feat: expand team-export surface to remaining team models

[SmittieC]

Product Description

no change — internal team-export tooling.

Technical Description

Follow-up to the export_apis branch: expands the team-export manifest from a handful of core models to the remaining team-owned surface (chats/messages, traces, evaluations, human annotations, analyses, notifications, cost tracking, custom actions, documents/files, events, generic-tagging).

Points worth a reviewer's attention:

• Manifest ordering is load-bearing. Entries are topologically ordered so every FK target is imported before its referrers. Generic-FK models (customtaggeditem, usercomment, score) are placed last because their content_object can point at many of the preceding models; usage_records follow pricing_rules.
• Generic FKs are now handled end-to-end: the serializer emits the content type as an app_label.model label (stable across servers) instead of the source ContentType pk, and the importer resolves the type by natural key and translates the object id through the FK store. Both sides read the GFK column names off the field itself (generic_fk_fields) rather than hardcoding. Coverage is guarded by test_generic_fk_target_models_are_synced.
• Multi-path team scoping. TEAM_PATH_REGISTRY values may now be a list of ORM paths (OR'd together, distinct() applied) for models reachable via more than one relation — e.g. EventAction has no team FK and is scoped through either of its two trigger one-to-ones.
• EXTRA_FILTERS adds post-team-scoping row exclusions; used to drop assistant-attached CustomActionOperation rows (assistants aren't synced) so the model's check constraint holds on import.
• New serializer field types for columns the default DRF mapping can't round-trip: pgvector columns → list of floats, FileField/ImageField → stored relative path (not /media/ URL, which Django's safe-join rejects on assign), and django_pydantic_field schema fields → plain JSON.
• Through-M2M exclusions: Collection.files / DocumentSource.files are excluded from serialization (their CollectionFile through rows carry the link) since .set() is rejected for explicit through models.
• annotationqueueaggregate matches an existing target row (recreated as a side effect of importing annotations) rather than creating a second, which the queue one-to-one forbids.

export.yml (the generated OpenAPI schema) is regenerated to match.

Migrations

N/A — no migrations.

Docs and Changelog

• This PR requires docs/changelog update

[coderabbitai]

📝 Walkthrough

Walkthrough

This PR extends the team data sync export/import system with generic foreign key support end-to-end. The export serializer gains custom field classes for file paths, pgvector values, and pydantic schemas, plus a factory that emits content-type labels (app_label.model) instead of raw PKs for generic FK fields. The sync manifest is expanded with new model entries, multi-path team scoping, and a generic_fk_fields() introspection helper. The importer resolves generic FK content types and object IDs during row import, adds queue-aggregate deduplication, and backfills ExperimentSession.last_activity_at after chat message import. New test factories cover all newly supported models, and tests validate each new behavior including a full manifest round-trip smoke test.

Estimated code review effort

🎯 5 (Critical) | ⏱️ ~120 minutes

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 62.22% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly summarizes the main change: expanding the team-export surface to remaining team models.
Description check	✅ Passed	The description matches the template and covers product, technical details, migrations, and changelog notes.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 2

🧹 Nitpick comments (1)

apps/utils/factories/evaluations.py (1)

199-199: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Drop the redundant computed_at factory value.
EvaluationRunAggregate.computed_at is auto_now_add=True, so factory.LazyFunction(timezone.now) is overwritten on create and only adds noise. If a deterministic timestamp is needed for tests, set it after creation with update() or a post-generation hook.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@apps/utils/factories/evaluations.py` at line 199, Remove the redundant
computed_at assignment from the evaluation factory so EvaluationRunAggregate
relies on its auto_now_add behavior instead of
factory.LazyFunction(timezone.now). Update the factory definition in the
evaluations factory class that builds EvaluationRunAggregate instances, and if
tests need a fixed timestamp, apply it after creation with an update call or a
post-generation hook rather than setting it during factory construction.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@apps/teams/export/importer.py`:
- Around line 277-281: The generic-FK handling in importer logic is silently
accepting missing translations by assigning None from
FKTranslationStore.get_target(), which can leave unresolved relations behind.
Update the gfk_pairs loop in the importer method that builds field_values so it
fails fast the same way regular FK resolution does: detect when get_target()
returns no translated target for the type_label and fk field, and raise the
unresolved-FK error instead of storing None. Keep the fix localized to the
generic relation translation path so ContentType lookup and valid target
assignment still behave the same.

In `@apps/teams/export/manifest.py`:
- Around line 81-82: The export manifest ordering in Manifest needs to be
adjusted so evaluation_datasets is processed before evaluation_messages. Update
the ManifestEntry sequence in manifest.py to place the evaluation_datasets entry
ahead of the evaluation_messages entry, since EvaluationMessage depends on
evaluationdataset__team and must not be imported before its related dataset
exists.

---

Nitpick comments:
In `@apps/utils/factories/evaluations.py`:
- Line 199: Remove the redundant computed_at assignment from the evaluation
factory so EvaluationRunAggregate relies on its auto_now_add behavior instead of
factory.LazyFunction(timezone.now). Update the factory definition in the
evaluations factory class that builds EvaluationRunAggregate instances, and if
tests need a fixed timestamp, apply it after creation with an update call or a
post-generation hook rather than setting it during factory construction.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 4b433f6d-f188-4e0c-b93b-e66fa8a50eed

📥 Commits

Reviewing files that changed from the base of the PR and between 5ae36f1 and 38fd0cd.

📒 Files selected for processing (19)

• api-schemas/export.yml
• apps/api/export/serializers.py
• apps/api/export/tests/test_serializers.py
• apps/api/export/tests/test_views.py
• apps/teams/export/importer.py
• apps/teams/export/manifest.py
• apps/teams/export/tests/test_importer.py
• apps/teams/export/tests/test_manifest.py
• apps/teams/management/commands/sync_team.py
• apps/utils/factories/analysis.py
• apps/utils/factories/annotations.py
• apps/utils/factories/custom_actions.py
• apps/utils/factories/documents.py
• apps/utils/factories/evaluations.py
• apps/utils/factories/experiment.py
• apps/utils/factories/files.py
• apps/utils/factories/human_annotations.py
• apps/utils/factories/notifications.py
• open-chat-studio.cs-export_apis

[codescene-delta-analysis]

Gates Failed
Enforce advisory code health rules (1 file with Complex Method, Complex Conditional)

Our agent can fix these. Install it.

Gates Passed
3 Quality Gates Passed

Reason for failure

Enforce advisory code health rules	Violations	Code Health Impact
test_importer.py	2 advisory rules	10.00 → 9.39	Suppress

See analysis details in CodeScene

Quality Gate Profile: Clean Code Collective
Install CodeScene MCP: safeguard and uplift AI-generated code. Catch issues early with our IDE extension and CLI tool.

[snopoke]

LGTM

[snopoke]

Does this work when using S3 storage?

[SmittieC]

Yes

[claude]

Code review

No issues found. Checked for bugs and CLAUDE.md compliance.