Follow-up to #39 (landed on develop via #65).
Articles captured before the #39 structured path existed are stored as text-only sources. They will NOT auto-upgrade to structured blocks on a normal re-run, because _needs_x_fetch treats an item that already has an x_article source as satisfied — the text-only → structured upgrade is --force-gated by the caching contract (an idempotent re-fetch intentionally produces no churn).
Do
- Run a backlog pass over already-captured Articles with
xbrain fetch-x --force so their bodies are re-fetched as structured blocks, then run media (inline-image download) and enrich/topics so the upgraded bodies propagate.
- Consider a scoped
--force (by date window / by item) so the pass doesn't needlessly re-fetch everything.
Depends on
Should run after the real-payload validation (the other #39 follow-up) — no point forcing a backlog re-fetch through an unvalidated parser.
Refs: #39, #65.
Follow-up to #39 (landed on
developvia #65).Articles captured before the #39 structured path existed are stored as text-only sources. They will NOT auto-upgrade to structured
blockson a normal re-run, because_needs_x_fetchtreats an item that already has anx_articlesource as satisfied — the text-only → structured upgrade is--force-gated by the caching contract (an idempotent re-fetch intentionally produces no churn).Do
xbrain fetch-x --forceso their bodies are re-fetched as structuredblocks, then runmedia(inline-image download) andenrich/topicsso the upgraded bodies propagate.--force(by date window / by item) so the pass doesn't needlessly re-fetch everything.Depends on
Should run after the real-payload validation (the other #39 follow-up) — no point forcing a backlog re-fetch through an unvalidated parser.
Refs: #39, #65.