feat(scrape): vision OCR of image-only doc pages (support matrices) (#5)

Co-authored-by: claude <[email protected]>
2026-07-23 09:48:38 -04:00
parent caf562b7c3
commit 698196bd63
10 changed files with 407 additions and 6 deletions
@@ -0,0 +1,6 @@
+# Optional: local vision OCR of image-only doc pages (scrape/vision.py).
+# Installed only in the refresh workflow, NOT baked into the server image —
+# the MCP server never runs OCR, so keeping Pillow out of the base image
+# keeps it lean (same reasoning as requirements-rerank.txt keeping out torch).
+# requests (the Ollama client) is already in requirements.txt.
+Pillow>=10.0