feat: add campaign copying scheduling and residual handling

This commit is contained in:
2026-08-07 14:54:04 +02:00
parent 696f8f6385
commit c2efd6b7bd
35 changed files with 3359 additions and 39 deletions
@@ -0,0 +1,363 @@
from __future__ import annotations
import calendar
import copy
import hashlib
import json
from collections.abc import Mapping
from datetime import UTC, datetime, timedelta
from zoneinfo import ZoneInfo, ZoneInfoNotFoundError
from sqlalchemy.orm import Session
from govoplan_campaign.backend.campaign.copying import campaign_copy_configuration
from govoplan_campaign.backend.db.models import (
Campaign,
CampaignSchedule,
CampaignScheduleOccurrence,
CampaignShare,
CampaignVersion,
)
from govoplan_campaign.backend.persistence.campaigns import (
create_campaign_version_from_json,
)
from govoplan_core.audit.logging import audit_event
RECURRENCE_KINDS = frozenset({"once", "daily", "weekly", "monthly"})
SCHEDULE_SOURCE_SCHEMA = "govoplan.campaign.schedule-source.v1"
def canonical_configuration_hash(value: Mapping[str, object]) -> str:
return hashlib.sha256(
json.dumps(
value,
ensure_ascii=True,
sort_keys=True,
separators=(",", ":"),
).encode("utf-8")
).hexdigest()
def campaign_schedule_source_snapshot(
*,
configuration: Mapping[str, object],
campaign_settings: Mapping[str, object],
mail_profile_policy: Mapping[str, object],
shares: list[Mapping[str, object]],
) -> dict[str, object]:
"""Seal every selected source domain so worker execution cannot drift."""
return {
"schema": SCHEDULE_SOURCE_SCHEMA,
"configuration": copy.deepcopy(dict(configuration)),
"campaign_settings": copy.deepcopy(dict(campaign_settings)),
"mail_profile_policy": copy.deepcopy(dict(mail_profile_policy)),
"shares": [copy.deepcopy(dict(item)) for item in shares],
}
def next_schedule_fire(
scheduled_for: datetime,
*,
recurrence_kind: str,
interval_count: int,
timezone_name: str,
) -> datetime | None:
if recurrence_kind == "once":
return None
if recurrence_kind not in RECURRENCE_KINDS:
raise ValueError(f"Unsupported campaign recurrence: {recurrence_kind}")
if interval_count < 1:
raise ValueError("Campaign recurrence interval must be positive")
try:
zone = ZoneInfo(timezone_name)
except ZoneInfoNotFoundError as exc:
raise ValueError(f"Unknown campaign schedule timezone: {timezone_name}") from exc
local = _as_utc(scheduled_for).astimezone(zone)
if recurrence_kind == "daily":
upcoming = local + timedelta(days=interval_count)
elif recurrence_kind == "weekly":
upcoming = local + timedelta(weeks=interval_count)
else:
month_index = local.year * 12 + local.month - 1 + interval_count
year, month_offset = divmod(month_index, 12)
month = month_offset + 1
day = min(local.day, calendar.monthrange(year, month)[1])
upcoming = local.replace(year=year, month=month, day=day)
return upcoming.astimezone(UTC)
def dispatch_due_campaign_schedules(
session: Session,
*,
tenant_id: str | None = None,
now: datetime | None = None,
limit: int = 50,
) -> dict[str, object]:
observed_at = _as_utc(now or datetime.now(UTC))
query = session.query(CampaignSchedule).filter(
CampaignSchedule.active.is_(True),
CampaignSchedule.next_fire_at.is_not(None),
CampaignSchedule.next_fire_at <= observed_at,
)
if tenant_id is not None:
query = query.filter(CampaignSchedule.tenant_id == tenant_id)
schedules = (
query.order_by(CampaignSchedule.next_fire_at.asc(), CampaignSchedule.id.asc())
.with_for_update(skip_locked=True)
.limit(max(1, min(limit, 250)))
.all()
)
result: dict[str, object] = {
"selected": len(schedules),
"prepared": 0,
"failed": 0,
"completed": 0,
"coalesced": 0,
"campaign_ids": [],
"operator_actions": [],
}
for schedule in schedules:
