopes_tenant_id"),
"core_credential_envelopes",
["tenant_id"],
unique=False,
)
op.create_index(
op.f("ix_core_credential_envelopes_scope_type"),
"core_credential_envelopes",
["scope_type"],
unique=False,
)
op.create_index(
op.f("ix_core_credential_envelopes_scope_id"),
"core_credential_envelopes",
["scope_id"],
unique=False,
)
op.create_index(
op.f("ix_core_credential_envelopes_credential_kind"),
"core_credential_envelopes",
["credential_kind"],
unique=False,
)
op.create_index(
op.f("ix_core_credential_envelopes_is_active"),
"core_credential_envelopes",
["is_active"],
unique=False,
)
op.create_index(
op.f("ix_core_credential_envelopes_created_by_user_id"),
"core_credential_envelopes",
["created_by_user_id"],
unique=False,
)
op.create_index(
op.f("ix_core_credential_envelopes_updated_by_user_id"),
"core_credential_envelopes",
["updated_by_user_id"],
unique=False,
)
op.create_index(
op.f("ix_core_credential_envelopes_deleted_at"),
"core_credential_envelopes",
["deleted_at"],
unique=False,
)
def downgrade() -> None:
inspector = sa.inspect(op.get_bind())
if "core_credential_envelopes" in inspector.get_table_names():
op.drop_table("core_credential_envelopes")
@@ -0,0 +1,129 @@
"""add generic resource ownership transfer state
Revision ID: d03a7b9c1e5f
Revises: c91f0a72be34
Create Date: 2026-07-30 00:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "d03a7b9c1e5f"
down_revision = "c91f0a72be34"
branch_labels = None
depends_on = None
def upgrade() -> None:
inspector = sa.inspect(op.get_bind())
if "core_ownership_transfers" in inspector.get_table_names():
return
op.create_table(
"core_ownership_transfers",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("tenant_id", sa.String(length=36), nullable=False),
sa.Column("resource_module", sa.String(length=100), nullable=False),
sa.Column("resource_type", sa.String(length=100), nullable=False),
sa.Column("resource_id", sa.String(length=255), nullable=False),
sa.Column("kind", sa.String(length=40), nullable=False),
sa.Column("status", sa.String(length=50), nullable=False),
sa.Column("current_owner_type", sa.String(length=40), nullable=False),
sa.Column("current_owner_id", sa.String(length=255), nullable=False),
sa.Column("target_owner_type", sa.String(length=40), nullable=False),
sa.Column("target_owner_id", sa.String(length=255), nullable=False),
sa.Column("initiated_by_type", sa.String(length=40), nullable=False),
sa.Column("initiated_by_id", sa.String(length=255), nullable=False),
sa.Column("owner_approved_by_type", sa.String(length=40), nullable=True),
sa.Column("owner_approved_by_id", sa.String(length=255), nullable=True),
sa.Column("target_accepted_by_type", sa.String(length=40), nullable=True),
sa.Column("target_accepted_by_id", sa.String(length=255), nullable=True),
sa.Column("reason", sa.Text(), nullable=True),
sa.Column("assurance_profile", sa.String(length=80), nullable=True),
sa.Column("required_approvals", sa.Integer(), nullable=False),
sa.Column("approvals", sa.JSON(), nullable=False),
sa.Column("decisions", sa.JSON(), nullable=False),
sa.Column("idempotency_key", sa.String(length=200), nullable=False),
sa.Column("canonical_request_hash", sa.String(length=64), nullable=False),
sa.Column("expires_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("execute_after", sa.DateTime(timezone=True), nullable=True),
sa.Column("completed_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("declined_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("cancelled_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("expired_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("revision", sa.Integer(), nullable=False),
sa.Column("metadata", sa.JSON(), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint("id", name=op.f("pk_core_ownership_transfers")),
sa.UniqueConstraint(
"tenant_id",
"resource_module",
"idempotency_key",
name="uq_core_ownership_transfer_idempotency",
),
)
op.create_index(
op.f("ix_core_ownership_transfers_tenant_id"),
"core_ownership_transfers",
["tenant_id"],
unique=False,
)
op.create_index(
op.f("ix_core_ownership_transfers_kind"),
"core_ownership_transfers",
["kind"],
unique=False,
)
op.create_index(
op.f("ix_core_ownership_transfers_status"),
"core_ownership_transfers",
["status"],
unique=False,
)
op.create_index(
"ix_core_ownership_transfer_resource",
"core_ownership_transfers",
[
"tenant_id",
"resource_module",
"resource_type",
"resource_id",
"status",
],
unique=False,
)
op.create_index(
"ix_core_ownership_transfer_expiry",
"core_ownership_transfers",
["status", "expires_at"],
unique=False,
)
def downgrade() -> None:
inspector = sa.inspect(op.get_bind())
if "core_ownership_transfers" not in inspector.get_table_names():
return
op.drop_index(
"ix_core_ownership_transfer_expiry",
table_name="core_ownership_transfers",
)
op.drop_index(
"ix_core_ownership_transfer_resource",
table_name="core_ownership_transfers",
)
op.drop_index(
op.f("ix_core_ownership_transfers_status"),
table_name="core_ownership_transfers",
)
op.drop_index(
op.f("ix_core_ownership_transfers_kind"),
table_name="core_ownership_transfers",
)
op.drop_index(
op.f("ix_core_ownership_transfers_tenant_id"),
table_name="core_ownership_transfers",
)
op.drop_table("core_ownership_transfers")
@@ -0,0 +1,232 @@
"""add runtime coordination and recovery evidence
Revision ID: e14b8c2d6f90
Revises: d03a7b9c1e5f
Create Date: 2026-08-01 00:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "e14b8c2d6f90"
down_revision = "d03a7b9c1e5f"
branch_labels = None
depends_on = None
def upgrade() -> None:
inspector = sa.inspect(op.get_bind())
tables = set(inspector.get_table_names())
if "core_runtime_nodes" not in tables:
op.create_table(
"core_runtime_nodes",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("installation_id", sa.String(length=100), nullable=False),
sa.Column("node_id", sa.String(length=200), nullable=False),
sa.Column("incarnation", sa.String(length=36), nullable=False),
sa.Column("role", sa.String(length=40), nullable=False),
sa.Column("software_version", sa.String(length=80), nullable=False),
sa.Column("composition_hash", sa.String(length=64), nullable=False),
sa.Column("queues", sa.JSON(), nullable=False),
sa.Column("state", sa.String(length=30), nullable=False),
sa.Column("started_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("last_heartbeat_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("drain_requested_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("drain_reason", sa.String(length=500), nullable=True),
sa.Column("stopped_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("metadata", sa.JSON(), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint("id", name=op.f("pk_core_runtime_nodes")),
sa.UniqueConstraint(
"installation_id",
"node_id",
name="uq_core_runtime_node_installation_node",
),
)
for column in (
"installation_id",
"node_id",
"incarnation",
"role",
"composition_hash",
"state",
):
op.create_index(
op.f(f"ix_core_runtime_nodes_{column}"),
"core_runtime_nodes",
[column],
unique=False,
)
op.create_index(
"ix_core_runtime_nodes_installation_state_heartbeat",
"core_runtime_nodes",
["installation_id", "state", "last_heartbeat_at"],
unique=False,
)
if "core_distributed_leases" not in tables:
op.create_table(
"core_distributed_leases",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("installation_id", sa.String(length=100), nullable=False),
sa.Column("resource_key", sa.String(length=255), nullable=False),
sa.Column("holder_node_id", sa.String(length=200), nullable=True),
sa.Column("holder_incarnation", sa.String(length=36), nullable=True),
sa.Column("fencing_token", sa.BigInteger(), nullable=False),
sa.Column("acquired_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("renewed_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("expires_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("metadata", sa.JSON(), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint("id", name=op.f("pk_core_distributed_leases")),
sa.UniqueConstraint(
"installation_id",
"resource_key",
name="uq_core_distributed_lease_resource",
),
)
for column in (
"installation_id",
"resource_key",
"holder_node_id",
"holder_incarnation",
):
op.create_index(
op.f(f"ix_core_distributed_leases_{column}"),
"core_distributed_leases",
[column],
unique=False,
)
op.create_index(
"ix_core_distributed_leases_expiry",
"core_distributed_leases",
["installation_id", "expires_at"],
unique=False,
)
if "core_recovery_operations" not in tables:
op.create_table(
"core_recovery_operations",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("installation_id", sa.String(length=100), nullable=False),
sa.Column("module_id", sa.String(length=100), nullable=False),
sa.Column("operation_type", sa.String(length=100), nullable=False),
sa.Column("resource_type", sa.String(length=100), nullable=True),
sa.Column("resource_id", sa.String(length=255), nullable=True),
sa.Column("mode", sa.String(length=40), nullable=False),
sa.Column("status", sa.String(length=40), nullable=False),
sa.Column("idempotency_key", sa.String(length=200), nullable=False),
sa.Column("request_sha256", sa.String(length=64), nullable=False),
sa.Column("plan", sa.JSON(), nullable=False),
sa.Column("backup_reference", sa.String(length=1000), nullable=True),
sa.Column("approval_reference", sa.String(length=1000), nullable=True),
sa.Column("lease_resource_key", sa.String(length=255), nullable=True),
sa.Column("holder_node_id", sa.String(length=200), nullable=True),
sa.Column("holder_incarnation", sa.String(length=36), nullable=True),
sa.Column("fencing_token", sa.BigInteger(), nullable=True),
sa.Column("checkpoint_count", sa.Integer(), nullable=False),
sa.Column("evidence_head_sha256", sa.String(length=64), nullable=True),
sa.Column("failure_summary", sa.Text(), nullable=True),
sa.Column("started_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("completed_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("recovery_started_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("recovered_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("revision", sa.Integer(), nullable=False),
sa.Column("metadata", sa.JSON(), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint("id", name=op.f("pk_core_recovery_operations")),
sa.UniqueConstraint(
"installation_id",
"module_id",
"idempotency_key",
name="uq_core_recovery_operation_idempotency",
),
)
for column in (
"installation_id",
"module_id",
"operation_type",
"resource_type",
"resource_id",
"mode",
"status",
):
op.create_index(
op.f(f"ix_core_recovery_operations_{column}"),
"core_recovery_operations",
[column],
unique=False,
)
op.create_index(
"ix_core_recovery_operations_status_updated",
"core_recovery_operations",
["installation_id", "status", "updated_at"],
unique=False,
)
op.create_index(
"ix_core_recovery_operations_resource",
"core_recovery_operations",
["module_id", "resource_type", "resource_id"],
unique=False,
)
if "core_recovery_checkpoints" not in tables:
op.create_table(
"core_recovery_checkpoints",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("operation_id", sa.String(length=36), nullable=False),
sa.Column("sequence", sa.Integer(), nullable=False),
sa.Column("status", sa.String(length=40), nullable=False),
sa.Column("kind", sa.String(length=80), nullable=False),
sa.Column("summary", sa.Text(), nullable=False),
sa.Column("evidence", sa.JSON(), nullable=False),
sa.Column("previous_sha256", sa.String(length=64), nullable=True),
sa.Column("checkpoint_sha256", sa.String(length=64), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(
["operation_id"],
["core_recovery_operations.id"],
name=op.f(
"fk_core_recovery_checkpoints_operation_id_core_recovery_operations"
),
ondelete="CASCADE",
),
sa.PrimaryKeyConstraint("id", name=op.f("pk_core_recovery_checkpoints")),
sa.UniqueConstraint(
"operation_id",
"sequence",
name="uq_core_recovery_checkpoint_sequence",
),
)
for column in ("operation_id", "status", "checkpoint_sha256"):
op.create_index(
op.f(f"ix_core_recovery_checkpoints_{column}"),
"core_recovery_checkpoints",
[column],
unique=False,
)
op.create_index(
"ix_core_recovery_checkpoints_operation_created",
"core_recovery_checkpoints",
["operation_id", "created_at"],
unique=False,
)
def downgrade() -> None:
inspector = sa.inspect(op.get_bind())
tables = set(inspector.get_table_names())
for table in (
"core_recovery_checkpoints",
"core_recovery_operations",
"core_distributed_leases",
"core_runtime_nodes",
):
if table in tables:
op.drop_table(table)
@@ -0,0 +1,110 @@
"""add controlled first-administrator enrollment evidence
Revision ID: f25c9d3e7a01
Revises: e14b8c2d6f90
Create Date: 2026-08-04 00:00:00.000000
"""
from __future__ import annotations
from alembic import op
import sqlalchemy as sa
revision = "f25c9d3e7a01"
down_revision = "e14b8c2d6f90"
branch_labels = None
depends_on = None
def upgrade() -> None:
inspector = sa.inspect(op.get_bind())
tables = set(inspector.get_table_names())
if "core_first_admin_enrollments" not in tables:
op.create_table(
"core_first_admin_enrollments",
sa.Column("installation_id", sa.String(length=100), nullable=False),
sa.Column("state", sa.String(length=24), nullable=False),
sa.Column("generation", sa.Integer(), nullable=False),
sa.Column("token_sha256", sa.String(length=64), nullable=True),
sa.Column("token_fingerprint", sa.String(length=16), nullable=True),
sa.Column("issued_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("expires_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("consumed_at", sa.DateTime(timezone=True), nullable=True),
sa.Column("consumed_account_id", sa.String(length=36), nullable=True),
sa.Column("consumed_membership_id", sa.String(length=36), nullable=True),
sa.Column("consumed_tenant_id", sa.String(length=36), nullable=True),
sa.Column("consumed_email", sa.String(length=320), nullable=True),
sa.Column("consumed_display_name", sa.String(length=255), nullable=True),
sa.Column("consumed_request_sha256", sa.String(length=64), nullable=True),
sa.Column("issue_reason", sa.String(length=500), nullable=True),
sa.Column("event_count", sa.Integer(), nullable=False),
sa.Column("evidence_head_sha256", sa.String(length=64), nullable=True),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.Column("updated_at", sa.DateTime(timezone=True), nullable=False),
sa.PrimaryKeyConstraint(
"installation_id",
name=op.f("pk_core_first_admin_enrollments"),
),
)
op.create_index(
op.f("ix_core_first_admin_enrollments_state"),
"core_first_admin_enrollments",
["state"],
unique=False,
)
op.create_index(
op.f("ix_core_first_admin_enrollments_expires_at"),
"core_first_admin_enrollments",
["expires_at"],
unique=False,
)
inspector = sa.inspect(op.get_bind())
tables = set(inspector.get_table_names())
if "core_first_admin_enrollment_events" not in tables:
op.create_table(
"core_first_admin_enrollment_events",
sa.Column("id", sa.String(length=36), nullable=False),
sa.Column("installation_id", sa.String(length=100), nullable=False),
sa.Column("sequence", sa.Integer(), nullable=False),
sa.Column("event_type", sa.String(length=80), nullable=False),
sa.Column("generation", sa.Integer(), nullable=False),
sa.Column("evidence", sa.JSON(), nullable=False),
sa.Column("previous_sha256", sa.String(length=64), nullable=True),
sa.Column("event_sha256", sa.String(length=64), nullable=False),
sa.Column("created_at", sa.DateTime(timezone=True), nullable=False),
sa.ForeignKeyConstraint(
["installation_id"],
["core_first_admin_enrollments.installation_id"],
name=op.f(
"fk_core_first_admin_enrollment_events_installation_id_core_first_admin_enrollments"
),
ondelete="CASCADE",
),
sa.PrimaryKeyConstraint(
"id",
name=op.f("pk_core_first_admin_enrollment_events"),
),
sa.UniqueConstraint(
"installation_id",
"sequence",
name="uq_core_first_admin_enrollment_event_sequence",
),
)
for column in ("installation_id", "event_type", "event_sha256"):
op.create_index(
op.f(f"ix_core_first_admin_enrollment_events_{column}"),
"core_first_admin_enrollment_events",
[column],
unique=False,
)
def downgrade() -> None:
inspector = sa.inspect(op.get_bind())
tables = set(inspector.get_table_names())
if "core_first_admin_enrollment_events" in tables:
op.drop_table("core_first_admin_enrollment_events")
if "core_first_admin_enrollments" in tables:
op.drop_table("core_first_admin_enrollments")
+343
View File
@@ -0,0 +1,343 @@
# GovOPlaN RBAC And Resource-Access Model
**Updated:** 2026-07-11
## Authorization Equation
An operation is permitted only when every applicable layer allows it:
```text
effective role/API-key capability
AND resource ownership/share access
AND workflow state
AND active governance/policy constraints
```
RBAC answers what an actor may do. ACLs answer which resource the actor may do
it to. Workflow state and policy decide whether the operation is currently
valid.
## Identity And Scope
```text
Account global login identity
+- User membership tenant-local identity
+- direct tenant roles
+- active group memberships
| +- inherited tenant roles
+- tenant-local API keys
Account
+- direct system-role assignments
```
A browser session has one active tenant membership. System privileges do not
silently grant tenant data access. API keys remain tenant-local and receive the
intersection of their configured scopes and their owner's live tenant scopes on
every request.
## Wildcards
```text
tenant:* every canonical tenant permission
system:* every canonical system permission
* legacy alias interpreted as tenant:* only
```
Tenant wildcards never grant system permissions.
## Canonical Tenant Permissions
Campaigns:
```text
campaign:read
campaign:create
campaign:update
campaign:copy
campaign:archive
campaign:delete
campaign:share
campaign:validate
campaign:build
campaign:review
campaign:send_test
campaign:queue
campaign:control
campaign:send
campaign:retry
campaign:reconcile
```
Recipients:
```text
recipients:read
recipients:write
recipients:import
recipients:export
```
Files:
```text
files:read
files:download
files:upload
files:organize
files:share
files:delete
files:admin
```
Reports and audit:
```text
reports:read
reports:export
reports:send
audit:read
```
Mail servers:
```text
mail_servers:read
mail_servers:use
mail_servers:test
mail_servers:write
mail_servers:manage_credentials
```
Tenant administration:
```text
admin:users:read
admin:users:create
admin:users:update
admin:users:suspend
admin:groups:read
admin:groups:write
admin:groups:manage_members
admin:roles:read
admin:roles:write
admin:roles:assign
admin:api_keys:read
admin:api_keys:create
admin:api_keys:revoke
admin:settings:read
admin:settings:write
admin:policies:read
admin:policies:write
```
## Canonical System Permissions
```text
system:tenants:read
system:tenants:create
system:tenants:update
system:tenants:suspend
system:accounts:read
system:accounts:create
system:accounts:update
system:accounts:suspend
system:roles:read
system:roles:write
system:roles:assign
system:access:read
system:access:assign
system:audit:read
system:settings:read
system:settings:write
system:governance:read
system:governance:write
```
`system:access:*` remains a read/assignment boundary for cross-tenant and
system access handling. It is not a separate primary UI area.
## Default Tenant Roles
- **Owner:** `tenant:*`. At least one active operational owner must remain.
- **Tenant administrator:** settings, policies, users, groups, roles, API keys,
and read access to campaigns/files/reports/audit. Real delivery remains
separately delegable.
- **Administrator:** all tenant permissions for upgraded installations.
- **Access administrator:** membership and assignment management within
delegation limits.
- **Campaign manager:** prepare, validate, and build campaigns; no review
approval or real delivery by default.
- **Reviewer:** inspect and approve prepared campaign messages.
- **Sender:** mock-test, queue, control, send, retry, and reconcile prepared
campaigns; can use/test approved mail profiles.
- **File manager:** managed file operations without campaign delivery rights.
- **Viewer:** read campaigns, recipients, files, and reports.
- **Auditor:** read campaigns, recipient evidence, reports, and audit records;
export detailed evidence.
## Default System Roles
- **System owner:** `system:*`, protected. At least one active account must
retain it.
- **System administrator:** all specific system permissions, editable and not
protected.
- **System auditor:** read-only system registry/settings/governance/audit role,
editable.
## Delegation Ceiling
For role definition, assignment, and API-key creation:
```text
requested scopes subset of actor delegateable scopes
```
Rules:
1. Tenant roles may contain tenant scopes only.
2. System roles may contain system scopes only.
3. Definition rights and assignment rights are separate.
4. Group definition and group membership management are separate.
5. API-key scopes are intersected with the owner's current effective scopes on
every request.
6. Suspended accounts, users, tenants, or groups stop contributing access
immediately.
7. Administrative updates are field-sensitive; a user with only status
authority cannot change role assignments.
## Campaign Ownership And ACLs
A campaign has exactly one owner:
```text
owner user OR owner group
```
Additional active shares may target users or groups with `read` or `write`.
Resolution:
- owner user: read and write;
- member of owner group: read and write;
- explicit read share: read;
- explicit write share: read and write;
- `tenant:*`: tenant-wide ACL bypass;
- ordinary campaign permission without ownership/share: no object access.
ACLs do not add capabilities. A write share still needs the specific permission
for update, validation, review, send, report, retry, or reconciliation.
## Files
The current file access model distinguishes tenant-level file capabilities from
space/folder/file ownership:
| Scope | Meaning |
| --- | --- |
| `files:read` | browse/read visible file spaces and metadata |
| `files:download` | download file content when ACL permits |
| `files:upload` | create files in writable spaces |
| `files:organize` | create folders, move files, and update metadata where ACL permits |
| `files:share` | share files/spaces according to owner and policy rules |
| `files:delete` | delete or retire files where ACL permits |
| `files:admin` | tenant-wide administration of user/group file spaces |
External file connections and spaces are additionally constrained by connector
policy and owner/group assignment in the files module.
## Resource Access Explanations
The access module exposes a diagnostic endpoint for explaining why a principal
can see or operate on a concrete resource:
```text
GET /api/v1/admin/access/resource-explanation
```
Required query values:
| Field | Meaning |
| --- | --- |
| `user_id` | Tenant membership to explain. The current module UIs pass the signed-in user. |
| `resource_type` | Module-owned type such as `file`, `folder`, or `campaign`. |
| `resource_id` | Stable module resource identifier. |
| `action` | Permission/action being explained, for example `files:file:read`. |
| `tenant_id` | Optional tenant override for system/admin contexts. |
The access module always contributes effective-scope provenance for the action.
Installed modules may add resource provenance by exposing a
`ResourceAccessExplanationProvider` through a capability consumed by access.
Providers should return only facts they own, using these provenance kinds:
| Kind | Meaning |
| --- | --- |
| `resource` | The concrete resource or an explicit not-found result. |
| `owner` | Matching user/group ownership. |
| `share` | Matching explicit user/group/tenant share. |
| `policy` | Administrative bypass or policy-derived grant. |
| `role` / `right` | Scope and role provenance from access itself. |
Files currently registers `files.access` and explains file assets plus folders.
Persisted folders use their database ID. Folder rows inferred from file paths use
a deterministic virtual ID:
```text
virtual-folder:v1:<tenant_id>:<owner_type>:<owner_id>:<base64url(normalized_path)>
```
The files provider validates virtual folder IDs before returning provenance: the
tenant and owner must match the resource ID, and at least one active file must
exist below the normalized folder path. This keeps virtual folder explanations
stable without forcing every inferred tree node to become a stored folder row.
Campaign currently registers `campaigns.access` and explains the campaign
ownership/sharing object itself. Finer-grained campaign sub-objects are tracked
separately in `govoplan-campaign#50` until their resource identifiers and access
rules are decided.
Cross-user resource explanation is a policy feature, not a module-local UI
detail. Until `govoplan-policy#6` is resolved, module UIs should default to
current-user explanation and avoid importing access-admin user-picking
components.
## Mail Servers
| Scope | Meaning |
| --- | --- |
| `mail_servers:read` | profile metadata and effective policy visibility |
| `mail_servers:use` | use an allowed profile for a campaign or message flow |
| `mail_servers:test` | run connectivity tests without revealing secrets |
| `mail_servers:write` | create/update profile metadata where policy allows |
| `mail_servers:manage_credentials` | create/replace SMTP/IMAP secrets or lower-level credentials where policy allows |
Reusable encrypted profiles exist. Effective usability is also constrained by
hierarchical mail-profile policy, ownership, allowed/forced profile sets,
credential inheritance mode, lower-level override switches, and allow/deny
patterns.
## Compatibility Aliases
Compatibility aliases may exist in backend code for upgraded installations, but
new UI and docs should use canonical scopes.
Current alias direction:
```text
* -> tenant:*
system:tenants:write -> create/update/suspend tenant scopes
system:access:write -> system access assignment/write scopes
```
A separate `retention:*` family is not currently canonical because retention is
managed through system settings and tenant policy scopes. Add it only if
retention operation duties need separation from general policy/settings
administration.
+162
View File
@@ -0,0 +1,162 @@
# Action, Effect, And Automation Layer
GovOPlaN needs an automation layer because administrative processes will not be
only linear screen flows. Workflows, schedules, imports, connectors, policies,
and external events all need to request governed actions without bypassing the
same safety rules that apply to human users.
The first implementation lives in `govoplan-workflow-engine` and Core contracts.
Create a separate `govoplan-automation` module only if action planning,
schedulers, rule execution, or cross-module automation become too broad for
workflow ownership.
## Layer Purpose
The automation layer should provide:
- a typed action catalogue
- a typed effect catalogue
- consequence preview before execution
- policy and permission checks
- idempotent execution
- audit and provenance records
- retry, quarantine, and manual exception handling
- system-actor execution without hiding responsibility
Automation is not a shortcut around module boundaries. It is a governed caller
of module capabilities.
## Action Definition
An `ActionDefinition` describes something a human or system actor can request.
The versioned runtime DTOs and provider protocol live in
`govoplan_core.core.automation`; domain modules implement the protocol and
Workflow resolves providers through module capabilities rather than importing
their implementations.
Recommended fields:
- `action_key`
- owning module
- input schema
- actor requirements and required scopes
- required capabilities
- policy checks
- risk level
- reversibility class: reversible, compensatable, corrective-only, or
irreversible
- expected effects
- idempotency key strategy
- recovery mode: atomic, compensating, snapshot restore, forward recovery, or
irreversible
- concrete verification steps which prove whether the effect occurred
- audit event names
- preview provider
Examples:
- create a case from a form submission
- assign a task
- generate a document from a template
- send a postbox message
- send an email notification
- append evidence to records
- create a payment request
- call an external connector
## Effect Definition
An `EffectDefinition` describes the expected and observed result of an action.
Recommended fields:
- `effect_key`
- affected module and resource references
- external system references where applicable
- created, changed, deleted, sent, notified, locked, or retained markers
- visibility and privacy classification
- audit event references
- rollback or compensation hints
- operator-facing explanation
Effects should be recorded even when execution fails partially. This makes
manual recovery and audit review possible.
## Execution Model
The runner should execute an action plan as follows:
1. Resolve actor context: human, delegated actor, or system actor.
2. Validate input schema.
3. Resolve required module capabilities.
4. Run permission and policy checks.
5. Generate a consequence preview.
6. Reserve or verify the idempotency key.
7. Create a durable recovery operation and acquire its execution fence.
8. Persist dispatch evidence before a non-atomic provider call.
9. Execute the owning module capability.
10. Verify the provider result and every announced effect using the action's
declared recovery checks.
11. Commit the local projection and verified recovery checkpoint together.
12. Emit events and audit records.
13. Mark the command complete, retryable, quarantined, or requiring manual
intervention.
The runner must never advance workflow state past a required side effect unless
the action definition explicitly allows asynchronous completion and the pending
state is visible.
For external and asynchronous effects, providers must preserve the distinction
between:
1. requested intent;
2. approved intent;
3. dispatched command;
4. possibly executed but unconfirmed outcome;
5. confirmed observed effect;
6. reconciled, corrected, or compensated outcome.
An API timeout after dispatch is not a failed effect and must not be retried as
an ordinary process failure or a fresh command. The runner records an unknown
outcome, releases its execution authority, and blocks continuation until an
operator or provider reconciliation proves either that the effect occurred or
that it is absent.
`ActionDefinition.recovery_mode` and `recovery_verification` are part of the
provider contract. The default is conservative forward recovery with explicit
provider-result and effect verification. Atomic mode is valid only when the
provider effect and its local projection share the same database transaction.
The actor context should retain the real identity/account,
represented function or party, delegation or power, and mandate/jurisdiction
references when applicable. Domain modules remain responsible for deciding
which of those references are required for their action.
## Failure States
Automation should use explicit failure states:
- `blocked`: policy, permission, missing capability, or invalid input prevents
execution.
- `retryable`: transient transport, timeout, rate-limit, or lock conflict.
- `quarantined`: unexpected response, schema mismatch, unsafe partial result,
or unknown external state.
- `manual_required`: human decision or correction is needed.
- `compensation_required`: a later action must correct an already observed
side effect.
These states should be visible in workflow, task, and admin diagnostics.
The contract names these states explicitly as `ActionExecutionState`, alongside
`pending`, `running`, and `completed`. A provider returns observed effects even
for partial failures; the runner, not the provider, owns durable attempts,
recovery decisions, and workflow advancement.
## Boundary
Core may own stable DTOs, registry contracts, and generic audit/event hooks.
`govoplan-workflow-engine` owns the first runner because workflow is the first
module that coordinates cross-module process actions.
Domain modules own their own action providers. For example, templates own
document generation actions, postbox owns postbox message actions, and
connectors own external handoff actions.
+59
View File
@@ -0,0 +1,59 @@
# Automation Contracts
Core defines provider-neutral automation contracts. It does not own domain
schedules, Workflow graphs, or Dataflow execution.
## Invocation Envelope
`AutomationInvocation` classifies a start as `manual`, `api`, `schedule`,
`event`, `workflow`, `dependency`, `retry`, or `backfill`. It carries opaque
trigger and delivery references, event identity, correlation and causation
IDs, scheduled time, requesting actor, and bounded metadata. Domain runs store
this envelope with their immutable definition revision.
## Current Authorization
An automated trigger must not persist a user session, bearer token, API key,
or a snapshot of all current permissions. It stores:
- tenant, account, and membership IDs;
- an opaque authorization reference;
- the minimum scopes required by the pinned definition and output target.
At delivery time the optional
`auth.automationPrincipalProvider` capability resolves current account,
membership, role, group, function, and delegation state. It intersects current
authorization with the stored grant. Missing, inactive, or reduced
authorization blocks the delivery before effects occur.
## Definition Governance
The optional `policy.definitionGovernance` capability evaluates `view`,
`edit`, `run`, `reuse`, `derive`, and `automate` for system, tenant, group, and
user definitions. A decision contains an ordered source path and effective
limits. Derived definitions pin their source revision and hash and retain
ancestor ceilings. Templates are reusable definitions and cannot run on their
own.
Without Policy, domain modules use a conservative tenant-local fallback:
local definitions remain viewable/editable and active complete flows may run;
inheritance, reuse, derivation, and automation are unavailable.
## Delivery Durability
Domain trigger implementations persist idempotent deliveries before running.
`emit_platform_event` binds event delivery to the producer's SQLAlchemy
transaction. When an enabled module provides `platform.eventOutbox`, the event
is stored in that transaction and a dispatcher may retry it across restarts and
workers. The Audit module provides the current SQL outbox implementation; the
Celery `govoplan.events.dispatch_outbox` task drains it through the Dataflow
event-ingestion capability and the local event bus.
The outbox capability remains optional so reduced module combinations can
start. Without it, Core queues events on the SQLAlchemy transaction and
publishes them to the process-local bus only after the outer commit. A rollback,
including a nested savepoint rollback, discards the corresponding events. This
fallback is suitable for local or non-critical reactions, but it is not a
durable multi-worker automation source. Deployments that rely on event-triggered
work must enable the outbox provider and run the `events` worker queue and
periodic dispatcher.
+26 -4
View File
@@ -24,6 +24,15 @@ network_access = false
[projects."/mnt/DATA/git/govoplan-core"]
trust_level = "trusted"
[projects."/mnt/DATA/git/govoplan-access"]
trust_level = "trusted"
[projects."/mnt/DATA/git/govoplan-organizations"]
trust_level = "trusted"
[projects."/mnt/DATA/git/govoplan-identity"]
trust_level = "trusted"
[projects."/mnt/DATA/git/govoplan-mail"]
trust_level = "trusted"
@@ -40,25 +49,36 @@ The broad writable root reduces approval churn. The explicit project trust entri
Each active repository has an `AGENTS.md` file. These files define ownership, module boundaries, and focused commands for Codex. Keep durable project conventions there instead of repeating them in every prompt.
Documentation is part of the completion criteria for every behavior change. The
owning module must update its manifest-driven `DocumentationTopic` contributions
for affected user and administrator workflows, settings, permissions,
limitations, and operational consequences. Feature documentation remains in the
feature module; the optional `govoplan-docs` module projects those contributions
without importing feature internals. Every module manifest must retain a static
user and administrator baseline even when runtime providers add configured-state
details.
Use `~/.codex/config.toml` for personal defaults, auth/runtime settings, writable roots, and trust decisions. Avoid checking in absolute-path writable roots or model preferences unless they are intentionally team-wide.
Use Gitea issues as the canonical backlog and state log. See `docs/GITEA_ISSUES.md` for label setup, TODO import, and Codex issue update commands. Durable docs should describe stable behavior; changing work state belongs on the issue.
## Focused Verification
Use the consolidated script after changes that touch module discovery, optional integrations, shared mail components, mailbox listing, or cross-module WebUI behavior:
```bash
cd /mnt/DATA/git/govoplan-core
./scripts/check-focused.sh
cd /mnt/DATA/git/govoplan
tools/checks/check-focused.sh
```
For smaller changes, prefer the narrow command named in the relevant `AGENTS.md` file. Examples:
```bash
cd /mnt/DATA/git/govoplan-core
./.venv/bin/python -m unittest tests.test_module_system
/mnt/DATA/git/govoplan/.venv/bin/python -m unittest tests.test_module_system
cd /mnt/DATA/git/govoplan-mail
/mnt/DATA/git/govoplan-core/.venv/bin/python -m unittest discover -s tests
/mnt/DATA/git/govoplan/.venv/bin/python -m unittest discover -s tests
cd /mnt/DATA/git/govoplan-core/webui
PATH=/mnt/DATA/git/govoplan-core/webui/node_modules/.bin:/home/zemion/.nvm/versions/node/v22.22.3/bin:$PATH /home/zemion/.nvm/versions/node/v22.22.3/bin/npm run test:module-permutations
@@ -70,4 +90,6 @@ PATH=/mnt/DATA/git/govoplan-core/webui/node_modules/.bin:/home/zemion/.nvm/versi
- Avoid broad recursive scans and full builds unless the change warrants them.
- Keep generated build/test folders ignored.
- Keep optional module behavior behind core registry/capability/module metadata boundaries.
- Run `tools/checks/check-manifest-shapes.py` after module behavior or manifest changes so user/admin documentation coverage remains complete.
- Create or update Gitea issues for TODOs, follow-ups, blockers, and feature requests instead of keeping local backlog files.
- Do not start persistent dev servers unless the user asks.
+44
View File
@@ -0,0 +1,44 @@
# GovOPlaN Compatibility Inventory
This inventory classifies compatibility paths covered by
`COMPATIBILITY_POLICY.md`. It is intentionally limited to behavior that changes
accepted data, imports, permissions, or migration state. Operational fallbacks
such as Redis degradation and language fallback are not compatibility paths.
## Database Bridges
| Path | Purpose | Retention | Removal |
| --- | --- | --- | --- |
| `govoplan_core.db.migrations.reconcile_legacy_create_all_schema` | Reconciles databases created before Alembic ownership was recorded. | At least one major release after runtime aliases are removed. | Review after `1.0`; keep release-baseline tests. |
| Migration table/column aliases in `govoplan_core.db.migrations` | Detect and reconcile pre-split table ownership and migration tracks. | All tagged `0.1.x` upgrade origins plus one major release cycle. | Remove only after the corresponding baseline leaves support. |
| Access and module migration backfills for legacy permission names | Converts persisted role assignments without dropping authority. | Same as the database upgrade origin that contains the old role. | Keep migrations immutable; remove only runtime expansion at `0.2`. |
## Portable-Schema Readers
| Path | Purpose | Retention | Removal |
| --- | --- | --- | --- |
| `govoplan_core.mail.config.normalize_split_transport_credentials` | Reads pre-split SMTP/IMAP credentials and emits the split representation. | Current and previous two configuration schema versions. | Version-gate once Mail writes an explicit current schema version; reject inputs older than the two-version window. |
| `ImapServerConfig.discard_legacy_enabled` | Reads the former nested IMAP `enabled` field without writing it. | Current and previous two configuration schema versions. | Remove with the oldest accepted Mail configuration schema. |
| `govoplan_core.core.configuration_packages` readers | Reads explicitly versioned configuration-package manifests. | Current and previous two schema versions. | Retire individual readers as their version leaves the window. |
## Runtime And API Aliases
| Path | Purpose | Retention | Removal |
| --- | --- | --- | --- |
| `govoplan_core.security.scope_aliases.LEGACY_SCOPE_ALIASES` | Expands pre-granular permission names. | Tagged `0.1.x` runtime/API window. | Remove at `0.2` after role backfills and migration notes are verified. |
| `govoplan_core.security.module_permissions.LEGACY_TO_MODULE_SCOPES` | Maps pre-module-split scopes to canonical owning-module scopes. | Tagged `0.1.x` runtime/API window. | Remove at `0.2`; keep database migration evidence for one major cycle. |
| `govoplan_core.privacy.retention` | Stable import facade delegating policy-owned behavior through a capability. | Tagged `0.1.x` import window. | Remove at `0.2` after all in-tree callers use the policy contract and release notes name the replacement. |
| Legacy tenant aliases in API response schemas | Preserves active-tenant response fields used by `0.1.x` clients. | Tagged `0.1.x` API window. | Remove at `0.2` with response-schema migration notes. |
| Optional fields in Poll response references | Accepts providers built against the earlier Poll contract. | Tagged `0.1.x` runtime contract window. | Remove or require a new interface version at `0.2`. |
| Legacy single-tenant summary providers | Allows modules without the batch provider introduced in `0.1.x`. | Tagged `0.1.x` module contract window. | Remove at `0.2` after manifests advertise the batch provider contract. |
| WebUI `react-router-dom` build alias | Resolves tagged `0.1.x` module source imports to Core's single `react-router` runtime so one composition never loads two router contexts. | Tagged `0.1.x` WebUI source window. | Remove at `0.2` after every supported module tag imports `react-router` directly. |
## Removed Paths
| Path | Reason | Removed |
| --- | --- | --- |
| `govoplan_core.core.module_installer._run_restart_command_legacy` | Private wrapper had no callers and never represented a persisted or published contract. | Current development line |
| Retired `govoplan_core.api.admin` and pre-split core model imports | In-tree callers and module packages use their owning modules; regression tests prohibit reintroduction. | Before `0.1.10` |
Every new compatibility path must be added here with its classification,
diagnostic, test owner, and planned removal release.
+69
View File
@@ -0,0 +1,69 @@
# GovOPlaN Compatibility Policy
This document defines the compatibility window that release tooling, module
owners, migration authors, import/export providers, and API maintainers must
preserve. It is the source of truth for deciding whether compatibility code can
be removed.
## Database Upgrades
- A released installation from every tagged `0.1.x` version is a supported
database upgrade origin.
- The recorded public release-baseline ledger starts at `v0.1.7`; earlier
`0.1.x` tags predate production installations. If an earlier tagged database
is encountered, the release must provide or document a compatibility bridge
instead of silently treating the database as a fresh installation.
- Released migration revision IDs and recorded release heads are immutable.
- Each release must prove an upgrade from every still-supported recorded
baseline, as well as a fresh installation, before its tag is published.
- Migration-only reconciliation needed by an old database remains available for
at least one subsequent major release cycle after the corresponding runtime
compatibility path is removed.
The release-baseline format and commands are documented in
`RELEASE_DEPENDENCIES.md`.
## Configuration And Export Schemas
- Writers emit only the current schema version.
- Readers accept the current schema version and the previous two schema
versions.
- Older input is rejected with a diagnostic that identifies its version and the
required staged upgrade or conversion path.
- A module-owned configuration provider must version its input and output
schema explicitly. It must not infer an old schema from missing fields once a
versioned schema has shipped.
- Round-trip and upgrade tests must cover all three readable versions before a
schema change is released.
This window applies to configuration packages, module-owned exports, and other
portable GovOPlaN configuration artifacts. Domain interchange standards with
their own compatibility rules remain governed by the owning module.
## Runtime And API Aliases
- Compatibility aliases must emit an explicit deprecation diagnostic and point
to the supported replacement.
- New callers must use the canonical contract. In-tree callers may not add new
uses of a deprecated alias.
- Runtime imports, request fields, response fields, routes, and scope aliases
carried for the `0.1.x` split line are retired at `0.2`, with migration notes.
- An alias may be removed earlier only when it never shipped in a tag or when a
security fix requires removal. The release notes must state the exception.
- Database reconciliation code is not a runtime/API alias and follows the
longer database window above.
## Removal Checklist
Compatibility code can be removed only when all of the following are true:
1. The path is inventoried as a database bridge, portable-schema reader, or
runtime/API alias.
2. Its minimum retention window has elapsed.
3. In-tree callers and published module manifests use the replacement.
4. Upgrade, import, or API regression tests cover the retained window.
5. Diagnostics and migration notes identify any staged action operators must
take.
If one condition is not met, version-gate the compatibility path and record its
planned removal release instead of deleting it.
+401
View File
@@ -0,0 +1,401 @@
# GovOPlaN Configuration Packages
Configuration packages are reusable, versioned preconfigurations for a working
GovOPlaN capability. They sit above modules: a module installs code,
migrations, routes, permissions, and capabilities; a configuration package wires
installed modules into a concrete operational setup.
Example: an application-handling package could configure a public portal form,
a case workflow, task creation, a mail template, payment processing, access
roles, audit evidence, and the interface bindings between those modules.
The guiding reference scenario is the government-operations permit journey in
`docs/GOVOPLAN_MASTER_ROADMAP.md`: a person applies through the portal,
uploads files, receives workflow-driven messages and appointment proposals, has
a case opened, gets a permit generated from a template, and completes payment.
Configuration packages are the mechanism that should make such processes
reusable and safely importable.
This document is durable architecture context and should be mirrored to the
Gitea wiki. Active implementation should be tracked in Gitea issues.
## Goals
- Import a preconfiguration from a trusted catalog or uploaded package.
- Detect which modules, module versions, and capabilities are required.
- Detect which configuration fragments must be created, updated, or bound.
- Explain import problems and offer actionable resolution paths.
- Ask the operator only for data needed to make the package work.
- Export an existing working configuration as a reusable package.
- Support signed catalogs so approved configurations can be shared, adapted,
re-exported, and provided to other installations.
## Vocabulary
| Term | Meaning |
| --- | --- |
| Module | Installable product code with manifest, migrations, routes, permissions, WebUI, events, and capabilities. |
| Configuration | Module-owned settings and resources that make behavior concrete: workflows, forms, templates, roles, policies, connector profiles, routing rules, and defaults. |
| Interface | A typed contract between modules: capabilities, commands, events, DTOs, data bindings, and UI extension points. |
| Data | Operator- or tenant-provided values needed for a working setup: legal text, sender addresses, account mappings, payment provider credentials, templates, defaults, and optionally reference data. |
The system should communicate these four layers explicitly:
```text
module = what can exist
configuration = what should exist here
interface = how configured parts connect
data = what the operator must provide for this deployment
```
## Package Classes
The same signed package mechanism supports several explicitly named classes:
| Class | Purpose |
| --- | --- |
| `reference` | Prove a bounded journey, degraded states, recovery, and target-environment acceptance. |
| `product` | Provide a reusable operating capability such as governed communication, service-to-decision, procurement/contracts, or governed BI. |
| `sector` | Specialize terminology, forms, rules, process baselines, controls, reports, and integrations for an institutional sector without forking modules. |
| `deployment` | Declare a supported infrastructure topology, operational assumptions, health gates, and recovery evidence. |
| `integration` | Declare external systems, provider bindings, source-authority modes, maturity, required health, and reconciliation behavior. |
Package class is metadata and validation context, not additional authority. A
sector package does not become a module and cannot write another module's
tables. Packages may extend other packages only through versioned fragments and
must preserve provenance and parent constraints.
