Classic298 e9cb194f46 fix: large spreadsheet upload takes minutes and stalls the server (#33770) ...

Uploading a big, sparsely filled .xlsx sent the document parser into a very slow table-detection step; a 28,000-row sheet took around 13 minutes, long enough for health checks to fail and pods to be restarted. Spreadsheets are now read as a whole sheet without that step, which took a 2,500-row sheet from 17.6 s to about 1 s with identical text. Sheets that hold several separate tables on one sheet now come out as one block instead of one block per table.

Fixes #33768

e9cb194f46 · 2026-11-05 21:47:16 +09:00

History

..

datalab_marker.py

…

external_document.py

…

external_web.py

…

local.py

…

main.py

fix: large spreadsheet upload takes minutes and stalls the server (#33770)

2026-11-05 21:47:16 +09:00

microsoft_web_iq.py

…

mineru.py

feat: support MinerU 4.1 servers (#33719)

2026-11-04 08:50:28 +09:00

mistral.py

refac

2026-11-02 08:48:51 +09:00

paddleocr_vl.py

…

pdf.py

refac

2026-10-25 07:30:34 +09:00

tavily.py

…

youtube.py

…