scheduled_for = _as_utc(schedule.next_fire_at or observed_at)
try:
with session.begin_nested():
campaign, version, skipped = _prepare_occurrence(
session,
schedule=schedule,
scheduled_for=scheduled_for,
observed_at=observed_at,
)
result["prepared"] = int(result["prepared"]) + 1
result["coalesced"] = int(result["coalesced"]) + skipped
result["campaign_ids"].append(campaign.id) # type: ignore[union-attr]
if not schedule.active:
result["completed"] = int(result["completed"]) + 1
except Exception as exc: # noqa: BLE001 - persist bounded operator evidence
session.add(
CampaignScheduleOccurrence(
tenant_id=schedule.tenant_id,
schedule_id=schedule.id,
scheduled_for=scheduled_for,
status="failed",
error=str(exc)[:4000],
)
)
schedule.active = False
schedule.last_error = str(exc)[:4000]
schedule.resource_revision += 1
session.add(schedule)
result["failed"] = int(result["failed"]) + 1
result["operator_actions"].append( # type: ignore[union-attr]
{
"schedule_id": schedule.id,
"campaign_id": schedule.campaign_id,
"reason": "draft_preparation_failed",
}
)
session.flush()
return result
def _prepare_occurrence(
session: Session,
*,
schedule: CampaignSchedule,
scheduled_for: datetime,
observed_at: datetime,
) -> tuple[Campaign, CampaignVersion, int]:
existing = (
session.query(CampaignScheduleOccurrence)
.filter(
CampaignScheduleOccurrence.schedule_id == schedule.id,
CampaignScheduleOccurrence.scheduled_for == scheduled_for,
)
.one_or_none()
)
if existing is not None:
raise RuntimeError("Campaign schedule occurrence was already recorded")
source_campaign = session.get(Campaign, schedule.campaign_id)
if source_campaign is None or source_campaign.tenant_id != schedule.tenant_id:
raise RuntimeError("Campaign schedule source is no longer available")
source_version = session.get(CampaignVersion, schedule.source_version_id)
if source_version is None or source_version.campaign_id != source_campaign.id:
raise RuntimeError("Campaign schedule source version is no longer available")
if canonical_configuration_hash(schedule.source_snapshot) != schedule.source_snapshot_hash:
raise RuntimeError("Campaign schedule source snapshot integrity check failed")
snapshot = _schedule_snapshot(schedule.source_snapshot)
sequence = schedule.occurrence_count + 1
external_id = _scheduled_external_id(
source_campaign.external_id,
schedule.id,
sequence,
)
local_date = scheduled_for.astimezone(ZoneInfo(schedule.timezone)).date().isoformat()
generated_name = f"{schedule.name} - {local_date}"
raw_json = campaign_copy_configuration(
snapshot["configuration"],
schedule.copy_options,
)
metadata = raw_json.get("campaign")
if not isinstance(metadata, dict):
raise RuntimeError("Campaign schedule snapshot has no campaign metadata")
metadata["id"] = external_id
metadata["name"] = generated_name
metadata["mode"] = "draft"
generated_campaign, generated_version = create_campaign_version_from_json(
session,
tenant_id=schedule.tenant_id,
user_id=schedule.created_by_user_id,
raw_json=raw_json,
source_filename=None,
source_base_path=schedule.source_base_path,
commit=False,
)
if bool(schedule.copy_options.get("include_policies", True)):
generated_campaign.settings = copy.deepcopy(snapshot["campaign_settings"])
if bool(schedule.copy_options.get("include_mail_profile", True)):
generated_campaign.mail_profile_policy = copy.deepcopy(
snapshot["mail_profile_policy"]
)
if bool(schedule.copy_options.get("include_shares", False)):
_copy_snapshot_shares(
session,
schedule=schedule,
generated_campaign=generated_campaign,
shares=snapshot["shares"],
)
occurrence = CampaignScheduleOccurrence(
tenant_id=schedule.tenant_id,
schedule_id=schedule.id,
scheduled_for=scheduled_for,