The contract enforces class-specific evidence. Reference packages require
target, recovery, security, operations, accessibility, privacy, and
documentation evidence. Deployment and integration packages require their
corresponding target/recovery/operations evidence, while integration packages
also name provider authority and minimum-maturity expectations. Preflight
blocks a missing, incompatible, or unhealthy provider. A derived package may
tighten parent module, capability, and provider requirements but cannot remove
or loosen them. Every non-documentation claim made by reference, deployment, or
integration packages carries a `sha256:<digest>` binding. Repository checks
recompute those hashes, while signed package verification protects the declared
manifest during transport.
## Package Model
A configuration package should be a signed, portable manifest plus module-owned
configuration fragments. The package is not a database dump. It should be
declarative, idempotent, tenant-aware, and explicit about secrets and external
systems.
Required package metadata:
- stable package id, name, version, description, publisher, and license
- supported GovOPlaN core/module compatibility bounds
- required modules with version constraints
- optional modules with conditional behavior
- required capabilities, commands, event subscriptions, and UI extension points
- configuration fragments grouped by owning module
- interface bindings between fragments
- data requirements to collect from the operator
- preflight checks and post-import health checks
- migration or transformation rules for older package versions
- provenance, export source metadata, and signature metadata
- package class and optional parent package/version constraints
- source-authority bindings and provider-operation expectations for every
external integration used by the package
Portable configuration schemas follow `COMPATIBILITY_POLICY.md`: providers
write only their current schema version and read that version plus the previous
two versions. Older input must produce a version-specific staged-upgrade
diagnostic.
Configuration fragments are interpreted only by the module that owns them. For
example, workflow imports workflow definitions; forms imports form schemas;
mail imports mail templates and delivery defaults; payments imports payment
profiles; access imports groups, roles, and permission assignments.
## Module Provider Contract
Each module that wants to participate should expose a configuration-package
provider capability. The core orchestrator should not understand module internals
or write module-owned tables directly.
Baseline provider responsibilities:
- publish JSON Schema or equivalent typed schemas for importable/exportable
fragments
- publish data requirements that can be rendered by a generic wizard
- validate fragments without applying them
- report missing dependencies, missing permissions, conflicts, and unsafe
changes
- produce a dry-run plan with create, update, bind, skip, and blocked actions
- apply fragments idempotently inside module-owned boundaries
- export selected module-owned configuration into portable fragments
- redact secrets and mark secret placeholders during export
- provide post-apply health checks and operator-facing diagnostics
Provider operations should be versioned. The first stable contract can be small:
```text
describe() -> schemas, supported fragment types, exported scopes
preflight(package_fragment, context) -> diagnostics, required_data, plan
apply(package_fragment, supplied_data, context) -> result, created_refs
export(selection, context) -> fragment, data_placeholders, warnings
health(import_result, context) -> diagnostics
```
The initial core contract lives in
`govoplan_core.core.configuration_packages`. Modules register providers through
module-specific capability keys and implement the `ConfigurationProvider`
protocol. The generic `configuration.provider` key names the contract and can be
used in package requirements; concrete providers such as `access.configuration`
are resolved by the admin package wizard. The first DTO surface includes package
manifests, module/capability requirements, module-owned fragments, diagnostics,
required operator data, dry-run plan items, apply results, export selections,
and export results.
The initial implementation includes provider-neutral orchestration helpers:
- `dry_run_configuration_package(...)`
- `apply_configuration_package(...)`
- `export_configuration_package(...)`
The first concrete provider is `govoplan_access.backend.configuration_provider`.
It supports access-owned `roles`, `groups`, and `group_role_assignments`
fragments and applies them idempotently. Mail and Files also register providers
for deployment configuration: Mail owns receipt-bound SMTP profiles and Files
validates the deployment-owned managed-storage binding.
### Deployment capability receipt
The installer mounts a bounded, non-secret infrastructure receipt at the path
named by `GOVOPLAN_DEPLOYMENT_CAPABILITIES_PATH`. Core validates that document
once for configuration-package context and exposes typed capability and
post-install-task records to providers. Invalid receipts fail closed. Endpoint
metadata is sanitized, and secret fields may cross this boundary only as
`env:VARIABLE_NAME` references.
Feature providers remain responsible for their own semantics:
- Mail can derive host and port from `mail.smtp`, collect missing non-secret
transport fields, and bind an existing credential-envelope id. It never
accepts or exports a username, password, token, or decrypted credential.
- Files compares `files.storage` with the effective runtime backend, endpoint,
trust marker, bucket, and presence of referenced environment secrets. Storage
remains deployment-owned, so the provider reports `skip` when they agree and
blocks drift instead of rewriting process environment or storage credentials.
- A system-scoped Mail profile requires system configuration authority. Tenant
scope is the conservative default.
- Existing Mail configuration is preserved unless a reviewed fragment
explicitly selects `on_conflict: update`. Reapplying an unchanged fragment is
a no-op.
Ops projects the same Core-validated receipt. It must not maintain a second
parser with different validation or secret-handling rules.
The admin wizard backend starts with these routes:
- `GET /api/v1/admin/configuration-packages/catalog`
- `POST /api/v1/admin/configuration-packages/dry-run`
- `POST /api/v1/admin/configuration-packages/apply`
- `POST /api/v1/admin/configuration-packages/export`
## Import Flow
1. Select a package from a trusted catalog or upload a package file.
2. Verify signature, channel, publisher trust, package compatibility, and schema.
3. Resolve required modules against installed modules and the module package
catalog.
4. Produce a dependency plan for modules that must be installed or upgraded.
5. Ask the operator for required data using a focused wizard.
6. Run dry-run preflight across all participating module providers.
7. Show problems grouped by severity, owner module, and resolution.
8. Apply module/package changes in a dependency-safe order.
9. Run post-import health checks and show the final working status.
10. Store import provenance, package version, supplied non-secret metadata, and
audit events.
The wizard should display everything necessary and nothing unnecessary. Generic
sections should cover package trust, dependency plan, required data, conflicts,
review, and result. Module-specific fields should appear only when the selected
package and installed modules require them.
## Problem Model
Import diagnostics should be structured and actionable, not free-text logs.
Every problem should include severity, owner, affected object, explanation, and
at least one suggested resolution when the system can infer it.
Common diagnostic types:
- missing module or incompatible module version
- missing capability or disabled optional integration
- missing operator-supplied data
- missing permission or insufficient administrator scope
- unresolved interface binding between modules
- conflicting existing configuration
- external service not reachable or credential test failed
- policy or tenant-governance violation
- unsafe overwrite, downgrade, or destructive change
- schema validation failure
- post-import health-check failure
Resolution actions can include install module, upgrade module, enable
integration, provide value, choose existing object, rename imported object,
skip optional fragment, replace existing configuration, or cancel import.
## Export Flow
Export should be possible from a working tenant or system configuration, but it
must be intentional about data boundaries.
1. Select export scope: system, tenant, module set, workflow, form, case type,
portal flow, or another domain object.
2. Ask participating modules to export owned fragments and data placeholders.
3. Classify exported material as configuration, reference data, sample data,
secrets, or deployment-local data.
4. Redact secrets by default and replace them with data requirements.
5. Let the operator choose whether to include reference/sample data.
6. Validate the assembled package by running the same preflight path against a
clean target context where possible.
7. Sign the package or produce an unsigned development package.
8. Optionally publish the package metadata to a configuration catalog.
Exported packages should record provenance: source GovOPlaN version, module
versions, exporter identity, timestamp, selected scope, redactions, and
validation status.
## Catalogs And Trust
Configuration catalogs should follow the existing module package catalog model:
a file-backed or remotely fetched JSON catalog with Ed25519 signatures, channel
gating, trusted key ids, and operator-controlled trust policy.
The initial catalog validator mirrors the module catalog environment model with
configuration-specific names:
```bash
GOVOPLAN_CONFIGURATION_PACKAGE_CATALOG=/srv/govoplan/configuration-catalogs/stable.json
GOVOPLAN_CONFIGURATION_PACKAGE_CATALOG_URL=https://govoplan.example/configuration-catalogs/stable.json
GOVOPLAN_CONFIGURATION_PACKAGE_CATALOG_CACHE=/srv/govoplan/runtime/configuration-catalog-cache/stable.json
GOVOPLAN_CONFIGURATION_PACKAGE_CATALOG_REQUIRE_SIGNATURE=true
GOVOPLAN_CONFIGURATION_PACKAGE_CATALOG_APPROVED_CHANNELS=stable,lts
GOVOPLAN_CONFIGURATION_PACKAGE_CATALOG_TRUSTED_KEYS_FILE=/srv/govoplan/trust/configuration-catalog-keyring.json
GOVOPLAN_CONFIGURATION_PACKAGE_CATALOG_SEQUENCE_STATE=/srv/govoplan/runtime/configuration-catalog-sequences.json
GOVOPLAN_CONFIGURATION_PACKAGE_CATALOG_ENFORCE_SEQUENCE=true
```
Catalog entries should include:
- package id, version, name, description, publisher, tags, and channel
- package artifact URL or repository ref
- signature metadata for the package artifact
- required modules and version ranges for quick compatibility display
- package category such as workflow, portal, case type, governance, connector,
report, or tenant baseline
- maturity flags such as official, verified, example, local, deprecated
The catalog verifies that a package is approved to view and import. The package
itself must still be signed and validated before use.
## Example Package
Application handling through a public portal:
Required modules:
- `portal` for the public user-facing entry point
- `forms` for the application form and validation
- `cases` for the case record and case type
- `workflow` for status transitions and automation
- `tasks` for internal task creation
- `templates` for mail/template rendering
- `mail` for delivery profile and outbound message sending
- `payments` for provider configuration and payment capture
- `access` for roles, groups, and permissions
- `audit` for import, case, mail, and payment evidence
Required configuration:
- portal route and access policy
- form schema, validation rules, confirmation texts, and attachments
- case type, fields, lifecycle states, and retention classification
- workflow steps, transitions, assignees, and completion triggers
- task template and queue/group assignment
- mail template and delivery rule
- payment product, amount rules, provider profile, and webhook binding
- access roles/groups for clerks, reviewers, supervisors, and operators
- audit event categories and evidence retention defaults
Required operator data:
- tenant or organizational unit
- service name and public contact details
- portal URL/path and legal imprint/privacy text
- mail sender profile and reply-to mailbox
- payment provider credentials or test-mode selection
- responsible groups or users for task assignment
- escalation contacts and deadlines
- template wording and localized text overrides
Potential problems:
- `payments` is missing or the provider capability is unavailable
- mail transport test fails for the selected sender profile
- imported role names conflict with existing tenant roles
- workflow references a task queue that does not exist
- public portal base URL is not configured
- required privacy/legal text is empty
- webhook endpoint cannot be validated from the payment provider
## Ownership
Core should own:
- package and catalog signature validation
- dependency resolution against installed modules and module catalogs
- orchestration of provider preflight/apply/export operations
- generic import/export APIs and audit envelope
- generic wizard shell and problem display components
Modules should own:
- schemas and semantics for their own fragments
- module-specific validation, apply, export, and health checks
- module-specific UI capabilities only when generic generated controls are not
sufficient
- redaction and classification of module-owned secrets or sensitive data
Access should own configuration fragments for users, groups, roles,
permissions, API keys, and principal mappings. It should not own workflow, mail,
payment, form, or portal semantics.
Admin should expose the operator-facing catalog, import, export, and history
screens through admin route contributions. The UI should keep the conceptual
layers visible: modules, configuration, interfaces, and data.
## Implementation Slices
1. Define package manifest and diagnostic schemas.
2. Add core configuration-package provider capability contracts.
3. Implement catalog validation and package signature verification.
4. Add dry-run orchestration against mock providers.
5. Add admin catalog/import wizard screens using provider data requirements.
6. Implement export/import providers for access-owned roles/groups first.
7. Add providers in workflow/forms/templates/mail/payments/portal/cases/tasks as
those modules mature.
8. Add round-trip tests: export a known setup, import into a clean tenant,
verify health checks.
9. Add documentation and field-level help for package authors and operators.
## Acceptance Criteria For Tracking Issue
- A Gitea tracking issue exists in `govoplan-core` for the cross-cutting
feature.
- The issue links module-specific follow-ups when implementation begins.
- The first implementation exposes typed contracts rather than direct
cross-module imports.
- A signed example catalog and unsigned development fixture exist.
- A dry-run import can identify required modules, missing data, and conflicts
before applying changes.
- An export can produce a package with secrets redacted into data requirements.
- Admin UI shows a focused wizard and actionable diagnostic list.
- Tests cover package validation, signature failure, dependency resolution,
provider preflight, export redaction, and import idempotency.
+73
View File
@@ -0,0 +1,73 @@
# Contextual Help Contract
GovOPlaN exposes context-sensitive help through `F1` and the titlebar help
control. The shell resolves a stable help identity from the focused control,
its containing surface, and the current route. The Docs module then projects
the best visible user or administrator topic for that identity.
## Resolution Order
The WebUI resolves help in this order:
1. an explicit `helpContextId` or `data-help-context-id` on the focused item
2. the focused shared control's `interfaceId`, `helpTopicId`, and label key
3. a containing dialog, card, administration section, or page surface
4. the current registered route, including dynamic module routes
5. a stable route-derived fallback when no explicit identity is available
Focused field and action contexts retain the page context as
`fallback_context`. This lets Docs show a field-specific topic when one exists
and otherwise open the owning page or module documentation instead of a generic
help page.
## Documentation Lookup
Static `DocumentationTopic` contributions announce exact contexts through
`metadata.help_contexts`. Core publishes that catalogue with the enabled module
manifest, allowing the shell to link directly to an exact topic when possible.
Docs still performs the authoritative audience, permission, configured-state,
and documentation-type filtering.
Core also maps explicit route, navigation, settings, and View surface IDs to
the module's static user or administrator documentation baseline. This makes a
page association complete by default and gives every derived field/action
context a useful fallback. Exact `metadata.help_contexts` remain the preferred
authoring mechanism for consequential or unfamiliar controls.
When there is no exact topic, Docs resolves the page fallback and then the first
visible topic owned by the module. If Docs is unavailable, the shell opens the
hosted documentation with the same context parameters.
## Authoring Controls
Core shared controls expose stable help metadata. Prefer these props rather
than adding custom `F1` listeners:
- `interfaceId` identifies a durable UI surface or action.
- `helpContextId` identifies a documentation context when it differs from the
interface identity.
- `helpModuleId` identifies the documentation-owning module when a shared
control is embedded in another module's page.
- `helpTopicId` links directly to a module-owned documentation topic.
- translated label keys provide deterministic field identities for ordinary
`FormField`, `ToggleSwitch`, search, date/time, email, button, dialog, and card
controls.
- `TableActionGroup` action definitions carry the same identities so focused
row actions can resolve consequence-specific help.
- `PageLayout` owns the page help scope and documentation identity for ordinary
headed pages. `WorkspaceLayout` owns the full-canvas workspace scope and its
labelled primary/content panes; pages inside it use `PageLayout` in
`workspace` mode and retain their own route-level help identity.
Module routes, public routes, settings sections, and administration sections
may also declare `helpContextId` and `helpTopicId`. Each module must keep a
static user/admin documentation baseline and should list its important route,
workflow, setting, permission, and limitation identities in
`metadata.help_contexts`.
## Boundary
Help identities describe presentation context; they are not authorization
claims. Opening help never bypasses route or documentation permissions. Docs
owns documentation projection, feature modules own their content, and Core owns
focus capture, context resolution, and fallback routing.
+74
View File
@@ -0,0 +1,74 @@
# DataGrid Sizing Contract
`DataGrid` turns every declared track into a deterministic pixel layout after
its container has a measurable width. The same contract is used on initial
layout, container resize, persisted-layout restore, and pointer resize.
## Column Declarations
- `width: number` or `Npx` is the preferred pixel width.
- `width: N%` is a preferred share of the measured container.
- `width: Nfr` shares residual width by fraction weight.
- `width: minmax(Npx, preferred)` combines a hard lower bound with any
supported preferred width.
- An omitted width and the legacy `fill` flag are one-fraction flexible tracks.
- `minWidth` is a hard floor. Header sort, filter, and resize controls may raise
the effective accessible floor.
- `maxWidth` bounds direct user growth and free/constrained compensation. In a
cover layout it is a preferred maximum: passive tracks may exceed it when
that is necessary to keep the table flush with its container.
## Layout Modes
| `initialFit` | `resizeBehavior` | Initial layout | Pointer resize |
| --- | --- | --- | --- |
| `container` | `cover` | Real columns fill the container. Hard-minimum excess scrolls horizontally. | Growth may create horizontal overflow. Shrink first consumes overflow, then grows resizable columns to the right; it stops before underflow. |
| `content` | `cover` | After measurement, the same cover invariant applies. | Same as cover above. |
| `container` | `constrained` | Real columns fill the container. | Peer compensation respects every hard min/max and stops the active resize when capacity is exhausted. |
| `content` | `free` | Declared content widths are retained. | Only the active column changes, so underflow or horizontal overflow is allowed. |
| `container` | `free` | The initial layout fills the container. | Later user resizing is free. A right-sticky column promotes this mode to cover so its edge remains stable. |
Sticky columns do not absorb ordinary cover residuals and are not resize
compensation targets. A last resizable column may grow into overflow. It may
shrink only by the current overflow, because shrinking farther would require a
blank filler track. Dragging farther past that stop does not bank width changes:
the column remains stopped until the pointer crosses the same boundary again.
## Persistence
Only the pixel layout resulting from an explicit user resize is persisted,
together with the container width at which the user selected it.
Persisted widths are keyed by a signature containing column IDs, declared
widths and bounds, resize affordances, sticky placement, initial fit, and resize
behavior. A changed signature discards the old override and recomputes the
declared layout.
Container reconciliation is suspended while a pointer drag is active. On
release, the already-rendered pixel layout becomes the persisted preference.
Reconciliation at that same container width never shrinks intentional user
overflow, so there is no drag-end snap. If the surrounding layout later
contracts, persisted tracks may shrink toward their hard minima. The layout
retains only the amount of horizontal overflow deliberately created by the
user; an exact-cover layout therefore remains exact-cover at narrower widths.
Legacy snapshots from the former hard-pixel persistence contract are discarded
once and recomputed from the declared column layout.
## Regression Matrix
`webui/tests/data-grid-sizing.test.ts` covers:
- pixel, percentage, fraction, `minmax`, omitted, and legacy-fill tracks;
- preferred max exhaustion without a synthetic filler column;
- hard-minimum horizontal overflow;
- fixed-only cover grids;
- persisted overrides under growth and viewport pressure;
- responsive contraction of persisted layouts without losing deliberate overflow;
- stale layout signatures;
- first and middle-column right-side compensation;
- last-resizable-column overflow, underflow stop, and reverse-pointer boundary;
- free, cover, and constrained resizing;
- cover-expanded tracks that already exceed preferred maxima; and
- preservation of the pointer layout across the commit fit.
`webui/tests/data-grid-actions.test.tsx` also verifies the rendered fixed-cover
shape and guards against reintroducing a synthetic buffer cell.
+91
View File
@@ -0,0 +1,91 @@
# Data-Subject Request Contract
This document defines the provider-neutral workflow for access and erasure
requests. It is an operational control and evidence mechanism. It does not
replace legal review, identity verification, retention policy, or the
institution's statutory response process.
## Ownership
Core owns the request aggregate, lifecycle API, optimistic concurrency,
provider discovery, export manifest, execution orchestration, and audit event
names. Modules that store subject-related data own their search, explanation,
retention, and mutation behavior through a `privacy.dsar.<module>` capability.
Core never scans module tables or guesses how a foreign resource may be
erased.
Access owns the first provider. It finds tenant memberships plus safe account,
identity, assignment, API-key, and session metadata. It does not export secret
hashes, session tokens, IP addresses, or browser fingerprints. Tenant-local
membership data can be anonymized and authentication material can be revoked.
Global accounts and identities require manual system-level review because they
may serve more than one tenant.
## Lifecycle
1. A privacy officer records a verified selector, purpose, legal basis, due
date, and internal reference.
2. Search invokes every available tenant capability independently. A provider
failure is isolated and recorded; it cannot turn an incomplete search into
a successful one.
3. The JSON export contains the request, records, provider runs, coverage,
retention reasons, execution evidence, and a SHA-256 manifest digest.
4. An erasure request produces stable provider-owned actions. Immutable
evidence generates an explicit non-executable `retain` decision.
5. Execution accepts only selected executable actions from the current plan.
It requires `If-Match`, the current resource revision, the dedicated erase
permission, and the exact `ERASE <request-id>` confirmation phrase.
6. Provider execution is idempotent. Completed or unchanged effects remain
durable in the request's execution evidence.
The API is rooted at
`/api/v1/admin/privacy/data-subject-requests`. Access exposes the independent
permissions `access:privacy:read`, `access:privacy:manage`,
`access:privacy:export`, and `access:privacy:erase`; the built-in privacy
officer role contains all four.
## Provider Rules
A provider must:
- enforce tenant ownership for every record and action;
- return stable, unique resource and action identities;
- avoid credentials, hashes, tokens, unnecessary telemetry, and unrelated
third-party data;
- distinguish mutable personal data from immutable institutional evidence;
- state a retention reason for immutable evidence;
- propose manual review instead of an automatic action when authority is
ambiguous or a resource spans tenants;
- return exactly one execution result per requested action;
- make execution idempotent and avoid committing the caller's transaction;
- keep all actual mutations inside the owning module.
Each active module without a DSAR provider is listed in coverage. This is a
deliberate fail-visible state, not proof that the module stores personal data.
An institution may call an export complete only after it has reviewed both the
provider runs and that coverage list.
## Retention And Evidence
Erasure and retention are separate decisions. Stable object IDs, authorization
history, function incumbency, formal decisions, delivery evidence, and audit
records may remain necessary for accountability. Providers expose those items
with a concrete reason and Core prevents them from being selected as executable
actions. Policy may further restrict an action, but it must never silently
loosen a provider's retention decision.
All lifecycle mutations and exports produce tenant audit events. The request
stores an evidence digest after every revision. This digest detects accidental
or unauthorized mutation of the aggregate; it is not a digital signature or a
substitute for signed recovery evidence.
## Current Limits
- Access is the first native provider. Other enabled modules appear in the
coverage list until they add a provider or an explicit no-subject-data
declaration is standardized.
- Verification of the requester's identity and statutory deadline escalation
remain institutional workflows outside this API.
- Global account or identity erasure is deliberately manual.
- Exports are JSON. A human-readable signed response package remains a later
Reporting/Templates integration.
+70
View File
@@ -0,0 +1,70 @@
# Dependency Audits
GovOPlaN keeps dependency vulnerability checks reproducible but separate from
the fast local smoke suite, because both Python and npm audits need network
metadata and can fail for newly disclosed advisories without a source change.
## Local Workflow
Install the whole-product development dependencies once from the meta
repository:
```bash
cd /mnt/DATA/git/govoplan
./.venv/bin/python -m pip install -r requirements-dev.txt
```
Run both backend and WebUI production audits:
```bash
cd /mnt/DATA/git/govoplan
bash tools/checks/check-dependency-audits.sh
```
The script runs:
- `tools/checks/check-dependency-hygiene.sh` for pip resolver consistency, stale
legacy editable package metadata, deprecated framework constants, and the
Starlette `TestClient` deprecation smoke when test dependencies are present
- `python -m pip_audit --progress-spinner off`
- `npm audit --omit=dev` in `webui`
For fast local checks without vulnerability metadata lookups, run:
```bash
cd /mnt/DATA/git/govoplan
CHECK_TESTCLIENT_DEPRECATIONS=1 \
bash tools/checks/check-dependency-hygiene.sh
```
This is also part of `govoplan/tools/checks/check-focused.sh`, so resolver drift and
deprecation regressions fail close to the code change that introduced them.
Override tool paths when testing from a disposable environment:
```bash
PYTHON=/tmp/govoplan-audit/bin/python \
NPM=/home/zemion/.nvm/versions/node/v22.22.3/bin/npm \
GOVOPLAN_CORE_ROOT=/mnt/DATA/git/govoplan-core \
bash /mnt/DATA/git/govoplan/tools/checks/check-dependency-audits.sh
```
## CI Workflow
`govoplan/.gitea/workflows/dependency-audit.yml` installs release dependencies from
tagged package refs, installs `pip-audit`, and runs the same script on pushes,
pull requests, and a weekly schedule.
The workflow intentionally uses release dependency refs instead of local
`file:` or editable sibling paths. Development lockfiles may keep local module
links, but release audit results should represent the installable product.
## Recording Results
When closing or triaging dependency-audit issues, add a short dated note under
`docs/audits/`. Record:
- the commands that were run
- whether Python and npm passed
- any advisories accepted as temporary risk
- follow-up issue links for required upgrades
+553
View File
@@ -0,0 +1,553 @@
# GovOPlaN Deployment Operator Guide
This guide defines the current install/runtime configuration contract and the
operator flow for a production-realistic self-hosted deployment. Keep secrets in
the deployment environment or a secret manager; do not commit populated `.env`
files.
## Runtime Configuration Contract
Worker and queue observability is provider-neutral. Runtime modules register a
bounded `RuntimeWorkStatusProviderRegistration` with Core; the Ops module
projects its sanitized status without importing Celery, Redis, or module job
implementations. Providers must use explicit `null` values for unsupported
queue depth, active/reserved work, failure count, and heartbeat evidence. An
unavailable metric must never be interpreted as zero or as proof of health.
The standard Core adapter reports the configured Celery/Redis runtime and
combines its bounded inspection result with registered worker heartbeat and
stale-threshold evidence.
Self-hosted installability follows the staged approach documented in
`SELF_HOSTED_INSTALLABILITY.md`: generate an explicit env template, validate it,
run production-like rehearsal with Compose-backed dependencies, then use the
installer CLI/daemon for package mutation under maintenance mode.
Generate a deployment-local template:
```bash
cd /mnt/DATA/git/govoplan
./.venv/bin/python -m govoplan_core.commands.config env-template \
--profile self-hosted \
--generate-secrets \
--output .env.self-hosted
```
Validate the active shell environment before migration or startup:
```bash
set -a
. .env.self-hosted
set +a
./.venv/bin/python -m govoplan_core.commands.config validate --profile self-hosted
```
### Required Runtime Identity
| Setting | Required outside dev | Purpose |
| --- | --- | --- |
| `APP_ENV` | yes | Runtime profile. Use `prod`, `staging`, or a deployment-specific value outside local development. |
| `MASTER_KEY_B64` | yes | Fernet key or base64 encoded 32-byte deployment root used for encrypted module secrets and, when enabled, the Encryption module's local server-envelope provider. Rotate only through an explicit provider-aware migration plan. |
| `DATABASE_URL` | yes | SQLAlchemy database URL for core and installed modules. SQLite is supported for dev/small installs; PostgreSQL is the preferred production target. |
| `ENABLED_MODULES` | yes | Comma-separated startup module set. Keep `tenancy,access` enabled; keep `admin` enabled for operator UI. |
Generate a local key for a new non-production environment:
```bash
python - <<'PY'
from cryptography.fernet import Fernet
print(Fernet.generate_key().decode())
PY
```
### Database And Migrations
| Setting | Default | Notes |
| --- | --- | --- |
| `DATABASE_URL` | `postgresql+psycopg://govoplan_dev@127.0.0.1:5432/govoplan_dev` | Local development and production-like profiles use PostgreSQL. Use `GOVOPLAN_DEV_DATABASE_BACKEND=sqlite` only for disposable SQLite runs. |
| `GOVOPLAN_MIGRATION_TRACK` | `release` | Use the release track for normal runtime and deployments. Use `dev` only for fresh/disposable databases that intentionally replay detailed development migrations. |
| `DEV_AUTO_MIGRATE_ENABLED` | `true` | Dev convenience only. Production should run migration commands explicitly during deployment. |
| `DEV_BOOTSTRAP_ENABLED` | `false` | Dev bootstrap only. `govoplan_core.devserver` and `govoplan/tools/launch/launch-dev.sh` default it to `true`; use controlled first-admin creation outside dev. |
| `FIRST_ADMIN_ENROLLMENT_TTL_SECONDS` | `1800` | Lifetime of a locally issued production enrollment credential. Allowed range: 60 seconds to 24 hours. |
| `FIRST_ADMIN_ENROLLMENT_FILE` | `/run/govoplan/first-admin-enrollment.json` | Local operator artifact. The command creates it with mode `0600` and never prints the secret. |
Operator rule: take a database backup before applying migrations or destructive
module retirement. For non-SQLite databases, configure deployment-specific
backup/restore hooks for the module installer.
### PostgreSQL Production Target
PostgreSQL is the primary development and production target. SQLite remains
supported only for tiny disposable profiles and unit-test style smoke runs.
Production/staging deployments should use a managed PostgreSQL database and
explicit migration commands.
Install the full release profile from the meta repository so core, modules, and
the `psycopg` driver are available:
```bash
cd /mnt/DATA/git/govoplan
./.venv/bin/python -m pip install -r requirements-release.txt
```
Example runtime database URLs:
```bash
export DATABASE_URL='postgresql+psycopg://govoplan:change-me@db.example.internal:5432/govoplan'
export GOVOPLAN_DATABASE_URL_PGTOOLS='postgresql://govoplan:change-me@db.example.internal:5432/govoplan'
```
Use the SQLAlchemy URL for GovOPlaN. Use the pg-tools URL for `pg_dump`,
`pg_restore`, and `psql`; these tools do not understand the
`postgresql+psycopg://` driver marker.
Bootstrap or upgrade the schema explicitly during deployment:
```bash
export APP_ENV=prod
export ENABLED_MODULES=tenancy,access,admin,policy,audit,campaigns,files,mail,calendar,docs,ops
./.venv/bin/python -m govoplan_core.commands.init_db \
--database-url "$DATABASE_URL"
```
Backup and restore-check before migration-bearing package changes:
```bash
pg_dump --format=custom \
--file "$PWD/runtime/govoplan-$(date +%Y%m%d%H%M%S).dump" \
"$GOVOPLAN_DATABASE_URL_PGTOOLS"
pg_restore --list "$PWD/runtime/govoplan-YYYYMMDDHHMMSS.dump" >/dev/null
```
Restore a checked backup to the target database:
```bash
pg_restore --clean --if-exists \
--dbname "$GOVOPLAN_DATABASE_URL_PGTOOLS" \
"$PWD/runtime/govoplan-YYYYMMDDHHMMSS.dump"
```
For local development, create the host database described in
`/mnt/DATA/git/govoplan/dev/postgres/README.md`, then run:
```bash
cd /mnt/DATA/git/govoplan
./.venv/bin/python -m govoplan_core.commands.init_db \
--database-url postgresql+psycopg://govoplan_dev@127.0.0.1:5432/govoplan_dev \
--with-dev-data
./.venv/bin/python -m govoplan_core.devserver --smoke --no-reload
tools/launch/launch-dev.sh
```
For disposable local validation against a throwaway PostgreSQL instance, use
the bundled PostgreSQL testbed:
```bash
cd /mnt/DATA/git/govoplan/dev/postgres
cp .env.example .env
docker compose --env-file .env up -d
cd /mnt/DATA/git/govoplan
set -a
. /mnt/DATA/git/govoplan/dev/postgres/.env
set +a
tools/checks/postgres-integration-check.py \
--database-url "$GOVOPLAN_POSTGRES_DATABASE_URL" \
--reset-schema
```
The integration check runs migrations and startup smoke checks across the
standard module permutations. It first requires the retirement atomicity proof,
using Files' real secret-owning provider and Audit's persistent recorder. That
proof uses only random, test-owned schemas and cleans them afterward; it does
not reset `public`. `--reset-schema` is destructive and belongs only on
throwaway databases. Do not pass `--skip-retirement-atomicity` when collecting
release evidence.
### Broker And Workers
| Setting | Default | Notes |
| --- | --- | --- |
| `REDIS_URL` | `redis://redis:6379/0` | Celery broker/result backend when async workers are enabled. |
| `CELERY_ENABLED` | `false` | Local/dev can send synchronously. Production campaign delivery should run workers and set this to `true`. |
| `CELERY_QUEUES` | `send_email,append_sent,notifications,mail,calendar,dataflow,workflow,postbox,events,idm,default` | Queue list expected by worker/process manager definitions. Keep this aligned with every enabled module task route; Ops reports missing worker queue consumers. |
| `CELERY_VISIBILITY_TIMEOUT_SECONDS` | `3600` | Maximum time before Redis may redeliver work left unacknowledged by a lost worker. Set this above the longest supported task duration; changing it requires a worker-loss acceptance drill. |
| `PLATFORM_EVENT_OUTBOX_MAX_ATTEMPTS` | `8` | Failed durable consumer deliveries are quarantined after this many attempts. |
| `PLATFORM_EVENT_OUTBOX_TERMINAL_RETENTION_DAYS` | `90` | Successful event envelopes older than this are removed by the daily retention task. Quarantined evidence is retained. |
Worker command:
```bash
python -m celery -A govoplan_core.celery_app:celery worker \
--queues send_email,append_sent,notifications,mail,calendar,dataflow,workflow,postbox,events,idm,default \
--loglevel INFO
```
Run Celery beat as a separately supervised process. Its built-in one-minute
schedule recovers Calendar outbox rows left behind by broker failures, process
crashes, and expired worker leases:
```bash
python -m celery -A govoplan_core.celery_app:celery beat --loglevel INFO
```
Before promoting a worker composition, run the repository worker-runtime drill
against the same Redis and Core build. It uses the bounded
`govoplan.worker.acceptance` task and records publish/consume, retry, warm
SIGTERM, and worker-loss redelivery evidence without accessing tenant data.
Production evidence must use the deployed queue configuration and a visibility
timeout that is longer than every supported business task.
### Storage
| Setting | Default | Notes |
| --- | --- | --- |
| `FILE_STORAGE_BACKEND` | `local` | Use `local` for dev/small deployments; use object storage when files must scale independently. |
| `FILE_STORAGE_LOCAL_ROOT` | `runtime/files` | Must live on durable storage and be backed up when `FILE_STORAGE_BACKEND=local`. |
| `FILE_STORAGE_LOCAL_FALLBACK_ROOTS` | empty | Read-only fallback roots for migrated local files. |
| `FILE_STORAGE_S3_ENDPOINT_URL` | empty | Object-store endpoint for the files module. |
| `FILE_STORAGE_S3_REGION` | empty | Object-store region. |
| `FILE_STORAGE_S3_ACCESS_KEY_ID` | empty | Secret; inject through deployment environment. |
| `FILE_STORAGE_S3_SECRET_ACCESS_KEY` | empty | Secret; inject through deployment environment. |
| `FILE_STORAGE_S3_BUCKET` | `files` | Managed-file object bucket. |
| `FILE_STORAGE_S3_DEPLOYMENT_MANAGED` | `false` | Reserved for the installer-owned `http://garage:3900` service. It does not authorize arbitrary internal or external S3 endpoints. |
| `FILE_ARCHIVE_MAX_ENTRIES` | `10000` | Maximum declared entries accepted during archive preview and confirmation. |
| `FILE_ARCHIVE_MAX_EXPANDED_BYTES` | `2147483648` | Maximum actual expanded bytes across selected archive content. |
| `FILE_ARCHIVE_MAX_EXPANSION_RATIO` | `100` | Maximum expanded-to-compressed archive ratio. |
| `FILE_ARCHIVE_PREVIEW_TTL_SECONDS` | `1800` | Lifetime of the sealed actor/archive/destination-bound preview token. |
Legacy `S3_*` settings remain for older storage paths but new deployments should
prefer `FILE_STORAGE_*`.
### HTTP, Cookies, And Base URLs
| Setting | Default | Notes |
| --- | --- | --- |
| `CORS_ORIGINS` | local dev origins | Set to the exact WebUI origins in staging/production. |
| `GOVOPLAN_TRUSTED_HOSTS` | empty | Exact API host names accepted by the application. Production-like validation requires an explicit list; narrowly scoped `*.example.org` entries are supported. |
| `FORWARDED_ALLOW_IPS` | Uvicorn default | Address or network of the trusted reverse proxy. Never use `*` in production-like deployments. |
| `AUTH_SESSION_COOKIE_NAME` | configured default | Change only through a controlled rollout because it logs users out. |
| `AUTH_CSRF_COOKIE_NAME` | configured default | Must match WebUI/API deployment. |
| `AUTH_COOKIE_SECURE` | `false` | Set `true` behind HTTPS. |
| `AUTH_COOKIE_SAMESITE` | `lax` | Use a stricter value only after testing login and CSRF flows. |
| `AUTH_COOKIE_DOMAIN` | empty | Set only when the API and WebUI intentionally share a parent domain. |
| `GOVOPLAN_HTTP_HSTS_SECONDS` | `31536000` in production, otherwise `0` | Emitted only for HTTPS requests. Set `0` while rehearsing a deployment that is not yet HTTPS-only. |
| `GOVOPLAN_HTTP_MAX_REQUEST_BODY_BYTES` | `536870912` (512 MiB) | Deployment hard ceiling; file and module APIs apply their own lower limits where appropriate. |
Interactive password login is enabled with fixed-window limits of 10 failures
per normalized identity and 100 failures per direct client over 900 seconds.
`AUTH_LOGIN_THROTTLE_*` settings change those limits. Counters use `REDIS_URL`
when Redis is reachable so replicas share state. Production-like startup fails
when throttling is enabled without `REDIS_URL`. Set
`GOVOPLAN_ALLOW_PROCESS_LOCAL_LOGIN_THROTTLE=true` only as an explicit
single-process risk acceptance. A bounded process-local fallback keeps
development and temporary Redis outages functional, with per-process
enforcement until Redis recovers; monitor Redis because protection is weaker
during that fallback.
### Outbound Connector Egress
| Setting | Default | Notes |
| --- | --- | --- |
| `GOVOPLAN_CONNECTOR_ALLOW_PRIVATE_NETWORKS` | `true` in dev/test, otherwise `false` | Deployment-wide decision. Set `true` only when pinned HTTP(S), DAV, SMTP, or IMAP transports must reach internal addresses. It does not enable an SDK transport that cannot pin every peer. |
| `GOVOPLAN_CONNECTOR_MAX_STRUCTURED_RESPONSE_BYTES` | `16777216` (16 MiB) | Maximum buffered JSON, XML, iCalendar, vCard, catalog, and connector error response. |
| `GOVOPLAN_CONNECTOR_MAX_FILE_TRANSFER_BYTES` | `536870912` (512 MiB) | Hard upper bound for a single remote file; module upload limits may be lower. |
| `GOVOPLAN_CONNECTOR_SECRET_ENV_ALLOWLIST` | empty | Comma-separated exact environment names usable by deployment-owned connector profiles. Tenant/API-managed profiles cannot select process variables, even when a name is listed. |
| `GOVOPLAN_CONNECTOR_CA_BUNDLE_ALLOWLIST` | empty | Comma-separated exact absolute CA bundle paths. Mount the same files at the same paths on every API and connector worker. |
Production-like configuration validation requires the private-network choice to
be explicit. HTTP connector downloads are streamed up to the configured bound,
and credential-bearing DAV redirects remain confined to their configured
origin.
The urllib, HTTPX/httpcore, SMTP, and IMAP transports resolve, validate, and
connect to the same approved address record while retaining the original host
for HTTP Host, TLS SNI, and certificate verification. Live SMB and S3 access
fails closed in both public-only and private-network deployments: the current
SDK transports cannot pin every initial and secondary peer or revalidate every
SDK-managed redirect/referral. An explicit IP endpoint does not bypass this
rule. Production deployments should still enforce the same decision at their
worker/container egress firewall or outbound proxy as a second boundary.
File connector TLS verification may be disabled only in dev/test. A custom CA
bundle must be an existing regular file whose resolved absolute path is listed
in `GOVOPLAN_CONNECTOR_CA_BUNDLE_ALLOWLIST`. Environment-backed file connector
credentials are supported only in deployment-owned connector JSON and require
their exact names in `GOVOPLAN_CONNECTOR_SECRET_ENV_ALLOWLIST`; UI/API profiles
must use encrypted stored credentials or a scoped secret-provider reference.
Public URLs are currently supplied by deployment/reverse-proxy configuration and
module settings. Do not hardcode them in core; configuration packages should ask
for portal, WebUI, postbox, and notification URLs when they become relevant.
Uvicorn applies `X-Forwarded-*` only from `FORWARDED_ALLOW_IPS`; keep that value
aligned with the reverse proxy and do not expose the application server directly
through the same trusted address range.
### Module Catalogs, Licenses, And Trust Roots
| Setting | Purpose |
| --- | --- |
| `GOVOPLAN_MODULE_PACKAGE_CATALOG_URL` or `GOVOPLAN_MODULE_PACKAGE_CATALOG` | Module package catalog source. |
| `GOVOPLAN_MODULE_PACKAGE_CATALOG_TRUSTED_KEYS_FILE` | Preferred production keyring path. |
| `GOVOPLAN_MODULE_PACKAGE_CATALOG_APPROVED_CHANNELS` | Comma-separated approved catalog channels, for example `stable`. The legacy singular name remains readable during migration. |
| `GOVOPLAN_LICENSE_TRUSTED_KEYS_FILE` | Trusted license issuer keyring path. |
| `GOVOPLAN_LICENSE_ENFORCEMENT` | Enables license enforcement when set to `true`. |
Trust roots are deployment-managed and should not be editable through the
running WebUI. When no catalog override is configured, the Admin package
directory uses GovOPlaN's public stable catalog and the trust anchor bundled
with the installed Core release. Production operators may still pin a newer or
institution-specific catalog/keyring explicitly with the settings above.
### Mail Test Credentials
Dedicated SMTP/IMAP test credentials belong to the mail/campaign test-bed
configuration, not the core runtime contract. Store them in a local ignored
`.env` file for the test bed or in CI secrets. Required values are:
- SMTP host, port, TLS mode, username, password, and envelope/from address.
- IMAP host, port, TLS mode, username, password, and append folder.
- At least one recipient mailbox that is safe for automated send tests.
## First Deployment Flow
1. Create an environment file or secret set with the runtime contract above.
2. Install the tagged core and module packages from meta `requirements-release.txt`.
3. Build the WebUI from `webui/package.release.json` or deploy a prebuilt
artifact from the same release tag.
4. Run database migrations with the target `DATABASE_URL`.
5. Create the first tenant and system owner through the controlled bootstrap:
```bash
python -m govoplan_core.commands.first_admin status
python -m govoplan_core.commands.first_admin issue \
--reason "initial production installation"
```
The issue command fails when an active system administrator already exists,
writes the random credential only to `FIRST_ADMIN_ENROLLMENT_FILE`, and does
not print it. Check `GET /api/v1/bootstrap/status`, then submit the account
and initial tenant fields to `POST /api/v1/bootstrap/first-admin` with the
secret in `X-GovOPlaN-Enrollment-Token`. The operation creates the protected
system owner and initial tenant-owner membership in one transaction and
retires the credential. A repeated identical request returns the same result
without creating another owner.
