# Archive Tools

Inspect, create, compare and safely extract archives locally.

Archive Tools is a standalone local-first application in the [add·ideas Toolbox](https://git.add-ideas.de/lotobo/toolbox-portal). Inputs are processed in the browser and are not uploaded.

## Version 0.1 scope

Binary reads use checked offset arithmetic and bounded cursors. gzip CRC-32 is
updated incrementally during decompression; temporary previews and downloads
use explicitly revoked shared Blob URL leases.

- Inspect ZIP and ZIP64 central directories, TAR/USTAR/PAX archives, gzip streams and tar.gz archives.
- Structurally inspect 7z start/next headers and RAR4/RAR5 plaintext headers under fixed header/count bounds. Header CRCs are checked before metadata is accepted; entry content is never decoded.
- List normalized paths, kinds, sizes, compression, timestamps, ZIP64 state, CRC-32 values and policy diagnostics.
- Preview bounded UTF-8 text, hexadecimal bytes and static JPEG/PNG/WebP images. HTML and SVG are shown as text, never rendered as active content.
- Select regular safe files, verify their bytes and CRC where available, and repackage them into a newly created ZIP. Archive paths are never restored directly to a filesystem.
- Read and create password-protected ZIP entries with local ZipCrypto or WinZip AES support. AES-256 is the recommended creation mode; legacy ZipCrypto is clearly labelled weak.
- Create deterministic unencrypted ZIP, TAR/PAX and tar.gz output from local files with sorted paths and normalized metadata. Encrypted ZIPs use fresh random salt and are intentionally not byte-reproducible.
- Compare two archive inventories by collision-safe path, entry kind, expanded size and CRC-32, independent of container format and timestamps.
- Download machine-readable inspection and comparison reports.

Bounded GNU TAR long-name and long-link records and concatenated gzip members are supported. For 7z, the start-header offsets/sizes and both CRC-32 fields are validated; a plain FilesInfo header can expose names and empty-file/directory markers, while encoded headers are reported without decoding. For RAR4/RAR5, block framing and header CRCs are validated and plaintext file headers expose names, sizes, compression method identifiers, encryption and split-volume flags. Encrypted headers stop inspection.

Split/multipart ZIPs, nested archive expansion, GNU sparse files, symlinks, hardlinks, devices and FIFOs remain unsupported for restoration. 7z/RAR codec chains, solid dictionaries, passwords, recovery data, decompression and extraction are explicitly unsupported; their inventories are evidence, not proof that packed file data is intact.

The application treats every imported file as untrusted, applies explicit count/path/expanded-byte/compression-ratio limits, and never executes imported active content. See [docs/ARCHITECTURE.md](docs/ARCHITECTURE.md) and [docs/PRIVACY-SECURITY.md](docs/PRIVACY-SECURITY.md).

## Safety limits

| Resource                   |                           Limit |
| -------------------------- | ------------------------------: |
| Source archive             |                         512 MiB |
| Entries inspected          |                          20,000 |
| Entries displayed at once  |                           5,000 |
| Expanded archive bytes     |                         512 MiB |
| One expanded entry         |                         256 MiB |
| Compression ratio          |                           200:1 |
| Path                       | 4,096 UTF-8 bytes / 64 segments |
| Creation                   |           5,000 files / 512 MiB |
| Concatenated gzip members  |                             256 |
| Password                   |               1,024 UTF-8 bytes |
| One 7z/RAR metadata header |                          16 MiB |
| RAR header blocks          |                          40,000 |
| Text / hex / image preview |        2 MiB / 128 KiB / 24 MiB |

These are application-level ceilings, not a promise that every device can comfortably process files near the maximum.

## Using the workspaces

1. **Inspect & extract** opens one archive, displays its safety inventory and previews eligible files. Select safe regular files and choose **Verify & download safe ZIP** to create a fresh archive.
2. **Create** accepts local files and produces reproducible ZIP, TAR or tar.gz output, or a locally encrypted AES-256/ZipCrypto ZIP. Browser file selection exposes relative paths only when the browser supplies them.
3. **Compare** opens two archives and reports same, changed, type-changed and one-sided entries.

## Development

Requires Node.js 22 and npm 11.

```sh
npm ci
npm run check
npm run test:browser
```

Runtime dependencies are exactly pinned. `npm audit` should be checked with every dependency update.

## Release

`npm run release:artifact` creates a deterministic `release/archive-tools-0.2.0.zip` and checksum sidecar.

## Licence

GPL-3.0-or-later. Runtime dependencies retain their own licences; see [THIRD_PARTY_NOTICES.md](THIRD_PARTY_NOTICES.md).