fix: classify publication origins explicitly

[Ciprian-LocalPulse]

Problem

Persisted publication responses previously exposed a fixed synthetic: false value. That made CI seed rows, manually shaped rows, and legacy records look equivalent to publications retrieved through an implemented provider path.

Change

• Add an explicit publication origin model: unknown, manual, provider, and synthetic.
• Normalize legacy rows conservatively as unknown, while SYN-*, SEED-*, and explicit synthetic payloads remain synthetic.
• Mark ordinary built-in PubMed search results as provider-origin before persistence, while parser-only and injected-transport paths remain unknown.
• Return origin and tri-state synthetic fields in publication list, detail, and history payloads.
• Label the CI seed payload as synthetic and document the API semantics.

Verification

• python -m ruff check --no-cache .
• python -m pytest -p no:cacheprovider -q
• python -m pytest --cov=openlongevity --cov-report=term-missing
• python scripts/audit_documentation.py --json
• git diff --check
• python -m compileall -q src

Local PostgreSQL integration tests are skipped when TEST_DATABASE_URL is absent; GitHub CI runs them against the workflow PostgreSQL service.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
openlongevity	Ready	Preview	Sep 22, 2026 11:51am UTC