If the artifact is lost or expires before use, a local operator may rotate
it only while no durable system administrator exists:
```bash
python -m govoplan_core.commands.first_admin recover \
--reason "expired installation handoff"
```
Issue and recovery write hash-chained Core evidence and an audit event. They
never enable or reuse `DEV_BOOTSTRAP_ENABLED`, `DEV_BOOTSTRAP_PASSWORD`, or
`DEV_BOOTSTRAP_API_KEY`.
6. Start the API service with `govoplan_core.server.app:app`.
7. Start workers when `CELERY_ENABLED=true`.
8. Start the WebUI/reverse proxy and verify CORS/cookie settings.
9. Open Admin > System > Modules, verify enabled modules, and save desired
module state if it differs from `ENABLED_MODULES`.
10. Run health checks:
```bash
curl -fsS http://127.0.0.1:8000/health
curl -fsS http://127.0.0.1:8000/api/v1/platform/modules
```
Authenticated health details require `system:settings:read`:
```bash
curl -fsS -H "X-API-Key: $GOVOPLAN_HEALTH_API_KEY" \
http://127.0.0.1:8000/health/details
```
## Production-Like Dev Profile
Use this profile to verify deployment behavior without publishing packages or
using real production credentials. The canonical launcher keeps API, worker, and
WebUI code in the editable repositories while Docker provides PostgreSQL and
Redis:
```bash
cd /mnt/DATA/git/govoplan
tools/launch/launch-production-like-dev.sh
```
The helper wrapper provides explicit lifecycle commands:
```bash
tools/launch/production-like-dev.sh validate-config
tools/launch/production-like-dev.sh seed
tools/launch/production-like-dev.sh start
tools/launch/production-like-dev.sh stop
tools/launch/production-like-dev.sh reset --yes
```
The launcher uses `govoplan/dev/production-like/.env` when present, otherwise
the checked in `.env.example`. It runs:
- PostgreSQL on `127.0.0.1:55433`
- Redis on `127.0.0.1:56379`
- explicit `ENABLED_MODULES`
- explicit migrations and `--with-dev-data` bootstrap
- API via the module-aware devserver
- a Celery worker for `send_email,append_sent,notifications,mail,calendar,dataflow,workflow,postbox,events,idm,default`
- WebUI through the Vite dev server
- durable local files under `runtime/production-like/files`
This profile validates explicit migration execution, config loading, module
discovery, route aggregation, local storage paths, Redis broker connectivity,
worker heartbeats, and health/readiness startup without real production
credentials. It does not replace a managed PostgreSQL/Redis/WebUI/worker
deployment test.
To stop PostgreSQL and Redis when the launcher exits:
```bash
GOVOPLAN_STOP_PROFILE_DEPENDENCIES_ON_EXIT=1 tools/launch/launch-production-like-dev.sh
```
## Module Install/Uninstall Operations
Use Admin > System > Modules for planning. The running API server validates and
queues install plans; it does not run pip/npm or restart itself from an HTTP
request. Package mutation belongs to the trusted installer CLI/daemon in an
operator shell while maintenance mode is active.
Preflight from the server shell:
```bash
govoplan-module-installer --format shell
```
Apply a prepared plan directly from a controlled shell:
```bash
govoplan-module-installer --apply --build-webui
```
For production-like runs, prefer supervised mode with migrations, database
backup/restore hooks, restart commands, and health checks:
```bash
govoplan-module-installer \
--supervise \
--migrate \
--restart-command 'systemctl restart govoplan-api' \
--restart-command 'systemctl restart govoplan-worker' \
--health-url http://127.0.0.1:8000/health \
--database-backup-command 'pg_dump --format=custom "$GOVOPLAN_DATABASE_URL_PGTOOLS" > "$GOVOPLAN_DATABASE_BACKUP_PATH"' \
--database-restore-check-command 'pg_restore --list "$GOVOPLAN_DATABASE_BACKUP_PATH" >/dev/null' \
--database-restore-command 'pg_restore --clean --if-exists --dbname "$GOVOPLAN_DATABASE_URL_PGTOOLS" "$GOVOPLAN_DATABASE_BACKUP_PATH"'
```
To let the admin UI submit work without executing package managers inside the
API process, run the daemon in a separate operator shell:
```bash
govoplan-module-installer \
--daemon \
--migrate \
--build-webui \
--database-backup-command 'pg_dump --format=custom "$GOVOPLAN_DATABASE_URL_PGTOOLS" > "$GOVOPLAN_DATABASE_BACKUP_PATH"' \
--database-restore-check-command 'pg_restore --list "$GOVOPLAN_DATABASE_BACKUP_PATH" >/dev/null' \
--database-restore-command 'pg_restore --clean --if-exists --dbname "$GOVOPLAN_DATABASE_URL_PGTOOLS" "$GOVOPLAN_DATABASE_BACKUP_PATH"' \
--health-url http://127.0.0.1:8000/health \
--restart-command '<restart govoplan server>'
```
The daemon claims one queued request at a time and writes request/run records
below `runtime/module-installer`. For process-manager one-shot usage or tests,
use `--daemon-once`. Check daemon status with:
```bash
govoplan-module-installer --daemon-status --format json
```
The installer uses a runtime lock, snapshots `pip freeze` plus WebUI package
files, writes a run record, and marks planned rows as applied only after all
commands succeed. With `--migrate`, SQLite databases are backed up through
SQLite's backup API; non-SQLite databases require
`--database-backup-command`, `--database-restore-check-command`, and
`--database-restore-command`.
Every non-dry run also owns the database-fenced
`core:module-lifecycle:deployment` recovery operation. The run record includes
its operation id and status. A supervised run reaches durable `succeeded` only
after restart and health verification. `recovery_required` or `outcome_unknown`
blocks another lifecycle mutation until the recorded operation is reconciled;
do not bypass this by deleting `install.lock`. See
[`MODULE_LIFECYCLE_RECOVERY.md`](MODULE_LIFECYCLE_RECOVERY.md).
Database hook commands receive:
- `GOVOPLAN_INSTALLER_RUN_DIR`
- `GOVOPLAN_DATABASE_URL`
- `GOVOPLAN_DATABASE_URL_PGTOOLS` for PostgreSQL tools
- `GOVOPLAN_DATABASE_BACKUP_PATH`
- `GOVOPLAN_DATABASE_BACKUP_METADATA`
Avoid embedding secrets directly in commands; prefer environment variables,
service credentials, or deployment-local secret injection.
Inspect installer history and lock state from the operator shell:
```bash
govoplan-module-installer --list-runs --format json
govoplan-module-installer --show-run <run-id> --format json
govoplan-module-installer --lock-status --format json
govoplan-module-installer --list-requests --format json
govoplan-module-installer --show-request <request-id> --format json
govoplan-module-installer --cancel-request <request-id> --format json
govoplan-module-installer --retry-request <request-id> --format json
```
Rollback uses the saved run snapshot:
```bash
govoplan-module-installer --rollback <run-id>
govoplan-module-installer --rollback <run-id> --database-restore-command '<override restore command>'
```
Uninstall is non-destructive by default. A planned uninstall row can set
`destroy_data: true` to request destructive module retirement. The module must
provide an automated retirement provider, and the installer snapshots the
database before dropping module-owned tables.
Run the rollback drill before relying on installer automation in a new
environment:
```bash
/mnt/DATA/git/govoplan/tools/checks/module-installer-rollback-drill.py \
--format json \
--evidence-path runtime/module-installer/restore-drill-evidence.json
```
The drill uses temporary SQLite databases and simulated package commands. It
does not install or uninstall real packages. It exercises:
- package command failure followed by supervised rollback;
- migration failure with a SQLite database snapshot;
- restart-command failure;
- health timeout after restart;
- destructive retirement executor failure with database rollback;
- PostgreSQL-style backup, restore-check, and restore hooks;
- daemon heartbeat, request queue claim/update, retry/cancel, and stale lock
detection/removal.
See `RELEASE_DEPENDENCIES.md` for release package refs, migration baseline
checks, catalog trust, signing, keyring, replay, and license operation.
## Operator Checklist
- Runtime secrets are injected outside git.
- `MASTER_KEY_B64` is set and backed up securely; restores of locally encrypted content fail closed without the exact matching key.
- Database backup and restore commands are tested.
- File/object storage is durable and backed up.
- `CORS_ORIGINS` and cookie settings match the deployed WebUI origin.
- Redis and workers are running before `CELERY_ENABLED=true`.
- Module catalog and license keyrings are pinned locally.
- Health endpoints are monitored.
- Test SMTP/IMAP credentials are non-production and isolated.
- Module installer rollback drill has passed in the deployment environment.
+70
View File
@@ -0,0 +1,70 @@
# GovOPlaN Documentation Map
This map defines the source-of-truth documents for the current repository docs.
Use it to avoid duplicating long procedures across architecture, release,
operator, and roadmap pages.
## Core Platform
| Topic | Canonical document | Notes |
| --- | --- | --- |
| Module architecture and kernel contracts | `MODULE_ARCHITECTURE.md` | Stable module contracts, API efficiency contracts, durable boundary decisions, lifecycle, and WebUI contribution rules. |
| Compatibility policy | `COMPATIBILITY_POLICY.md` | Supported database upgrade origins, portable-schema read/write windows, runtime/API alias retirement, and compatibility-code removal criteria. |
| RBAC and resource access | `ACCESS_RBAC_MODEL.md` | Current permission, role, API-key, and resource-access model. |
| Governance hierarchy | `GOVERNANCE_MODEL.md` | System, tenant, user/group, campaign policy inheritance and admin UI structure. |
| Policy decision DTOs and provenance | `POLICY_CONTRACTS.md` | Shared explain/provenance shape; module-specific policy docs should link here. |
| Events and audit trace context | `EVENTS_AND_AUDIT.md` | Event dispatch semantics and audit payload conventions. |
| Action/effect automation layer | `ACTION_EFFECT_AUTOMATION_LAYER.md` | Action/effect contracts, consequence preview, runner semantics, and module boundary for automation. |
| External references and integration maturity | `EXTERNAL_REFERENCES_AND_INTEGRATION_MATURITY.md` | Stable external identity and cumulative connector maturity; configured source authority is defined by the meta target architecture. |
| Institutional context and governed references | `INSTITUTIONAL_CONTEXT_CONTRACT.md` | Shared temporal, actor/representation, institution, mandate, service, party, decision, evidence, legal-basis, information-governance, presentation, and geo DTO/provider contracts. |
| Provider-neutral record filing | `RECORDS_FILING_CONTRACT.md` | Exact source-revision identity, current source authorization, idempotent filing, capability discovery, and ownership boundary. |
| Ticket routing and Case escalation | `TICKET_INTEGRATION_CONTRACTS.md` | Optional fail-open routing, replay-safe Case handoff, authorization, evidence, and ownership boundaries. |
| Temporal data read context | `TEMPORAL_DATA_CONTEXT.md` | Valid-time and recorded-time titlebar selection, HTTP/cache contract, security boundary, and module-adoption rule. |
| Cross-module information governance adoption | `INFORMATION_GOVERNANCE_ADOPTION.md` | Manifest evidence and enforcement rules for temporal browsing, purpose-aware access, retention, and institutional context. |
| Data-subject access and erasure requests | `DATA_SUBJECT_REQUESTS.md` | Provider-owned search and mutation, explicit coverage, governed export, retained evidence, permissions, and idempotent execution. |
| Context-sensitive F1 help | `CONTEXTUAL_HELP_CONTRACT.md` | Focus, route, module-manifest documentation contexts, Docs projection, and hosted fallback. |
| Semantic documentation subjects | `SEMANTIC_DOCUMENTATION_SUBJECTS.md` | Stable configured-artifact identity, safe provider discovery, revision review, authorization, and lifecycle semantics. |
| German localization and help quality gate | `LOCALIZATION_AND_HELP_QUALITY.md` | German reference locale, new-installation default, catalog completeness, automatic page associations, and explicit-help review priorities. |
| Postbox E2EE target architecture | `POSTBOX_E2EE_ARCHITECTURE.md` | Strategic encrypted postbox/mailbox model, key ownership, role mailbox semantics, and retraction limits. |
| Shared state, runtime coordination, and recovery | `STATE_AND_RECOVERY_CONTRACT.md` | State profiles, object storage, node registration/drain, fenced leases, migration ordering, and recovery evidence. |
| Module lifecycle recovery | `MODULE_LIFECYCLE_RECOVERY.md` | Installer/live-graph recovery modes, deployment fence, evidence, retry blocking, and operator reconciliation. |
## Release And Operations
| Topic | Canonical document | Notes |
| --- | --- | --- |
| Runtime configuration and operator flow | `DEPLOYMENT_OPERATOR_GUIDE.md` | Production/staging configuration, migrations, backups, installer operation, and rollback drill. |
| Self-hosted installability | `SELF_HOSTED_INSTALLABILITY.md` | Packaging decision, generated env templates, config validation, production-like dev stack commands, and boundary gate. |
| Release dependencies and catalogs | `RELEASE_DEPENDENCIES.md` | Release package refs, migration baselines, release lockfiles, catalog trust/licensing, catalog publishing, and release checklist. |
| Dependency vulnerability audits | `DEPENDENCY_AUDITS.md` | Local and CI audit commands plus dated audit result notes. |
| Remote WebUI bundle design | `REMOTE_WEBUI_BUNDLES.md` | Experimental controlled-deployment design; normal releases still use package builds. |
| WebUI loading and bundle budgets | `WEBUI_BUNDLE_BUDGETS.md` | Installed-module lazy boundaries, enforced initial/async budgets, and baseline measurements. |
## Product And Module Planning
| Topic | Canonical document | Notes |
| --- | --- | --- |
| Product roadmap and module routing | `GOVOPLAN_MASTER_ROADMAP.md` | Product-level sequencing, implementation gates, issue routing, and missing-module decisions. |
| Stable platform ideas | `govoplan/docs/strategy/PLATFORM_CORE_IDEAS.md` | Cross-product thesis, canonical distinctions, product experience rule, maturity rule, and decision test. |
| Current cross-product reconciliation | `govoplan/docs/strategy/STRATEGY_STATUS.md` | The only current prose status source; generated evidence and Gitea remain authoritative inputs. |
| Institutional governance target | `govoplan/docs/architecture/INSTITUTIONAL_GOVERNANCE_TARGET_ARCHITECTURE.md` | Cross-product semantic layers, source-authority modes, candidate Mandates/Services/Parties/Decisions boundaries, and migration sequence. |
| UI/UX decisions | `UI_UX_DECISION_LEDGER.md` | Binding guided-UI decisions, open decisions, impact index, and review checklist. |
| Core interface migration | `INTERFACE_PATTERN_MIGRATION.md` | Core-owned settings, credential, retention, lifecycle, and shared-component evidence for the product pattern language. |
| Interface ethics and design doctrine | `INTERFACE_ETHICS_AND_DESIGN_DOCTRINE.md` | Product-level doctrine for context, decision, consequence, contestability, responsibility, and traceability. |
| Public-sector integration posture | `PUBLIC_SECTOR_INTEGRATION_STRATEGY.md` | Strategy index; executable target inventory lives in `govoplan-connectors`. |
| Configuration packages | `CONFIGURATION_PACKAGES.md` | Package model, provider contract, import/export flow, and tracking slices. |
## Workflow Docs
| Topic | Canonical document | Notes |
| --- | --- | --- |
| Gitea issues and wiki sync | `GITEA_ISSUES.md` | Issue labels, imports, wiki mirroring, and Codex state updates. |
| Codex local workflow | `CODEX_WORKFLOW.md` | Local agent setup and focused verification commands. |
## Cross-Repo Rule
Core docs may keep strategy, kernel contracts, and routing decisions. Module
repositories should own executable module behavior, concrete API/UI contracts,
and operator notes for their own module. When content spans both, keep the
durable decision in core and link to the module document for implementation
details.
+44
View File
@@ -0,0 +1,44 @@
# Durable Recovery Operations
Modules must use `begin_durable_recovery_operation` for work whose effects can
outlive the caller's SQLAlchemy transaction. The helper commits the canonical
request hash, recovery plan, precondition evidence, running state, and lease
fence before the caller mutates object storage, a queue, a filesystem, or an
external provider.
Each later checkpoint is written through an independent database session. A
business-transaction rollback therefore cannot erase evidence of an earlier
effect. Successful completion requires concrete verification checks and a valid
hash chain. Compensation likewise records recovery-required, recovering, and
verified-recovered checkpoints rather than reporting an ordinary failure.
A definitive pre-effect or provider rejection records terminal `rejected`
evidence instead of being mislabeled as success, atomic rollback, or recovery
work.
If a runtime disappears, another runtime may claim the operation only after the
lease expires. The takeover records both fences. A stale compensatable operation
becomes recovery-required; a stale forward-only or irreversible external effect
becomes outcome-unknown; a database-only atomic operation is recorded failed
because its transaction rolled back. Takeover never re-executes the original
request automatically.
Evidence and metadata may contain opaque references, digests, counts, and
provider result codes. They must never contain credentials or resolved secrets.
Ops is the platform surface for unresolved operation status; owning modules must
provide the reconciliation action and business-level explanation.
Database-only operations must use the durable handle's atomic terminal methods
when their module rows and final recovery checkpoint belong to one invariant.
Those methods stage the terminal checkpoint and lease release in the caller's
SQLAlchemy transaction, then commit the domain rows and recovery evidence
together. A failed commit rolls both back and leaves the previously durable
`running` record available for stale-fence handling; modules must not commit
their domain state first and close an `atomic` recovery record afterwards.
An owning module may reconcile an `outcome_unknown` provider effect through the
claimed durable handle's `resolve_unknown` method. External evidence that the
effect occurred records verified success. Evidence that it did not occur moves
the operation through recovery-required and recovering to verified recovered,
so any later attempt must use a new deliberate idempotency key. The method does
not infer provider state and requires the same terminal verification structure
and hash-chain checks as ordinary completion.
+211
View File
@@ -0,0 +1,211 @@
# Events And Audit
GovOPlaN uses a small kernel event contract first, not a broad command bus.
Commands remain module-owned application-service methods or API endpoints until
there is a concrete need for durable asynchronous command orchestration. Events
are facts about completed work and are safe for audit, projections, optional
module reactions, and operator diagnostics.
## Production Transport Decision
The production transport is a **transactional database outbox plus retrying
dispatcher**:
- Use `govoplan_core.core.events.PlatformEvent` for domain and platform events.
- Call `emit_platform_event(session, event)` to bind event delivery to the
domain transaction.
- The optional `platform.eventOutbox` capability persists events atomically.
The Audit module provides the current SQL implementation.
- Without an outbox provider, Core publishes to `EventBus` only after the outer
transaction commits. This preserves reduced installations but is not durable
across process failure or multiple workers.
- Use `EventBus` as the in-process dispatch contract for module reactions
invoked by the outbox dispatcher or for non-critical fallback reactions.
- Use the shared `audit_event` / `audit_from_principal` helper for audited
module actions. The helper persists the audit row and transactionally emits a
governed `PlatformEvent` whose `type` is the audit action.
- Use `record_change` for module delta feeds. It persists the change-sequence
row and transactionally emits a generic module change event such as
`mail.profile.updated`.
- Drain the outbox with the Celery `govoplan.events.dispatch_outbox` task on the
`events` queue. The periodic schedule also retries pending rows.
- The dispatcher invokes Dataflow event ingestion when that capability is
active, then publishes to the process-local bus.
- Treat Redis/Celery as worker/job infrastructure, not as the authoritative
first event transport. PostgreSQL remains authoritative until dispatch is
recorded.
- Keep the dispatch implementation pluggable behind the `PlatformEvent`
envelope so a future message broker can be added without changing event
producers.
- Keep commands out of the kernel until workflows need retryable, durable,
operator-visible command records.
This keeps the first contract small, PostgreSQL-friendly, auditable, and
recoverable after process crashes. It also avoids making Redis a correctness
dependency for deployments that only need synchronous mail/tests or light
background work.
## Dispatch Semantics
Event producers should write their domain state and outbox event in the same
database transaction wherever possible. Handlers must be idempotent because the
outbox dispatcher can retry after a crash or timeout.
The current outbox stores:
- `event_id`, `event_type`, `module_id`
- `correlation_id`, `causation_id`
- `classification`, serialized event `payload`
- `status`, `attempts`, `next_attempt_at`
- `dispatched_at`, `last_error`, timestamps
Handlers must be idempotent: a worker may complete an external effect and fail
before marking its outbox row dispatched. Anything that must survive process
failure, restart, package update, or worker redeployment requires the outbox
provider and dispatcher.
## Trace IDs
Every `PlatformEvent` has:
- `event_id`: unique ID for that event.
- `correlation_id`: stable ID for the whole request, workflow, or job.
- `causation_id`: the event ID or external operation ID that caused this
event.
- `actor`: optional typed actor reference, for example user, API key, system
actor, delegated actor, or installer daemon.
- `tenant`: optional tenant reference when the event is tenant-scoped.
- `subject`: optional typed subject reference for the person, organization,
account, case, campaign, or other entity the event is about.
- `resource`: optional typed resource reference for the object changed or
observed by the event.
- `classification`: payload sensitivity, currently `public`, `internal`,
`confidential`, or `restricted`.
- `payload`: JSON-serializable module-owned event details.
The FastAPI app factory creates an event context for every request. It accepts
`X-Correlation-ID` or `X-Request-ID` when the value is a compact safe trace ID,
otherwise it generates a new ID. Responses include `X-Correlation-ID`.
Audit logging reads the current event context and stores trace IDs in
`details._trace`. Callers can also pass explicit `correlation_id` and
`causation_id` to `audit_event` or `audit_from_principal`. The same trace is
copied into the emitted `PlatformEvent`, so audit rows and event subscribers can
be correlated without route-specific glue code.
Admin and lifecycle code should use the compact operational detail shape
documented in `govoplan-audit/docs/AUDIT_TRACE_CONTEXT.md`. The core
`audit_operation_context` helper preserves `module_id`, `request_id`, `run_id`,
`outcome`, and `_trace` while applying the shared audit redaction pass to
additional detail values.
## Audit MVP Boundary
`govoplan-audit` owns:
- the `audit_log` table and audit API routes
- audit route contribution through its module manifest
- audit retention behavior in cooperation with policy/retention settings
- future audit sink/export capability implementations
`govoplan-core` owns:
- `AuditEvent` and `AuditSink` protocol contracts
- request and event trace context
- the compatibility `audit_event` helper while routes are still migrating
- retention orchestration that calls module capabilities
Feature modules should record audit facts through the shared helper or a future
`audit.sink` capability. They should not import audit storage internals. Module
state changes that only need delta-feed visibility should go through
`record_change`; route-level business actions should still record a semantic
audit action.
## Initial Domain Event Inventory
Access:
- `access.account.created`
- `access.account.updated`
- `access.membership.created`
- `access.membership.updated`
- `access.group.created`
- `access.group.updated`
- `access.role.created`
- `access.role.updated`
- `access.role.deleted`
- `access.api_key.created`
- `access.api_key.revoked`
- `access.session.created`
- `access.session.revoked`
Tenancy:
- `tenant.created`
- `tenant.updated`
- `tenant.suspended`
- `tenant.resumed`
- `tenant.deletion_requested`
- `tenant.erasure_completed`
Policy:
- `policy.system.updated`
- `policy.tenant.updated`
- `policy.user.updated`
- `policy.group.updated`
- `policy.campaign.updated`
- `policy.effective_policy.changed`
Files:
- `files.file.uploaded`
- `files.file.renamed`
- `files.file.deleted`
- `files.file.frozen`
- `files.share.created`
- `files.share.revoked`
- `files.connector.imported`
- `files.connector.access_denied`
Mail:
- `mail.profile.created`
- `mail.profile.updated`
- `mail.profile.credentials_rotated`
- `mail.profile.tested`
- `mail.message.sent`
- `mail.message.send_failed`
- `mail.imap.appended`
- `mail.imap.append_failed`
- `mail.mailbox.message_seen`
Campaign:
- `campaign.created`
- `campaign.version.created`
- `campaign.validated`
- `campaign.built`
- `campaign.reviewed`
- `campaign.queued`
- `campaign.send_started`
- `campaign.recipient_attempted`
- `campaign.recipient_delivered`
- `campaign.recipient_failed`
- `campaign.paused`
- `campaign.resumed`
- `campaign.cancelled`
- `campaign.report.exported`
## Event Payload Rules
- Payloads must be JSON-serializable.
- Use stable IDs, not ORM objects.
- Include tenant ID when tenant-scoped.
- Include actor/principal, subject, and resource references only as typed DTOs
or primitive IDs.
- Set `classification` to the highest sensitivity needed by the envelope or
payload, not the lowest sensitivity of any individual field.
- Do not include secrets, raw message bodies, full recipient lists, or file
content.
- Put large evidence in owning module storage and reference it by ID.
@@ -0,0 +1,67 @@
# External References And Integration Maturity
GovOPlaN integrations use a shared external-reference contract instead of
storing connector-specific URLs and identifiers in every module.
An external reference identifies an object by:
- external system instance
- object type
- stable external object ID
- optional connector configuration
- canonical HTTP(S) URL without embedded credentials
- optional source version, ETag, observation time, and non-secret metadata
The identity key is `system:object_type:object_id`. A GovOPlaN object may retain
multiple references, but one reference must never silently change its identity.
Moving or escalating work creates a new object and an explicit relationship; it
does not rewrite either object's history.
## Integration Maturity
Maturity is cumulative:
1. `discover`: identify configured external systems and their health.
2. `link`: retain and open stable external references.
3. `search`: include authorized external objects in GovOPlaN search.
4. `read`: display authoritative external content.
5. `publish`: create or update external content from GovOPlaN.
6. `synchronize`: reconcile changes in both directions with conflict handling.
7. `migrate`: perform a governed, verifiable transfer into GovOPlaN.
8. `replace`: provide the native operational capability without the external tool.
Connectors must declare and document the maturity they actually implement.
`synchronize` requires durable cursors, idempotency, provenance, conflict
handling, deletion semantics, and observable failures. A link-only connector
must not imply that GovOPlaN holds an authoritative copy.
## Source Authority Is A Separate Dimension
Integration maturity states what an adapter is capable of doing. It does not
decide which system owns truth for a configured object or field group. A
binding separately selects one of the source-authority modes defined by the
[institutional governance target architecture](../../govoplan/docs/architecture/INSTITUTIONAL_GOVERNANCE_TARGET_ARCHITECTURE.md):
- `native_authoritative`
- `external_authoritative`
- `external_mirror`
- `governed_sync`
- `governance_overlay`
- `linked_reference`
A connector can therefore support `synchronize` while a tenant deliberately
uses it only as an external mirror. Conversely, a native GovOPlaN object may
retain link-only references to several external systems. Authority may be
narrowed by tenant, organization, service, object type, object, field group, or
process step and must be visible in provenance and configuration preflight.
## Domain Ownership
- Domain modules own native GovOPlaN objects and their authorization.
- Connectors own protocols, credentials, discovery, transport, and sync state.
- Search owns indexing and result aggregation, but source modules remain
responsible for authorization.
- Core owns only the stable DTOs and extension contracts.
The Python contract is
`govoplan_core.core.external_references.ExternalObjectReference`.
+16
View File
@@ -0,0 +1,16 @@
# Gitea Issues And Wiki Workflow
The shared GovOPlaN Gitea issue, label, and wiki workflow tooling moved to the
meta repository.
Use:
```bash
cd /mnt/DATA/git/govoplan
tools/gitea/gitea-sync-labels.py --help
tools/gitea/gitea-sync-wiki.py --help
```
Canonical documentation:
- `/mnt/DATA/git/govoplan/docs/project/GITEA_ISSUES.md`
+229
View File
@@ -0,0 +1,229 @@
# GovOPlaN Governance Model
**Updated:** 2026-07-09
## Governance Rule
System policy is authoritative for tenants and all lower levels. Each lower
level may only narrow what it inherits:
```text
system
-> tenant
-> user or group owner
-> campaign
```
Lower levels do not widen privileges, allowed profiles, retention durations, or
credential rights granted by a higher level.
## Administration Structure
GovOPlaN separates system administration from scoped configuration:
```text
ADMINISTRATION
- Modules
- Packages
- Maintenance
- Changes
GLOBAL
- Tenants
- Roles
- Groups and users
- File connectors
- Mail servers
- API keys
- Retention
TENANT
- Roles
- Groups and users
- File connectors
- Mail servers
- API keys
- Retention
GROUP
- File connectors
- Mail servers
- API keys
- Retention
USER
- File connectors
- Mail servers
- API keys
- Retention
```
System access scopes remain in the backend for assignment/read boundaries, but
the UI should present the configuration hierarchy rather than a separate
"system access" concept.
## Tenant Governance
System settings define tenant defaults and whether tenants may narrow selected
options. Tenant overrides can only restrict:
- custom groups;
- custom roles;
- tenant API keys.
The backend enforces that tenant governance cannot widen system-denied
privileges.
## Mail-Profile Governance
Mail server profiles may exist at these scopes:
```text
system
tenant
user
group
campaign
```
Effective campaign profile availability follows campaign ownership. A campaign
owned by a user resolves through system, tenant, that user, and campaign
policy. A group-owned campaign resolves through system, tenant, that group, and
campaign policy.
Policy semantics:
- higher levels define the maximum available profile set;
- lower levels can further restrict the set;
- forced profiles mean the lower level must choose from the forced set;
- a forced set with one profile effectively enforces that profile;
- campaign-level profile creation is allowed only if the effective policy
permits it;
- SMTP/IMAP credentials use one inheritance decision per protocol: lower levels
must inherit profile credentials, may inherit profile credentials, or must
provide local credentials;
- the lower-level override switch for `smtp_credentials.inherit` and
`imap_credentials.inherit` controls whether descendants may change that
inheritance decision;
- deny patterns always win over allow patterns;
- empty or `*` allowlist means allow all except denied;
- non-empty allowlist means at least one allow rule must match and no deny rule
may match.
Pattern targets:
```text
SMTP hostname
IMAP hostname
envelope sender
From header
recipient domains
```
Ownership transfer is intentionally deferred as a two-step workflow: original
owner initiates, new owner accepts and reselects/repairs the mail profile if
their effective policy requires it.
## File-Connector Governance
File connector profiles and credentials are separated. Profiles describe
external endpoints; credentials bind authentication material and policy to a
scope. Concrete linked folders appear as file spaces in the files module.
Governance follows the same inheritance shape as mail:
- system and tenant policy can permit, require, or forbid lower-level
connections/credentials;
- user or group ownership controls which spaces appear to principals;
- spaces inherit endpoint and credential policy from their connector profile;
- connector health and credential tests must not reveal plaintext secrets.
## Retention Governance
Retention policy is hierarchical:
```text
system -> tenant -> user/group -> campaign
```
Managed fields:
- raw campaign JSON retention days;
- generated EML retention days;
- stored report detail retention days;
- mock mailbox retention days;
- audit detail retention days.
Rules:
- system may set concrete defaults or unlimited retention;
- system exposes allow-limiting toggles per field;
- tenants, users/groups, and campaigns may only shorten inherited retention
where the parent allows limiting;
- blank lower-level values inherit;
- mock mailbox retention is currently system-level because mock mailbox records
do not yet carry tenant/campaign ownership metadata;
- dry-run/apply retention actions report affected classes before destructive
cleanup.
## Role Definitions And Assignments
### System Roles
System roles define instance-wide permissions. `system:*` is stored as one
wildcard and displayed as granting the full system catalogue. System owner is
protected.
### Tenant Roles
Tenant roles can be system-governed templates or tenant-local definitions,
subject to system tenant-governance settings and actor delegation ceilings.
Wildcard counts are expanded against the canonical tenant catalogue.
## Audit Access
Audit access remains scope-separated:
```text
system audit -> system:audit:read
tenant audit -> active tenant + audit:read
```
Audit pages use server pagination, filtering, and bounded grids.
## Tenant Switching
Tenant switching preserves the current URL when possible and falls back when a
route/resource is not accessible in the new tenant context.
The tenant selector is hidden for ordinary single-tenant accounts and visible
for multi-tenant or system tenant-management contexts.
## Administration DataGrid Contract
Admin lists use bounded container grids:
- one flexible fill column;
- fixed total table width;
- compact action/status/count columns;
- resizable text/date columns;
- no intrinsic content growth;
- sticky headers where needed;
- server pagination for audit.
## Deferred Work
- DataGrid sizing and resize behavior remains explicitly deferred. The current
bounded-grid contract above is binding, but further layout changes should be
handled as a dedicated, isolated UI debt item because the component is shared
and brittle.
- real SMTP/IMAP test-bed verification and operator runbook;
- recipient import with column mapping;
- session/device revocation UI;
- backup/restore, monitoring, and update procedures;
- additional module providers and signed human-readable response packages for
the implemented DSAR workflow described in `DATA_SUBJECT_REQUESTS.md`;
- campaign ownership transfer workflow;
- policy impact analysis before delete/disable/unshare/change;
- LDAP/OIDC/SAML provisioning;
- destructive tenant erasure orchestration.
+742
View File
@@ -0,0 +1,742 @@
# GovOPlaN Master Roadmap
This roadmap is the technical and module-sequencing companion for GovOPlaN as
a modular platform for administrative operations. It translates the
cross-product outcome horizons into dependency waves without turning every
possible public-sector need into an immediate implementation track.
Use this document for technical sequencing, module routing, and implementation
gates. Issues are the active backlog; this document is durable architecture
planning context and should be mirrored to the Gitea wiki.
The meta repository's
[GovOPlaN Roadmap](https://git.add-ideas.de/GovOPlaN/govoplan/src/branch/main/docs/strategy/ROADMAP.md)
describes the corresponding cross-product stakeholder visions, configurable
service and operating configurations, connected outcome stories, and
capability horizons. The selected five-stage delivery sequence and its gates
are in the meta repository's
[Reference Journey Program](https://git.add-ideas.de/GovOPlaN/govoplan/src/branch/main/docs/strategy/REFERENCE_JOURNEY_PROGRAM.md).
The semantic target, source-authority modes, and reconciliation with the
implemented platform are in the meta repository's
[Institutional Governance Target Architecture](https://git.add-ideas.de/GovOPlaN/govoplan/src/branch/main/docs/architecture/INSTITUTIONAL_GOVERNANCE_TARGET_ARCHITECTURE.md).
Those product documents are canonical; this Core roadmap remains their
technical sequencing and module-routing companion.
## Product Thesis
GovOPlaN should become a configurable operations platform for public
institutions. It should help an institution model real administrative
procedures, connect existing systems, keep durable evidence, and explain the
configured system to users.
The goal is not to replace every existing system. GovOPlaN should connect
existing systems, provide better workflows where the current landscape is weak,
and make administrative processes configurable, auditable, and reusable.
The product should first provide a reliable administrative spine:
- identities, roles, tenants, policy, audit, and governance
- forms, files, cases, workflow, tasks, templates, and records
- postboxes, notifications, mail, portal, appointments, and booking
- identity trust, role-bound postboxes, and eventually encrypted administrative
communication
- configuration packages that assemble modules into repeatable procedures
- docs that explain the configured system, not the full theoretical product
Domain modules should come after the spine can run a reference procedure end to
end.
The platform should be a governance-capable runtime for modules, connectors,
configuration, and administrative decisions. The kernel must stay free of domain
semantics while still providing the contracts needed for modules to explain what
they do, what they require, which effects they create, and how operators can
verify or reverse those effects.
## Design Principles
- Modules must stay independently installable, enableable, and disableable.
- Cross-module behavior should use core-mediated capabilities, commands,
events, DTOs, and UI contribution points rather than direct imports.
- Configuration packages should turn installed modules into concrete, reusable
administrative processes.
- Operators should be able to configure the platform through the UI.
- Every powerful configuration path needs preflight, preview, audit, rollback,
RBAC, and policy checks.
- Context, decision, consequence, and traceability must be visible together for
consequential actions. The full doctrine lives in
`INTERFACE_ETHICS_AND_DESIGN_DOCTRINE.md`.
- Automation must use governed action/effect contracts, not hidden side
effects. The first automation layer is defined in
`ACTION_EFFECT_AUTOMATION_LAYER.md`; its first runner lives in
`govoplan-workflow-engine`. Create a separate automation module only if the
scheduler/action runtime outgrows workflow coordination.
- Encrypted postboxes are a strategic target. Early postbox, access, and
identity-trust contracts should stay compatible with the E2EE architecture in
`POSTBOX_E2EE_ARCHITECTURE.md`.
- Integration should be a first-class product path: connect to existing
systems, consume their data, and publish governed outputs back to them.
- GovOPlaN should scale from a small local installation to a larger deployment
with separately scalable web, API, worker, storage, and database components.
## User Experience Direction
GovOPlaN should expose the full power of the platform without forcing
non-technical users to face every field, flag, and internal representation at
once. The default experience should feel guided, explainable, and calm. Expert
depth should remain available, but it should be layered behind deliberate
interaction patterns.
Core UX rules:
- Use progressive disclosure. Common decisions stay visible; advanced,
hazardous, or rarely used options live in collapsed panels, secondary steps,
or explicit advanced areas.
- Do not use raw JSON as the primary configuration UI. Every configurable value
should have an appropriate control, validation, and plain-language help.
Import/export and diagnostics may show JSON as a secondary artifact.
- Prefer guided flows over option dumps. Connector setup, package import,
module installation, policy changes, and destructive operations should use
wizards that explain what is happening, why it matters, and what will happen
next.
- Discover values when the system can infer them. For example, a Nextcloud file
connection should start with the base URL, discover the WebDAV endpoint, and
fill technical fields for review instead of asking the user to know them
upfront.
- Make explanations always available without making every screen verbose.
Inline helper text should be short; richer explanations should be reachable
through expandable help, tooltips, side panels, or review steps.
- Explain blocked actions in actionable language. A disabled control or failed
step should say what is missing, who can fix it, and where to go, for example
"A system administrator must allow this provider" or "Configure the provider
in Settings > File Providers before linking a folder here."
- Reuse visual language and placements consistently. Similar configuration,
policy, connection, credential, review, and confirmation flows should share
components, button placement, modal behavior, problem lists, and empty/error
states.
- Use modals and step flows for focused creation/editing where they reduce page
clutter. Reserve large always-open pages for overview, comparison, and
repeated administration work.
- Treat diagnostics as product UX. Validation results, preflight blockers,
policy explanations, permission denials, and missing capabilities should be
understandable to a non-technical operator before exposing internal details.
This is a product quality gate. New admin/configuration surfaces should not be
considered complete if they expose all options at once, require JSON editing,
hide why an action is unavailable, or use a one-off layout where a shared
pattern exists.
## Focus Rules
1. Build one selected reference journey stage at a time; a later capability
cluster is not an active program merely because it appears below.
2. Do not implement a module because the repository exists.
3. Do not add module-to-module imports for optional behavior.
4. Every new domain module must justify its own semantics beyond `cases`,
`workflow`, `tasks`, `forms`, and `files`.
5. Prefer connector first when an external specialist system is likely to remain
the system of record.
6. Keep active work in Gitea issues; keep durable context in docs and synced
wiki pages.
7. A module moves from scaffold to implementation only when it has an owner,
reference journey, boundary notes, capability contracts, and testable MVP.
## Capability Map
| Capability | Likely owner |
| --- | --- |
| Public application entry point | `govoplan-portal` |
| Structured forms and validation | `govoplan-forms` |
| Uploaded files and managed storage | `govoplan-files` |
| Case record and lifecycle | `govoplan-cases` |
| Workflow runtime, transitions, and automation | `govoplan-workflow-engine` |
| Workflow definition editing | optional `govoplan-workflow` |
| Action/effect catalogue and automation runner | first `govoplan-workflow-engine`; possible future `govoplan-automation` only if it outgrows workflow |
| Internal work queues and tasks | `govoplan-tasks` |
| Appointment proposals and booking | `govoplan-appointments`, `govoplan-calendar` |
| Postbox, email, and notifications | `govoplan-postbox`, `govoplan-mail`, `govoplan-notifications` |
| Canonical subjects and account links | `govoplan-identity` |
| Organizational structures, units, and functions | `govoplan-organizations` |
| Identity-to-function assignments and directory synchronization | `govoplan-idm` |
| Identity trust, device keys, and encrypted postbox key contracts | `govoplan-identity-trust`, `govoplan-access`, `govoplan-postbox` |
| Service directory presentation | `govoplan-portal` |
| Versioned institutional service definition | shared service contract first; candidate `govoplan-services` after reuse proof |
| Mandate, jurisdiction, and institutional authority | shared mandate contract first; candidate `govoplan-mandates` after reuse proof |
| Procedure-local parties and representation | shared party contract first; candidate `govoplan-parties` after reuse proof |
| Formal institutional decisions | shared decision contract first; candidate `govoplan-decisions` after reuse proof |
| Permit/document generation | `govoplan-templates`, `govoplan-dms` |
| Payment capture and accounting handoff | `govoplan-payments`, `govoplan-ledger` |
| Authentication projection, roles, permissions, acting context | `govoplan-access` |
| Tenants, policy, and audit | `govoplan-tenancy`, `govoplan-policy`, `govoplan-audit` |
| External software integration | `govoplan-connectors` |
| Governed data/register catalogue | `govoplan-datasources` |
| Recurring extraction and transformation | `govoplan-dataflow` over Datasources and Connectors contracts |
| Reports, BI, and management visibility | `govoplan-reporting` |
## Configuration And Safety Target
The long-term target is that operators configure the platform through the UI
instead of editing files for normal operation.
UI-managed configuration should include:
- module installation, enablement, lifecycle state, and health
- tenants, users, groups, roles, policies, and permissions
- connectors, credentials, secret references, and external service tests
- workflows, forms, templates, task queues, schedules, and notifications
- configuration package import/export and environment-specific data collection
- retention, audit, privacy, maintenance mode, and safety controls
- deployment-visible settings such as public URLs, mail senders, storage
profiles, queues, and worker capabilities
Safety controls should include dry-run plans, field-level validation, policy
explanations, two-person approval for destructive changes, versioned
configuration history, rollback paths, audit events, and maintenance-mode
guards.
The initial safety metadata contract lives in
`govoplan_core.core.configuration_safety`. It classifies known configuration
fields as UI-managed or deployment-managed, assigns risk levels, marks secret
handling as reference-only or env-only, and declares dry-run, policy
explanation, audit, approval, rollback-history, maintenance-mode, and RBAC
requirements. Admin UI editors should consume this metadata before exposing
powerful settings.
The initial executable guardrail path is `plan_configuration_change(...)`,
exposed through:
- `GET /api/v1/admin/configuration-safety`
- `POST /api/v1/admin/configuration-safety/plan`
The planner reports missing scopes, dry-run requirements, maintenance-mode
requirements, two-person approval status, secret-reference violations,
rollback-history requirements, policy explanations, and audit event names before
an editor applies a high-impact configuration change.
## Reference Journeys
The active sequence is selected. Workflow Engine and its optional editor are
now available foundations, but a reference journey does not depend on Workflow
unless its package explicitly composes and proves it.
### Journey 1: Campaign Demonstration Composition
Campaign is the first complete proof of modular composition. Campaign owns
intent, recipient snapshots, personalization, execution state, and delivery
evidence. Mail owns reusable profiles, credentials, protocol policy, and
provider execution; Campaign stores only a selected profile reference. Files
owns storage, connector profiles, file policy, and provenance.