status="prepared",
generated_campaign_id=generated_campaign.id,
generated_version_id=generated_version.id,
)
session.add(occurrence)
schedule.occurrence_count = sequence
schedule.last_fired_at = scheduled_for
schedule.last_campaign_id = generated_campaign.id
schedule.last_error = None
next_fire = next_schedule_fire(
scheduled_for,
recurrence_kind=schedule.recurrence_kind,
interval_count=schedule.interval_count,
timezone_name=schedule.timezone,
)
coalesced = 0
while next_fire is not None and next_fire <= observed_at:
next_fire = next_schedule_fire(
next_fire,
recurrence_kind=schedule.recurrence_kind,
interval_count=schedule.interval_count,
timezone_name=schedule.timezone,
)
coalesced += 1
if (
next_fire is None
or sequence >= schedule.max_occurrences
or (schedule.ends_at is not None and next_fire > _as_utc(schedule.ends_at))
):
schedule.active = False
schedule.next_fire_at = None
else:
schedule.next_fire_at = next_fire
schedule.resource_revision += 1
session.add(schedule)
audit_event(
session,
tenant_id=schedule.tenant_id,
user_id=schedule.created_by_user_id,
action="campaign.schedule.draft_prepared",
object_type="campaign_schedule",
object_id=schedule.id,
details={
"source_campaign_id": source_campaign.id,
"source_version_id": source_version.id,
"scheduled_for": scheduled_for.isoformat(),
"generated_campaign_id": generated_campaign.id,
"generated_version_id": generated_version.id,
"occurrence": sequence,
"coalesced_missed_intervals": coalesced,
"delivery_started": False,
},
commit=False,
)
return generated_campaign, generated_version, coalesced
def _copy_snapshot_shares(
session: Session,
*,
schedule: CampaignSchedule,
generated_campaign: Campaign,
shares: object,
) -> None:
if not isinstance(shares, list):
raise RuntimeError("Campaign schedule share snapshot is invalid")
for source in shares:
if not isinstance(source, Mapping):
raise RuntimeError("Campaign schedule share snapshot is invalid")
target_type = str(source.get("target_type") or "")
target_id = str(source.get("target_id") or "")
permission = str(source.get("permission") or "read")
if not target_type or not target_id:
raise RuntimeError("Campaign schedule share snapshot is incomplete")
session.add(
CampaignShare(
tenant_id=schedule.tenant_id,
campaign_id=generated_campaign.id,
target_type=target_type,
target_id=target_id,
permission=permission,
created_by_user_id=schedule.created_by_user_id,
)
)
def _schedule_snapshot(value: object) -> dict[str, object]:
if not isinstance(value, Mapping) or value.get("schema") != SCHEDULE_SOURCE_SCHEMA:
raise RuntimeError("Campaign schedule source snapshot schema is invalid")
configuration = value.get("configuration")
settings = value.get("campaign_settings")
mail_policy = value.get("mail_profile_policy")
shares = value.get("shares")
if (
not isinstance(configuration, Mapping)
or not isinstance(settings, Mapping)
or not isinstance(mail_policy, Mapping)
or not isinstance(shares, list)
):
raise RuntimeError("Campaign schedule source snapshot is incomplete")
return {
"configuration": dict(configuration),
"campaign_settings": dict(settings),
"mail_profile_policy": dict(mail_policy),
"shares": shares,
}
def _scheduled_external_id(source: str, schedule_id: str, sequence: int) -> str:
suffix = f"-scheduled-{schedule_id[:8]}-{sequence}"
return f"{source[:255 - len(suffix)]}{suffix}"
def _as_utc(value: datetime) -> datetime:
if value.tzinfo is None:
return value.replace(tzinfo=UTC)
return value.astimezone(UTC)
__all__ = [
"RECURRENCE_KINDS",
"SCHEDULE_SOURCE_SCHEMA",
"campaign_schedule_source_snapshot",
"canonical_configuration_hash",
"dispatch_due_campaign_schedules",
"next_schedule_fire",
]