Isolate per-file extraction failures in canvas and edX content syncs - #3752

Closed

mbertrand wants to merge 12 commits into

mainfrom

mattbert/canvas-per-file-fault-isolation

Commits on Aug 10, 2026

Fall back to tika when OCR extraction raises
mbertrand
committed
9615f19
Skip and record files whose extraction raises in process_olx_path

mbertrand
and
claude
committed
a63be45
Retain contentfiles for files whose extraction failed during canvas sync
mbertrand
committed
4baf077
Retain tutor problem files whose extraction failed
mbertrand
committed
4f4bfad
Keep failed-extraction contentfiles published through the staleness pass
mbertrand
committed
358c44e

Commits on Aug 11, 2026

Log OCR-to-tika fallback at warning level

mbertrand
and
claude
committed
3f85471
Treat pypdf-invalid PDFs as extraction failures

mbertrand
and
claude
committed
83d6d1d
Retain edX contentfiles whose extraction failed; don't mark all-failed archives empty

mbertrand
and
claude
committed
062324e
Retry canvas archives where every file failed; add end-to-end failure tests

mbertrand
and
claude
committed
cc8d4c6
Make failure accumulator params keyword-only and document the contract

mbertrand
and
claude
committed
d6131b9
Pin course resource type in edX retention tests

mbertrand
and
claude
committed
a152bbc
Mock search-index plugin hook in edX retention test

mbertrand
and
claude
committed
97cd4e7