The technical gate is a pinned Campaign/Mail/Files composition with central
UI, adaptive user/admin/operator/integration documentation, target SMTP/IMAP
and file-provider evidence, and explicit test/send/resend/retry/reconciliation
semantics. Readers must not receive backend paths, worker claims, secrets, or
raw provider diagnostics.
### Journey 2: Function-Bound Postbox Delivery
Postbox accepts delivery to an addressable postbox or a function in an
organizational unit. Organizations owns units and functions, Identity owns
subjects, IDM owns identity-to-function assignments and upstream sync, and
Access resolves current roles, delegation, acting context, and permission.
Campaign consumes a typed delivery-target capability without importing Postbox
or identity internals. Reassignment changes future access without moving the
message; vacancy or ambiguous acting context fails visibly; delivery, access,
and correction remain auditable.
### Journey 3: Data-Backed Templates, Reports, And Deep Launch
An authenticated user follows an opaque, short-lived launch reference from HIS
or another specialist system. GovOPlaN re-authorizes the actor, resolves a
curated data context server-side, displays source/freshness/version, and renders
one reproducible document and report.
Templates owns definition/version/schema/rendering, Reporting owns source
selection/parameters/execution/export, Files owns generated bytes, and
connectors own protocol access. URLs do not carry credentials, arbitrary SQL,
or trusted raw personal data. Retries are idempotent and generation evidence
connects source, snapshot/reference, transformation, definition, parameters,
output checksum, actor, and policy.
### Journey 4: Governed University BI Path
Starting from the Journey 3 source contract, one bounded university dataset is
catalogued, staged by snapshot or watermark, validated, transformed through a
versioned lineage graph, and exposed as a policy-aware analytical data product.
The result must preserve official-key mappings, organizational and reporting
date semantics, quality findings, quarantine/replay, transparent calculation,
and reproducible promotion between development, test, and production.
Reporting consumes the product. Datasources owns the governed source and
materialization lifecycle; Dataflow owns typed transformation/run lineage;
Connectors owns external transport. The concrete path must now prove those
implemented boundaries and expose any missing contracts instead of recreating
them inside Reporting or a producing domain module.
### Journey 5: Collaborative Document Lifecycle
An uploaded or generated artifact becomes a DMS document. Files continues to
own bytes; DMS owns identity, versions, renditions, editing sessions, locks,
comments, review, approval, comparison, and recovery; a collaboration connector
owns provider-specific protocol behavior; Records owns later classification,
hold, archive, and disposal.
The gate requires no silent lost updates, short-lived and currently authorized
editing sessions, idempotent authenticated callbacks, visible uncertain saves,
immutable accepted renditions, and a Records-ready handoff with stable content
and provenance.
## Capability Dependency Waves
The waves below remain a dependency and ownership catalogue for the wider
product vision. They are not the active delivery order. The five selected
journeys above and the meta roadmap decide what is implemented now; other
clusters remain dormant until a selected journey consumes them or they are
explicitly reprioritized.
### Wave 0: Platform Spine
Goal: make the platform safe to configure and extend.
Refine:
- `govoplan-core`: module discovery, capabilities, events, migrations, release
catalog, configuration package runtime, WebUI shell.
- `govoplan-identity`: canonical identities and account links.
- `govoplan-organizations`: organizational structures, units, and functions.
- `govoplan-idm`: identity-to-function assignments, directory synchronization,
preview, conflicts, and reconciliation.
- `govoplan-access`: sessions, API keys, users, groups, roles, memberships,
function-to-role projection, delegation, acting context, and RBAC decisions.
- `govoplan-tenancy`: tenant lifecycle and tenant boundaries.
- `govoplan-identity-trust`: initial trust contracts for device keys, public key
directory, assurance, and later encrypted postbox key access.
- `govoplan-policy`: policy sources, policy decisions, retention inputs.
- `govoplan-audit`: audit sink, audit routes, trace context, retention
cooperation.
- `govoplan-admin`: UI-managed configuration with preflight, rollback, audit,
and approval controls.
- `govoplan-dashboard`: configurable user home assembled from module-provided
widgets, with core providing only a minimal fallback when the module is absent.
- `govoplan-docs`: configured-system documentation and evidence-aware help.
- `govoplan-ops`: deployment profiles, health checks, worker split, sizing
assumptions.
Do not expand domain scope in this wave. The output is a dependable platform
surface for later modules.
Exit criteria:
- module enablement and capability lookup are stable
- configuration package preflight works for at least one simple package
- audit and policy decisions are visible in admin flows
- access distinguishes identity, account, function, role, and right in durable
contracts
- action/effect automation contracts are specified before hidden side effects
spread across modules
- docs can show installed/enabled modules and configured routes
### Wave 1: Permit-To-Payment MVP
Goal: one complete public-administration process.
Create or refine in this order:
1. `govoplan-forms-runtime`: form submissions, drafts, validation, attachments,
and submission evidence.
2. `govoplan-portal`: public authenticated and unauthenticated entry points.
3. `govoplan-files`: upload, evidence links, file spaces, and permissioned
access.
4. `govoplan-cases`: case record, status, assignments, deadlines, and case
evidence.
5. `govoplan-workflow-engine`: state machine, transitions, commands, module
handoff, and resumable execution; optional `govoplan-workflow` supplies the
editor.
6. `govoplan-tasks`: work queues, assignments, due dates, and follow-ups.
7. `govoplan-templates`: permit/decision document generation.
8. `govoplan-postbox`, `govoplan-mail`, `govoplan-notifications`: applicant and
internal communication, with the postbox model compatible with later E2EE
and role-bound access.
9. `govoplan-calendar`, `govoplan-appointments`, `govoplan-booking`: appointment
or booking handoff for the reference process.
10. `govoplan-payments`, `govoplan-ledger`, `govoplan-xrechnung`: payment
capture, payment evidence, and accounting/e-invoice handoff.
Exit criteria:
- a permit-to-payment package can be installed in a local demo
- every step has audit evidence
- every module integration uses capabilities, events, or DTOs
- user-facing documentation describes only the configured process
### Wave 2: Booking And Resource Operations
Goal: make time, capacity, and resource allocation a reusable product area.
Create or refine in this order:
1. `govoplan-booking`: booking rules, capacity, quotas, waitlists, cancellation,
attendance, no-shows, and approvals.
2. `govoplan-resources`: rooms, equipment, vehicles, counters, labs, capacity,
availability, and maintenance blocks.
3. `govoplan-calendar`: event and availability integration.
4. `govoplan-appointments`: appointment-specific booking surfaces.
5. `govoplan-facilities`: buildings, rooms, maintenance, access zones,
inspections, and defects.
6. `govoplan-assets`: inventory, assignment, lifecycle, maintenance, and handover
evidence.
Reference journey: reserve a room/resource for a training or appointment and
preserve all booking evidence.
Exit criteria:
- bookable resources are not hardcoded into appointments
- booking decisions are explainable and auditable
- resources can be blocked by maintenance or facility status
### Wave 3: Training And Certificates
Goal: support university-style and internal public-sector training without
building a full learning-management system first.
Create or refine in this order:
1. `govoplan-learning`: course planning, course booking, participant lists,
trainers, attendance, evaluations, and learning records.
2. `govoplan-certificates`: participation confirmations, certificates,
verification, revocation, and evidence links.
3. `govoplan-booking`, `govoplan-resources`, `govoplan-calendar`: reuse booking
and room/resource allocation.
4. `govoplan-templates`, `govoplan-files`, `govoplan-records`: certificate
generation and durable retention.
Reference journey: plan a course, book resources, register participants, track
attendance, and issue a certificate.
Exit criteria:
- attendance can produce certificate eligibility
- certificates can be verified later
- course operations do not depend on university-specific assumptions
### Wave 4: Report-To-Resolution Operations
Goal: cover internal support and public issue reporting.
Create or refine in this order:
1. `govoplan-tickets`: public/internal reports, requests, incidents, queues,
location, evidence, and triage.
2. `govoplan-helpdesk`: service desk tickets, queues, SLAs, assignments,
escalation, and resolution evidence.
3. `govoplan-facilities` and `govoplan-assets`: issue handoff to maintenance or
asset lifecycle.
4. `govoplan-cases`, `govoplan-workflow`, `govoplan-tasks`: escalation into
formal administrative matters.
5. `govoplan-reporting`: workload, SLA, recurring defects, and status reports.
Reference journey: report a facility issue, triage it, assign work, resolve it,
and report recurring defects.
Exit criteria:
- issue reporting is not just a generic form inbox
- helpdesk tickets can remain lightweight unless formal case handling is needed
- operational metrics exist without custom SQL
### Wave 5: Records, DMS, Transparency
Goal: make evidence legally and organizationally durable.
Create or refine in this order:
1. `govoplan-dms`: document lifecycle, collaboration, versions, locks, approvals,
and DMS connectors.
2. `govoplan-records`: file plans, classification, retention schedules, disposal
holds, archive handoff, and records evidence.
3. `govoplan-policy` and `govoplan-audit`: retention policy, legal hold, audit
traceability.
4. `govoplan-search`: permissioned cross-module discovery.
5. `govoplan-transparency`: FOI/public-records requests, redaction, publication,
and disclosure evidence.
Reference journey: close a case, classify records, apply retention, and answer a
transparency request.
Exit criteria:
- files, documents, and records are not treated as the same concept
- transparency disclosure can redact and explain evidence
- retention and archive handoff are policy-governed
### Wave 6: Finance, Procurement, Contracts, Grants
Goal: cover the back-office procedures around spending, obligations, funding,
and revenue without replacing a finance system too early.
Create or refine in this order:
1. `govoplan-procurement`: purchase requests, approvals, vendor comparison,
tender references, goods receipt, and handoff.
2. `govoplan-contracts`: contract register, obligations, renewals, reminders,
and responsible units.
3. `govoplan-grants`: funding programs, applications, eligibility, awards,
milestones, claims, and reporting duties.
4. `govoplan-payments`, `govoplan-ledger`, `govoplan-xrechnung`: payment,
accounting, and e-invoice handoff.
5. `govoplan-erp`: integration with the actual system of record, not a full ERP
replacement unless later justified.
Reference journey: purchase or grant approval through obligation tracking and
financial handoff.
Exit criteria:
- ERP remains an integration point unless GovOPlaN must own semantics
- contracts and grants produce obligations, deadlines, and reporting tasks
- financial evidence links back to cases, files, and audit
### Wave 7: Institutional Governance And Participation
Goal: support public bodies, municipalities, ministries, and universities in
formal coordination.
Create or refine in this order:
1. `govoplan-committee`: committees, boards, councils, agendas, minutes,
decisions, voting, and follow-up tasks.
2. `govoplan-consultation`: hearings, public consultations, stakeholder
feedback, formal comments, response matrices, and publication workflows.
3. `govoplan-campaign`: communication campaigns, recipients, templates, review,
sending control, and reports.
4. `govoplan-addresses`: persons, organizations, contacts, distribution lists,
and recipient import/export.
5. `govoplan-portal` and `govoplan-transparency`: public participation and
publication surfaces.
Reference journey: prepare a committee decision, publish consultation material,
collect feedback, document the decision, and assign follow-up tasks.
Exit criteria:
- decisions create durable tasks and records
- consultations can publish evidence without exposing restricted material
- campaign and address behavior stays optional and capability-based
### Wave 8: Integration, Data, And Reporting
Goal: connect the platform to real institutional landscapes.
Refine:
- `govoplan-connectors`: integration catalog, connector metadata, adapter
lifecycle, test harnesses.
- `govoplan-idm`, `govoplan-identity-trust`: identity provider, directory, and
assurance integration.
- `govoplan-fit-connect`, `govoplan-xoev`, `govoplan-xta-osci`: public-sector
transport and standards integration.
- `govoplan-reporting`: operational reporting, scheduled outputs, exports, and
dashboard data.
- `govoplan-search`: permissioned cross-module discovery.
Refine the existing owners:
- `govoplan-datasources`: governed data/register catalog, live/cached/static
sources, staging, immutable materializations, freshness, quality, legal and
organizational context, and provenance. Connector profiles and credentials
remain in Connectors.
- `govoplan-dataflow`: typed transformations, validation, lineage, manual,
scheduled and event-triggered runs, reusable definitions, and publication
outputs.
- `govoplan-projects`: native projects, portfolios, milestones, goals,
dependencies, capacity, outcomes, and external OpenProject references;
Connectors owns OpenProject transport and synchronization.
Reference journey: monthly data extraction, transformation, validation, approval,
publication, and reporting.
Recurring extraction/transformation should be delivered as a configuration
package across Connectors, Datasources, Dataflow, Workflow Engine, Reporting,
Files, and Templates. The package should register sources, declare schemas,
define mapping/validation versions, schedule runs, produce previewable diffs,
write governed outputs, and preserve lineage, hashes, operator actions, and
audit evidence.
Exit criteria:
- connector catalog exists before building many adapters
- datasource and dataflow ownership remains provider-neutral and is proved by
the recurring transformation package
- reporting consumes governed sources with provenance
## Implementation Gates
Before a module receives implementation work, it needs:
- one reference journey step it owns
- ownership and boundary notes
- initial manifest and permission list
- capability contracts or API DTOs
- audit and policy expectations
- Gitea issues for MVP tasks
- docs page that can sync to the wiki
- focused tests that can run from the core environment
Before a module becomes release-included, it needs:
- installable Python package metadata where applicable
- WebUI package metadata where applicable
- module manifest entry point
- release catalog entry
- smoke test or permutation test
- no required imports from optional modules
## Technical Dependency Order Summary
Use this active order while respecting the ownership and implementation gates
in the capability waves:
1. Keep the platform/release spine green and extend connector, identity,
external-effect, provenance, documentation, focused-view, recovery, and
version contracts only as the current journey requires.
2. Complete and package Campaign with Mail-owned profiles, Files, target
delivery/recovery, central UI, and adaptive documentation.
3. Implement function-bound Postbox delivery through
OrganizationsIdentityIDMAccess and consume it from Campaign through a
typed capability.
4. Implement one data-backed Templates/Reporting path and safe HIS-style deep
launch.
5. Extend that concrete source into one governed university analytical data
product and use it to harden the existing Datasources/Dataflow ownership,
quality, lineage, and promotion contracts.
6. Implement Files-backed DMS versions and one provider-neutral collaborative
editing lifecycle, then connect Records handoff.
7. Maintain already integrated Calendar/Scheduling/Poll and other foundations;
activate another capability cluster only when the current journey needs it
or the product roadmap explicitly reprioritizes it.
8. Extend Workflow Engine and the optional editor only through stable actions
and one demonstrated package at a time.
## Deliberate Deferrals
Defer these until a reference journey proves the need:
- full ERP replacement
- unsupported breadth in native project management before the Projects/OpenProject
boundary is proved in a reference journey
- unbounded Dataflow operators or execution engines without golden-flow,
quality, lineage, resource-limit, and recovery evidence
- every possible public-sector protocol adapter
- rich LMS behavior beyond training administration
- full qualified digital signing/trust services beyond the identity-trust and
encrypted-postbox contracts needed for early architecture safety
- mobile apps
- AI assistants embedded into workflows
These may become important, but they should not distract from the first complete
administrative journeys.
## Module And Integration Routing
This table maps current module and integration ideas to existing GovOPlaN
repositories or to explicit missing-module decisions.
| Idea | Owner | Tracking |
| --- | --- | --- |
| Government operations backbone reference model | `govoplan-core` | `GovOPlaN/govoplan-core#213` |
| Permit-to-payment configuration package | `govoplan-core` plus participating modules | `GovOPlaN/govoplan-core#214` |
| Fully UI-managed configuration with safety controls | `govoplan-admin`, `govoplan-core`, `govoplan-policy`, `govoplan-access`, `govoplan-audit` | `GovOPlaN/govoplan-core#218` |
| Access as a module | `govoplan-access` | `GovOPlaN/govoplan-access#7` |
| Interface ethics and decision-consequence doctrine | `govoplan-core` plus all UI-owning modules | `GovOPlaN/govoplan-core#227` |
| Action/effect automation layer | `govoplan-workflow-engine`; possible future `govoplan-automation` only after boundary proof | `GovOPlaN/govoplan-workflow#1` |
| E2EE role/function postbox architecture | `govoplan-postbox`, `govoplan-identity-trust`, `govoplan-access`, `govoplan-policy`, `govoplan-audit` | `GovOPlaN/govoplan-postbox#15`, `GovOPlaN/govoplan-identity-trust#1` |
| Identity, account, function, role, right semantic model | `govoplan-access` | `GovOPlaN/govoplan-access#9` |
| Role-based service directory presentation | `govoplan-portal`; reusable service definition is a shared contract and candidate `govoplan-services` | `GovOPlaN/govoplan-portal#1` |
| Unified inbox across tasks, postbox, notifications, and portal | `govoplan-core` coordination plus owning modules | `GovOPlaN/govoplan-tasks#1`, `GovOPlaN/govoplan-notifications#1` |
| OpenProject API / project management connector | `govoplan-connectors` | `GovOPlaN/govoplan-connectors#1` |
| Native project-management module | `govoplan-projects`; OpenProject transport remains connector-owned | `GovOPlaN/govoplan-projects#1`, `GovOPlaN/govoplan-connectors#12` |
| Datasources for databases, CSV, files, APIs | `govoplan-datasources`, with external protocol and credential ownership in Connectors | `GovOPlaN/govoplan-datasources#1` |
| Dataflow for pipelines, BI, publication | `govoplan-dataflow` over Datasources and module-owned providers | `GovOPlaN/govoplan-dataflow#1` |
| Monthly datasource and transformation workflows | configuration package across Connectors, Datasources, Dataflow, Workflow Engine, Reporting, Files, and Templates | `GovOPlaN/govoplan#8`, `GovOPlaN/govoplan-dataflow#17` |
| Templates for letters, emails, forms, reports | `govoplan-templates`, separate from reporting | `GovOPlaN/govoplan-core#190`, `GovOPlaN/govoplan-templates#1` |
| Reporting and BI | `govoplan-reporting`, separate from templates | `GovOPlaN/govoplan-core#190`, `GovOPlaN/govoplan-reporting#1` |
| File connectors: Nextcloud, Seafile, SMB, NFS | `govoplan-files` | `GovOPlaN/govoplan-files#15` |
| Public-sector software integration catalogue | `govoplan-connectors` with core strategy index | `GovOPlaN/govoplan-core#191`, `GovOPlaN/govoplan-connectors#2` |
| Public-sector integration landscape catalogue | `govoplan-connectors` with core tracking | `GovOPlaN/govoplan-core#215` |
| Cases module concept | `govoplan-cases` | `GovOPlaN/govoplan-core#174` |
| Workflow runtime/editor split | `govoplan-workflow-engine` runtime plus optional `govoplan-workflow` editor | `GovOPlaN/govoplan-workflow#12`, `GovOPlaN/govoplan-workflow#13` |
| Connectors module concept | `govoplan-connectors` | `GovOPlaN/govoplan-core#176` |
| Adrema-style address and distribution-list management | `govoplan-addresses` | `GovOPlaN/govoplan-addresses#1` |
| Consume sources and become a governed source | Connectors acquires, Datasources governs/materializes, Dataflow transforms/publishes | `GovOPlaN/govoplan-connectors#3`, `GovOPlaN/govoplan-datasources#3` |
| Governed connector configuration, dry-run, and simulation runtime | `govoplan-connectors` | `GovOPlaN/govoplan-connectors#6` |
| Terminfindung and meeting scheduling polls | `govoplan-scheduling`; calendar primitives remain in calendar | `GovOPlaN/govoplan-core#193`, `GovOPlaN/govoplan-scheduling#1` |
| Terminplaner and calendar primitives | `govoplan-calendar` | `GovOPlaN/govoplan-core#193`, `GovOPlaN/govoplan-calendar#1` |
| Terminbuchung appointment booking | `govoplan-appointments` | `GovOPlaN/govoplan-core#193`, `GovOPlaN/govoplan-appointments#1` |
| Collaborative documents | `govoplan-dms` | `GovOPlaN/govoplan-dms#1` |
| Forms | `govoplan-forms` for definitions and `govoplan-forms-runtime` for submissions/runtime behavior | `GovOPlaN/govoplan-core#194`, `GovOPlaN/govoplan-forms#1` |
| RSS consume and emit | `govoplan-connectors` | `GovOPlaN/govoplan-connectors#4` |
| LDAP, Active Directory, OpenDesk identity | `govoplan-idm` | `GovOPlaN/govoplan-idm#1` |
| OpenDesk stack integration map | integration profile across IDM/access, mail/calendar, files/DMS, and connectors; not a monolithic module | `GovOPlaN/govoplan-core#195`, `GovOPlaN/govoplan-connectors#5` |
| Open-Xchange mail/groupware | `govoplan-mail` | `GovOPlaN/govoplan-mail#5` |
| Open-Xchange calendar | `govoplan-calendar` | `GovOPlaN/govoplan-calendar#2` |
| Scalability profiles and autoscaling readiness | `govoplan-ops`, `govoplan-core` | `GovOPlaN/govoplan-core#217` |
| Hardware sizing matrix and requirements calculator | `govoplan-ops`, `govoplan-core` | `GovOPlaN/govoplan-core#219` |
| Collaboration suite integration strategy | `govoplan-connectors`, `govoplan-dms`, `govoplan-workflow`, `govoplan-tasks`, `govoplan-appointments`, `govoplan-calendar` | `GovOPlaN/govoplan-core#220` |
| Install/runtime configuration contract | `govoplan-core`, later `govoplan-ops` | `GovOPlaN/govoplan-core#19` |
| Installer/deployment operator flow | `govoplan-core`, later `govoplan-ops` | `GovOPlaN/govoplan-core#26` |
| Production-like deployment documentation | `govoplan-core`, later `govoplan-ops` | `GovOPlaN/govoplan-core#28` |
Boundary rationale lives in `MODULE_ARCHITECTURE.md`. Current decisions:
- templates and reporting are separate modules
- RSS/source consume-publish starts in Connectors; governed source identity and
snapshots belong to Datasources and transformations belong to Dataflow
- calendar, scheduling, and appointments are three separate modules
- forms definitions and forms runtime are separate responsibilities
- OpenDesk is an integration profile across modules, not a monolithic module
- OpenProject transport is connector-owned; native portfolio/project semantics
belong to Projects
- public-sector integration strategy stays in core; executable catalogue work
lives in connectors
- encrypted postbox and identity-trust are strategic contracts, not mail-module
behavior
- automation starts in Workflow Engine and may split into a dedicated module
only after the runner becomes broader than workflow
- Mandates, Services, Parties, and Decisions begin as shared semantic contracts;
create repositories only after independent persistence, lifecycle, security,
and multiple-consumer evidence passes the repository threshold in the
institutional governance target architecture
Core keeps the strategy index in
`PUBLIC_SECTOR_INTEGRATION_STRATEGY.md`: integration postures, default
ownership, and prioritization rules. `govoplan-connectors` owns the detailed
target inventory and connector entry shape in
`govoplan-connectors/docs/PUBLIC_SECTOR_INTEGRATION_CATALOGUE.md`.
Release composition and tag-only repository handling are documented in
`RELEASE_DEPENDENCIES.md`.
## Next Practical Work
The active cross-product story is
[`GovOPlaN/govoplan#14`](https://git.add-ideas.de/GovOPlaN/govoplan/issues/14).
Module repositories own implementation issues; do not clone their state here.
Immediate issue buckets:
- fail-closed connector destination pinning and private-network deployment
control for every real transport
- Mail-profile-only Campaign authoring/build/delivery and safe legacy failure
- immediate audited secret deletion when a provider/profile is removed
- Campaign central-component and role-safe UI acceptance
- adaptive Campaign, Mail, and Files task/process/admin/operator/integration/
security/acceptance documentation
- target SMTP/IMAP, file-provider, queue/reconciliation, install/upgrade, and
restore proof for the pinned Campaign reference composition
Once that gate is demonstrable, activate the existing Postbox model, access,
API, inbox, and Campaign integration issues. Templates/Reporting, governed BI,
and DMS collaboration remain durable selected direction, but should be
decomposed only as the preceding stage stabilizes or a bounded independent
contract can be implemented without pre-deciding target-system choices.
+131
View File
@@ -0,0 +1,131 @@
# Information Governance Adoption
## Platform Rule
Temporal browsing, purpose-aware access, retention, and institutional context
are platform-wide information-governance dimensions. Every module receives the
same contract by default. A module may claim `partial` or `enforced` only with
repository-owned object scope, evidence, and limitations; it may claim
`not_applicable` only when the dimension genuinely does not apply.
Historical business data is always authorized under the current security
state. No module may use a historical permission, membership, role, function
assignment, or policy projection to weaken present-day access.
Platform-wide adoption is tracked in
[GovOPlaN #40](https://git.add-ideas.de/GovOPlaN/govoplan/issues/40), with
temporal reads detailed in
[GovOPlaN #39](https://git.add-ideas.de/GovOPlaN/govoplan/issues/39).
## Manifest Declaration
`ModuleManifest.information_governance` publishes four dimensions:
- `temporal_browsing`;
- `purpose_aware_access`;
- `retention`;
- `institutional_context`.
Each dimension declares:
- adoption: `not_applicable`, `contract_only`, `partial`, or `enforced`;
- object types covered;
- repository-local test/documentation evidence;
- the remaining limitation for `contract_only` or `partial`.
The default is intentionally `contract_only`. It applies the platform rule
without pretending that existing domain queries and effects already enforce
it. `reference_ready`, `supported`, and `lts` modules cannot retain an
applicable dimension below `enforced`.
## Read Contract
For every persistent domain object, the owner classifies the read:
1. **Current-only:** historical semantics do not exist and the API says so.
2. **Valid-time:** select facts effective now or at the requested instant.
3. **Bitemporal:** additionally select only revisions known by `recorded_at`.
4. **All-validity:** return effective revisions in a bounded history view.
The Core temporal middleware supplies the request context. Owners apply it in
repositories or query helpers, include it in cache keys, return evaluated
context, and test current/at/all plus recorded-time boundaries. Search,
reporting, exports, selectors, counts, and drill-through must use the same
projection as the owning list/detail API.
## Purpose-Aware Access Contract
Permission establishes a technical action ceiling. Purpose-aware access asks
whether this actor, represented capacity, case/work item, legal basis, and
declared use may access this object now.
- A client-supplied purpose is an assertion, never authority by itself.
- The owner or Policy capability validates the purpose and returns explainable
provenance.
- Sensitive access can require case assignment, mandate, reason capture,
approval, or break-glass evidence.
- Search, selectors, reporting, exports, background jobs, and connectors apply
the same decision.
- Audit records the validated purpose identifier and decision reference, not
unnecessary content.
## Retention Contract
Every persistent object declares an owner, retention class or policy reference,
trigger, start instant, hold behavior, review/disposition action, and evidence.
Retention is not a generic timestamp deletion job.
- Domain owners enumerate and execute their own effects through a typed
retention provider.
- Policy resolves inherited ceilings and simulation.
- Records owns record disposition; Files owns byte/object effects; Audit owns
audit-detail behavior; external providers declare their own effect and
recovery semantics.
- Dry-run, legal hold, exact revision, idempotency, outcome unknown,
reconciliation, correction, and destruction evidence are mandatory for
consequential removal.
## Institutional Context Contract
Consequential objects and effects carry the relevant tenant, institution,
organization unit, function, mandate/jurisdiction, service/case/work item,
party/representation, decision, and record references. Context is minimized to
what the operation needs. Organizational membership is not itself permission
or mandate.
Events, automation intents, audit evidence, records, and external effects retain
the same governed context envelope or an exact reference to it. Consumers must
not reconstruct authority later from mutable current structures.
## Adoption Order
1. Inventory every domain list/detail/search/export/effect and classify all
four dimensions.
2. Migrate institutional owners first: Access, IDM, Organizations, Mandates,
Services, Parties, Cases, Approvals, Committee, Decisions, Voting, and
Records.
3. Migrate communication and content: Addresses, Distribution Lists, Campaign,
Postbox, Mail, Calendar, Files, Templates, and Forms Runtime.
4. Migrate data projections: Connectors, Datasources, Dataflow, Reporting,
Search, Risk Compliance, and Dashboard.
5. Migrate workflow/task/background/provider operations and prove that no
asynchronous path drops context.
6. Advance manifest claims only after owner tests and browser/reference-journey
evidence pass.
The generated platform inventory reports adoption counts and module details.
Gitea tracks individual migrations; the declaration is evidence and a maturity
gate, not a substitute for implementation.
## Definition Of Enforced
A dimension is `enforced` only when:
- all declared object types and public reads/effects use it;
- list/detail/count/search/export/worker behavior is consistent;
- cache and pagination semantics cannot cross contexts;
- absence, invalid values, and inaccessible referenced context fail safely;
- tests cover current, historical, unauthorized, replay, and module-absence
combinations appropriate to the dimension;
- user/admin documentation explains behavior and limitations;
- the manifest cites those tests and docs.
+162
View File
@@ -0,0 +1,162 @@
# Institutional Context And Governed References
GovOPlaN consequential work must retain enough context to answer who acted,
for whom, through which function, under which mandate and jurisdiction, using
which rule and evidence versions, and with which requested and observed effect.
The shared contract lives in `govoplan_core.core.institutional`.
Core owns reference shapes and provider protocols only. It does not own shared
Mandate, Service, Party, Decision, evidence, or geography tables. Domain modules
own persistence and authorization; optional capabilities resolve the references.
Interactive reads use the separate platform temporal-data context documented in
`TEMPORAL_DATA_CONTEXT.md`; it never changes current authorization or supplies
mutation dates.
## Envelope
`GovernedContextEnvelope` version 1 carries:
- a tenant and `TemporalRevision` with validity, recording, supersession, and
change reason;
- the real account or service account and represented account, function,
procedure party, assignment, delegation/power, and mandate;
- institution, organization unit, function, task, mandate, jurisdiction,
service, case, party, work item, workflow, approval, decision, and record
references;
- versioned legal bases and evidence references;
- information classification, purposes, retention/holds, minimization, and
disclosure state;
- external-source authority, maturity, freshness, health, and conflict state;
- language, accessibility, channel, explanation, and availability references.
Every institutional reference includes the owner module, tenant, stable object
identity, optional version/effective instant, and a protected display label.
Cross-tenant references are rejected. Safe serialization omits labels,
inspection URLs, formal reasoning, operative results, and conditions unless a
caller explicitly requests the protected projection.
## Semantic Providers
The first provider-neutral capabilities are:
- `mandates.resolver`: resolve competence for a task/authority type at an
effective instant and return the governing Mandate definition and evidence;
- `services.definitions`: obtain versioned institutional service definitions;
- `parties.resolver`: obtain effective procedure-local parties and powers of
representation without copying Identity or Organizations subjects; and
- `decisions.registry`: record and retrieve formal Decisions under optimistic
revision control.
The Mandate, Service, Party/representation, and Decision DTOs have strict
mapping round-trips so they can cross capability, event, package, and storage
boundaries without shared ORM models. Their lifecycle states are explicit:
Mandates distinguish draft/active/suspended/replaced/retired, Services retain
publication state, Parties retain effective representation and revocation, and
Decisions retain correction, revocation, and supersession references.
The DTOs are a repository threshold, not a mandate to create four modules.
Independent persistence, lifecycle, security/operations behavior, release
reason, reuse, and tests are still required before extraction.
Mandate resolution is deterministic: Core filters candidates by tenant,
effective interval, active state, task and authority type, stable
organization/function identity, jurisdiction coverage, and subject type. A
result is competent only when exactly one matching Mandate remains and it has
no unresolved conflicts. Evidence from matching definitions is deduplicated
and retained in the explanation result. `revise_mandate_definition` applies
optimistic concurrency and the allowed activation, suspension, replacement,
and retirement transitions while leaving the previous revision immutable.
`revise_formal_decision` provides the equivalent lifecycle primitive for
formal outcomes. Every accepted transition requires a new recorded revision
and change reason, links `supersedes_ref` to the prior version, updates the
authority envelope to the new version, and records explicit correction or
revocation provenance. Terminal and backward transitions fail closed. Each
Decision also records whether responsibility was human, human-reviewed
automation, or an automated service account acting under mandate. Automation
preparation/recommendation references remain inspectable without being
mistaken for the responsible outcome.
Procedure-party corrections use `revise_procedure_party`: the stable party
identity is retained, a new revision and reason are required, stale writes are
rejected, and revoked/expired/superseded assignments are terminal.
`revoke_party_representation` separately records when a limited power ceased
to authorize actions. This allows consuming procedures to evaluate historical
delivery or representation authority without rewriting Identity,
Organizations, or Addresses records.
Service templates and package/tenant specializations use
`derive_service_restriction`. The derived definition retains an explicit
parent-version reference, cannot extend the parent's validity, audience,
channels, or publication ceiling, and cannot remove inherited prerequisites,
required evidence, legal bases, or bindings. This is the fail-closed semantic
rule; configuration-package signature and provenance checks remain the package
transport rule.
`ServiceAvailabilityRequirement` represents module, capability, mandate,
policy, connector, maintenance, audience, and configuration prerequisites with
an explicit unavailable-or-hidden failure mode and explanation reference. The
optional `services.availability` evaluator returns policy-scoped boolean
assessments, reason codes, and evidence. Unknown consequential requirements
fail closed; a reference itself never grants access.
## Service Launch
`ServiceLaunchRequest` and `ServiceLaunchResult` define the owner-neutral
boundary between Portal entry and a case, form, or workflow runtime effect.
The request carries the exact published Service definition, exact selected
binding, tenant, acting identity, timezone-aware request time, bounded
parameters, and idempotency key. The result must retain that exact Service and
binding, a same-tenant target reference, optional same-tenant evidence, and
only a relative or credential-free HTTP(S) destination.
`service_launch_capability(kind)` maps bindings to owner capabilities:
- `case` -> `cases.service_launcher`
- `form` -> `forms_runtime.service_launcher`
- `workflow` -> `workflow_engine.service_launcher`
`FormDefinition`, `FormFieldDefinition`, and `forms.definitions` provide the
owner-neutral exact-schema boundary used by Forms Runtime. Definitions carry an
exact tenant/revision, field types/options/constraints/defaults, publication,
draft, attachment, signature, policy, and handoff requirements. Forms owns
those immutable definitions; Forms Runtime persists instances and validation
evidence. A form Service binding uses `<form-id>/<revision>` and the launcher
rejects missing, superseded, unpublished, cross-tenant, or invalid definitions.
Portal may discover and invoke those capabilities but cannot write owner
tables. The owner must revalidate its definition/binding and current
authorization, produce its normal audit/event state, and make replay after an
ambiguous response safe. If the capability is absent, the service is
explainably unavailable. URL-only entries pass through the same launch-time
availability check and destination validation. Forms Runtime now supplies the
definition-aware form launcher when both Forms and Forms Runtime are active;
otherwise Portal continues to fail closed.
## Propagation
`PlatformEvent`, `ActionExecutionRequest`, and `AuditEvent` can carry the
envelope. Audit persistence stores only its safe projection; platform-event
outbox serialization preserves it across asynchronous delivery. A module must
not invent a parallel context dictionary when the shared fields apply.
## First Proof
Committee's `committee.decision_path` capability is the first bounded proof. It
requires one effective, conflict-free Mandate covering the organization unit
function, and jurisdiction, an approval reference, fact evidence, versioned legal bases,
operative result, and reasoning. It emits a reconstructable `FormalDecision`,
including requested/observed effects and information governance. If a Decision
registry is installed it persists there; Committee does not take ownership of
the generic Decision lifecycle.
## Compatibility And Security
- Contract version changes follow Core compatibility policy.
- Unknown tenant or reference-kind combinations fail closed.
- Datetimes that affect authority must be timezone-aware.
- Protected labels, reasoning, evidence inspection links, and source details
remain subject to the owning module's access policy.
- References do not grant access to their targets.
- Evidence and audit payloads must contain stable references/checksums, not
plaintext secrets.
@@ -0,0 +1,102 @@
# GovOPlaN Interface Ethics And Design Doctrine
This document captures the product-level design doctrine for GovOPlaN. It is
more durable than an individual screen design and should guide admin,
configuration, workflow, policy, portal, and operational UI decisions.
GovOPlaN is meant to support administrative responsibility. The interface must
therefore make context, decision, consequence, and traceability visible enough
that users can act deliberately instead of being pushed through opaque
automation.
## Core Doctrine
1. Context, decision, and consequence belong together.
2. Decisions should not silently happen.
3. Transparency comes before convenience when rights, duties, records, money,
access, or legal effects are involved.
4. Explicit state is preferable to implicit state.
5. Navigation is not consent.
6. Responsibility cannot be delegated to the system.
7. Traceability is part of the action, not a later reporting feature.
8. Context loss is a product defect.
These rules do not mean every screen should become verbose. They mean the
interface must expose the right explanation at the moment of decision and keep
technical detail available without making it the default surface.
## Decision Surface Contract
Any action that changes records, rights, policies, retention, communication,
payments, external systems, or workflow state should answer these questions
before execution:
- What object, person, organization, or process is affected?
- Which authority or role allows the actor to do this?
- What will change immediately?
- What downstream effects may happen?
- Can the action be undone, superseded, or only corrected later?
- What evidence or audit entry will be created?
- Which policy, configuration, or missing capability blocks the action?
- Who can resolve a blocker?
The answer may be shown through inline labels, a review step, a side panel, a
problem list, or a confirmation dialog. The important point is that consequence
and responsibility are not hidden behind a generic submit button.
## Contestability
Administrative decisions are often contestable or reviewable. GovOPlaN should
therefore preserve the path from input to decision:
- source data and attachments
- workflow state and task assignment
- policy decisions and source path
- actor and delegation context
- generated document/template version
- external handoff result
- notification or postbox delivery evidence
- retention and record classification state
Where a user sees a decision, they should be able to reach the provenance that
explains how the system got there. This is especially important for denials,
locks, calculated defaults, generated documents, payment state, retention
state, and access decisions.
## Anti-Patterns
Avoid these patterns in GovOPlaN interfaces:
- Magical buttons that execute multiple side effects without preview.
- Process tunnels that hide where the user is in an administrative procedure.
- Silent automation that changes external systems without an audit-visible
command record.
- Friendly hiding that removes complexity at the cost of obscuring authority,
consequence, or accountability.
- Disabled controls without actionable explanation.
- Configuration screens that ask users to edit raw JSON as the normal path.
## Automation Rule
Automation must use the same governed action surface as a human actor. The
system may execute actions as a system actor, but it must still run through
policy checks, capability contracts, audit, idempotency, and failure handling.
When an automated decision is not clear, GovOPlaN should create a manual
exception, task, or review item instead of guessing silently.
## Relationship To UI Components
Shared components should make this doctrine easy to follow:
- preflight and problem-list components for blockers
- policy source path and effective decision displays
- action review panels for consequence preview
- audit/provenance links on decision outputs
- guided dialogs for risky configuration
- disabled-action explanations with actor and next step
- confirmation dialogs that distinguish reversible, corrective, and destructive
actions
The UI/UX decision ledger defines concrete implementation rules. This doctrine
defines why those rules exist.
+53
View File
@@ -0,0 +1,53 @@
# Core Interface Pattern Migration
This document records the Core-owned part of the product-wide interface
pattern-language rollout. The normative product grammar and complete route
inventory live in the `govoplan` meta repository. Core owns reusable behavior;
domain modules own their compositions.
## Core Surfaces
| Surface | Pattern | Consequence and provenance contract | Evidence |
| --- | --- | --- | --- |
| User settings | Two-zone settings workspace with typed controls and unsaved-change protection | Save actions distinguish busy, unchanged, and test-in-progress states; contextual help resolves through Docs or the hosted fallback | `SettingsPage.tsx`, `test-core-interface-patterns.mjs` |
| Reusable credentials | Repeated administration with an adaptive create/edit dialog, optional password generator, and destructive confirmation | Secret values are write-only; generated candidates use the browser cryptographic API without a weak fallback and do not replace the field until explicitly confirmed; scope/permission blockers name the required action, responsible actor, and destination; unavailable row actions remain keyboard-explainable | `CredentialEnvelopeManager.tsx`, shared `PasswordField`, `PasswordGeneratorDialog`, `ActionBlockerHint`, `Button`, `TableActionGroup`, and `ConfirmDialog` |
| Retention policy | Effective-policy editor with inherited source paths and typed, narrowing-only controls | Parent locks and missing write authority are explicit; the save action distinguishes locks, missing target, loading, clean draft, and active save | `RetentionPolicyManagement.tsx`, policy logic tests, `test-core-interface-patterns.mjs` |
| Module lifecycle | Guided operator projection over durable installer-queue evidence | Preflight, handoff, progress, stale evidence, recovery, and rollback consequences remain visible | Admin module lifecycle tests and the Core installer-queue contract |
| Shared page frame | Domain-neutral headed page layout used by Core and optional modules | Standalone, workspace, and embedded modes make inset and scroll ownership explicit; sticky heading, rich descriptions, route actions, notices, loading, narrow-layout collapse, and contextual-help identity are centralized; composite administration workspaces may delegate the visible heading to their contributed panel while retaining the same frame; `AdminPageLayout` composes the contract | `PageLayout.tsx`, `page-layout.test.tsx`, Core Settings, Access administration, Docs, Mail bounce processing, Dashboard, Ops, Campaign, and `check-shared-webui-layouts.py` |
| Full-canvas workspace | Navigation/content and list/detail canvases that own pane geometry and scrolling | Navigation and split variants, primary-pane width, pane-owned or contained scrolling, responsive stacking or navigation collapse, pane labels, and contextual-help identity are centralized without encoding domain navigation | `WorkspaceLayout.tsx`, `workspace-layout.test.tsx`, Core Settings, Access administration, Docs, Organizations, Campaign, Templates, Approvals, and `check-shared-webui-layouts.py`; the raw-workspace exception baseline is empty |
| Full-height module frame | Outer module landmark and viewport/container sizing | `WorkspaceFrame` centralizes surface, overflow, box sizing, accessible naming, help identity, and application-viewport height so modules do not copy the `100vh - shell` frame | `WorkspaceFrame.tsx`, `layout-primitives.test.tsx`, Dataflow, Workflow, Datasources, Distribution Lists, Notifications, Tasks, Scheduling, Forms, Portal, Projects, Records, and Reporting |
| Responsive action toolbar | Domain-neutral action and filter grouping for pages, workspaces, editors, and overlays | Density, surface, grouping, flexible space, accessible naming, toolbar help identity, and responsive wrapping are centralized while modules retain action wording, authority, and consequence | `ActionToolbar.tsx`, `layout-primitives.test.tsx`, WYSIWYG, Calendar, Files, Forms, Templates, and the product-wide primitive check |
| Semantic page and pane action bars | Overview, collection, detail, editor, and workspace intent declared independently from frame geometry; full-canvas panes add workspace/collection/detail/editor scope | Core renders leading Reload from a guarded descriptor; editor persistence owns clean, dirty, invalid, saving, failed, and conflict feedback plus guarded Discard and far-right Save; destructive actions occupy an explicit named boundary; read-only surfaces do not invent Save | `PageActionBar.tsx`, `WorkspaceActionBar.tsx`, `PAGE_LAYOUT_USAGE_GUIDELINES.md`, component and browser conformance, every headed page and full-canvas workspace, and the discovery-based `check-shared-webui-layouts.py` |
| Catalogue and state composition | Search/filter bars, selectable navigation lists, count badges, and empty/blocked/error panels | Width, surface, wrap, selection geometry, title/description truncation, numeric emphasis, state sizing, tone and action placement are centralized; modules retain query behavior, object state and consequences | `FilterBar.tsx`, `SelectionList.tsx`, `CountBadge.tsx`, `StatePanel.tsx`, `layout-primitives.test.tsx`, and list/detail modules across Cases, Committee, Dataflow, Forms, Notifications, Portal, Postbox, Projects, Records, Reporting, Tasks, Templates, and Workflow |
| Content and form grids | Equal-column content, field, and native-form geometry | Explicit 14 columns, standard gaps, item spans, alignment, and named narrow/workspace/standard/wide collapse points replace generic and module-prefixed copies; unequal domain tracks remain local | `ContentGrid.tsx`, `layout-primitives.test.tsx`, Core dashboard/settings/mail, Calendar dialogs, Forms editor, Datasources, Postbox, Campaign, administration surfaces, and the product-wide primitive check |
| Content sections | Repeated editor/detail section surfaces | Border, surface, compact/default density, stacked flow and block rhythm are centralized without encoding section contents | `ContentSection.tsx`, `layout-primitives.test.tsx`, Datasources, Distribution Lists, Templates, Dataflow, and Workflow |
| Form sections | Reusable heading/description/action/content grouping inside forms | Plain, separated, and panel variants centralize hierarchy and narrow action placement without moving validation, permissions, values, or domain wording into Core | `FormSection.tsx`, `layout-primitives.test.tsx`, Addresses contact editing, and Quick Access preferences |
| Metric groups and drill-downs | Reusable responsive grouping around metric cards with an explicit optional detail affordance | Fixed one-to-five and auto-fit columns, minimum card widths, density, block/inset/zero spacing, and named collapse points replace the product-wide `metric-grid` class and cross-module dashboard overrides; typed link or in-page drill-downs name their destination while summary-only, non-enumerable, derived, or privacy-suppressed values remain inert | `MetricGrid.tsx`, `MetricCard.tsx`, `metric-card.test.tsx`, `layout-primitives.test.tsx`, Core and Dashboard summaries, administration, Campaign, Ops, Files, Search, and dashboard widgets |
| Description lists | Semantic property and fact presentation | Stacked and inline variants, one-to-five list columns, density, term width, wrapping, and responsive collapse replace both `admin-details-grid` and `detail-list`; `DescriptionItem` preserves native `dt`/`dd` anatomy | `DescriptionList.tsx`, `layout-primitives.test.tsx`, Access and Tenancy administration, Audit, Policy, Campaign reports/imports, Docs, Settings, Ops, and Reporting |
| Dialog anatomy | Shared outer dialog plus composable body and footer regions | Size and administration variants, body padding, descriptions, notices, fixed action wrapping, native form flow, and section grouping are centralized; focus trapping and stack lifecycle remain unchanged | `Dialog.tsx`, `DialogAnatomy.tsx`, `dialog-focus.test.tsx`, `layout-primitives.test.tsx`, Addresses, Calendar, Records, Datasources, Distribution Lists, Files, and Templates |
| Definition-editor visuals | Reusable graph palette, canvas chrome, node icon/port geometry, empty overlay and floating activity state | Core owns visual and responsive anatomy while node/edge types, validation, execution, provenance and workflow semantics remain in Dataflow or Workflow | `DefinitionPalette.tsx`, `DefinitionNodeIcon.tsx`, `FloatingStatus.tsx`, shared definition styles, Dataflow and Workflow structure/build checks |
| Shared configuration primitives | Cross-module component contract | Dialog focus, blocker structure, disabled-action focus, route/page/field/action F1 help, unsaved changes, confirmation, loading, alerts, problem lists, and policy provenance are centralized | Core component tests, `CONTEXTUAL_HELP_CONTRACT.md`, and module-permutation build |
## Boundary
Files and Mail are the first two external consumers of the layered
server/credential/policy pattern. Their own repositories retain provider
discovery, transport behavior, authorization, and migration evidence. Remaining
module surfaces are tracked by bounded module-owned issues under GovOPlaN #11;
they are not reasons to add sibling-private behavior to Core.
Raw JSON remains permitted only for diagnostics, expert inspection,
interchange, or conflict evidence. It is not a primary Core configuration
editor.
New headed pages use `PageLayout`; full-canvas modules use `WorkspaceFrame`
and, where applicable, `WorkspaceLayout`, so Core owns the frame and pane
scrolling. Module CSS continues to own unequal domain content layout, never the
shared page, workspace, toolbar, state, list, filter, metric, section, or graph
chrome. Retired copies and module-local component definitions are rejected by
`check-shared-webui-primitives.py`. That check also requires standard dialog
widths to use `Dialog size` and keeps every remaining domain-specific width in
a reviewed, decrease-only exception baseline. The companion layout check now
has zero raw page-frame and zero raw workspace exceptions, discovers semantic
consumers without a hand-maintained route list, requires semantic action bars
on `WorkspaceFrame` routes, and rejects ad-hoc panel-header toolbars.
+97
View File
@@ -0,0 +1,97 @@
# Localization And Contextual Help Quality
## Reference Language
German (`de`) is GovOPlaN's first-class reference target. Every translation key
used by a shipped WebUI must exist in German and English. German completeness is
a release gate; English remains the source-code fallback language so existing
literal labels and external developer APIs do not change semantics.
New installations and tenants default to German. Existing system, tenant, and
user preferences are preserved. The available-language and policy model can
still select another default or disable a package at the relevant scope.
Explicit high-risk help content and browser acceptance are tracked in
[Core #284](https://git.add-ideas.de/GovOPlaN/govoplan-core/issues/284).
The platform inventory recognizes both inline locale objects and generated
catalogs declared as `const de` / `const en`. Its strict mode requires both
locales and reports `de` explicitly as the reference locale.
## Help Resolution
Every focusable field and action receives a stable derived F1 identity from the
shared shell, even when the component has no dedicated help text. Resolution
falls back from field/action to dialog or page and then to the module's visible
documentation baseline.
Backend manifests publish explicit topic associations first. Core additionally
associates declared route, navigation, settings, and View surface IDs with the
module's static user or administrator documentation baseline. Feature modules
should still add exact `metadata.help_contexts` entries for consequential,
unfamiliar, policy-controlled, destructive, security-sensitive, or legally
meaningful fields and actions.
The shared retention-policy editor exposes explicit contexts for each stored
data category, audit-detail control, lower-level override switch, target
selector, reload, and save action. The Policy module owns the matching German
administrator guidance. Retention execution surfaces use separate contexts for
dry-run, destructive apply, confirmation, and outcome review so F1 opens the
consequence and recovery guidance closest to the focused control.
Shared controls may set `helpModuleId` when their documentation owner differs
from the containing page; the retention editor uses this to resolve Policy help
from both administration and Campaign surfaces.
The shared reusable-credential manager keeps Access as its documentation owner
and publishes exact contexts for credential kind, secret replacement/removal,
module and server restrictions, lower-scope visibility, activation, save, and
irreversible deletion. This ensures F1 explains secret custody and the effect on
dependent connections from system, tenant, group, user, and personal surfaces.
The source inventory treats literal `helpContextId` and
`data-help-context-id` declarations as authored help associations, including a
native control nested in `FormField`. Dynamic context expressions remain
separate evidence and generic derived fallbacks remain in the richer-help
candidate queue.
The generated `help_review_candidates` list is therefore a content-depth queue,
not a list of controls on which F1 cannot work. It should prioritize:
1. effect, deletion, delivery, retention, disclosure, encryption, and recovery;
2. identity, representation, mandate, institutional context, and purpose;
3. valid-time versus recorded-time selection;
4. provider authority, synchronization, conflict, and outcome unknown;
5. fields whose consequences are not evident from their label.
The shared browser conformance journey mounts the production Help menu and
resolver. It proves that F1 uses the focused control rather than only the page,
maps an exact retention action to Policy-owned administrator documentation,
retains the page context as fallback for derived actions, exposes an accessible
modal at narrow widths, closes with Escape, and restores focus to the triggering
control. Module journeys should add their own exact high-risk mappings; they do
not need to reimplement the keyboard or dialog mechanics.
## Verification
```bash
cd /mnt/DATA/git/govoplan
/mnt/DATA/git/govoplan/.venv/bin/python \
tools/inventory/platform-interface-inventory.py \
--strict --strict-declarations --strict-endpoints
```
The check must report:
- reference locale `de` present and complete;
- no used key missing from `de` or `en`;
- every field has a resolvable F1 context;
- no duplicate stable IDs;
- no undeclared public WebUI surface;
- no stale runtime route or endpoint declaration.
Browser acceptance is part of the focused workspace gate and can be run alone:
```bash
cd /mnt/DATA/git/govoplan-core/webui
npm run test:conformance
```
File diff suppressed because it is too large Load Diff
+66
View File
@@ -0,0 +1,66 @@
# Module Lifecycle Recovery
## Migration Revision Namespace
All enabled module migration directories are assembled into one Alembic graph. Revision IDs are therefore global across Core and every module even though each module owns a separate `migrations/versions` directory. Core validates literal revision declarations before constructing the graph and rejects duplicates with both file paths. A module must assign a new globally unique revision ID; reusing another module's ID can otherwise make Alembic treat an unrelated schema change as already applied or report an ancestor/head overlap.
When correcting a collision that has already reached a database, first verify the schema objects that identify which migration actually ran. Rename the unapplied migration, or transactionally translate the corresponding `alembic_version` row when the applied owner is unambiguous. Never add both colliding IDs as heads or blindly stamp the database.
Package changes and live module-graph changes use Core's durable recovery
ledger. The local `install.lock` still prevents duplicate work in one runtime
directory; the database lease `core:module-lifecycle:deployment` is the
deployment-wide authority across API, installer, worker, and scheduler nodes.
## Declared Boundaries
| Operation | Recovery mode | Completion condition |
| --- | --- | --- |
| `module-lifecycle.pre-migration` | compensation | package, WebUI, manifest, and desired-graph evidence match |
| `module-lifecycle.post-migration` | forward recovery | migration tasks, manifests, desired graph, restart, and health are verified |
| `module-retirement.destroy-data` | snapshot restore | a hashed, restore-checked backup exists and retirement state is verified |
| `module-runtime.apply-graph` | compensation | hooks, capability contexts, active graph, and workflow contributions match |
The installer prepares the recovery operation before it captures the database
snapshot. A full database restore therefore retains the prepared operation and
its fence instead of erasing the fact that a mutation was attempted. Backup
artifacts are hashed and sized before any package, migration, or retirement
effect starts.
Every command boundary records the command source and canonical hashes of the
redacted command/result records. Credentials, database URLs, command output,
and package-registry secrets are never copied into recovery evidence.
## Failure And Retry Rules
- A conclusive failure before effects is terminal `failed`.
- A command or compensatable effect that started but did not complete is
`recovery_required`.
- A lost or unexpected outcome after a migration/external boundary is
`outcome_unknown`.
- A verified package/database rollback becomes `recovered`.
- A supervised install becomes `succeeded` only after restart and all configured
health probes succeed.
An unresolved lifecycle operation blocks every later lifecycle mutation on the
same deployment fence, even after its execution lease is released. Operators
must inspect the checkpoint chain and run record, restore or complete the
declared recovery path, and explicitly reconcile the operation. A new install
must not be used as an implicit retry.
Live graph changes use the same fence. A non-migrating hook or registry failure
restores the prior in-process graph and records verified compensation. A failure
after migrations begin remains unresolved because restoring the process-local
registry does not reverse database schema effects.
## Operator Evidence
The installer run record contains the recovery operation id, mode, plan hash,
and current lifecycle status. The Ops recovery view is authoritative for the
durable state and evidence-chain result. Keep both the run directory and the
state-service backup evidence until the operation is terminal and the normal
retention policy permits removal.
Run the module installer rollback drill and recovery-runtime test matrix before
enabling lifecycle mutation in a new deployment. Shared-state deployments must
still use immutable release images; the ledger does not make in-place package
mutation across replicas safe.
+158
View File
@@ -0,0 +1,158 @@
# Page Layout and Action Guidelines
This document defines the binding composition grammar for headed GovOPlaN
pages. Core owns the reusable anatomy; each module owns its domain actions,
wording, authorization, consequences, and data state.
## Required Page Frame
- Use `PageLayout` for every headed standalone, workspace, or embedded page.
- Declare exactly one semantic `archetype`; do not infer page intent from the
`mode`, which controls geometry and scroll ownership only.
- Use `WorkspaceFrame` for a full-height module surface and
`WorkspaceLayout` only where navigation/content or list/detail panes are
genuinely part of the interaction.
- Put page-wide feedback in `PageLayout` notices. Use `DismissibleAlert` for a
recoverable warning or failure and `StatePanel` when the entire surface is
loading, empty, unavailable, or blocked.
- Do not reproduce shared page padding, heading, toolbar, form-grid, section,
table, dialog, or breakpoint CSS in a module.
## Product Side Rail
Module manifests contribute stable navigation surface identifiers, labels,
paths, icons, and default order. Core owns the side-rail composition and the
shared `NavigationPreferenceEditor`; modules must not fork this editor or
persist their own rail ordering.
Navigation preferences are layered in this order: module defaults, system,
tenant, then user. Each higher layer may reorder or change visibility. System
and tenant administrators may lock an entry visible; a lower layer can still
move that entry, but cannot hide it. Personal preferences cannot create locks.
An unset preference inherits the complete lower layer, while “Use inherited
order” removes the current layer rather than copying its values. Unknown item
identifiers remain harmless so uninstalling, disabling, or later reinstalling
a module does not corrupt the rail.
The platform module response projects module, system, and tenant layer states
alongside the effective user state. Editors must initialize from the layer
immediately below the scope they edit, so a system or tenant administrator's
personal preference is never promoted accidentally. Preference saves refresh
the platform module projection. View policy, permissions, and tenant module
entitlements remain independent final visibility gates; changing rail
preferences never grants access.
## Semantic Page Archetypes
| Archetype | Use when |
| --- | --- |
| `overview` | The page summarizes health, metrics, or several peer areas without owning one primary collection or draft. |
| `collection` | The primary object is a searchable/listable collection and Create, when available, applies to that collection. |
| `detail` | The page primarily presents one record, report, or immutable projection. |
| `editor` | The page owns one explicit draft with Save and Discard behavior. |
| `workspace` | The page coordinates several panes, stages, or task-local operations that cannot honestly be reduced to one record or draft. |
The archetype remains stable for the current interaction. A page may switch
from `overview` to `editor` when the user explicitly enters configuration
mode. It must not call a page an editor merely because a dialog or an inline
filter is editable.
## Page Action Rules
Pass one `PageActionBar` to the `PageLayout` `actions` slot. Full-canvas
workspaces use the same contract through `WorkspaceActionBar`, with an explicit
`workspace`, `collection-pane`, `detail-pane`, or `editor-pane` scope. The
variant makes the surface's intent inspectable and preserves the same keyboard
and visual order across modules. `ActionToolbar` remains the lower-level
component for section-local controls; it is not a substitute for a semantic
page or pane action bar.
| Page kind | Leading group | Trailing group |
| --- | --- | --- |
| Overview | Reload when refreshable, then context | Help, then ordinary primary actions |
| Collection | Reload when refreshable, then collection context such as export | Help, then Create at the far right |
| Detail | Reload when refreshable, then object context | Help, ordinary primary actions, then a separated destructive group |
| Editor | Reload only when refresh is a distinct safe operation, then context | Dirty state, Help, ordinary primary actions, separated destructive actions, Discard, then Save at the far right |
| Workspace | Reload when the coordinated projection can become stale, then task context | Help, ordinary primary actions, then a separated destructive group |
Reload means re-fetch or re-evaluate the current surface. A page declaring
`refreshable` must provide it, and a non-refreshable page must not use Reload as
a synonym for Cancel, Reset, or Discard. Reload never silently destroys a dirty
draft. Create is a collection-wide action and is not duplicated in a
persistent side panel. Save is present only where the page owns an editable
draft; a read-only detail page must not display a disabled or inert Save merely
to fill the slot.
Editor bars always keep Discard and Save visible. Their required `state`
projection is one of `clean`, `dirty`, `invalid`, `saving`, `save-failed`, or
`conflict`, and the central component announces it through a live status label.
Clean and saving states disable both persistence actions; invalid disables Save
while retaining Discard. Failed saves and conflicts keep the draft recoverable
and allow an authorized retry after the module has shown the owning error or
conflict evidence. A module may add a more specific validation, policy, or
permission blocker. The editor must register its draft with
`useUnsavedDraftGuard` (or a shared hook that uses the same registration
contract), so browser unload, route navigation, section changes, Reload, and
the explicit Discard path cannot silently lose work.
Reload is rendered by Core from a descriptor rather than passed as arbitrary
button markup. It can project `current`, `stale`, `reloading`, or
`reload-failed`; `loading` is the shorthand for `reloading`. A failed refresh
must preserve usable loaded data, expose its stale/failure state, and leave
Reload available for recovery. Reload goes through the same unsaved-navigation
guard as route changes.
Destructive page actions use `destructiveActions`; never put a danger action in
`contextActions` or the ordinary primary group. Core renders a persistent
visual and semantic boundary before this group. In an editor it precedes the
Discard/Save pair, keeping Save in the final keyboard and visual position.
`PageActionBar` controls non-editor placement and owns the standard editor
persistence buttons. Other actions continue to use central
`Button`, `IconButton`, or `TableActionGroup` components. When an action is
visible but unavailable because of permission, target, policy, state, or
validation, keep it in its stable slot and supply `disabledReason`. Do not
silently hide a normally applicable action.
## Forms and Dialogs
- Compose forms from `FormLayout`/`FormGrid`, `FormSection`, and `FormField`.
- Use `FieldLabel` through `FormField` for every field that is not genuinely
self-explanatory; record justified omissions in the owning UI ledger.
- Use `Dialog`, `DialogForm`, `DialogSection`, and `DialogActions` for modal
work. A dialog can be domain-specific while its anatomy remains central.
- Use `useUnsavedDraftGuard` for explicit Discard and guarded navigation on an
editable page or dialog.
- Explain irreversible or operationally consequential actions before the
commit button, including reversibility and durable evidence.
## Collections and Details
- Use `FilterBar` for collection query controls and `DataGrid` for tabular
collections. Keep a single ordered `TableActionGroup` action set per table.
- Use `MetricGrid`/`MetricCard` for summary measures, `Card` or
`ContentSection` for logical sections, and `DescriptionList` for labelled
facts.
- Add a `MetricCard.drilldown` only when the displayed measure has a useful,
authorized underlying collection or detail. Name the destination explicitly
(for example, “Review failed deliveries”) and preserve the current scope and
filters in its `href` or action. The card itself remains non-interactive so
the action is visible and keyboard-predictable. Derived, privacy-suppressed,
non-enumerable, or purely informational aggregates remain plain metrics;
when an ordinarily available drill-down is temporarily blocked, keep its
action and provide `disabledReason`.
- Preserve loaded data after a refresh failure and mark it stale; offer Reload
as the recovery action. Distinguish initial loading, empty, unavailable,
permission-blocked, conflict, success, and retry states.
## Review Evidence
Every new or changed page or workspace pane must have structural evidence for
its frame, semantic archetype/scope and slot order, refresh declaration, shared
component usage, stable disabled actions, dirty guard, destructive boundary,
and module-owned help identity. Type checks enforce conditional Reload and
editor persistence props. The product check discovers all consumers, rejects
undeclared archetypes and `ActionToolbar` panel-header copies, and requires
semantic actions for every `WorkspaceFrame` route. Browser conformance confirms
keyboard order, lifecycle changes, accessibility, destructive separation,
narrow wrapping, and screenshot geometry.
+185
View File
@@ -0,0 +1,185 @@
# GovOPlaN Policy Contracts
GovOPlaN has several policy families that are moving out of core into owning
modules. The shared kernel contract keeps their decision and provenance shape
consistent while each module still owns its domain rules.
## Current Policy Inventory
| Policy area | Current owner | Runtime surface | Notes |
| --- | --- | --- | --- |
| Privacy retention | `govoplan-policy` implementation and routes, with compatibility helpers in core | `/api/v1/admin/privacy-retention/policies/{scope}` and `/explain`; capability `policy.privacyRetention` | System, tenant, user, group, and campaign sources merge into the effective retention policy. Parent locks block lower-level widening. |
| Mail profile policy | `govoplan-mail` | `/api/v1/mail/policies/{scope}` | Uses the same source-step path format for system, tenant, owner, and campaign provenance. |
| RBAC/access policy | `govoplan-access` | access capabilities in `govoplan_core.core.access` | Permission decisions should use access capability contracts. Explain responses should adopt `PolicyDecision` when an API-level explanation is added. |
| Governance defaults | `govoplan-admin` plus `govoplan-access` materializer | admin settings, governance template routes, access materialization capability | System governance can block tenant-local groups, roles, and API keys. |
| Delegation and ownership policy | access/campaign/mail/files modules | capability checks and owner-scoped APIs | Source provenance should use this contract when policies become externally explainable. |
| Definition governance | `govoplan-policy` | capability `policy.definitionGovernance` | Resolves view, edit, run/start, reuse, derive, and automate for system, tenant, group, and user Dataflow/Workflow definitions. |
| Function assignment governance | `govoplan-policy` | capability `policy.functionAssignmentGovernance` | Returns current review steps, delegation depth/validity ceilings, and explicit timed-escalation targets consumed by IDM. |
## Policy Decision
The shared DTO lives in `govoplan_core.core.policy.PolicyDecision`.
```json
{
"allowed": false,
"reason": "Parent retention policy locks lower-level changes.",
"source_path": [
{
"scope_type": "system",
"scope_id": null,
"path": "system",
"label": "System",
"applied_fields": ["allow_lower_level_limits"],
"policy": {}
}
],
"requirements": ["raw_campaign_json_retention_days"],
"details": {
"blocked_fields": ["raw_campaign_json_retention_days"]
}
}
```
`allowed` is the effective answer for the checked action. `reason` is a stable,
human-readable summary. `source_path` lists the policy sources that explain the
answer. `requirements` lists machine-readable blockers or prerequisites, and
`details` carries domain-specific structured context.
Every source step should be concrete enough for an operator to understand the
decision without knowing internal merge rules. Use real scope labels such as
`System`, `Tenant`, `Owner user`, `Group`, or a campaign/profile name. Include
the stable `path`, the fields applied by that step, and the local policy
fragment that caused them. This lets UIs render explanations like
`System: Allow > Tenant: Deny without override` without additional lookups.
If a policy family cannot expose the full local fragment for security reasons,
it must still include a redacted structured value that identifies the applied
field and the effective allow/deny or lock state.
## Source Path Format
Policy source paths are stable string identifiers for provenance steps:
- `system`
- `<scope_type>:<url-encoded-scope-id>`
Supported scope types are `system`, `tenant`, `user`, `group`, and `campaign`.
Examples:
- `tenant:4a45b4fe-1d86-43ce-9d10-6022333f4d4b`
- `campaign:campaign%2Fwith%20space`
Use `policy_source_path()` and `parse_policy_source_path()` instead of building
or splitting these strings manually.
## Retention Explain Endpoint
`GET /api/v1/admin/privacy-retention/policies/{scope_type}/explain` returns:
- `scope_type` and optional `scope_id`
- `decision`, using the shared `PolicyDecision` shape
- `effective_policy`
- optional `parent_policy`
- `effective_policy_sources`
- `parent_policy_sources`
- `blocked_fields`
The endpoint is read-only. Enforcement remains in the existing policy write
path. For lower-level scopes, `blocked_fields` is derived from the parent
policy's `allow_lower_level_limits`; clients can use it to disable local
controls before attempting a write.
The retention implementation lives in `govoplan-policy`
(`govoplan_policy.backend.retention`). Core keeps
`govoplan_core.privacy.retention` only as a compatibility facade for older
imports. Effective/scoped retention behavior dispatches through the
`policy.privacyRetention` capability; core does not import policy implementation
code as a hidden fallback when the module is disabled or no runtime is active.
New backend code should import policy-owned retention behavior from
`govoplan-policy` or request the capability, not add new implementation logic
to core.
The retention API DTOs live in `govoplan_core.privacy.schemas`.
`PrivacyRetentionPolicyItem`, `PrivacyRetentionPolicyPatchItem`,
`RETENTION_POLICY_FIELD_KEYS`, and `default_allow_lower_level_limits()` are
platform contracts because admin, access compatibility, and policy routes expose
the same stable retention payload shape. The policy engine's internal
`PrivacyRetentionPolicy` and `PrivacyRetentionPolicyPatch` models stay in
`govoplan-policy`, because they carry implementation validators and merge
behavior that are not generic API contracts.
Tenant administration DTOs remain owned by `govoplan-tenancy`; access keeps
matching compatibility DTOs only for its legacy admin surface. Admin overview
responses remain module-local because the same counters are exposed from
different menu contexts and are not yet a separately versioned platform API.
## Definition Governance
Dataflow and Workflow submit a `DefinitionGovernanceRequest` using only stable
scope, principal, status, definition-kind, and limit fields. Policy returns a
standard `PolicyDecision`. System definitions may be inherited as read-only;
group and user definitions are visible only in matching contexts. Templates
may be viewed and derived but never run or automated. A derived definition
passes its pinned ancestor limits back through the request context, and Policy
applies those limits as ceilings rather than defaults that can be broadened.
When the capability is absent, modules must not silently emulate cross-scope
inheritance. Their conservative fallback is limited to local tenant
definitions and disables reuse, derivation, and automation.
## Function Assignment Delegation And Escalation
`FunctionAssignmentGovernanceDecision` is the versioned cross-module contract
for request/grant review. In addition to the required holder, authority, and
recipient steps, it returns `delegation_allowed`,
`maximum_delegation_depth`, `maximum_delegated_validity_days`, and typed
`FunctionAssignmentEscalationRule` entries. Each escalation entry binds one
review step to an exact target function and timeout.
The decision is a current ceiling, not durable authorization. IDM must recheck
the complete assignment-source chain and all recorded decisions before final
application. An elapsed timeout creates explicit state and evidence; it must
never be interpreted as approval or as permission to silently substitute an
approver. Missing providers, malformed rules, invalid chains, or tightened
limits fail closed with an explainable reason.
## Bounded Impact-Subject Providers
Policy impact previews discover optional subject providers through capability
names beginning with `policy.impactSubjects.`. The suffix is the stable
provider ID; for example, Views contributes `policy.impactSubjects.views`.
Providers implement `PolicyImpactSubjectProvider` and receive a
`PolicyImpactPopulationRequest` containing the active tenant, policy family,
an explicit selector, actor scopes, detail-disclosure decision, and a limit of
at most 500. They return `PolicyImpactSubjectBatch` with unique opaque subject
references and an explicit `complete`, `sampled`, `truncated`, or `unavailable`
state. An unavailable batch must explain the gap, and a total may never be
smaller than the returned subject count.
Core does not scan module data or evaluate domain policy. The owning module
selects and permission-filters its candidates; Policy compares the current and
proposed decisions and controls response disclosure. A caller must select one
or more provider populations explicitly. This preserves optional-module
boundaries and prevents a seemingly harmless preview from becoming an
unbounded platform query. Providers must not include credentials, secrets, or
unfiltered cross-tenant labels in subject attributes.
## Frontend Contract
Policy UIs must:
- render effective source provenance when `effective_policy_sources` is present
- display a field-level path when the source data is shown next to a specific
setting, using concrete source labels and stop at the first non-overridable
deny/lock
- disable local field controls when the parent policy sets that field's
lower-level limit to `false`
- avoid sending locked fields or re-enable attempts in save payloads
- show inherited values separately from local overrides
- require a current impact preview before enabling a governed high-impact save,
preserve its proposal hash on commit, and explain incomplete population
coverage rather than presenting unavailable providers as zero impact
The core WebUI helper `privacyRetentionParentAllowsField()` centralizes the
field-lock decision used by the retention editor and its lightweight module
tests.
+214
View File
@@ -0,0 +1,214 @@
# Postbox End-To-End Encryption Architecture
This document records the encryption boundary for GovOPlaN postboxes. Postbox
now implements the server-side contracts for three selectable profiles:
unencrypted content, institution-managed server envelopes, and externally
produced E2EE envelopes. The E2EE contract is operational—the server rejects
plaintext and retains ciphertext, signed manifests, wrapped keys, and digest
evidence—but a reviewed browser/device client and private-key custody provider
remain separately deployed responsibilities.
The core principle is that a postbox can become a trusted administrative
communication channel without requiring the server to see plaintext content.
The server may route, store, authorize, audit, retain, and expire messages while
message bodies and attachments remain client-encrypted.
## Goals
- asynchronous encrypted delivery for internal and portal-facing postboxes
- personal, organizational, role-bound, and function-bound postboxes
- attachments encrypted with the message
- access based on current role/function membership when configured
- honest retraction and expiry semantics
- auditable key access, delivery, and fetch events
- support for external recipients without platform accounts
- replaceable identity and trust providers
## Envelope Model
The target model is envelope encryption:
- Generate one random data encryption key per message or attachment set.
- Encrypt content with an authenticated encryption algorithm.
- Wrap the data encryption key for each authorized recipient or role mailbox.
- Store only ciphertext, wrapped keys, signed manifests, and governed metadata
on the server.
Algorithm choices should remain replaceable behind a crypto profile. The first
profile should prefer standard, reviewed primitives such as HPKE for key
wrapping and AEAD encryption for content.
## Product Profiles And Default
The content-protection policy is configurable per exact Postbox or immutable
template revision:
- `server_envelope_v1` is the recommended default. An institution-selected
Encryption vault controls server-readable envelopes and their migration
evidence. It is not end-to-end encryption.
- `external_e2ee_v1` is server-blind. An approved client or producer supplies
the ciphertext reference, signed manifest, wrapped recipient keys, key epoch,
and SHA-256 plaintext digest. GovOPlaN has no private key that can decrypt it.
- `plaintext_v1` stores clear content for institutions that explicitly choose
that boundary.
Operational metadata—including subject, routing, participants,
classifications, timestamps, attachment references, receipts, and retention
state—remains visible under every profile. Administrators therefore choose a
content-protection boundary, not a metadata-anonymity profile.
The standard policy grants new incumbents history since assignment, uses key
rewrapping for ordinary rotation and content re-encryption after compromise,
requires two-person institutional recovery and dual-control hand-over,
emergency, export, and destruction, requires strong external identity, and
limits vacancy escalation to metadata. Deployments may select other policy
values rather than inheriting a decision from GovOPlaN.
## Governed Profile Changes
A profile transition applies to new messages immediately and increments the
Postbox key epoch. Retained history can remain under the previous profile or be
migrated. The transition ledger records source and target profiles/vaults,
authority route, consent and key-holder evidence, quorum, reason, immutable
configuration snapshot, per-message source and target digest, and outcome.
Plaintext and managed-envelope migrations can use the server-side Encryption
capability. Managed decrypt, export, and re-encryption operations also create
Encryption migration records so old envelopes are disposed of through the
governed provider contract. Any transition to or from E2EE pauses each retained
message for an approved client transform. The client must return plaintext or
ciphertext as appropriate, plus evidence and the original content digest;
Postbox verifies digest continuity before changing the stored representation.
Leaving E2EE requires user-consent evidence, while changing managed history
requires institutional key-holder evidence. Dual control can require both.
This transition mechanism cannot revoke plaintext already decrypted, copied,
printed, or exported. Administrators must explicitly acknowledge that residual
disclosure before a transition is accepted.
## Identity And Device Keys
The platform should distinguish:
- account identity
- tenant membership
- role/function assignment
- device key
- postbox binding
Identity providers and directories can authenticate users and provide membership
facts, but they must not see postbox private keys or message plaintext.
The trust layer should provide:
- public key directory
- account or identity signing keys
- per-device encryption keys
- device registration and revocation
- key rotation and epoch tracking
- recovery policy hooks
Recovery must be organizationally governed. A server-held universal plaintext
key would defeat the E2EE claim; any escrow, threshold recovery, or emergency
grant needs an explicit assurance profile, authority/quorum, audit trail, and
user-visible consequence.
## Role And Function Postboxes
Role-bound access needs special handling. A postbox can be bound to an
organizational unit and a role or function. Current members can access current
messages according to policy; former members should lose access to not-yet
fetched material when revocation is still technically enforceable.
The target design should support role encryption keys or an equivalent
rewrapping service:
- sender encrypts the content key for the role/function postbox
- access service verifies current membership and required assurance
- trust service rewraps the content key to the actor's current device key
- audit records the key access decision and fetch event
Key epochs are required when role membership changes. Older messages may remain
readable according to policy, but new access must use the current epoch.
The function-bound container exists independently of membership. It may remain
vacant and continue to receive ciphertext without falling back to an unrelated
personal mailbox. Zero, one, or several incumbents are valid states. Each
incumbent receives an independently auditable, device-bound wrapped-key path;
the postbox is never copied into their account ownership.
A new assignment or hand-over rotates the function/postbox key epoch. Envelope
encryption permits the normal rotation path to rewrap per-message data keys
rather than rewrite large ciphertext objects; a security policy may require
full content re-encryption for selected compromise or cryptographic-profile
events. The history available to a new incumbent must be selected policy (all
retained history, a bounded historical window, or assignment-time content) and
recorded with the grant.
Delegation is a time-bounded represented-function grant, not a copy or
substitution of the postbox. Expiry or withdrawal stops future key release and
actions. It cannot revoke plaintext already decrypted, printed, exported, or
captured outside the platform. Multiple simultaneous incumbents and delegates
remain distinguishable in key-fetch and action evidence.
Postbox content and signed manifests are immutable. Correction or replacement
creates a linked new object/version; it never silently substitutes ciphertext
or evidence that another actor may already have inspected.
## External Recipients
External recipients may need one-time or time-limited access without a full
platform account. The target model should support capability links or invitation
tokens that are:
- scoped to specific message or attachment resources
- time-limited
- optionally one-time
- protected by an out-of-band secret, passphrase, or stronger external identity
proof
- revocable before key fetch
- fully audited
## Retraction Semantics
GovOPlaN should be honest about retraction.
Before a recipient fetches a key or decrypts content, the system can revoke
tokens, remove wrapped-key access, expire links, and delete ciphertext according
to retention policy.
After a recipient has decrypted or copied plaintext, the system cannot make the
recipient forget it. The platform can only record access, revoke future access,
notify parties, and apply legal or organizational controls.
The UI must explain this distinction whenever it offers expiry, retraction, or
message withdrawal.
## Server Responsibilities
The server remains important even when content is encrypted:
- store ciphertext and signed manifests
- store routing and policy metadata
- enforce access before key release or rewrapping
- provide public key directory access
- emit notifications without plaintext content
- record audit events
- enforce retention and expiry where possible
- expose diagnostics for delivery and key-access failures
## Module Ownership
- `govoplan-postbox` owns postbox bindings, postbox messages, message metadata,
and postbox UI.
- `govoplan-identity-trust` owns device keys, public key directory, key epochs,
and assurance integration.
- `govoplan-access` owns current identity, membership, function, delegation, and
permission decisions.
- `govoplan-policy` owns retention, retraction, and security policy decisions.
- `govoplan-audit` owns durable audit traces.
- `govoplan-files` owns managed file storage when encrypted postbox attachments
are backed by file objects.
No module should import another module's internals to decrypt content. All
interaction must use capabilities, DTOs, and audited service contracts.
+348
View File
@@ -0,0 +1,348 @@
# Public-Sector Integration Strategy
GovOPlaN should integrate with the existing public-sector software landscape
before deciding to replace specialist workflows. This document is the core
strategy index. The executable connector catalogue lives in
`govoplan-connectors/docs/PUBLIC_SECTOR_INTEGRATION_CATALOGUE.md`.
The canonical cumulative maturity model and external-object DTO are documented
in [EXTERNAL_REFERENCES_AND_INTEGRATION_MATURITY.md](EXTERNAL_REFERENCES_AND_INTEGRATION_MATURITY.md).
## Strategy Labels
Use one or more of these labels for every external system family:
- `integrate`: GovOPlaN talks to the system through a stable API/protocol.
- `link`: GovOPlaN stores external references and opens the external system for
source-of-truth work.
- `import`: GovOPlaN consumes data or files into governed module storage.
- `synchronize`: GovOPlaN keeps selected records aligned both ways or through a
source-of-truth rule.
- `replace selected workflow`: GovOPlaN may own a narrow workflow where the
external product is weak, but does not replace the whole product family.
- `no first-class support`: GovOPlaN only stores manual references unless a
deployment project creates a specific connector.
## Initial Classification
| System family | Examples | Default strategy | Likely owner |
| --- | --- | --- | --- |
| File providers | SMB/CIFS, WebDAV, Nextcloud, Seafile, SFTP, S3 | integrate, import, link | `govoplan-files`, connector inventory in `govoplan-connectors` |
| Project/task management | OpenProject, Jira, Redmine, Microsoft Planner | link, synchronize selected records, replace selected workflow only after proof | `govoplan-connectors`, later `govoplan-tasks` or workflow modules |
| Identity providers | LDAP, Active Directory, OIDC, SAML, OpenDesk IDM | integrate, synchronize | `govoplan-idm`, `govoplan-access` |
| Mail and groupware | IMAP/SMTP, Open-Xchange, Exchange/M365, CalDAV/CardDAV | integrate, link | `govoplan-mail`, `govoplan-calendar`, `govoplan-connectors` |
| DMS/e-file/archive | d.velop/d.3, enaio, ELO, Fabasoft, CMIS, VIS/eAkte | link, import, synchronize selected metadata | `govoplan-dms`, `govoplan-files`, `govoplan-records` |
| ERP/finance/procurement | SAP, MACH, Infoma, DATEV, procurement feeds | export, import, synchronize; do not replace by default | `govoplan-erp`, `govoplan-procurement`, `govoplan-ledger`, `govoplan-payments` |
| Public-sector transport | FIT-Connect, XTA/OSCI, Peppol access points | integrate, publish, receive | dedicated protocol modules plus `govoplan-connectors` inventory |
| Standards registries | XRepository, XOE/V catalogues | link, import metadata/cache | `govoplan-connectors`, `govoplan-xoev` |
| Publication/data exchange | RSS, open-data APIs, API feeds, CSV/Excel drops | consume, publish, transform | proposed `govoplan-datasources`, proposed `govoplan-dataflow`, `govoplan-reporting` |
| Collaboration suites | Matrix, Jitsi, BigBlueButton, Nextcloud Talk, Collabora/OnlyOffice | integrate, link; native behavior only for governed evidence | `govoplan-connectors`, `govoplan-dms`, `govoplan-workflow` |
| Specialist Fachverfahren | register-specific and domain-specific systems | link first; integrate/import when a real project supplies contracts | domain module or deployment-specific connector |
## Landscape Catalogue
This catalogue is intentionally implementation-oriented. Each entry records the
first API/auth/data assumptions needed to turn an inventory entry into a
connector or module issue.
### Citizen And Service Portals
- Strategy: integrate/link first; replace selected intake workflow only when a
GovOPlaN portal package owns the complete journey.
- Protocol/API surface: REST/JSON APIs, form submission webhooks, OIDC/SAML
login, eID interfaces where available, file-upload callbacks, case-status
callbacks.
- Auth model: OIDC/SAML service clients, signed webhook secrets, tenant-scoped
API keys, later eID/trust-provider handoff.
- Data shape: applicant identity reference, application form payload,
attachment references, consent declarations, status events, receipt IDs.
- Deployment assumptions: externally reachable HTTPS, reverse proxy, portal DMZ
separation, strict CSRF/origin settings, large upload path.
- Risks: personal data exposure, duplicate identity mapping, partial
submissions, upload malware, inconsistent portal status models.
- MVP test path: submit a test application with one file, create a form
submission/case/task stub, return a receipt and status reference.
- Owner/priority: `govoplan-portal`, `govoplan-forms-runtime`,
`govoplan-files`, Wave 1.
### DMS, E-File, Records, And Archives
- Strategy: link/import/synchronize selected metadata; do not replace the DMS by
default.
- Protocol/API surface: CMIS, WebDAV, vendor REST APIs, S3/object archive
staging, file-plan export/import, archive handoff APIs.
- Auth model: service accounts, OAuth/OIDC where supported, mTLS for regulated
archives, secret references for vendor tokens.
- Data shape: document ID, version, file-plan/classification code, retention
metadata, owner/case reference, external URL, checksum, lock/legal-hold state.
- Deployment assumptions: usually internal network or VPN, strict storage
quotas, existing retention policies, archive immutability requirements.
- Risks: record duplication, broken legal hold, permission mismatch, version
drift, destructive retention/export mistakes.
- MVP test path: create a connector inventory entry, test read-only metadata
lookup, link one GovOPlaN file/case evidence item to an external document.
- Owner/priority: `govoplan-dms`, `govoplan-records`, `govoplan-files`,
`govoplan-connectors`, Wave 2/5.
### ERP, Finance, Procurement, And Accounting
- Strategy: export/import/synchronize selected records; replacement only by
narrow domain decision.
- Protocol/API surface: vendor REST/SOAP APIs, CSV/XML batch exchange, SFTP,
XRechnung/Peppol, XBestellung/procurement feeds, payment reconciliation files.
- Auth model: service accounts, client certificates, mTLS, SFTP keys, token
references, environment-specific account separation.
- Data shape: debtor/creditor reference, payment request, invoice, order,
budget/cost-center code, booking status, receipt/evidence reference.
- Deployment assumptions: batch windows, finance-system approval workflows,
test tenants often separated from production by vendor process.
- Risks: financial posting errors, double export, tax/legal data retention,
inconsistent master data, irreversible accounting handoff.
- MVP test path: dry-run export of one payment/accounting handoff file with
checksum, validation report, and no remote posting.
- Owner/priority: `govoplan-payments`, `govoplan-ledger`,
`govoplan-xrechnung`, `govoplan-erp`, `govoplan-procurement`, Wave 1/6.
### Identity, IAM, And Directory Services
- Strategy: integrate/synchronize; access remains GovOPlaN's local
authorization boundary.
- Protocol/API surface: LDAP, Active Directory, SCIM, OIDC, SAML, OpenDesk IDM
APIs, group membership sync, account deactivation feeds.
- Auth model: bind accounts, service clients, OIDC/SAML metadata, SCIM tokens,
certificate-backed clients where required.
- Data shape: account, user, group, membership, role claim, tenant/org-unit
mapping, status, external directory ID.
- Deployment assumptions: directory is usually internal; identity provider may
be organization-wide and not GovOPlaN-owned.
- Risks: privilege escalation through group mapping, stale memberships, account
collision, deprovisioning latency, tenant-boundary mistakes.
- MVP test path: read-only directory profile test, map one external group to a
tenant group, show a dry-run membership diff.
- Owner/priority: `govoplan-idm`, `govoplan-access`, Wave 1.
### Groupware, Mail, Calendar, And Collaboration
- Strategy: integrate/link; native behavior only where GovOPlaN needs governed
evidence or process state.
- Protocol/API surface: IMAP/SMTP, CalDAV/CardDAV, Open-Xchange APIs, Microsoft
Graph/EWS, Matrix APIs, Jitsi/BigBlueButton APIs, Collabora/OnlyOffice
integration points.
- Auth model: service accounts, delegated OAuth/OIDC, app passwords, mailbox
credentials, groupware-specific tokens, secret references.
- Data shape: mailbox folder/message references, event/free-busy data, meeting
URL, chat room ID, participant list, document-editing session reference.
- Deployment assumptions: often internal/existing tenant infrastructure; mail
and calendar may be separate from identity even in OpenDesk-style stacks.
- Risks: mail credential exposure, calendar privacy, double invitations, room
booking conflicts, chat/document data escaping retention rules.
- MVP test path: profile test for mailbox/calendar reachability, read-only
folder/free-busy lookup, create a non-production event/message draft.
- Owner/priority: `govoplan-mail`, `govoplan-calendar`,
`govoplan-connectors`, Wave 1/2.
#### Collaboration-suite boundary and hand-offs
Collaboration remains connector-first. The product named below never changes
which GovOPlaN module owns the administrative meaning of the work:
| External family | Initial posture | GovOPlaN semantic owner | Connector-owned boundary |
| --- | --- | --- | --- |
| Collabora Online, OnlyOffice, Nextcloud Office | Link an externally edited document and its editing session; import a governed rendition only when required | DMS owns document/version, lock, review, approval, retention, and collaboration-session evidence; Files owns stored bytes | Discovery, endpoint health, WOPI/vendor session exchange, callbacks, and provider object references |
| Matrix, Mattermost, Rocket.Chat, Nextcloud Talk | Create or link a room/thread for a governed work context; do not mirror all conversation history by default | The initiating Case, Workflow, or Task owns the work-context link and disposition; DMS/Records own retained evidence deliberately captured from it | Room/thread creation, membership synchronization, webhook/event normalization, and stable external links |
| Jitsi and BigBlueButton | Provision or link a conference for an existing appointment/event | Appointments owns booking intent; Calendar owns event, attendee, invitation, and time semantics | Conference provisioning, join/moderator references, provider lifecycle, and bounded attendance/result callbacks |
| OpenProject and comparable project suites | Link first, then publish or synchronize selected work packages | Tasks owns GovOPlaN task state; Workflow owns orchestration; Cases own case state and evidence references | Project/work-package lookup, publish/synchronize transport, webhooks, version tokens, and external URLs |
| Cross-suite activity streams | Consume normalized, bounded events only for an authorized work context | The receiving module decides whether an event changes state or becomes evidence; Audit records the GovOPlaN operation | Provider subscriptions, cursor/checkpoint handling, signature validation, event normalization, and replay protection |
Native collaboration behavior is justified only when GovOPlaN must own the
semantic state, authorization decision, audit evidence, retention/legal-hold
rule, or configuration-package fragment. Endpoint profiles, tokens, health,
protocol clients, provider IDs, retries, and webhook transport remain in
Connectors (or the owning protocol connector). A feature module consumes a
Core capability/DTO and must still start and fail explicitly when that optional
connector is absent; it never imports a provider client.
The minimum hand-off sequences are:
1. **Appointment to conference:** Appointments confirms the booking intent;
Calendar creates or updates the event and invitations; an optional
conference connector provisions the room idempotently and returns an
opaque join reference. Calendar stores that reference with the event, not
the provider credential.
2. **Case or Workflow to collaborative document:** the initiating module asks
DMS for a governed document/session; DMS requests an optional office-suite
connector session and retains version, lock, approval, and callback
evidence. The Case/Workflow keeps only the DMS reference.
3. **Case, Workflow, or Task to chat:** the semantic owner requests a room or
thread with an idempotency key and bounded membership intent. The connector
returns an external reference; capturing messages as evidence requires an
explicit DMS/Records action and policy decision.
4. **Task or Workflow to project suite:** Tasks supplies the task payload and
Workflow supplies correlation; the OpenProject connector publishes or
reconciles the work package and returns versioned external-reference and
retry/conflict evidence. Neither consumer writes connector tables.
Every executable collaboration connector must pass the common connector
contract checks plus a provider-focused minimum proof:
- optional-module startup and partial compositions work without the provider;
- profile health uses secret references and redacts credentials and remote
response bodies;
- tenant/resource authorization is checked before discovery, provisioning,
lookup, synchronization, or evidence capture;
- dry-run/simulation performs no remote mutation and explains unsupported
operations;
- create/publish calls are idempotent, retries preserve the same external
reference, and outcome-unknown or version conflicts remain reconcilable;
- callbacks/webhooks verify authenticity, tenant/profile binding, replay
protection, and bounded payloads;
- disable/retire behavior revokes new use while preserving non-secret audit and
external-reference evidence;
- Collabora/OnlyOffice prove discovery plus one non-production editing-session
round trip; Matrix/Mattermost/Rocket.Chat prove room lookup/create plus one
authenticated bounded event; Jitsi/BigBlueButton prove conference
provision/cancel; OpenProject proves project/work-package lookup, idempotent
publish, and conflict handling.
These are connector acceptance tests, not a claim that those connectors are
already implemented. Their implementation state remains in the owning
connector issues and catalogue.
### Payment And Public Cashier Systems
- Strategy: integrate/export/import; keep the payment provider or cashier as
source of settlement truth.
- Protocol/API surface: payment provider APIs, redirect/callback flows,
reconciliation files, SEPA/export formats, cash-register/cashier interfaces.
- Auth model: provider API keys, signed webhooks, client certificates, mTLS,
tenant-specific merchant accounts.
- Data shape: payment intent, amount/currency, payer reference, provider
transaction ID, settlement status, receipt, refund/cancellation reference.
- Deployment assumptions: public callback URLs, strict environment separation,
PCI-sensitive providers, finance reconciliation cadence.
- Risks: duplicate charges, callback replay, amount mismatch, refund workflow
gaps, evidence-retention mistakes.
- MVP test path: sandbox payment intent, signed callback verification, receipt
evidence link, reconciliation dry-run.
- Owner/priority: `govoplan-payments`, `govoplan-ledger`, Wave 1/6.
### Reporting, BI, Open Data, And Publication
- Strategy: consume/publish/transform; native reporting owns GovOPlaN views, not
every external BI product.
- Protocol/API surface: SQL read replicas, CSV/Excel export/import, REST APIs,
RSS/Atom, open-data APIs, SFTP/WebDAV publication targets.
- Auth model: read-only DB users, API tokens, SFTP keys, OAuth clients, public
anonymous publication profiles where appropriate.
- Data shape: dataset metadata, schema/version, report parameters, generated
file references, publication URL, freshness/lineage, validation results.
- Deployment assumptions: publication can be public or internal; generated
datasets need retention and provenance.
- Risks: leaking restricted data, stale publications, schema drift, expensive
queries, untraceable manual transformations.
- MVP test path: publish one report/export as a governed file plus RSS/Atom
entry with checksum, timestamp, and permission check.
- Owner/priority: `govoplan-reporting`, `govoplan-connectors`,
`govoplan-datasources`, and `govoplan-dataflow`, Wave 2. Reporting owns
presentation/publication, Connectors owns transport, Datasources owns the
governed source/materialization catalogue, and Dataflow owns transformations.
### Public-Sector Protocols And Registries
- Strategy: integrate/publish/receive; protocol modules own protocol semantics.
- Protocol/API surface: FIT-Connect, XTA/OSCI, XRepository, XOE/V, XRechnung,
XBestellung, Peppol, registry-specific Fachverfahren APIs.
- Auth model: certificates, mTLS, service accounts, destination credentials,
protocol-specific trust anchors and key rotation.
- Data shape: transport envelope, payload schema/version, destination IDs,
receipt/acknowledgement, message status, standard-specific metadata.
- Deployment assumptions: regulated trust chains, test/prod endpoint
separation, formal onboarding, strict logging and retention expectations.
- Risks: invalid schemas, failed delivery receipts, certificate expiry, wrong
destination routing, protocol version drift.
- MVP test path: validate a sample payload against a schema, test endpoint
reachability in sandbox, store receipt/evidence reference.
- Owner/priority: `govoplan-fit-connect`, `govoplan-xoev`,
`govoplan-xrechnung`, `govoplan-xta-osci`, `govoplan-connectors`, Wave 1/2.
### File Providers And Shared Storage
- Strategy: integrate/import/link; files module owns GovOPlaN file semantics.
- Protocol/API surface: SMB/CIFS, WebDAV, Nextcloud, Seafile, SFTP, S3,
local/object storage profiles.
- Auth model: service accounts, user credentials, app tokens, OAuth where
supported, secret references, environment variables for deployment-managed
credentials.
- Data shape: file ID/path, provider object ID, checksum, MIME type, size,
version/ETag, owner, permission snapshot, imported file reference.
- Deployment assumptions: internal networks, large files, existing shares,
variable permissions, provider-specific rate limits.
- Risks: permission mismatch, stale imports, overwrites, duplicate files, path
traversal, storage growth.
- MVP test path: profile test, list folder, import one file into governed
storage, keep provider reference and checksum.
- Owner/priority: `govoplan-files`, `govoplan-connectors`, Wave 0/1.
### Project, Task, And Case-Adjacent Systems
- Strategy: connector-first for OpenProject/Jira/Redmine; native module only
when GovOPlaN owns project semantics.
- Protocol/API surface: OpenProject API v3, webhooks, Jira/Redmine REST APIs,
Microsoft Graph for Planner/Project where applicable.
- Auth model: API tokens, OAuth/OIDC apps, webhook secrets, service accounts.
- Data shape: project ID, work package/task ID, status, assignee reference,
external URL, version/lock token, publish/sync trace.
- Deployment assumptions: external project tool remains source of truth for
broad project management; GovOPlaN links selected records.
- Risks: task duplication, bidirectional sync conflicts, permission mismatch,
over-mirroring comments/attachments.
- MVP test path: OpenProject profile test, project/work-package lookup,
external-reference round-trip.
- Owner/priority: `govoplan-connectors`, later `govoplan-tasks`/workflow/cases
consumers, Wave 0/2.
### Specialist Fachverfahren
- Strategy: link first; integrate/import only when a deployment project supplies
concrete contracts and a domain owner.
- Protocol/API surface: vendor APIs, CSV/XML batch imports, SFTP, database
views, message queues, protocol-specific transports.
- Auth model: usually service accounts, VPN, mTLS, SFTP keys, or vendor tokens.
- Data shape: domain-specific record IDs, status, applicant/person references,
file/evidence references, case/status events.
- Deployment assumptions: strongly local/vendor-specific, often no stable test
API, data model differs by jurisdiction.
- Risks: brittle vendor contracts, legal source-of-truth ambiguity, high
customization cost, migration expectations.
- MVP test path: inventory entry and manual external-reference link; require a
project-specific connector issue before automation.
- Owner/priority: domain module or deployment-specific connector, case by case.
## Prioritization Rules
1. Start with connectors that unblock Wave 0 or Wave 1 reference journeys.
2. Prefer open standards and self-hosted/open-source APIs where they are common
in public-sector deployments.
3. Treat inventory-only entries as useful because operators need a map of their
software landscape even before automation exists.
4. Keep connector code in the owning connector/protocol module. Domain modules
consume capabilities, DTOs, external references, and events through core.
5. Every executable connector needs health diagnostics, secret-reference
handling, lifecycle state, audit events, and retirement behavior.
## Connector Catalogue Handoff
`govoplan-connectors` owns the detailed catalogue entry shape:
- connector type key
- category and owner module
- supported directions and trigger modes
- credential and secret handling
- health check and diagnostics payload
- external-reference shape
- required capabilities and optional module combinations
- lifecycle support
Core should only keep strategy, routing, and cross-module architecture notes.
Connector implementation and public-sector target inventory belong in
`govoplan-connectors`.
-291
View File
@@ -1,291 +0,0 @@
# Multi Seal Mail - Current RBAC and Resource-Access Model
**Updated:** 2026-06-16
**Current migration head:** `f5a6b7c8d9e0`
## Authorization Equation
An operation is permitted only when every applicable layer allows it:
```text
effective role/API-key capability
AND resource ownership/share access
AND workflow state
AND active governance/policy constraints
```
RBAC answers what an actor may do. ACLs answer which resource the actor may do it to. Workflow state and policy decide whether the operation is currently valid.
## Identity and Scope
```text
Account global login identity
+- User membership tenant-local identity
+- direct tenant roles
+- active group memberships
| +- inherited tenant roles
+- tenant-local API keys
Account
+- direct system-role assignments
```
A browser session has one active tenant membership. System privileges do not silently grant tenant data access. API keys remain tenant-local and receive the intersection of their configured scopes and their owner's live tenant scopes on every request.
## Wildcards
```text
tenant:* every canonical tenant permission
system:* every canonical system permission
* legacy alias interpreted as tenant:* only
```
Tenant wildcards never grant system permissions.
## Canonical Tenant Permissions - 53
### Campaigns
```text
campaign:read
campaign:create
campaign:update
campaign:copy
campaign:archive
campaign:delete
campaign:share
campaign:validate
campaign:build
campaign:review
campaign:send_test
campaign:queue
campaign:control
campaign:send
campaign:retry
campaign:reconcile
```
### Recipients
```text
recipients:read
recipients:write
recipients:import
recipients:export
```
### Files
```text
files:read
files:download
files:upload
files:organize
files:share
files:delete
files:admin
```
### Reports and Audit
```text
reports:read
reports:export
reports:send
audit:read
```
### Mail Servers
```text
mail_servers:read
mail_servers:use
mail_servers:test
mail_servers:write
mail_servers:manage_credentials
```
### Tenant Administration
```text
admin:users:read
admin:users:create
admin:users:update
admin:users:suspend
admin:groups:read
admin:groups:write
admin:groups:manage_members
admin:roles:read
admin:roles:write
admin:roles:assign
admin:api_keys:read
admin:api_keys:create
admin:api_keys:revoke
admin:settings:read
admin:settings:write
admin:policies:read
admin:policies:write
```
## Canonical System Permissions - 18
```text
system:tenants:read
system:tenants:create
system:tenants:update
system:tenants:suspend
system:accounts:read
system:accounts:create
system:accounts:update
system:accounts:suspend
system:roles:read
system:roles:write
system:roles:assign
system:access:read
system:access:assign
system:audit:read
system:settings:read
system:settings:write
system:governance:read
system:governance:write
```
`system:access:*` remains as a compatibility/read and assignment boundary for cross-tenant/system access handling. It is not a separate primary UI area.
## Default Tenant Roles
- **Owner:** `tenant:*`. At least one active operational owner must remain.
- **Tenant administrator:** settings, policies, users, groups, roles and API keys plus read access to campaigns/files/reports/audit. Real delivery remains separately delegable.
- **Administrator (legacy):** all tenant permissions for upgraded installations.
- **Access administrator:** membership and assignment management within delegation limits.
- **Campaign manager:** prepare, validate and build campaigns; no review approval or real delivery by default.
- **Reviewer:** inspect and approve prepared campaign messages.
- **Sender:** mock-test, queue, control, send, retry and reconcile prepared campaigns; can use/test approved mail profiles.
- **File manager:** managed file operations without campaign delivery rights.
- **Viewer:** read campaigns, recipients, files and reports.
- **Auditor:** read campaigns, recipient evidence, reports and audit records; export detailed evidence.
## Default System Roles
- **System owner:** `system:*`, protected. At least one active account must retain it.
- **System administrator:** all specific system permissions, editable and not protected.
- **System auditor:** read-only system registry/settings/governance/audit role, editable.
## Delegation Ceiling
For role definition, assignment and API-key creation:
```text
requested scopes subset of actor delegateable scopes
```
Rules:
1. Tenant roles may contain tenant scopes only.
2. System roles may contain system scopes only.
3. Definition rights and assignment rights are separate.
4. Group definition and group membership management are separate.
5. API-key scopes are intersected with the owner's current effective scopes on every request.
6. Suspended accounts, users, tenants or groups stop contributing access immediately.
7. Administrative updates are field-sensitive; a user with only status authority cannot change role assignments.
## Campaign Ownership and ACLs
A campaign has exactly one owner:
```text
owner user OR owner group
```
Additional active shares may target users or groups with `read` or `write`.
Resolution:
- owner user: read and write;
- member of owner group: read and write;
- explicit read share: read;
- explicit write share: read and write;
- `tenant:*`: tenant-wide ACL bypass;
- ordinary campaign permission without ownership/share: no object access.
ACLs do not add capabilities. A write share still needs the specific permission for update, validation, review, send, report, retry or reconciliation.
## Sensitive Recipient Boundary
Recipient-complete campaign JSON, message data and job detail require `recipients:read`. Recipient edits require `recipients:write`; exports require `recipients:export`; import is reserved for the dedicated recipient import/list workflow.
## Files
| Permission | Operations |
|---|---|
| `files:read` | list, search, inspect, resolve metadata |
| `files:download` | download file bytes and generated ZIP archives |
| `files:upload` | upload files and ZIP contents |
| `files:organize` | create folders, rename, move, copy and bulk rename |
| `files:share` | create/revoke file shares |
| `files:delete` | delete/hide files and folders subject to retention |
| `files:admin` | tenant-wide administration of user/group file spaces |
## Mail Servers
| Permission | Boundary |
|---|---|
| `mail_servers:read` | profile metadata and effective policy visibility |
| `mail_servers:use` | select an approved profile without reading secrets |
| `mail_servers:test` | run server-side connection tests |
| `mail_servers:write` | define/edit profiles in allowed scopes |
| `mail_servers:manage_credentials` | create/replace SMTP/IMAP secrets or campaign-level credentials where policy allows |
Reusable encrypted profiles now exist. Effective usability is also constrained by hierarchical mail-profile policy, ownership, allowed/forced profile sets, credential inheritance mode, the lower-level override switch for that mode, and allow/deny patterns.
## Sessions, API Keys and CSRF
- Browser login creates an HttpOnly session cookie and a separate readable CSRF cookie.
- Unsafe cookie-authenticated requests require matching CSRF cookie/header and stored CSRF hash.
- API keys remain supported for CLI/automation and do not use browser CSRF.
- Login responses still expose a compatibility session token in the response body; the WebUI does not persist it.
## Legacy Compatibility
Runtime aliases remain only for names that are no longer canonical, including:
```text
campaign:write
attachments:read
attachments:write
admin:users
admin:users:write
admin:api_keys:write
admin:settings
system:tenants:write
system:access:write
```
Canonical scopes are not widened by runtime alias expansion after migration.
## Deferred Permission Families
Add these only with their corresponding implemented features:
```text
templates:*
address_books:*
recipient_lists:*
connectors:*
dsar:*
system:monitoring:read
system:backups:run
system:backups:restore
system:updates:apply
system:updates:rollback
```
A separate `retention:*` family is not currently canonical because retention is managed through system settings and tenant policy scopes. Add it only if retention operation duties need separation from general policy/settings administration.
+94
View File
@@ -0,0 +1,94 @@
# Records Filing Contract
Core exposes a small provider-neutral contract for filing exact source
revisions into an institutional record. Core does not own records semantics,
source-object authorization, or source bytes. `govoplan-records` owns filing
orchestration and chronology; each source module owns resolution of its exact
revision.
## Capability Names
- `records.filing` is supplied by the enabled Records module.
- `records.source.<module>` is supplied by an enabled source module, for
example `records.source.files` or `records.source.cases`.
- `records.archive.<provider>` is supplied by an enabled archive-transfer
adapter. Discovery does not imply conformance or current health.
Callers discover capabilities through the module registry. They must not
import optional source-module internals.
## Exact Source Identity
`RecordSourceLocator` identifies one tenant, source module, resource type,
resource ID, and immutable source revision. A source provider must:
1. reject cross-tenant resolution;
2. require a non-empty purpose;
3. re-evaluate the caller's current module and object authorization;
4. resolve exactly the requested revision, never a mutable "current" alias;
5. return safe display/provenance metadata and a SHA-256 digest when the source
has stable bytes or a canonical snapshot;
6. fail closed when the revision is missing, quarantined, corrupt, or no longer
authorized.
Historical Records browsing never revives historical access rights. The
source's current authorization decision remains authoritative when filing.
## Filing Semantics
`RecordFilingRequest` binds the exact source to a record, purpose, filing
reason, relationship, institutional context, and idempotency key. Records must
persist source identity and resolution evidence together with the filing actor,
represented capacity, valid time, recorded time, and immutable chronology.
An idempotency key may replay only an identical request. A conflicting reuse
must fail. Filing does not transfer ownership of source content and must not
silently copy mutable source state.
## Versioning
The Python DTOs and protocols live in `govoplan_core.core.records`. The
manifest interface `records.filing` starts at `1.0.0`. Incompatible DTO or
behavior changes require a new interface version and release impact analysis;
additional optional metadata remains backward compatible.
## Initial Providers
- Files resolves an exact managed `FileVersion`, verifies current Files access
and blob integrity, and returns its stored content digest.
- Cases resolves an exact immutable case revision after current case access and
returns a digest of the canonical revision snapshot.
Provider-specific selection UI belongs to the source module. The generic
Records dialog remains a diagnostic/manual fallback for exact identifiers.
## Archive Transfer Boundary
`RecordTransferPackage` binds a stable package ID, record revision, provider
profile, canonical manifest, and manifest SHA-256. An archive provider exposes
`RecordArchiveProviderState` before dispatch and accepts only a
`RecordArchiveTransferRequest` for a declared healthy profile. Its receipt must
identify the same package and provider and return one bounded outcome:
`accepted`, `rejected`, or `outcome_unknown`.
An unknown outcome is never retry-safe. Callers must retain the intent and
reconcile it against the provider before another effect. Provider state also
declares authority mode, freshness, limitations, and whether the provider is a
simulation. Credentials, transport configuration, archive-specific package
schemas, and custody semantics remain provider-owned.
Records includes `records.archive.simulation` to prove package and receipt
handling. The simulation is explicitly non-conformant, transfers no custody,
and cannot be used as evidence of an archive handoff. A real provider requires
a selected target/profile, provider-specific recovery declaration, and target
test evidence.
## Form Evidence Boundary
Form attachments use the separate provider-neutral contract in
`govoplan_core.core.form_evidence`. Forms Runtime requests short-lived,
purpose-bound upload grants and re-inspects the exact provider-owned evidence
before final submission. The provider keeps byte storage, quarantine,
classification, and retention ownership; Forms Runtime stores only immutable
evidence references and bounded verification results. This contract is not an
alternative path for Records filing or archive custody.
+916 -28
View File
@@ -1,71 +1,959 @@
# GovOPlaN Release Dependencies
Release installs must not depend on sibling checkout paths. Local development can keep editable installs and `file:` WebUI links, but release packaging should resolve modules from tagged git refs or from a package registry.
This document owns release package composition, signed package catalogs,
license checks, catalog publishing, migration baselines, and the final release
checklist.
## Backend
Operator runtime configuration and module install/uninstall execution live in
`DEPLOYMENT_OPERATOR_GUIDE.md`.
## Backend Packages
Release installs must not depend on sibling checkout paths. Local development
can keep editable installs and `file:` WebUI links, but release packaging must
resolve modules from tagged git refs or from a package registry.
Local development:
```bash
cd /mnt/DATA/git/govoplan-core
cd /mnt/DATA/git/govoplan
./.venv/bin/python -m pip install -r requirements-dev.txt
```
Release install from a core checkout plus tagged module repositories:
Release install from the meta checkout plus tagged module repositories:
```bash
cd /mnt/DATA/git/govoplan-core
cd /mnt/DATA/git/govoplan
./.venv/bin/python -m pip install -r requirements-release.txt
```
`.[server]` is resolved relative to the current working directory. If you create the virtualenv elsewhere, still run the install command from the core checkout:
`../govoplan-core[server]` is resolved relative to the meta requirements file.
If you create the virtualenv elsewhere, still run the install command from the
meta checkout:
```bash
cd /mnt/DATA/git/govoplan-core
cd /mnt/DATA/git/govoplan
/tmp/govoplan-release-test/bin/python -m pip install -r requirements-release.txt
```
`requirements-release.txt` pins the module repositories to the release tag. Update those refs when cutting a release:
`requirements-release.txt` pins the module repositories to the release tag.
Update those refs when cutting a release:
```text
govoplan-files git@git.add-ideas.de:add-ideas/govoplan-files.git v0.1.1
govoplan-mail git@git.add-ideas.de:add-ideas/govoplan-mail.git v0.1.1
govoplan-campaign git@git.add-ideas.de:add-ideas/govoplan-campaign.git v0.1.1
govoplan-tenancy git@git.add-ideas.de:GovOPlaN/govoplan-tenancy.git v0.1.8
govoplan-organizations git@git.add-ideas.de:GovOPlaN/govoplan-organizations.git v0.1.8
govoplan-identity git@git.add-ideas.de:GovOPlaN/govoplan-identity.git v0.1.8
govoplan-idm git@git.add-ideas.de:GovOPlaN/govoplan-idm.git v0.1.8
govoplan-access git@git.add-ideas.de:GovOPlaN/govoplan-access.git v0.1.8
govoplan-admin git@git.add-ideas.de:GovOPlaN/govoplan-admin.git v0.1.8
govoplan-policy git@git.add-ideas.de:GovOPlaN/govoplan-policy.git v0.1.8
govoplan-audit git@git.add-ideas.de:GovOPlaN/govoplan-audit.git v0.1.8
govoplan-dashboard git@git.add-ideas.de:GovOPlaN/govoplan-dashboard.git v0.1.8
govoplan-addresses git@git.add-ideas.de:GovOPlaN/govoplan-addresses.git v0.1.8
govoplan-files git@git.add-ideas.de:GovOPlaN/govoplan-files.git v0.1.8
govoplan-mail git@git.add-ideas.de:GovOPlaN/govoplan-mail.git v0.1.8
govoplan-campaign git@git.add-ideas.de:GovOPlaN/govoplan-campaign.git v0.1.8
govoplan-calendar git@git.add-ideas.de:GovOPlaN/govoplan-calendar.git v0.1.8
govoplan-poll git@git.add-ideas.de:GovOPlaN/govoplan-poll.git v0.1.8
govoplan-scheduling git@git.add-ideas.de:GovOPlaN/govoplan-scheduling.git v0.1.8
govoplan-notifications git@git.add-ideas.de:GovOPlaN/govoplan-notifications.git v0.1.8
govoplan-evaluation git@git.add-ideas.de:GovOPlaN/govoplan-evaluation.git v0.1.8
govoplan-docs git@git.add-ideas.de:GovOPlaN/govoplan-docs.git v0.1.8
govoplan-ops git@git.add-ideas.de:GovOPlaN/govoplan-ops.git v0.1.8
```
## WebUI
## WebUI Packages
Local development uses `webui/package.json`, which may point at sibling module checkouts while active development is happening.
Local development uses `webui/package.json`, which may point at sibling module
checkouts while active development is happening.
Release WebUI installs should use `webui/package.release.json`. It points module dependencies at the same tagged git repositories. After the module tags referenced there exist, generate the committed release lockfile without touching the development package files:
Release WebUI installs should use `webui/package.release.json`. It points
module dependencies at the same tagged git repositories. After the module tags
referenced there exist, generate the committed release lockfile without
touching the development package files:
```bash
cd /mnt/DATA/git/govoplan-core
scripts/generate-release-lock.sh
cd webui
cd /mnt/DATA/git/govoplan
tools/release/generate-release-lock.sh
cd /mnt/DATA/git/govoplan-core/webui
PATH=/home/zemion/.nvm/versions/node/v22.22.3/bin:$PATH /home/zemion/.nvm/versions/node/v22.22.3/bin/npm run build
```
The module repositories include root-level npm package manifests so git installs can resolve `@govoplan/files-webui`, `@govoplan/mail-webui`, and `@govoplan/campaign-webui` from repository roots even though their source lives below `webui/src`.
Module repositories with a frontend include root-level npm package manifests
so the `@govoplan/*-webui` dependencies in `webui/package.release.json` can be
resolved from repository roots even though their source lives below
`webui/src`.
The normal release path is automated by `scripts/push-release-tag.sh`: it bumps or accepts the target version, updates Python/WebUI/module manifest versions, commits/tags/pushes the module repositories first, regenerates `webui/package-lock.release.json`, and then commits/tags/pushes core. If the working tree has already been bumped, pass the current version explicitly:
### Release Lockfile Strategy
The supported backend release composition is the set pinned in the meta
repository's `requirements-release.txt`. The supported frontend composition
is the independently buildable module set pinned in
`webui/package.release.json`; backend-only modules do not need a frontend
package entry. Keep one committed full-product release lockfile at
`webui/package-lock.release.json`, generated from
`webui/package.release.json` in a clean release workspace. Development
`package-lock.json` may continue to point at local `file:` dependencies.
Frontend module permutations are regression-tested through
`GOVOPLAN_WEBUI_MODULE_PACKAGES` and temporary build output, not through
committed lockfiles for every possible combination. If a smaller composition
becomes a separately shipped product, add an explicit release manifest and
lockfile pair for that product, for example
`package.release.files-mail.json` and `package-lock.release.files-mail.json`,
generated in a clean release workspace from tagged git dependencies.
## Release Tag Script
The normal release path is automated by `/mnt/DATA/git/govoplan/tools/release/push-release-tag.sh`: it bumps
or accepts the target version, updates Python/WebUI/module manifest versions,
commits/tags/pushes the module repositories first, regenerates
`webui/package-lock.release.json`, and then commits/tags/pushes core. If the
working tree has already been bumped, pass the current version explicitly:
```bash
cd /mnt/DATA/git/govoplan-core
scripts/push-release-tag.sh --version 0.1.2
cd /mnt/DATA/git/govoplan
tools/release/push-release-tag.sh --version 0.1.8
```
### Release lockfile strategy
`/mnt/DATA/git/govoplan/tools/release/generate-release-catalog.py` reads installed/discovered
`ModuleManifest` objects while writing catalog entries. When a manifest is
available, the catalog entry uses the manifest version, points package refs at
`v<manifest.version>`, and copies `provides_interfaces` /
`requires_interfaces` from the manifest. It also copies module migration order
and `migration_tasks` metadata when present. If a manifest cannot be
discovered, the entry falls back to the release version passed with `--version`
and omits interface and migration-task metadata. This keeps the catalog aligned
with independently versioned module packages instead of relying on a hardcoded
compatibility table.
The supported release composition currently is the full Multi Seal Mail product: core plus files, mail, and campaign. Keep one committed full-product release lockfile at `webui/package-lock.release.json`, generated from `webui/package.release.json` in a clean release workspace. Development `package-lock.json` may continue to point at local `file:` dependencies.
The script also includes GovOPlaN roadmap/scaffold module repositories that do
not yet have package metadata. Those repositories are committed, tagged, and
pushed with the same release tag, but they are tag-only until they contain
`pyproject.toml`, module manifests, or WebUI packages. Tag-only repositories
are not listed in `requirements-release.txt` or `webui/package.release.json`.
Frontend module permutations are regression-tested through `GOVOPLAN_WEBUI_MODULE_PACKAGES` and temporary build output, not through committed lockfiles for every possible combination. If a smaller composition becomes a separately shipped product, add an explicit release manifest and lockfile pair for that product, for example `package.release.files-mail.json` and `package-lock.release.files-mail.json`, generated in a clean release workspace from tagged git dependencies.
Current tag-only module repositories:
- `govoplan-appointments`
- `govoplan-cases`
- `govoplan-connectors`
- `govoplan-dms`
- `govoplan-erp`
- `govoplan-fit-connect`
- `govoplan-forms`
- `govoplan-identity-trust`
- `govoplan-ledger`
- `govoplan-payments`
- `govoplan-portal`
- `govoplan-postbox`
- `govoplan-reporting`
- `govoplan-search`
- `govoplan-tasks`
- `govoplan-templates`
- `govoplan-workflow-engine` (headless definitions, migrations, and execution)
- `govoplan-workflow` (optional authoring and inspection WebUI)
- `govoplan-xoev`
- `govoplan-xrechnung`
- `govoplan-xta-osci`
## Catalog Trust And Licensing
GovOPlaN module install and uninstall must remain operator-controlled. The
running server may plan and validate package changes, but package mutation is
performed by the separate installer daemon or an operator shell during
maintenance mode.
`addideas-govoplan-website` is the public static distribution surface for
official catalog resources:
- signed module package catalogs, grouped by release channel
- public catalog keyrings
- public license verification keyrings
- examples and operator-facing download paths
`govoplan-core` is the verifier and orchestrator:
- fetches a local or remote module catalog
- verifies catalog signatures against configured trusted keys
- enforces approved release channels
- rejects expired or not-yet-valid catalogs
- records accepted catalog sequence numbers for replay protection
- checks catalog entry license feature requirements before planning installs
- writes installer plans and request records
Feature and platform modules own their package artifacts, manifests, migration
metadata, retirement providers, and optional lifecycle behavior.
Core accepts either a local catalog file or a remote URL:
```bash
GOVOPLAN_MODULE_PACKAGE_CATALOG=/srv/govoplan/catalogs/stable.json
GOVOPLAN_MODULE_PACKAGE_CATALOG_URL=https://govoplan.example/catalogs/v1/channels/stable.json
GOVOPLAN_MODULE_PACKAGE_CATALOG_CACHE=/srv/govoplan/runtime/catalog-cache/stable.json
```
If both file and URL are set, the URL wins. The cache is used when a remote
fetch fails, so an operator can still inspect the last known catalog. A cached
catalog must still pass signature, freshness, channel, and replay validation.
If neither source is configured, the Admin package directory discovers the
official public stable catalog at
`https://govoplan.add-ideas.de/catalogs/v1/channels/stable.json`. Core verifies
that fallback against the public key pinned in the installed Core package. An
explicit deployment catalog always takes precedence; a configured source that
is unavailable or invalid fails closed instead of silently falling back.
An official catalog is a JSON object with:
- `catalog_version`
- `channel`
- `sequence`
- `generated_at`
- `not_before` when delayed activation is needed
- `expires_at`
- `modules`
- `signatures`
Each module entry can declare:
- backend package name and pinned install reference
- WebUI package name and pinned install reference
- `artifact_integrity` for each package, including the HTTPS registry URL,
filename, byte size, SHA-256, package identity, source tag, and source commit
- `source`, binding the repository and immutable tag/commit identity, with
optional HTTPS repository and revision links
- `availability`, either `available` or `withdrawn`; a withdrawn entry must
carry an operator-readable `availability_reason` and cannot be planned
- `configuration_requirements` and an optional HTTPS `release_notes_url` for
prerequisites and release-specific operator guidance
- display metadata and tags
- `license_features`, the feature entitlements required to plan that install
- `dependencies` and `optional_dependencies`, the module ids expected in the
target module set
- `migration_safety`, one of `automatic`, `requires_review`, `forward_only`,
or `destructive`
- `migration_notes`, operator-facing data/migration guidance for review,
forward-only, or destructive changes
- `migration_after` and `migration_before`, explicit module ids used to order
module-owned migration heads when a release needs a live-data sequencing rule
- `migration_tasks`, constrained live-data tasks that run around Alembic
migration phases. Each task declares `task_id`, `phase`, `summary`,
`task_version`, `safety`, `idempotent`, and optionally `timeout_seconds`.
The allowed phases are `pre_migration_check`, `pre_migration_prepare`,
`post_migration_backfill`, and `post_migration_verify`.
- `current_version_min` and `current_version_max_exclusive`, the installed
version window from which this catalog target may be applied directly
- `bridge_release` and `bridge_notes`, marking a target as an intermediate
compatibility release in a staged update path
- `allow_downgrade` and `allow_same_version`, explicit opt-ins for reviewed
rollback or package-refresh plans
- `recovery_tested` and `recovery_notes`, documenting the rehearsal for
forward-only or destructive data changes
- `provides_interfaces`, named interface contracts exported by this module
- `requires_interfaces`, named interface contracts and version ranges required
by this module
Core validates these fields before exposing the directory. Admin derives a
read-only catalog state from the installed package set, catalog dependency
closure, named-interface providers, current-version window, availability, and
generic license policy. This is an early operator diagnostic; trusted installer
preflight remains the authoritative mutation gate.
The signature is Ed25519 over canonical JSON with both `signature` and
`signatures` removed. Core accepts the legacy single `signature` field and the
new `signatures` array.
When `APP_ENV` is `prod` or `production`, module package catalog signature
verification is required by default unless
`GOVOPLAN_MODULE_PACKAGE_CATALOG_REQUIRE_SIGNATURE=false` is set explicitly.
### Independent Module Versions And Interface Ranges
Modules do not need to ship on the same version number. Release catalogs should
pin each package to the exact backend/WebUI ref being installed and declare any
cross-module API compatibility through named interfaces.
Provider shape:
```json
"provides_interfaces": [
{ "name": "files.campaign_attachments", "version": "1.4.0" }
]
```
Requirement shape:
```json
"requires_interfaces": [
{
"name": "files.campaign_attachments",
"version_min": "1.0.0",
"version_max_exclusive": "2.0.0"
}
]
```
`version_min` is inclusive. `version_max_exclusive` is exclusive, so the range
above means `>= 1.0.0` and `< 2.0.0`. Use this for SemVer major-version
compatibility lines. Set `"optional": true` only when the module can operate
without that interface being present. If a provider is installed but its
version is outside the declared optional range, activation is still blocked.
Catalog validation normalizes these fields and warns when catalog entries do
not satisfy each other's ranges. Registry activation and installer preflight
perform the blocking checks against the discovered installed manifests before
the desired module set is activated.
The admin module-management UI shows catalog warnings in the package catalog
section and repeats them as warning-level installer preflight issues while a
package install is planned.
Install-plan items carry a `source` field. Manually entered items use
`source: "manual"`; entries planned from the package catalog use
`source: "catalog"`. Catalog-sourced items also carry a `catalog` metadata
object with the validation snapshot used when the item was planned: catalog
source/path, source type, cache path, channel, sequence, generated/validity
timestamps, signature state, trusted key id, and cache state where available.
Catalog provenance changes preflight severity:
- catalog-sourced installs and updates require a configured, valid package
catalog before activation
- invalid, untrusted, expired, not-yet-valid, replayed, or unapproved-channel
catalogs block catalog-sourced installs and updates
- the same catalog validation failures remain warnings for manual install
plans, so operators can still use offline or emergency package refs
- valid-catalog warnings, such as intentionally unsigned local catalogs when
signature enforcement is disabled, remain warnings
- a saved catalog plan must match the currently validated entry exactly;
altered package refs, artifact identities, channel, sequence, trust state, or
signing-key identity block the run and require replanning
- a trusted remote artifact is downloaded before mutation into a private
SHA-256-addressed installer cache, checked for exact size and digest, and
passed to `pip` or npm only as that verified local file
- selected catalog entries with unsatisfied non-optional named interface ranges
block activation before the installer runs
- selected catalog entries whose target dependencies are neither installed nor
planned block activation before the installer runs
- catalog update targets older than the installed module version block unless
the catalog entry declares `allow_downgrade: true`
- catalog update targets equal to the installed module version block unless the
catalog entry declares `allow_same_version: true`
- catalog update targets with a `current_version_min` /
`current_version_max_exclusive` window block when the installed version is
outside that window; publish and apply a bridge release instead
- catalog entries marked `forward_only` or `destructive` block activation until
the plan row has an explicit data-safety acknowledgement
- catalog entries marked `forward_only` or `destructive` also block unless the
catalog entry declares `recovery_tested: true` and either the catalog entry or
operator plan row contains recovery notes
- catalog entries marked `destructive` also require catalog migration notes or
operator notes describing the cleanup or retirement plan
### Update Paths
Package updates are target-state operations, not one-module-at-a-time runtime
toggles. The safe unit of planning is a desired module version set plus a
catalog validation snapshot. The installer may install multiple packages into
the environment before activation, then validate the discovered manifests and
activate the resulting set together.
Install-plan rows support explicit `install`, `update`, and `uninstall`
actions. Catalog planning writes `update` when the module is already installed.
Preflight resolves the target set from installed manifests plus the planned
catalog entries. Unplanned catalog entries are not treated as installed. When a
catalog entry would satisfy a missing dependency or named interface, preflight
blocks activation with a companion-update issue; the admin catalog planner adds
those companion rows automatically when it can resolve them from the current
catalog. The preflight response also includes a structured `target_plan` summary
with each planned module's action, current version, catalog target version,
package refs, migration-safety level, current-version update window, bridge
metadata, recovery metadata, and acknowledgement state.
Database migrations are planned against that same target module set. When the
installer is run with `--migrate`, it calls `govoplan_core.commands.init_db`
with the target enabled modules rather than the pre-update startup module list,
so newly installed module migration directories are discovered before
activation. Preflight also returns a structured migration plan. Its step order is
derived from:
- manifest and catalog `migration_after` / `migration_before` declarations
- module dependencies and optional dependencies when both modules are in the
target plan
- named interface provider/consumer relationships when both sides are in the
target plan
Preflight blocks cycles in that ordering graph. It also blocks non-idempotent
module migration tasks, forward-only/destructive tasks without operator
acknowledgement, and installed manifest tasks that declare no executor.
Catalog-only task executors are marked as pending because they can only be
confirmed after the target package is installed. The migrator runs pre-migration
tasks, upgrades the ordered module heads first, finishes with Alembic `heads`,
and then runs post-migration tasks, so Alembic's revision graph remains
authoritative while GovOPlaN still gives operators a module-aware live-data
order.
This avoids circular "upgrade A first / upgrade B first" traps: named interface
requirements are solved against the target set, not against each intermediate
package-install moment. If the target set cannot satisfy all non-optional
interfaces and module dependencies at once, the plan is invalid. Operators
should add the necessary module updates to the same plan instead of trying to
force an order.
Live data upgrades need an even stricter rule:
- migrations must be idempotent and ordered by module migration metadata
- destructive schema/data changes need an explicit retirement or cleanup plan,
not an automatic package update side effect
- cross-module data migrations must be compatible with both the old and target
provider interface until activation finishes
- rollback must restore the package set and database state together, or be
documented as forward-only with a tested recovery procedure
- if two modules require mutually incompatible live-data states, the catalog
must publish an intermediate compatibility release rather than a circular
update chain
The release catalog is the first safety gate for this. Generated catalogs mark
modules with registered migrations as `requires_review` by default. Release
authors should keep that value for ordinary reversible migrations, raise it to
`forward_only` when database rollback requires restoring a snapshot, and raise
it to `destructive` when the update removes or irreversibly rewrites persisted
data. Forward-only and destructive entries must include `recovery_tested: true`
and recovery notes after a verified restore or forward-recovery rehearsal. The
admin install-plan UI exposes the safety level and lets operators record an
explicit acknowledgement; preflight keeps acknowledged forward-only/destructive
changes visible as warnings.
In practice, circular dependencies are avoided by designing interfaces with
compatibility windows and by publishing bridge releases. A bridge release keeps
the old interface while introducing the new one, allowing dependent modules to
move first; a later release can retire the old interface after every dependent
module has a compatible target version. Use `current_version_min` and
`current_version_max_exclusive` to make those direct-update windows explicit in
the catalog, and set `bridge_release: true` on intermediate targets that exist
primarily to carry installations safely across a compatibility gap.
Trusted catalog keys are configured locally:
```bash
GOVOPLAN_MODULE_PACKAGE_CATALOG_TRUSTED_KEYS_FILE=/srv/govoplan/trust/catalog-keyring.json
GOVOPLAN_MODULE_PACKAGE_CATALOG_TRUSTED_KEYS='{"release-key-1":"<base64 public key>"}'
```
For development or tightly controlled deployments, a keyring can be read from a
URL and cached:
```bash
GOVOPLAN_MODULE_PACKAGE_CATALOG_TRUSTED_KEYS_URL=https://govoplan.example/catalogs/v1/keyring.json
GOVOPLAN_MODULE_PACKAGE_CATALOG_TRUSTED_KEYS_CACHE=/srv/govoplan/runtime/catalog-cache/keyring.json
```
Production installations should pin the trusted keyring locally or ship it
through deployment configuration. Fetching trusted keys from the same public
origin as the catalog is convenient, but that origin must not become the only
trust root.
## Dependency Audits
Dependency vulnerability checks are documented in
[`DEPENDENCY_AUDITS.md`](DEPENDENCY_AUDITS.md). The local audit runner is:
```bash
cd /mnt/DATA/git/govoplan
bash tools/checks/check-dependency-audits.sh
```
The Gitea workflow in `govoplan/.gitea/workflows/dependency-audit.yml` runs the same
check against release dependency refs on pushes, pull requests, and a weekly
schedule.
Keyring entries support:
- `key_id`
- `public_key` or `public_key_base64`
- `status`: `active`, `next`, `retired`, `revoked`, or `disabled`
- `not_before`
- `not_after`
Rotation process:
1. Add the next public key to the local trusted keyring with status `next`.
2. Publish catalogs signed by both current and next keys.
3. Upgrade installations so the next key is locally trusted.
4. Promote the next key to `active`.
5. Retire the old key only after every supported installation trusts the new
key.
6. Mark a compromised key `revoked` and publish a higher sequence catalog
signed by an uncompromised key.
Use replay state in production:
```bash
GOVOPLAN_MODULE_PACKAGE_CATALOG_SEQUENCE_STATE=/srv/govoplan/runtime/catalog-sequences.json
GOVOPLAN_MODULE_PACKAGE_CATALOG_ENFORCE_SEQUENCE=true
```
Core records the accepted sequence per channel after a catalog entry is planned
from the admin interface. With strict sequence enforcement, a previously
accepted sequence is rejected; without strict enforcement, only older sequences
are rejected. Catalogs should always expire.
The sequence state file is operational state, not a trust root. Keep it on
persistent storage and include it in normal backups:
```json
{
"channels": {
"stable": {
"last_sequence": 42,
"accepted_at": "2026-07-07T12:00:00Z",
"key_id": "release-key-1",
"source": "https://govoplan.example/catalogs/v1/channels/stable.json"
}
}
}
```
If the file is lost, restore it from backup. If no backup exists, reconstruct
each channel from the highest sequence already accepted in installer run
records, release records, or the currently deployed module package set. Do not
lower `last_sequence` to make an older catalog pass; publish a new higher
sequence catalog when the accepted point is uncertain.
If the file is corrupted, copy it aside for incident review, validate the
current signed catalog with channel and freshness enforcement, then rewrite the
state with the known accepted sequence. Keep
`GOVOPLAN_MODULE_PACKAGE_CATALOG_REQUIRE_SIGNATURE=true` and approved-channel
checks enabled during recovery. Temporarily disabling
`GOVOPLAN_MODULE_PACKAGE_CATALOG_ENFORCE_SEQUENCE` allows revalidating the same
sequence, but older sequences remain rejected once the reconstructed
`last_sequence` is in place.
Approved channels are deployment policy:
```bash
GOVOPLAN_MODULE_PACKAGE_CATALOG_APPROVED_CHANNELS=stable,lts
GOVOPLAN_MODULE_PACKAGE_CATALOG_REQUIRE_SIGNATURE=true
```
The admin UI can display other catalog metadata, but core rejects catalogs from
unapproved channels when validation is configured.
Catalog entries can require license features:
```json
"license_features": ["module.mail", "support.standard"]
```
Core checks those requirements against an offline license file before allowing
the entry into the install plan.
Official open-source GovOPlaN entries do not declare license features. The
license contract remains generic for external catalogs, deployment presets,
configuration/package directories, and support offerings; it gates only an
entry that explicitly asks for a feature.
```bash
GOVOPLAN_LICENSE_FILE=/srv/govoplan/license.json
GOVOPLAN_LICENSE_ENFORCEMENT=true
GOVOPLAN_LICENSE_TRUSTED_KEYS_FILE=/srv/govoplan/trust/license-keyring.json
```
License files are JSON objects with:
- `license_id`
- `subject`
- `features`
- `valid_from`
- `valid_until`
- `signature`
Issue or renew a license from an operator/release shell that has the Ed25519
private key:
```bash
govoplan-module-installer \
--issue-license /srv/govoplan/license.json \
--license-id customer-2026-07 \
--license-subject "Example Municipality" \
--license-feature module.mail \
--license-feature support.standard \
--license-valid-until 2027-07-31T23:59:59Z \
--license-signing-key-id license-issuer-1 \
--license-signing-private-key /srv/govoplan/secrets/license-issuer-1.pem \
--format json
```
Validate an imported license without exposing secrets:
```bash
govoplan-module-installer \
--validate-license /srv/govoplan/license.json \
--license-trusted-key license-issuer-1="<base64 public key>" \
--require-trusted-license \
--license-required-feature module.mail \
--format json
```
The CLI and admin module catalog panel report the license id, subject,
validity window, signing key id, signed/trusted state, available features, and
missing entitlements for the configured package catalog. They do not expose
private signing material.
License enforcement can run in observe-only mode by leaving
`GOVOPLAN_LICENSE_ENFORCEMENT` unset. In that mode, missing or invalid license
data is surfaced as a warning but does not block planning.
Renewal is an ordinary re-issuance with a new `license_id`, extended
`valid_until`, and the full intended feature set. Import the renewed JSON to
`GOVOPLAN_LICENSE_FILE`, keep the previous file for audit, and validate it
before setting enforcement.
Revocation is handled through the trusted license keyring. Mark a compromised
or invalid issuer key as `revoked` or `disabled`, publish or deploy the updated
keyring, then reissue affected licenses with an active key. Installations that
run with `GOVOPLAN_LICENSE_ENFORCEMENT=true` reject licenses signed only by a
revoked key after the local keyring is updated.
Emergency fallback is deliberately explicit. Operators can temporarily unset
`GOVOPLAN_LICENSE_ENFORCEMENT` to keep package planning observable while a
license or keyring is recovered. Record the change in the operational incident
log, keep catalog signature and channel enforcement enabled, and restore
license enforcement after a trusted renewal validates successfully.
Licensing is intentionally separate from open-source code licensing. The
catalog/license mechanism can govern support channels, official release
eligibility, hosted update access, professional support, or commercial
entitlements without changing the source license of the repositories.
Production-grade distribution still needs remote registry/git artifact
resolution before package-manager apply, a hardened catalog publishing pipeline
that writes to `addideas-govoplan-website`, and automated key rotation and
emergency revocation drills.
## Release Catalog Publishing
GovOPlaN release catalogs are published to `addideas-govoplan-website` as
static JSON and verified by `govoplan-core` before installer plans are
accepted. Private signing keys must stay outside all git repositories. Public
keyrings are published with the website.
Create the first catalog signing key on the release machine:
```bash
cd /mnt/DATA/git/govoplan
KEY_DIR="$HOME/.config/govoplan/release-keys"
mkdir -p "$KEY_DIR"
./.venv/bin/python tools/release/generate-catalog-keypair.py \
--key-id release-key-1 \
--private-key "$KEY_DIR/release-key-1.pem" \
--public-key "$KEY_DIR/release-key-1.pub" \
--keyring "$KEY_DIR/catalog-keyring.json"
```
Keep `release-key-1.pem` private. The generated keyring contains only public
material.
Generate the signed catalog into `addideas-govoplan-website`:
```bash
cd /mnt/DATA/git/govoplan
KEY_DIR="$HOME/.config/govoplan/release-keys"
tools/release/publish-release-catalog.sh \
--version <x.y.z> \
--sequence 202607071340 \
--catalog-signing-key "release-key-1=$KEY_DIR/release-key-1.pem" \
--build-web
```
This writes:
- `/mnt/DATA/git/addideas-govoplan-website/public/catalogs/v1/channels/stable.json`
- `/mnt/DATA/git/addideas-govoplan-website/public/catalogs/v1/keyring.json`
The wrapper validates the catalog with core using the generated public keyring.
For normal module/core releases, first audit and record migration baselines,
then tag and push the module/core repos. Finally publish the website catalog:
```bash
cd /mnt/DATA/git/govoplan
./.venv/bin/python tools/release/release-migration-audit.py --strict
```
```bash
cd /mnt/DATA/git/govoplan
KEY_DIR="$HOME/.config/govoplan/release-keys"
tools/release/publish-release-catalog.sh \
--version <x.y.z> \
--catalog-signing-key "release-key-1=$KEY_DIR/release-key-1.pem" \
--build-web \
--commit \
--tag \
--push
```
The website tag is `catalog-v<x.y.z>`. The public URL is:
```text
https://govoplan.add-ideas.de/catalogs/v1/channels/stable.json
```
The public keyring URL is:
```text
https://govoplan.add-ideas.de/catalogs/v1/keyring.json
```
`/mnt/DATA/git/govoplan/tools/release/push-release-tag.sh` can publish the web catalog after module and core
tags have been pushed. It runs the migration release audit in automatic mode:
warning-only before the first recorded migration baseline, strict after a
baseline exists. Add `--strict-migration-audit` when you want to force strict
mode explicitly:
```bash
cd /mnt/DATA/git/govoplan
KEY_DIR="$HOME/.config/govoplan/release-keys"
tools/release/push-release-tag.sh \
--bump subversion \
--strict-migration-audit \
--publish-web-catalog \
--catalog-signing-key "release-key-1=$KEY_DIR/release-key-1.pem" \
--build-web-catalog
```
Use `--catalog-signing-key` more than once during a key rotation window. The
catalog will contain multiple signatures and the public keyring will include the
corresponding public keys.
## Release Doctor
Use `/mnt/DATA/git/govoplan/tools/release/release-doctor.py` before release preparation and again before
publishing. It inspects repository state, local versions, release tags,
migration audit state, release package refs, stable catalog/keyring state, and
optionally public catalog availability. The default mode is read-only and
offline. Status output is intentionally concise:
```bash
cd /mnt/DATA/git/govoplan
./.venv/bin/python tools/release/release-doctor.py status --target-version <x.y.z>
```
Use `--details` when the full findings should be printed directly:
```bash
./.venv/bin/python tools/release/release-doctor.py status --target-version <x.y.z> --details
```
For concise guidance, print only suggested next commands:
```bash
./.venv/bin/python tools/release/release-doctor.py next --target-version <x.y.z>
```
For an operator-guided session, run interactive mode. The doctor asks which
suggested command to run; mutating commands require typing `RUN` before they
execute:
```bash
./.venv/bin/python tools/release/release-doctor.py interactive --target-version <x.y.z>
```
Interactive mode starts with a compact summary and waits for an action. It can
open the full report in the configured pager, run one suggested command, repair
Git `safe.directory` dubious-ownership blocks after explicit confirmation, push
all clean repositories that are ahead of their upstream, or commit and push all
dirty repositories. The dirty-repository bulk action asks for a commit message,
skips repositories that are behind upstream, and requires typing
`COMMIT AND PUSH` before it stages, commits, and pushes.
Add `--online` when remote tag and public catalog/keyring reachability should be
checked. Add `--json` when a CI job or another tool should consume the report.
On a GovOPlaN installation that should consume the official stable catalog:
```bash
GOVOPLAN_MODULE_PACKAGE_CATALOG_URL=https://govoplan.add-ideas.de/catalogs/v1/channels/stable.json
GOVOPLAN_MODULE_PACKAGE_CATALOG_CACHE=/srv/govoplan/runtime/catalog-cache/stable.json
GOVOPLAN_MODULE_PACKAGE_CATALOG_REQUIRE_SIGNATURE=true
GOVOPLAN_MODULE_PACKAGE_CATALOG_APPROVED_CHANNELS=stable
GOVOPLAN_MODULE_PACKAGE_CATALOG_TRUSTED_KEYS_FILE=/srv/govoplan/trust/catalog-keyring.json
GOVOPLAN_MODULE_PACKAGE_CATALOG_SEQUENCE_STATE=/srv/govoplan/runtime/catalog-sequences.json
GOVOPLAN_MODULE_PACKAGE_CATALOG_ENFORCE_SEQUENCE=true
```
For production, copy the public keyring into deployment configuration and pin it
locally. Do not rely on a URL-fetched keyring as the only trust root.
`stable.json` includes a top-level `core_release` section for operator/update
tooling. Core is intentionally not listed as a normal module entry because it
must not be added to saved enabled-module state. Core upgrades should remain an
operator-supervised package update with restart and health checks.
Key rotation for published catalogs:
1. Generate the next private key outside git.
2. Run `/mnt/DATA/git/govoplan/tools/release/publish-release-catalog.sh` with both signing keys.
3. Publish the web catalog/keyring.
4. Roll the new public keyring into installations.
5. Stop signing with the old key after the supported fleet trusts the new key.
6. Mark compromised keys as revoked in the public keyring and publish a higher
sequence catalog signed by a trusted uncompromised key.
## PostgreSQL Release Check
Release candidates should pass a disposable PostgreSQL migration and startup
smoke check before tagging or publishing catalogs. Start the local testbed,
then run the permutation check from the core checkout:
```bash
cd /mnt/DATA/git/govoplan/dev/postgres
cp .env.example .env
docker compose --env-file .env up -d
cd /mnt/DATA/git/govoplan
set -a
. /mnt/DATA/git/govoplan/dev/postgres/.env
set +a
tools/checks/postgres-integration-check.py \
--database-url "$GOVOPLAN_POSTGRES_DATABASE_URL" \
--reset-schema
```
The script checks migrations and `/health` startup for core-only, files-only,
mail-only, campaign-only, campaign+files, campaign+mail, and full-product
module sets. `--reset-schema` is destructive and must only be used against a
throwaway database. Before those permutations, the required Core proof runs in
random, test-owned schemas without modifying `public`. It exercises Files' real
credential-owning retirement provider and proves that credential scrubbing,
non-secret audit
insertion, and table retirement commit together; database-injected audit and
DDL failures roll the entire unit back. A 500 ms PostgreSQL `lock_timeout` and
captured backend process IDs also prove that each `DROP TABLE` uses the
installer Session connection instead of waiting through a second connection.
The meta check enables the release-gate flag so missing PostgreSQL configuration
or full-stack test packages are a hard failure; ordinary Core-only test discovery
skips this integration proof. Do not pass `--skip-retirement-atomicity` when
collecting release evidence.
## Migration Baselines
Release migration support follows `COMPATIBILITY_POLICY.md`: every tagged
`0.1.x` installation is a supported upgrade origin, released revision IDs are
immutable, and migration-only reconciliation remains available for at least one
subsequent major release cycle after the matching runtime compatibility path is
removed.
Development migrations may be small and numerous while a feature is moving.
GovOPlaN keeps those detailed migrations on an explicit development track and
publishes reviewed release shortcuts on the release track. Before a stable
release, unreleased development migrations may be squashed into a release-level
baseline or release-to-release upgrade migration. After a release tag has
shipped, released migration revision IDs are immutable.
The release policy is:
- unreleased development migrations live in `dev_versions` and are not deleted
when a release shortcut is added;
- released migrations live in `versions` and are never rewritten or deleted;
- future release shortcuts are additive. A new release-to-release migration
starts from the previous recorded release heads, so installations can upgrade
through sane release steps instead of replaying every development revision;
- each stable release records the public migration head revisions in
`docs/migration-release-baselines.json`;
- `heads` records the Alembic dependency-leaf graph heads for a full graph,
while `owner_heads` records the latest migration revision per migration
owner for subset installs and review;
- fresh installations should apply release-level baselines/upgrades, not
unreleased create-then-rename churn;
- release-to-release schema changes should be folded into one reviewed
migration per migration owner where practical.
The default runtime track is `release`:
```bash
cd /mnt/DATA/git/govoplan
GOVOPLAN_MIGRATION_TRACK=release ./.venv/bin/python -m govoplan_core.commands.init_db
```
Use `dev` only for local/disposable development databases that intentionally
need the detailed chain:
```bash
cd /mnt/DATA/git/govoplan
GOVOPLAN_MIGRATION_TRACK=dev ./.venv/bin/python -m govoplan_core.commands.init_db
```
Production, release checks, operator install flows, and the normal development
launcher should stay on the `release` track unless a fresh/disposable database
is intentionally being used to test the detailed development chain.
Audit the current graph during release preparation:
```bash
cd /mnt/DATA/git/govoplan
./.venv/bin/python tools/release/release-migration-audit.py
```
Audit the detailed development graph separately when a squash is prepared:
```bash
./.venv/bin/python tools/release/release-migration-audit.py --track dev
```
Generate the reviewed/manual squash checklist:
```bash
./.venv/bin/python tools/release/release-migration-audit.py --squash-plan
```
After the release migrations have been reviewed and the graph is final, record
the release baseline:
```bash
./.venv/bin/python tools/release/release-migration-audit.py --record-release <x.y.z>
```
Use strict mode to verify that the current heads are recorded:
```bash
./.venv/bin/python tools/release/release-migration-audit.py --strict
```
`/mnt/DATA/git/govoplan/tools/release/push-release-tag.sh` runs the audit by default in automatic mode:
non-strict while no release baseline exists, strict after the first baseline is
recorded. Pass `--warn-migration-audit` for an explicit non-strict audit,
`--strict-migration-audit` to force strict mode, or `--skip-migration-audit`
only for emergency/manual release work.
The first public baseline is v0.1.7. It intentionally adds release-track
shortcuts for the unreleased v0.0.0 -> v0.1.7 development chains while keeping
the detailed chains on the `dev` track. No production installations existed
before that baseline, so pre-v0.1.7 development revisions are not release
upgrade targets. Future release-to-release changes must start from a recorded
release baseline and add a new release-track step-up instead of replacing prior
release shortcuts. The tracking issue is
`GovOPlaN/govoplan-core#223`.
## Related Operator Documents
- `DEPLOYMENT_OPERATOR_GUIDE.md`: runtime environment, explicit migrations,
backup/restore commands, module installer daemon/supervisor operation, and
rollback drills.
- `REMOTE_WEBUI_BUNDLES.md`: experimental browser-loaded module bundles for
controlled deployments; normal releases use package builds.
## Release Checklist
- Keep Python package versions, WebUI package versions, and git tags aligned.
- Tag core, files, mail, and campaign repositories together.
- Update `requirements-release.txt` and `webui/package.release.json` when the release tag changes.
- Generate the committed full-product release lockfile from `package.release.json` with `scripts/generate-release-lock.sh`.
- Add separate release manifest/lockfile pairs only for module compositions that are shipped as their own products.
- Tag core, access, admin, tenancy, policy, audit, files, mail, campaign,
calendar, and scaffold module repositories together.
- Update meta `requirements-release.txt` and core `webui/package.release.json` when the
release tag changes.
- Generate the committed full-product release lockfile from
`package.release.json` with `/mnt/DATA/git/govoplan/tools/release/generate-release-lock.sh`.
- Run `/mnt/DATA/git/govoplan/tools/release/release-migration-audit.py --strict` after recording a release
baseline.
- Run the PostgreSQL release check against a disposable database.
- Publish the signed catalog through the release catalog publishing flow above.
- Add separate release manifest/lockfile pairs only for module compositions
that are shipped as their own products.
- Do not commit local sibling paths into release manifests.
+161
View File
@@ -0,0 +1,161 @@
# Remote WebUI Bundle Loading
GovOPlaN WebUI modules normally ship through the core WebUI package graph:
install a tagged npm/git dependency, run `npm install`, rebuild the shell, and
restart or reload the served assets. Remote WebUI bundles are an experimental
future path for controlled deployments where a backend-enabled module is not in
the local WebUI package graph and the operator wants the shell to load its
frontend without rebuilding core.
This design defines the target guardrails. The current rebuild/reload path
remains the default production path.
## Current Rebuild Path
The supported release path is:
1. Install or remove backend and WebUI package dependencies through the trusted
installer CLI/daemon.
2. Snapshot package state before mutation.
3. Run `npm install` and optionally `npm run build`.
4. Restart or reload the served WebUI assets.
5. Use installer rollback if package apply, restart, or health checks fail.
Strengths:
- package manager and lockfile semantics stay conventional
- CSP can stay strict because all code is served as built assets
- rollback restores package files and lockfiles
- local development uses sibling workspace dependencies naturally
Costs:
- frontend changes require a rebuild/reload
- hot enabling a module with frontend code is not possible in the running shell
- failed rebuilds happen at install time, not at lazy module-load time
## Remote Bundle Path
Remote loading is only for modules that are enabled by the backend and absent
from `virtual:govoplan-installed-modules`. The backend manifest exposes:
- `FrontendModule.asset_manifest`
- `asset_manifest_integrity`
- `asset_manifest_signature`
- `asset_manifest_public_key_id`
- `asset_manifest_contract_version`
The WebUI shell fetches the asset manifest, verifies manifest integrity and/or
signature, validates the manifest contract, fetches the entry bundle, verifies
the entry integrity, imports the bundle, validates the exported
`PlatformWebModule`, and applies backend metadata before registering routes,
navigation, and UI capabilities.
Unsigned and unhashed manifests are skipped. Entries without integrity are
skipped. A module id mismatch is skipped.
## Asset Manifest Contract
Contract version `1`:
```json
{
"contractVersion": "1",
"moduleId": "files",
"entry": "./files-webui.remote.js",
"entryIntegrity": "SHA-256-<base64-or-hex-digest>",
"moduleExport": "default"
}
```
The backend manifest carries the manifest-level trust metadata. The remote
asset manifest carries the concrete entry URL and entry digest.
## Compatibility Checks
Before a remote bundle is accepted:
- backend module must be enabled
- local WebUI module with the same id must be absent
- manifest contract must be supported by the shell
- manifest `moduleId`, exported module `id`, and backend module id must match
- backend metadata remains authoritative for label, version, dependencies,
nav, and runtime UI capability exposure
- future contract versions must declare required shell capabilities so old
shells fail closed
Remote bundles must not broaden backend permissions. They can only contribute
routes/nav/capabilities that the backend metadata and existing permission
checks allow.
## CSP
The current implementation imports verified entry bytes through a blob URL. A
production CSP for this mode must explicitly allow:
- `connect-src` for the approved asset-manifest and bundle origins
- `script-src` for `blob:` only when remote loading is enabled
- no `unsafe-inline` requirement for remote modules
If an installation cannot allow `blob:` scripts, the follow-up implementation
should use signed same-origin module assets with static URLs instead of blob
imports. Remote loading must be disableable by configuration so strict-CSP
deployments can keep the rebuild path only.
## Cache Invalidation
The shell cache key is:
```text
<module-id>:<asset-manifest-url>:<backend-module-version>
```
Release catalogs should publish immutable manifest and entry URLs or change at
least one cache-key component on every release. Emergency rollback can point the
backend manifest at a previous immutable manifest URL or lower the enabled
module version after the backend package rollback.
Browsers may still cache remote responses. Approved asset servers should use:
- long cache lifetimes only for content-addressed immutable assets
- short cache lifetimes or explicit revalidation for channel/latest manifest
aliases
- `Cache-Control: no-store` for emergency override manifests
## Rollback
Remote WebUI rollback is metadata rollback, not package-manager rollback:
- if the backend package install rolls back, the previous backend frontend
metadata returns
- if only remote assets are bad, publish a new manifest URL or revert the
backend/frontend metadata to the last known good manifest
- failed remote loading must degrade by omitting the remote module frontend,
not by breaking the shell
Installer run records should eventually include remote asset manifest URLs,
integrity, signature key ids, and load-test results when a plan enables a
remote frontend.
## Local And Development Behavior
Local development should keep using workspace/file dependencies and Vite. Remote
loading is useful for integration testing release artifacts, not for day-to-day
module UI development.
Development deployments may use unsigned catalogs and local asset servers only
when signature/integrity enforcement is intentionally disabled. The remote
loader itself still requires an integrity hash or a verifiable signature.
## Follow-Up Slices
1. Add a server-side remote-bundle policy flag and surface whether remote
loading is enabled in platform metadata.
2. Add CSP documentation/config generation for strict rebuild-only mode versus
controlled remote-bundle mode.
3. Add an installer/catalog preflight that validates remote WebUI asset
manifests and records verified identity in installer run records.
4. Add Playwright coverage for a signed test remote bundle, failed digest, bad
module id, cache-key refresh, and fallback when the module is unavailable.
5. Add release tooling to emit immutable remote asset manifests with digest,
signature, key id, and rollback metadata.
+26
View File
@@ -0,0 +1,26 @@
# Search event indexing contract
Core defines, but does not implement, the optional Search indexing boundary.
Feature modules register `SearchSourceProvider` implementations for bounded
backfills and live authorization checks. A provider may additionally implement
`SearchEventSourceProvider` to translate a committed `PlatformEvent` into one
or more authoritative `SearchIndexChange` values.
When the Search index-writer capability is active, the platform event worker
uses the durable consumer identity `search.indexing.v1`. It accepts only public
and internal events, passes the outbox delivery key to each event-capable
source, and then advances a bounded batch of queued index changes in the same
worker transaction. Stable change IDs make delivery replay idempotent.
The boundary has three non-negotiable rules:
- a source may emit changes only for its registered module, provider, resource
type, and event tenant;
- Search validates every upsert document before queueing it and rejects secret
metadata keys;
- an index ACL is only a candidate filter. Resources marked for authorization
recheck are returned only after the owning source explicitly allows the
current principal at query time.
Search and its worker remain optional. Core-only startup and feature-module
operation do not require the Search package.
+15
View File
@@ -0,0 +1,15 @@
# Security Audit Toolchain
The shared GovOPlaN security audit toolbox moved to the meta repository.
Use:
```bash
cd /mnt/DATA/git/govoplan
tools/checks/security-audit/run.sh --mode ci --scope govoplan
tools/checks/security-audit/run.sh --mode full --scope govoplan
```
Canonical documentation:
- `/mnt/DATA/git/govoplan/docs/operations/SECURITY_AUDIT.md`
+86
View File
@@ -0,0 +1,86 @@
# Self-Hosted Installability
GovOPlaN uses a staged self-hosted installability path.
## Packaging Decision
The early packaging approach is a staged combination:
1. Generate an explicit environment template and validate it before startup.
2. Use a Compose-backed production-like development profile for local rehearsal.
3. Use the deployment operator guide as the runbook for migrations, workers,
backups, health checks, and module installer rollback drills.
4. Use the module installer CLI/daemon for package mutation once the runtime is
already installed and under maintenance mode.
This keeps first installation understandable while still preserving the later
goal of install/update/uninstall through signed catalogs and the installer
daemon. The API server must not run package managers from request handlers.
## Config Bootstrap
Generate a self-hosted template:
```bash
cd /mnt/DATA/git/govoplan
./.venv/bin/python -m govoplan_core.commands.config env-template \
--profile self-hosted \
--generate-secrets \
--output .env.self-hosted
```
Validate the current shell environment:
```bash
set -a
. .env.self-hosted
set +a
./.venv/bin/python -m govoplan_core.commands.config validate --profile self-hosted
```
The command reports all known blockers at once. Production-like/self-hosted
profiles require explicit `APP_ENV`, `DATABASE_URL`, `MASTER_KEY_B64`,
`ENABLED_MODULES`, `CORS_ORIGINS`, `GOVOPLAN_TRUSTED_HOSTS`, and a deployment-wide decision for
`GOVOPLAN_CONNECTOR_ALLOW_PRIVATE_NETWORKS`. Production rejects SQLite, development
bootstrap, insecure auth cookies, and unsigned catalog trust roots when a
catalog source is configured.
Connector process-secret names and custom CA files are deployment-owned through
the exact, default-empty `GOVOPLAN_CONNECTOR_SECRET_ENV_ALLOWLIST` and
`GOVOPLAN_CONNECTOR_CA_BUNDLE_ALLOWLIST`; tenant/API configuration cannot widen
either boundary.
## Production-Like Dev Stack
Use the local production-like wrapper for repeatable rehearsal:
```bash
cd /mnt/DATA/git/govoplan
tools/launch/production-like-dev.sh validate-config
tools/launch/production-like-dev.sh seed
tools/launch/production-like-dev.sh start
```
Stop Docker dependencies:
```bash
tools/launch/production-like-dev.sh stop
```
Reset all profile data:
```bash
tools/launch/production-like-dev.sh reset --yes
```
The start command delegates to `tools/launch/launch-production-like-dev.sh`, which
runs API, worker, and WebUI in the foreground. Stop those processes with
`Ctrl+C` in the launcher terminal.
## Module Boundary Gate
`govoplan/tools/checks/check_dependency_boundaries.py` is part of the focused verification
path. It checks backend imports and WebUI package/source imports so modules do
not grow hidden runtime dependencies on each other. Feature modules should
integrate through core capabilities, backend APIs/events, route contributions,
or explicit UI extension points.
+118
View File
@@ -0,0 +1,118 @@
# Semantic Documentation Subjects
## Purpose And Ownership
The semantic-documentation subject contract lets an optional module expose the
configured artifacts that administrators may document: for example a form, a
form field, a workflow, or a workflow state. It is a discovery and resolution
contract, not a second configuration API.
The module that owns an artifact also owns its subject provider, authorization,
identity, revision, route, and lifecycle semantics. Docs may discover those
providers through Core and attach authored documentation to their stable
references. Docs must not import the feature module, read its tables, or copy
configuration content into a generic index.
This contract is additive to manifest `DocumentationTopic` contributions and
configured-state `documentation_providers`. Every providing module must retain
static user and administrator documentation baselines. The baselines explain
the feature even when the provider is disabled, unavailable, or has no
configured subjects.
## Identity And Versioning
`SemanticDocumentationSubjectReference` identifies a subject with:
- owning module and tenant;
- a module-defined subject kind and stable identifier;
- an optional typed nested anchor, such as `field/registration-number`;
- the revision and canonical fingerprint observed when documentation was
authored or reviewed.
The `stable_key` derives only from identity. A rename or configuration revision
therefore does not detach existing documentation. A nested anchor has its own
identity so a field can be documented independently from its form.
Providers must resolve an old reference as one of:
- `available`: the observed revision/fingerprint is still current;
- `changed`: the same stable subject has changed and may need review;
- `superseded`: another stable reference replaced it;
- `missing`: the subject was removed or is no longer resolvable;
- `temporarily_unavailable`: the provider cannot currently determine state.
Absence is not authorization. A provider returns `None` when the principal may
not learn whether a subject exists. Core also rejects cross-tenant list and
resolution requests before calling a provider.
## Safe Projection
Descriptors contain only bounded, explicit presentation fields: localized
labels and descriptions, breadcrumbs, a local route, audience,
classification, and required scopes. They must not contain credentials,
personal data, arbitrary provider metadata, configuration payloads, or the
authored documentation itself. Routes are application-local and are still
subject to normal route authorization.
The fingerprint is a review signal, not a concurrency token or a content hash
that callers may use to reconstruct configuration. Providers should calculate
it from the smallest canonical JSON projection whose semantic changes require
documentation review. Volatile timestamps and secrets must be excluded.
## Provider Registration
A provider is registered under its exact module-scoped capability name:
```python
from govoplan_core.core.modules import CapabilityDocumentation
from govoplan_core.core.semantic_documentation import (
SEMANTIC_DOCUMENTATION_SUBJECT_CONTRACT_VERSION,
semantic_documentation_subject_capability,
)
capability = semantic_documentation_subject_capability("forms")
manifest = ModuleManifest(
id="forms",
# ...
capability_factories={capability: build_semantic_subject_provider},
capability_documentation={
capability: CapabilityDocumentation(
label="Form semantic subjects",
summary="Lists authorized configured forms and fields for Docs.",
contract_version=SEMANTIC_DOCUMENTATION_SUBJECT_CONTRACT_VERSION,
documentation_types=("admin", "user"),
)
},
documentation=(admin_baseline, user_baseline),
)
```
The capability is `documentation.semantic_subjects.<module_id>`. Registry
validation rejects a mismatched owner, missing capability documentation, a
wrong contract version, or missing static baselines.
`list_semantic_documentation_subjects` performs authorized, paginated discovery
across installed providers. `resolve_semantic_documentation_subject` targets
one owner without loading another feature module. Providers must apply the
current tenant and principal on every call and must not infer visibility from a
previous list result.
## Lifecycle And Integration Rules
- Keep subject and anchor identifiers stable across display-name and route
changes.
- Return `superseded` only with the replacement reference; do not silently
rewrite stored references.
- Return a reason code for missing or temporarily unavailable subjects without
exposing sensitive detail.
- Reauthorize both discovery and resolution. Stored documentation references
confer no access to a live artifact.
- Treat a changed fingerprint as a request for editorial review. It does not
automatically invalidate or publish authored documentation.
- Removing a feature module leaves references resolvable as provider
unavailable. Docs can preserve history without importing the module.
Forms, Workflow, and later modules should implement their subject providers in
their own repositories. Docs owns the authored semantic-documentation records,
review workflow, and projection UI.
+162
View File
@@ -0,0 +1,162 @@
# State And Recovery Contract
## State Profiles
Core accepts three runtime state profiles:
| Profile | Runtime placement | Durable storage |
| --- | --- | --- |
| `local` | One development process set | Local filesystem is permitted. |
| `host-shared` | Replicas on one host | One host-shared volume plus PostgreSQL and Redis. |
| `shared` | Independent hosts | PostgreSQL, Redis, and S3-compatible object storage. |
All replicas in one installation use one stable
`GOVOPLAN_INSTALLATION_ID`, one immutable module composition, and identical
database, broker, encryption-key, and object-storage bindings. Core rejects
replicas with the `local` profile and rejects `shared` without PostgreSQL,
Redis, S3, and a non-default installation identifier.
`FILE_STORAGE_S3_ENDPOINT_TRUSTED=true` is an explicit deployment trust
declaration for a clean HTTPS S3 origin. It does not authorize a
user-controlled connector endpoint and it is separate from installer-managed
Garage's exact endpoint trust.
## Object Storage
`govoplan_core.core.object_storage` is the shared backend contract for durable
module artifacts. It provides bounded read/write/list/stat/delete operations
for local and S3-compatible storage. Modules own their object-key namespace and
business metadata; Core does not interpret module files.
`stat` and `list_objects` return object size plus a UTC `modified_at` value when
the backend can prove it. Reconciliation and retention code may use that value
for conservative grace periods, but must treat a missing timestamp as
ineligible for automatic deletion rather than guessing an age.
Rules for modules:
- Store only opaque object keys in business records, never local absolute
paths.
- Use node-local directories only for temporary materialization.
- Verify expected size and digest before consuming consequential artifacts.
- If object creation precedes database commit, compensate successfully created
objects on failure.
- If object deletion fails, retain the database reference and report a retryable
failure rather than claiming deletion.
- Define an orphan-inventory strategy for hard process loss between object
creation and metadata commit.
The Files module delegates its backend implementation to this Core contract.
Campaign generated EML artifacts use a Campaign-owned object prefix and are
read by workers through the same shared backend.
## Runtime Nodes And Leases
API and worker incarnations register in `core_runtime_nodes` with role,
software version, module-composition hash, queues, start time, and heartbeat.
The registration identity includes a process incarnation so a stale process
cannot update a replacement's row.
Worker metadata also records the orchestrator pool and declared concurrency.
Every Celery prefork child disposes the SQLAlchemy pool inherited from its
parent and creates a process-local pool before handling work. Deployment
rendering must therefore budget one database pool for the worker parent and
each child. Ops compares active queue ownership, software versions, and the
order-independent module-composition hash with the graph loaded by the API.
Drain is durable operator intent:
- an API enters not-ready state after observing drain;
- a worker cancels queue consumers after observing drain;
- cancellation returns an eligible draining node to active state;
- clean shutdown marks the matching incarnation stopped.
Coordination loss also fails closed. An API reports
`coordination_unavailable` from `/health/ready` until a heartbeat succeeds
again. A worker cancels its local queue consumers on any heartbeat or database
failure and only resumes them after its existing incarnation heartbeats
successfully. It never re-registers from the heartbeat path, so a stale worker
cannot reclaim a node identity from its replacement.
`core_distributed_leases` provides installation/resource uniqueness, expiry,
holder incarnation, and monotonically increasing fencing tokens. Lease expiry
does not make an old process harmless by itself: code performing an effect must
assert its exact fence before commit. `govoplan_core.commands.fenced_run`
renews a lease around a subprocess and terminates the child when the lease is
lost. The deployment profiles use it for the singleton scheduler.
## Migration Ordering
Only `govoplan_core.commands.init_db` mutates schema. On PostgreSQL it acquires
a deterministic installation/track advisory lock before pre-migration tasks,
Alembic, and post-migration tasks. The lock is session-scoped and therefore
released if the migration process dies.
Runtime roles run `govoplan_core.commands.wait_for_database`. It waits until
the database has exactly the configured, dependency-resolved Core/module
Alembic heads and never upgrades schema. Cross-module `depends_on` revisions
therefore do not leave runtime roles waiting for a branch marker Alembic has
correctly consumed. This permits a migration Job and runtime Deployments to be
submitted together while keeping startup fail-closed.
Runtime coordination records the installed `govoplan-core` distribution
version for API, worker and scheduler roles. FastAPI/OpenAPI metadata versions
are presentation metadata and must not be used as deployable software identity;
mixing the two would create a false version-skew readiness failure.
## Recovery Ledger
`govoplan_core.core.recovery` provides a durable operation and evidence
contract. Recovery modes are:
- `atomic`: one database transaction, no external effect;
- `compensation`: explicit inverse actions;
- `snapshot_restore`: separately verified backup reference;
- `forward_recovery`: repair/resume the current version;
- `irreversible`: explicit approval, no automated recovery claim.
Every plan requires verification steps. Mode-specific evidence is mandatory.
Operations bind an idempotency key to a canonical request hash, may bind a
runtime fencing token, and emit append-only hash-chained checkpoints. Backup
and approval references are part of the hashed plan evidence. Every low-level
state transition and checkpoint append revalidates the operation's recorded
fence while holding the operation row lock. Plaintext secrets are rejected
from metadata and evidence.
The state machine makes partial and uncertain outcomes visible. A non-atomic
running operation cannot transition directly to ordinary failure, and success
or recovery requires explicit verified checks. Ops projects states requiring
attention, but module behavior gains this guarantee only after it adopts the
ledger around its own side effects.
## Recovery Boundary
Application/configuration rollback and database rollback are not equivalent.
Once an incompatible migration starts, old code may be unsafe even if its image
is available. Deployment automation must switch to forward recovery unless a
coordinated and verified database/object/key backup is restored.
Core does not create production database backups. The deployment owner must
provide backup, retention, encryption, restore verification, and recovery-point
coordination for PostgreSQL, object storage, and encryption keys. The canonical
operator procedure is documented in
`govoplan/docs/operations/RECOVERY_AND_ROLLBACK_GUARANTEES.md`.
## Verification
Focused contracts are covered by:
```sh
/mnt/DATA/git/govoplan/.venv/bin/python -m pytest -q \
tests/test_object_storage.py \
tests/test_runtime_coordination.py \
tests/test_runtime_agents.py \
tests/test_fenced_run.py \
tests/test_migration_lock.py \
tests/test_wait_for_database.py \
tests/test_recovery_guarantees.py
```
Production acceptance additionally requires multi-node failure and coordinated
restore drills against the actual PostgreSQL, Redis, object-store, ingress, and
secret-provider topology.
-179
View File
@@ -1,179 +0,0 @@
# Multi Seal Mail - Current System and Tenant Governance Model
**Updated:** 2026-06-16
**Current migration head:** `f5a6b7c8d9e0`
## Governance Rule
System policy is authoritative for tenants and all lower levels. Each lower level may only narrow what it inherits:
```text
system
-> tenant
-> user or group owner
-> campaign
```
Lower levels do not widen privileges, allowed profiles, retention durations or credential rights granted by a higher level.
## Administration Structure
```text
SYSTEM
- Settings
- Retention
- Mail servers
- Tenants
- Users
- Groups
- System roles
- Tenant roles
- Audit
TENANT
- Settings boundary
- Users
- Groups
- Roles
- API keys
- Mail servers
- Retention
- Audit
USER
- User mail
- User retention
GROUP
- Group mail
- Group retention
```
There is no separate System access page. Compatibility access scopes remain in the backend for assignment/read boundaries.
## Tenant Governance
System settings define tenant defaults and whether tenants may narrow selected options. Tenant overrides can only restrict:
- custom groups;
- custom roles;
- tenant API keys.
The backend enforces that tenant governance cannot widen system-denied privileges.
## Mail-Profile Governance
Mail server profiles may exist at these scopes:
```text
system
tenant
user
group
campaign
```
Effective campaign profile availability follows campaign ownership. A campaign owned by a user resolves through system, tenant, that user and campaign policy. A group-owned campaign resolves through system, tenant, that group and campaign policy.
Policy semantics:
- higher levels define the maximum available profile set;
- lower levels can further restrict the set;
- forced profiles mean the lower level must choose from the forced set;
- a forced set with one profile effectively enforces that profile;
- campaign-level profile creation is allowed only if the effective policy permits it;
- SMTP/IMAP credentials use one inheritance decision per protocol: lower levels must inherit profile credentials, may inherit profile credentials, or must provide local credentials;
- the lower-level override switch for `smtp_credentials.inherit` and `imap_credentials.inherit` controls whether descendants may change that inheritance decision;
- deny patterns always win over allow patterns;
- empty or `*` allowlist means allow all except denied;
- non-empty allowlist means at least one allow rule must match and no deny rule may match.
Pattern targets:
```text
SMTP hostname
IMAP hostname
envelope sender
From header
recipient domains
```
Ownership transfer is intentionally deferred as a two-step workflow: original owner initiates, new owner accepts and reselects/repairs the mail profile if their effective policy requires it.
## Retention Governance
Retention policy is hierarchical:
```text
system -> tenant -> user/group -> campaign
```
Managed fields:
- raw campaign JSON retention days;
- generated EML retention days;
- stored report detail retention days;
- mock mailbox retention days;
- audit detail retention days.
Rules:
- system may set concrete defaults or unlimited retention;
- system exposes allow-limiting toggles per field;
- tenants, users/groups and campaigns may only shorten inherited retention where the parent allows limiting;
- blank lower-level values inherit;
- mock mailbox retention is currently system-level because mock mailbox records do not yet carry tenant/campaign ownership metadata;
- dry-run/apply retention actions report affected classes before destructive cleanup.
## Role Definitions and Assignments
### System roles
System roles define instance-wide permissions. `system:*` is stored as one wildcard and displayed as granting the full system catalogue. System owner is protected.
### Tenant roles
Tenant roles can be system-governed templates or tenant-local definitions, subject to system tenant-governance settings and actor delegation ceilings. Wildcard counts are expanded against the canonical tenant catalogue.
## Audit Access
Audit access remains scope-separated:
```text
system audit -> system:audit:read
tenant audit -> active tenant + audit:read
```
Audit pages use server pagination, filtering and bounded grids.
## Tenant Switching
Tenant switching preserves the current URL when possible and falls back when a route/resource is not accessible in the new tenant context.
The tenant selector is hidden for ordinary single-tenant accounts and visible for multi-tenant or system tenant-management contexts.
## DataGrid Contract in Administration
Admin lists use bounded container grids:
- one flexible fill column;
- fixed total table width;
- compact action/status/count columns;
- resizable text/date columns;
- no intrinsic content growth;
- sticky headers where needed;
- server pagination for audit.
## Still Deferred
- real SMTP/IMAP test-bed verification and operator runbook;
- recipient import with column mapping;
- Seafile/external connector governance;
- system/tenant/group/user file-space hierarchy and external storage hierarchy;
- session/device revocation UI;
- backup/restore, monitoring and update procedures;
- DSAR workflows and evidence bundle verifier;
- campaign ownership transfer workflow;
- policy impact analysis before delete/disable/unshare/change;
- LDAP/OIDC/SAML provisioning;
- destructive tenant erasure orchestration.
+21
View File
@@ -0,0 +1,21 @@
# Tabular Source Preview Contract
Core defines provider-neutral DTOs for optional tabular source providers. A
source declares whether it is live, cached, file-backed, or static; its schema
and immutable fingerprint; structured health; and the exact projection,
pagination, filter, aggregation, and sorting operations that the provider can
push down. Consumers must not infer pushdown support from a provider name.
Every preview request carries independent row, byte, and elapsed-time budgets.
A provider may tighten these values but must return its effective limits,
returned byte count, elapsed milliseconds, truncation state, and structured
diagnostics. Equivalent fields on the Datasources read request and result
preserve that evidence when a live source is consumed through the catalogue.
A row that cannot fit within the byte budget fails explicitly rather than
leaking a partial value. Timeout, stale fingerprint, unavailable source, and
authorization failures remain distinct provider-neutral errors.
Connector health and preview diagnostics must contain no credentials, endpoint
userinfo, row values, or unbounded remote error bodies. A Datasource origin
preserves this contract so registration and staging do not erase source mode,
health, pushdown, or preview-limit evidence.
+28
View File
@@ -0,0 +1,28 @@
# Template And Generated Artifact Capability Contracts
Core defines provider-neutral contracts for optional template libraries and
generated artifact storage. Core does not render templates or store generated
files itself.
## Templates
- `templates.catalog` lists typed, versioned template references and checks a
consumer's available fields, usage, and output format.
- `templates.renderer` accepts a `TemplateRenderRequest` containing pinned
input data and returns immutable render evidence plus an artifact reference.
The DTOs contain identifiers, hashes, plain mappings, and scalar metadata. They
do not expose Template ORM models or require Campaign, Distribution Lists,
Addresses, Reporting, Forms, or Mail.
## Generated Artifacts
`files.artifact_store` accepts a `ManagedArtifactWriteRequest` and returns a
`ManagedArtifactRef`. Producers supply bytes, a safe filename/content type,
idempotency key, and non-secret provenance. Files owns path normalization,
authorization, versions, storage, and download behavior.
Consumers must discover both contracts through the module registry and degrade
only the unavailable path. A template renderer may return a bounded download
when Files is absent. A caller must not infer successful external delivery from
successful rendering or artifact persistence.
+68
View File
@@ -0,0 +1,68 @@
# Temporal Data Context
GovOPlaN exposes one read context for data validity and system knowledge. The
calendar control in the authenticated titlebar applies that context to
supported list and detail reads for the current account and tenant.
## Two Independent Axes
- **Valid time** answers when a fact applied in the represented domain.
- **Recorded time** answers what the system had recorded by a particular
instant.
The default is data valid now under the latest recorded state. `At time`
selects a valid-time instant. `All` removes the valid-time interval filter but
still uses the selected recorded state. The optional recorded-state cutoff can
be combined with any valid-time mode, which keeps correction history distinct
from changes in real-world validity.
An interval is half open: `valid_from <= instant < valid_to`. A revision belongs
to a recorded-state snapshot when `recorded_at <= cutoff` and it was not
superseded at or before that cutoff.
## Security And Mutation Rules
The temporal data context is a read projection, not an authorization context.
Authentication, permissions, active delegations, tenant boundaries, module
policy, and maintenance controls are always evaluated under current security
state. A historical projection never restores an expired permission.
The context also does not supply mutation dates. Writes continue to target the
current lifecycle revision and must carry their explicit valid/effective dates,
expected revision, reason, and evidence where the owning contract requires
them. A screen showing historical data must not silently turn a normal edit
into a historical correction.
## HTTP Contract
Core accepts these request headers:
| Header | Meaning |
| --- | --- |
| `X-Govoplan-Validity-Mode` | `current`, `at`, or `all` |
| `X-Govoplan-Valid-At` | Timezone-aware ISO 8601 instant required by `at` |
| `X-Govoplan-Recorded-At` | Optional timezone-aware system-knowledge cutoff |
Invalid or naive timestamps fail with HTTP 400. Responses expose the resolved
mode and evaluated instant. Conditional JSON responses vary by all three
request headers, and the shared WebUI API client includes them in request
deduplication and conditional-cache keys.
## Module Adoption
Revision-owning modules apply
`govoplan_core.db.temporal.apply_temporal_revision_filter` only to read queries
that are meant to follow the platform context. Explicit version references and
explicit resolver `effective_at` arguments take precedence. Current-row
lookups used for optimistic concurrency, authorization, routing, effects, or
other mutations must remain explicit and context-independent.
The initial bitemporal adoption covers Decisions, Mandates, Parties, and
Services. Their immutable revisions have indexed valid, recorded, and
superseded timestamps. Modules with effective-dated security records or
recorded-only revision histories require separate display-query adoption so
the global selector cannot affect current authorization or execution.
The WebUI selection is stored in session storage per account and tenant. A
change remounts the active module route so existing page loaders issue a fresh
request. Returning both axes to their defaults removes the stored selection.
+56
View File
@@ -0,0 +1,56 @@
# WebUI Theme Contract
GovOPlaN supports `system`, `light`, and `dark` as persisted user preferences.
`system` follows `prefers-color-scheme` live; it is not resolved permanently at
save time. Core applies the resolved mode through `data-theme` on the document
root and exposes the selected preference through `data-theme-preference`.
Each user may also choose a validated `default`, `civic_blue`, `forest`, or
`plum` accent palette. Core applies it through `data-palette`; every module
inherits the result through semantic tokens without module-specific CSS.
## Ownership
- Core owns semantic CSS tokens, native `color-scheme`, preference persistence,
the Settings selector, and the shared shell.
- Modules consume semantic tokens such as `--surface`, `--text`, `--line`, and
the status token families. They may define domain aliases whose values resolve
to shared tokens.
- Palette defaults form a provenance chain: system, tenant, then an explicit
user choice. Invalid stored values are ignored. Reset means inheritance and
does not copy the current parent value into the child scope.
- A policy lock is separate from the default. A system lock wins over every
child scope; otherwise a tenant lock suppresses a personal override. The
authenticated profile reports the effective palette, source, inherited
palette, and lock state.
- Advanced personal overrides are a separately governed surface. The system
must opt in, a tenant may inherit or block that decision, and palette locks
always suppress overrides. Changing either policy requires
`admin:policies:write` in addition to the owning settings permission.
## Palette safety and scope
The Settings preview shows the chosen or inherited accent in every applicable
light/dark preview before Save. Presets are checked for WCAG AA contrast in the
theme contract. When policy permits, the shared advanced editor can atomically
override accent, surface, and semantic status pairs for both modes. Every
foreground/background pair must meet WCAG AA contrast, and success,
information, warning, and danger colors must remain distinct. Invalid stored
documents fail closed and are not partially applied.
Import and export use the exact versioned JSON schema `schema_version: "1"`.
Both `light` and `dark` must contain every supported token exactly once as a
six-digit hex value. Import changes only the local draft; Save persists the
whole document. Removing overrides returns to palette and policy inheritance.
The system default is disabled so upgrades do not unexpectedly admit arbitrary
branding. Tenant `null` means inherit, `false` blocks, and `true` is accepted
only while the system permits overrides.
Do not introduce fixed foreground/background colors in a module merely to make
one mode look correct. Add or reuse a semantic Core token, then define both
light and dark values. Bitmap content and externally authored HTML are exempt,
but their surrounding controls must still use the shared tokens.
`npm run test:theme-contract` verifies root mode/palette behavior, preset and
custom-override validation/application, and representative
Campaign, Calendar, Files, and Mail token consumption. The check runs before a
production WebUI build.
+24
View File
@@ -0,0 +1,24 @@
# Shared fixed-window throttling
Core exposes `govoplan_core.core.throttling` for security-sensitive endpoints
that need bounded fixed-window counters. Subjects are SHA-256 hashed before they
become store keys. A configured Redis instance provides atomic counters shared
across API workers. Development and temporary Redis outages use a bounded
process-local fallback. Production-like startup rejects an enabled login
throttle without `REDIS_URL` unless
`GOVOPLAN_ALLOW_PROCESS_LOCAL_LOGIN_THROTTLE=true` explicitly acknowledges the
single-process limitation. When Redis fails at runtime, local attempts are still
mirrored so losing the distributed store does not reset the active worker's
protection window.
Callers define one or more `ThrottleDimension` values with a controlled
namespace, a subject and a positive limit. They must call `check` before an
expensive verifier, `record` after a failed attempt, and may `reset` the relevant
dimension after successful verification. A blocked decision includes a
`retry_after_seconds` value suitable for an HTTP `Retry-After` header.
The first consumer is Scheduling's anonymous participation password challenge.
Its namespace is `poll-participation-password`; its subject combines tenant,
scheduling request and Poll's non-secret invitation-token fingerprint. Access's
login throttle predates this primitive and should be migrated onto it in a
separate compatibility-preserving slice.
+59
View File
@@ -0,0 +1,59 @@
# Ticket Integration Capability Contracts
Core owns two narrow, optional contracts that let the Tickets module compose
with policy and formal-procedure modules without importing either one. Tickets
remains the authority for operational ticket identity, lifecycle, assignment,
comments, links, and immutable history.
## Capability Names
- `tickets.routing` optionally supplies a `TicketRoutingProvider`.
- `tickets.case_escalation` optionally supplies a
`TicketCaseEscalationProvider`.
Both contracts are version 1 and are defined in
`govoplan_core.core.tickets`. Registry helpers return `None` when a capability
is absent or has the wrong shape, so optional-module absence is normal runtime
state rather than a startup failure.
## Routing
Tickets sends a bounded, tenant-scoped `TicketRoutingRequest` containing the
ticket reference, type, priority, title, receive time, optional queue hint, and
non-secret attributes. The provider returns its identity and may return a queue
reference, timezone-aware service target, human-readable explanation, and
bounded metadata.
The provider is advisory. Tickets snapshots any returned queue and target into
its own record and history. An absent provider, a no-match plan, or an absent
queue must not prevent ticket intake; authorized staff can route manually.
Providers must not persist a second ticket lifecycle.
## Case Escalation
Tickets sends a `TicketCaseEscalationCommand` with stable tenant, ticket, and
display references, the requested Case type, actor-visible handoff note,
timezone-aware occurrence time, and an idempotency key. The provider returns a
stable Case identifier, number, bounded application-relative URL, replay flag,
and bounded metadata.
Providers must:
- recheck tenant and Case-creation authorization;
- reject an absent or inactive requested Case type;
- make identical retries resolve the same Case;
- preserve the Ticket reference in governed Case context; and
- return only an application-relative path, never an untrusted external URL.
Tickets records the result and its own escalation evidence. Cases remains the
authority for the formal procedure; Tickets remains the authority for the
operational request. Creating a Case does not merge or silently close either
lifecycle.
## Failure And Transaction Semantics
Capability calls receive the caller's active persistence session so a concrete
provider can participate in the same unit of work. Authorization and validation
errors fail the requested routing/escalation mutation explicitly. The caller
must still apply its own permission checks, tenant boundary, replay protection,
and immutable evidence rules.
+401
View File
@@ -0,0 +1,401 @@
# GovOPlaN UI/UX Decision Ledger
This ledger records product UI/UX decisions that affect admin, settings,
configuration, connector, policy, and module-management surfaces. It is a
binding design reference: future implementation should follow these decisions
unless the decision is explicitly revised here and affected screens are updated
to match.
Active tracking issue: `GovOPlaN/govoplan-core#225`.
## Operating Rule
GovOPlaN must expose advanced platform capability without presenting the user
with every option at once. Non-technical users should be able to complete common
workflows through guided, plain-language flows. Expert and diagnostic detail may
exist, but it must be deliberately layered.
The ethical design doctrine in `INTERFACE_ETHICS_AND_DESIGN_DOCTRINE.md` is the
normative baseline for this ledger. A screen can be visually quiet and still be
ethically complete only when it preserves context, consequence,
contestability, responsibility, and traceability at the point of action.
## Binding Decisions
| ID | Decision | Status | Applies To |
| --- | --- | --- | --- |
| UX-001 | Use progressive disclosure by default. Common decisions stay visible; advanced or hazardous options live in collapsed panels, later wizard steps, or explicit advanced sections. | Accepted | Admin, settings, connector setup, policy editors, module operations |
| UX-002 | Raw JSON is not a primary configuration editor. Every normal configuration path needs typed controls, validation, and help text. JSON may be shown for import/export, diagnostics, or expert inspection only. | Accepted | All admin/configuration UIs |
| UX-003 | Prefer guided workflows over option dumps for setup and risky changes. Use wizards for connector setup, configuration package import, module install/uninstall, destructive actions, and policy changes with broad impact. | Accepted | Connectors, package import, module lifecycle, governance/policy |
| UX-004 | Discover technical values when possible. Ask for the smallest user-known input, then discover and prefill technical fields for review. | Accepted | File connectors, mail/groupware, public URLs, future external providers |
| UX-005 | Disabled actions and failed steps must explain why they are unavailable, who can fix them, and where to go next. Silent disabled states are not acceptable for primary actions. | Accepted | All primary actions |
| UX-006 | Explanations must be available but quiet. Use short inline text and expose richer explanations through help affordances, side panels, expandable sections, or review steps. | Accepted | All complex forms and flows |
| UX-007 | Creation/editing should prefer modals or focused step flows when it reduces page clutter. Overview and comparison screens remain full-page. | Accepted | Settings/admin surfaces |
| UX-008 | Similar concepts must use shared placement and components: server/credential/policy rows, problem lists, review steps, advanced panels, confirmation modals, and empty/error states. | Accepted | Core WebUI and module WebUIs |
| UX-009 | Preflight and diagnostics are product UX. Validation, policy, permission, dependency, and capability failures must be written for operators before exposing internal details. | Accepted | Installer, connectors, policy, package import |
| UX-010 | Context, decision, and consequence must be visible together for actions that affect rights, duties, records, money, communication, retention, external systems, or workflow state. | Accepted | Workflow, admin, portal, policy, connector, records, payments |
| UX-011 | Navigation is not consent. Route changes, panel switches, and passive selection must not execute consequential actions without an explicit action surface. | Accepted | All WebUI surfaces |
| UX-012 | Automated actions must remain inspectable. The UI must show the system actor, trigger, policy result, observed effects, and failure/manual-intervention state when automation changes administrative state. | Accepted | Workflow, automation, connectors, tasks, audit |
| UX-013 | Contestable decisions must expose provenance. Denials, locks, generated outputs, calculated defaults, policy decisions, access decisions, and retention decisions need a reachable source path. | Accepted | Policy, access, templates, workflow, retention, records |
| UX-014 | Retraction, expiry, undo, rollback, and delete controls must state the real limit of the operation. Corrective or future-only actions must not be described as if they undo already observed effects. | Accepted | Postbox, files, records, installer, workflow, payments |
| UX-015 | Core owns the platform appearance contract. Modules must use shared CSS tokens and shared controls for theme-aware UI; they must not define independent light/dark palette systems. | Accepted | Core shell and all module WebUIs |
| UX-016 | Full-page create/edit surfaces keep `Discard` and the named `Save …` action in the upper-right page action cluster, with Save at the far right. Their position remains stable through validation and loading states. | Accepted | All full-page create/edit surfaces |
| UX-017 | Table row actions use icon-only controls in a stable rightmost column and intent order: inspect/open, edit, copy, transfer/share/download, retry/restore, destructive action last. Every icon requires a translated accessible name and tooltip. | Accepted | All structured tables |
| UX-018 | A collapsible card containing only one table gives the table the card's full available body, without decorative inner wrappers, duplicate padding, max-widths, or nested scrolling. | Accepted | List, detail, workflow, and configuration surfaces |
| UX-019 | Focused-view precedence is manual session pin, current-task suggestion, user default, role/tenant default, then the full interface. The active source and a full-interface escape remain visible; a suggested view never changes authorization or implies consent. | Accepted | Shell, modules, future workflow composition |
| UX-020 | Centrally exported Core components are mandatory wherever their contract covers the interaction. A custom reusable control, presentation primitive, or module-local substitute requires explicit product-owner authorization, a narrowly specific purpose, and documented rationale and scope; it must not duplicate a central component. | Accepted | Core WebUI and all module WebUIs |
| UX-021 | A collection-wide create action belongs in that collection's page heading and is not duplicated in a persistent side panel. When a side panel is the creation surface, it is present for the creation view only. | Accepted | List-detail, directory, and create surfaces |
| UX-022 | Use central `Card` components for logical sections, `DataGrid` for tabular row collections and their ordered actions, and `ToggleSwitch` for boolean settings. Repeatable people/contact editors use one structured row per person with name, email address, and actions; free-form address parsing is reserved for an explicitly designed bulk-import flow. | Accepted | All WebUI forms and collection editors |
| UX-023 | `FieldLabel` is the standard label/help surface for every field that is not self-explanatory. Any field rendered without it must be recorded in the omission register below, including its accessible-name source and rationale. Users may hide inline help markers through their persisted interface preference; the field label itself remains visible. | Accepted | All Core and module forms |
| UX-024 | Explicit `Discard` actions and dirty in-application navigation use the shared `UnsavedChangesProvider` dialog. A page registers save/discard behavior with `useUnsavedDraftGuard`; its Discard button calls `requestDiscard`, and route changes use `useGuardedNavigate` or `requestNavigation`. | Accepted | All create/edit surfaces |
| UX-025 | `window.alert` and the global `alert` function are prohibited. A narrowly necessary exception requires product-owner authorization and an entry in the alert exception register before implementation. | Accepted | All WebUI code |
| UX-026 | A table defines one stable ordered action set. A row-level unavailable action remains in its normal position and is disabled, preferably with `disabledReason`; structurally irrelevant actions are omitted for the entire table. Empty rows reserve the same slots so their Add action stays in the normal left-most action position. | Accepted | All structured tables |
| UX-027 | The platform icon rail keeps its brand header and utility footer visible. Only the module-navigation region scrolls when installed and permitted modules exceed the available viewport height. | Accepted | Core WebUI shell |
| UX-028 | Maintenance and offline states use their established textual warning banners centered in the titlebar. They must not recolor the shell or add decorative status icons. Global search is a right-side command, so warnings do not replace its trigger or overlay. | Accepted | Core WebUI shell |
| UX-029 | Recoverable page and module errors use the central compact `DismissibleAlert` presentation with an explicit recovery action where one exists. Full-height workspaces must overlay page feedback instead of allowing an alert to become a stretched workspace row. | Accepted | Core and module WebUIs |
| UX-030 | At narrow widths, the titlebar uses separate context and command rows. Context selectors remain horizontally reachable, search retains its compact trigger, and language/help/notification/account commands remain fixed icon controls without overlap. Shared content padding contracts so domain workspaces retain usable width. | Accepted | Core WebUI shell and all module workspaces |
| UX-031 | Public controls and extension contributions use stable, module-namespaced interface identities. Shared controls expose `interfaceId` and `helpTopicId`; generated source anchors are inventory evidence, not a substitute for an explicit ID when documentation, policy, or automation refers to the control. | Accepted | Core and module WebUIs |
| UX-032 | `F1` resolves help from the focused field or action, then its dialog/section/page and registered route. Focused contexts retain the page fallback; Docs applies audience and permission filtering and falls back to visible module documentation. | Accepted | Core shell, Docs, and all module WebUIs |
| UX-033 | Global search is the left-most titlebar command, immediately before language selection. Its icon, `F3`, and `Ctrl`/`Cmd`+`K` all open the same permission-aware search overlay; the titlebar does not reserve a persistent query field. | Accepted | Core shell and Search WebUI |
| UX-034 | Every headed `PageLayout` declares one of `overview`, `collection`, `detail`, `editor`, or `workspace` independently from its standalone/workspace/embedded geometry. Its actions use the matching semantic `PageActionBar`: a refreshable page must provide Reload in the leading slot; collections keep Create far right; read-only pages do not invent Save. | Accepted | Core and all module WebUIs |
| UX-035 | Editor action bars expose clean, dirty, and saving state; always retain Discard immediately before the far-right Save; centrally disable both while clean or saving; and participate in the unsaved-change navigation guard. Danger actions occupy the explicit separated destructive group after ordinary actions and before editor persistence. | Accepted | Core and all module WebUIs |
## Confirmed Implementation Decisions
These decisions were accepted on 2026-07-09 for the first implementation slice.
They shape reusable components and screen structure. Future changes must revise
this section and update affected screens.
### DUE-001: Primary Admin Configuration Shell
Decision: admin/configuration surfaces should standardize on a two-zone layout.
- Left or top area: searchable overview/list, status, and primary actions.
- Main area: selected item summary and common settings.
- Modal/wizard: create, connect, edit, test, review, and confirm actions.
- Collapsed advanced panels: rarely used technical fields.
Use this for file connectors and mail servers first, then migrate policy,
retention, API keys, and module operations.
### DUE-002: Wizard Step Model
Decision: standard wizard steps and names use this baseline:
1. Choose type or scope.
2. Enter essentials.
3. Discover or test.
4. Configure ownership and policy.
5. Review changes and blockers.
6. Save or submit for operator action.
Not every wizard needs every step, but flows should use these names and order
where applicable.
This applies to explicit assisted setup, onboarding, import, preflight, and
risky multi-step operations. It is not the default shape for ordinary create or
edit dialogs.
### DUE-003: Explanation Placement
Decision: richer explanations should use this placement model:
- Short helper text below labels only when it prevents common mistakes.
- Tooltips for icon-only controls and compact terms.
- Expandable "Why?" or "Details" blocks for contextual explanations.
- Right-side detail panel or review step for preflight/provenance/diagnostics.
Avoid permanently visible paragraphs inside dense admin cards.
### DUE-004: Advanced Options Contract
Decision: advanced options are fields that are needed for control, compatibility,
or diagnostics but are not part of the common setup path.
- protocol-specific endpoints, ports, path overrides, TLS/signing toggles,
timeout/retry tuning, raw headers, migration/destructive flags, and fallback
compatibility settings are advanced.
- names, descriptions, provider type, base URL, ownership, policy mode, and
basic credentials are not advanced.
Advanced fields must still be editable through typed controls.
### DUE-005: Blocker Language Contract
Decision: all disabled or blocked primary actions should use a shared structured
reason object.
Shape:
- `summary`: short plain-language reason.
- `details`: optional explanation.
- `required_action`: what needs to happen.
- `actor`: who can do it, for example system administrator or tenant admin.
- `target`: where to go or which setting/capability is missing.
- `technical_details`: optional expandable developer/operator data.
For simple missing required fields, highlight the fields and put the structured
reason in a compact hover/focus bubble on the disabled primary action instead of
adding a large persistent warning block.
### DUE-006: First Migration Surface
Decision: convert file connectors first, then mail servers. They share the same
server/credential/policy model, are high-value, and will prove the reusable
patterns quickly.
### DUE-007: Adaptive Create/Edit Forms
Decision: ordinary create/edit dialogs should show the full editable state in an
adaptive form, not force a linear wizard.
- The user chooses a type, provider, credential mode, or policy mode.
- The dialog immediately adjusts to show only the fields relevant to that state.
- Editing an existing object uses the same field grouping and layout as creating
it, so users can recognize the state they configured.
- Discovery and test actions must give visible feedback in the same dialog:
directly usable, discovered alternative, credentials needed/rejected, or not
found with the next action the user can take.
- Wizard shells remain available for assisted setup, first-run guidance,
imports, discovery-heavy flows, and operational preflight workflows.
### DUE-008: Platform Theme Contract
Decision: the WebUI shell exposes a small, stable appearance contract based on
shared CSS tokens and persisted user preference selection.
- Core applies `system`, `light`, and `dark` preferences at the document root.
- Core applies validated user accent presets through `data-palette`; palette
values change semantic tokens globally and never require module CSS changes.
- Core owns shared tokens such as `--bg`, `--bar`, `--panel`, `--surface`,
`--line`, `--line-dark`, `--text`, `--text-strong`, `--muted`, semantic
status colors, radii, shadows, and disabled-control colors.
- Modules must style new UI with these tokens and shared controls. Module-local
CSS may tune layout and spacing, but it must not introduce a separate
appearance system.
- Appearance controls live in user settings. A personal palette wins over
unlocked tenant and system defaults; system and tenant locks take precedence.
Advanced personal token overrides additionally require system opt-in and may
be narrowed by tenant policy. Their versioned import/export document is
validated and applied all-or-nothing in both light and dark modes.
- Visual preview in settings is illustrative; it must reflect token families,
not become a second theme implementation.
### DUE-009: Central Component And Exception Contract
Decision: module interfaces are compositions of the components exported by
`@govoplan/core-webui`. When Core already owns the matching interaction, using
the central component is required rather than preferred.
A route or domain-specific page composed from central components is ordinary
module composition. A new reusable UI control, presentation primitive, or
module-local substitute is a custom component. Before one is implemented, the
product owner must explicitly authorize it and the owning decision or issue must
record:
- its single, narrowly defined purpose and intended consumers
- why central components or their composition cannot meet that purpose
- the permitted scope and the boundary it must not grow beyond
- accessibility, reachable states, theme behavior, and test expectations
- whether the component remains domain-owned or is a candidate for Core
Custom components must not duplicate, fork, or cosmetically replace a central
component. An existing local implementation does not grant an exception. If a
central contract later covers the need, migrate to it unless the product owner
explicitly retains the exception.
### DUE-010: Scheduling Request Reference Composition
Decision: Scheduling requests provide a concrete reference application of the
universal placement and component rules.
- The persistent left panel stacks `My scheduling requests` and `Scheduling
requests for me`; it is list context, not a second creation affordance.
- The left panel's `Scheduling requests` header owns one `Add` action. It opens
the shared view/create/edit surface in the right main panel.
- Basic information, Calendar integration, candidate slots, and participants
use the central `Card` component as four logical sections.
- Candidate slots and participants use the central `DataGrid`, including its
standard row-action placement and order.
- Calendar integration uses the central `ToggleSwitch`, with its dependent
controls shown when enabled.
- Each participant is edited as one structured row with name, email address,
and actions. The normal editor does not parse a free-form list of addresses;
that interaction requires a separate, explicitly designed bulk-import flow.
Equivalent list/create/edit surfaces use the same underlying rules. These are
not Scheduling-local component variants.
### DUE-011: Field Help, Discard, And Table Action Contracts
Decision: the central components own these interactions; modules compose them
instead of reproducing their behavior.
- `FormField` and `ToggleSwitch` already render `FieldLabel`. Direct field
compositions use `FieldLabel` explicitly when the meaning or limitation is
not self-explanatory.
- `help` content is contextual guidance, not the accessible name. The persisted
`show_inline_help_hints` user preference hides only the `InlineHelp` marker by
applying `ui-hide-help-hints` at the document root.
- Shared action-bearing components accept an optional disabled reason. In
particular, `MailServerSettingsPanel` forwards protocol-specific test
blockers into the shared focusable disabled-action tooltip; modules provide
the domain-specific required field, permission, or in-progress reason.
- A dirty editor registers once with `useUnsavedDraftGuard`. An explicit
Discard button calls `useUnsavedChanges().requestDiscard(afterResolve)`; SPA
navigation uses `useGuardedNavigate` or `requestNavigation`. Both paths show
the same shared unsaved-changes dialog. A browser tab/window unload remains a
browser-controlled confirmation because browsers do not permit a custom
modal at that boundary.
- `TableActionGroup` receives the table's stable action set. Use `disabled` and
`disabledReason` for row state; omit an action only when that action does not
belong to the table. `minimumSlots` reserves trailing positions for an empty
row. `DataGridEmptyAction` does this for the standard add/move/remove layout.
- A paginated `DataGrid` has exactly one query owner. Client mode receives the
complete logical row set and applies filtering and sorting before slicing a
page. Server mode receives only the loaded page, requires `onQueryChange`,
and the backend applies every emitted filter/sort before pagination while
returning `totalRows` for the filtered result. Server list filters declare
their complete option domain instead of deriving it from the loaded page.
External filter affordances such as summary-count shortcuts update the
grid's `query` contract; the grid header controls and backend query therefore
always display and execute the same filter state.
- Feedback and confirmation use `Dialog`, `ConfirmDialog`, or
`DismissibleAlert`. They never fall back to `window.alert`.
### DUE-012: Rich HTML Editing Contract
Decision: modules that edit persisted HTML use the central
`WysiwygEditor` exported from `@govoplan/core-webui/wysiwyg`.
- The dedicated subpath is intentional: the editor and its engine remain a
shared Core contract without adding their code to module combinations that
never consume rich-text editing.
- Consumers provide controlled HTML and domain-specific token labels. The
editor owns visual/source switching, formatting, links, images, safe URL
handling, and atomic inline token rendering; it does not own template
semantics or persistence.
- Existing HTML outside the supported visual subset opens in source mode.
Rendering the value must not rewrite it, and users receive an explicit
warning before choosing the visual surface.
- Domain placeholders remain their original serialized text. Atomic token
presentation is an editing aid only, so backend renderers and existing
templates do not need a new storage format.
#### FieldLabel Omission Register
Every Core field surface that intentionally does not render `FieldLabel` is
listed here. Module repositories keep an equivalent register in their durable
UI documentation until a central cross-repository audit is available.
| Core scope | Why `FieldLabel` is omitted | Accessible/context label source |
| --- | --- | --- |
| `PasswordField`, `ColorPickerField`, `DateField`, `TimeField`, and `DateTimeField` input internals | These are label-neutral composite primitives and are placed inside `FormField`/`FieldLabel` by the consuming form. Rendering another label inside the primitive would duplicate it. `PasswordField` may opt into the shared cryptographic generator; the candidate dialog is subordinate to the enclosing field and commits only through its explicit Use action. | Enclosing label; a direct consumer must pass an accessible name and record that direct composition here. |
| `ToggleSwitch` native checkbox | The shared component already renders its visible text through `FieldLabel`; the native input must not render a second label. | The enclosing native label and derived `aria-label`. |
| `FileDropZone` hidden file input | The input is an implementation detail of the labelled keyboard-operable drop target. | Drop target text and `inputLabel`/`aria-label`. |
| `AdminSelectionList` and `DataGrid` list-filter checkboxes | Each option is self-explanatory and already enclosed by its visible option label. | Enclosing native option label. |
| `EmailAddressInput` compact Name and Email fields | These two conventional fields are self-explanatory in the compact address popover; richer address guidance belongs to the enclosing field. | Visible native labels; the free-form editor also has a descriptive `aria-label`. |
| `DataGrid` page-size, filter, and inline cell editors | The surrounding column header/filter heading supplies field context; repeating a labelled help marker in every cell would add noise. | Column header, filter heading/native label, or generated cell `aria-label`. |
| Retention-policy value controls | `PolicyRow` owns the field label, help, effective value, and provenance for its control. | The containing `PolicyRow` label/help contract. |
#### Alert Exception Register
No `window.alert` or global `alert` exception is authorized.
## Implementation Sequence
| Phase | Scope | Output |
| --- | --- | --- |
| 0 | UX inventory | List every admin/settings/configuration surface, classify it, and record whether it violates a binding decision. |
| 1 | Core primitives | Shared wizard shell, advanced panel, help affordance, blocker callout, problem list, review step, and discovery/test result components. |
| 2 | File connectors | Adaptive create/edit for provider/server, credential, discovery/test, ownership/policy, plus optional assisted wizard later. |
| 3 | Mail servers | Same adaptive pattern as files, adapted to server/credential/policy and test-send/test-login behavior. |
| 4 | Policy/retention editors | Effective value first, provenance visible, override/edit in modal, blocked edits explained. |
| 5 | Module/package operations | Step-based install/uninstall flow with preflight, maintenance, daemon handoff, migration, and rollback explanation. |
| 6 | Remaining settings/admin screens | Apply inventory findings by priority and remove one-off layouts. |
## Current Surface Inventory
This is the phase-0 inventory baseline. It should be extended as each screen is
converted or reviewed.
| Surface | Repository | UX State | Next Action |
| --- | --- | --- | --- |
| File connector settings | `govoplan-files` | Migrated to the shared adaptive server/credential/policy pattern with provider discovery, typed controls, actionable blockers, consequence-aware removal, and module-owned verification evidence. | Continue only through bounded Files-owned follow-ups. |
| Mail server settings | `govoplan-mail` / `govoplan-core` | Migrated to the same layered profile/server/credential/policy pattern, including focused connection tests, unsaved-state handling, contextual help, and permission/target blockers. | Continue only through bounded Mail-owned follow-ups. |
| Connector policy/effective rows | `govoplan-core`, module UIs | Effective-policy direction exists, but many editors still expose broad option sets. | Put effective value first, move overrides into modal, and explain blocked edits. |
| Admin module management | `govoplan-admin` | Has preflight concepts, but operational choices are still technical and dense. | Convert install/uninstall/package changes to operator wizards. |
| Configuration packages | `govoplan-admin` | Catalog/import work exists, but package editing can still drift toward technical fields. | Add guided import/review/problem-list flow. |
| Retention and privacy | `govoplan-core` | Typed effective-policy editor exposes source paths, narrowing semantics, platform locks, permission/target blockers, and explicit clean/loading/save states. | Broader governed-change review remains module-owned where a policy change requires approval. |
| API keys | `govoplan-access` / admin UI | Security-sensitive creation needs least-privilege guidance. | Add scoped creation wizard with expiry/owner review. |
| User settings | `govoplan-core` | Simple typed sections use unsaved-change guards, quiet result feedback, contextual help, explicit busy/clean disabled-action reasons, and an effective appearance source. Palette selection and light/dark preview are shared with system and tenant administration. | Keep bounded; new contributed sections must satisfy the checklist. |
## Impact Index
| Surface | Current Risk | Expected Pattern |
| --- | --- | --- |
| File connectors | Too many technical fields and unclear setup order. | Adaptive create/edit form with optional assisted wizard, discovery/test, credential binding, policy review, and advanced protocol panel. |
| Mail servers | Similar server/credential/policy concepts risk diverging from files. | Same tree/list and adaptive server/credential/policy model as file connectors. |
| Connector credentials | Security-sensitive details can overwhelm users. | Separate credential flow with secret-reference language and test result explanation. |
| Policy and effective settings | Users need to know why a value is inherited or locked. | Effective row first, source/provenance, local override action, actionable blocked reason. |
| Module install/uninstall | Operationally risky, currently inherently technical. | Operator wizard with preflight, maintenance, daemon handoff, review, and rollback explanation. |
| Configuration packages | Could become package JSON editing. | Package catalog/import wizard using provider data requirements and problem lists. |
| Retention/privacy | High-risk settings need explanation and provenance. | Layered editor with plain-language consequences and review. |
| Automation/workflow commands | Hidden side effects would undermine accountability. | Action/effect preview, system-actor display, command record, retry/quarantine/manual states, and audit links. |
| Postbox and encrypted communication | Retraction and access can be misunderstood. | Honest key-fetch/decryption state, expiry limits, recipient/device access provenance, and delivery evidence. |
| API keys | Security-sensitive creation and scope selection. | Scoped creation wizard, least-privilege suggestions, clear expiry/owner explanation. |
| User settings | Needs clarity and persistence across profile/interface/preferences. | Simple settings sections with immediate feedback, explicit inherit/reset semantics, effective-source provenance, and no double-click navigation traps. |
## Review Checklist
Every new or changed admin/configuration surface should answer:
- What is the common path, and is it visible without noise?
- Which fields are advanced, and are they collapsed by default?
- Is every configuration value editable through typed controls?
- Does the flow avoid JSON as the primary editor?
- Does the screen explain disabled actions and failed validation in plain
language?
- Does it say who can fix a blocker and where?
- Does a module-localized blocker pass its translated row labels through the
shared `ActionBlockerHint` contract instead of reproducing the component?
- Does longer field or blocker guidance use a stable `DocumentationHelpLink`
topic/context reference, with hosted fallback when the optional Docs module
is absent?
- Does it reuse existing core patterns for wizard steps, problem lists, modals,
help, and review?
- Is there a review or preflight step before broad, destructive, or risky
changes?
- Does the action surface show consequence, reversibility, and audit evidence
when rights, duties, records, money, communication, external systems, or
workflow state are affected?
- Does the surface use every applicable central Core component? If it contains
a custom component, is the product-owner authorization, narrow purpose,
rationale, scope, and non-duplication evidence recorded?
- Is a collection-wide create action in the collection heading rather than
duplicated in a persistent side panel?
- Are logical sections, tabular collections, boolean settings, and repeatable
people/contact rows composed with `Card`, `DataGrid`, `ToggleSwitch`, and one
structured row per person respectively?
- Does every non-self-explanatory field use `FieldLabel`, and is every omission
recorded with its rationale and accessible-name source?
- Do explicit Discard and dirty navigation use the shared unsaved-changes
registration/dialog rather than a page-local confirmation?
- Does every row retain the table's action set in the same order, disabling
unavailable actions and reserving the same empty-row slots?
- Is feedback rendered with a central dialog/alert component, with no
unauthorized `window.alert` or global `alert` call?
- If automation is involved, can the user see the trigger, system actor,
observed effects, and failure/manual-intervention state?
- Are technical details available without being the first thing the user sees?
## Revision Rule
When a UX decision changes:
1. Update this ledger.
2. Update the affected shared components.
3. Update existing screens listed in the impact index.
4. Add or update Gitea issues for any remaining surfaces that still follow the
old decision.
5. Sync the wiki.
+71
View File
@@ -0,0 +1,71 @@
# WebUI Loading And Bundle Budgets
The Core WebUI host owns the loading boundary for installed module packages.
Vite discovers configured packages at build time, but emits an asynchronous
loader for each package's `src/module.ts` contribution descriptor. At runtime,
Core imports only descriptors whose backend manifests are enabled and identify
the matching `frontend.package_name`.
The direct descriptor entry is intentional. A package root may re-export pages
for consumers; importing that barrel as module wiring can cause those pages to
be evaluated before navigation. Route pages and substantial panels should use
`React.lazy`, and Core wraps routes in the shared loading/error boundary.
## Enforced Budgets
`webui/bundle-budget.json` contains the production limits:
| Measurement | Raw limit | Gzip limit |
| --- | ---: | ---: |
| Initial JavaScript static import closure | 512 KiB | 160 KiB |
| Largest individual asynchronous JavaScript chunk | 384 KiB | 110 KiB |
`npm run build` writes a Vite manifest, measures the entry and its recursive
static imports, writes `dist/bundle-metrics.json`, and fails when either budget
is exceeded. `npm run test:module-permutations` applies the same gate to every
permutation and records the collected results in
`dist/module-permutation-bundle-metrics.json`. In CI, each result is also added
to the step summary.
Budgets are limits, not targets. A change that approaches a limit should add a
new lazy boundary or remove unnecessary entry code instead of raising the
limit without measurement and review.
## 2026-07-30 Baseline
Measurements use the same full-product source tree and Node 22 runtime. The
post-change build additionally includes the Search module in the default and
full-product sets.
| Initial-load measurement | Before | After | Reduction |
| --- | ---: | ---: | ---: |
| JavaScript assets in initial static closure | 1 | 1 | 0% |
| Raw JavaScript | 1,387,043 B | 453,769 B | 67.3% |
| Gzip level 9 | 364,767 B | 141,725 B | 61.1% |
| Brotli quality 11 | 254,797 B | 106,701 B | 58.1% |
| Parse proxy median | 18.776 ms | 7.750 ms | 58.7% |
| Parse proxy p95 | 21.980 ms | 8.739 ms | 60.2% |
The parse proxy constructs a fresh `node:vm` `SourceTextModule` from the entry
source 30 times with a randomized source marker. It is useful for a controlled
before/after comparison, but is not enforced in CI because absolute timings
vary across runners. Transfer budgets use deterministic raw and gzip byte
counts.
The first budgeted full-product build reported:
- initial JavaScript: 453,769 B raw / 141,725 B gzip;
- largest async chunk: `CampaignWorkspace`, 353,724 B raw / 98,142 B gzip.
## Verification
```bash
cd /mnt/DATA/git/govoplan-core/webui
npm run build
npm run check:bundle-budget
npm run test:module-permutations
```
The build gate also catches accidental eager imports: a page pulled into the
entry closure consumes the initial budget, while an oversized page or module
descriptor consumes the asynchronous chunk budget.
+12
View File
@@ -0,0 +1,12 @@
# WebUI Module Package Layout
Core discovers a module contribution from `src/module.ts` when `node_modules`
links directly to a module's `webui` package. Tagged release dependencies are
installed from repository-root packages and expose the same contribution at
`webui/src/module.ts`. The Vite registry accepts both layouts and imports the
contribution descriptor directly so route-level lazy loading is preserved.
A release package is invalid if neither entry exists. The module-permutation CI
matrix builds source-linked and installed release compositions; it must not fall
back to a package root barrel because that would eagerly pull module pages into
the shell bundle.
@@ -0,0 +1,71 @@
# Dependency Audit - 2026-07-09
Commands:
```bash
cd /mnt/DATA/git/govoplan
GOVOPLAN_CORE_ROOT=/mnt/DATA/git/govoplan-core bash tools/checks/check-dependency-audits.sh
```
Status: remediated.
Initial result:
- Python audit failed: 24 advisories were reported across `cryptography`,
`pip`, `python-multipart`, `pyzipper`, and `starlette`.
- npm production audit passed: `npm audit --omit=dev` reported 0
vulnerabilities.
Python findings:
| Package | Installed | Advisory count | Minimum reported fix |
| --- | ---: | ---: | --- |
| `cryptography` | `44.0.0` | 5 | `48.0.1` |
| `pip` | `26.0.1` | 3 | `26.1.2` |
| `python-multipart` | `0.0.17` | 7 | `0.0.31` |
| `pyzipper` | `0.3.6` | 1 | `0.4.0` |
| `starlette` | `0.41.3` | 8 | `1.3.1` |
Private GovOPlaN packages were skipped by `pip-audit` because they are not
published on PyPI. That is expected for local editable development installs.
The remediation below upgrades those dependencies and records the compatibility
checks run against core, files, and campaign behavior.
## Remediation
Remediation applied on 2026-07-09:
- upgraded core's FastAPI floor to `fastapi>=0.139,<1`, resolving Starlette to
`starlette==1.3.1`
- upgraded core's cryptography floor to `cryptography>=48.0.1,<50`, resolving
to `cryptography==49.0.0`
- declared the files module upload parser dependency as
`python-multipart>=0.0.31,<1`, resolving to `python-multipart==0.0.32`
- upgraded the campaign ZIP dependency to `pyzipper>=0.4,<1`, resolving to
`pyzipper==0.4.0`
- upgraded the local audit environment to `pip==26.1.2`
- removed the obsolete local mailer-module editable install
from the audit environment so the audit reflects the split module product
Post-remediation result:
- `bash tools/checks/check-dependency-audits.sh`: passed, no known Python
vulnerabilities found and npm production audit reported 0 vulnerabilities.
- `python -m pip check`: passed.
- `bash tools/checks/check-module-matrix.sh`: passed.
- `python -m unittest tests.test_api_smoke`: passed.
- campaign encrypted/plain ZIP smoke with `pyzipper==0.4.0`: passed.
Notes:
- `pip-audit` still reports private GovOPlaN packages as skipped because they
are not published on PyPI. That is expected for editable local development
installs.
- `httpx2>=2.5,<3` is included in development requirements so Starlette's
`TestClient` uses the non-deprecated backend. `httpx==0.28.1` remains in dev
requirements for tests that mock connector HTTP responses directly.
- Split-module API routers now use Starlette's renamed
`HTTP_422_UNPROCESSABLE_CONTENT` status constant. This preserves the 422
status code while avoiding the deprecated `HTTP_422_UNPROCESSABLE_ENTITY`
alias.

Some files were not shown because too many files have changed in this diff Show More