Skip to content

Add OCR fallback for scanned PDFs - #34

Merged
hashwnath merged 2 commits into
mainfrom
feat/ocr-fallback-scanned-pdfs
Oct 18, 2026
Merged

hashwnath merged 2 commits into
mainfrom
feat/ocr-fallback-scanned-pdfs

Conversation

@hashwnath

Copy link
Copy Markdown
Member

Scanned PDFs (image-only, no text layer) were returning empty chunks because
page.get_text("text") only works on digitally embedded text.

Now when a page yields fewer than 20 characters, we try PyMuPDF's built-in OCR
via get_textpage_ocr(). This uses Tesseract under the hood but doesn't add
any new Python dependencies. If Tesseract isn't installed, it logs a warning
and moves on gracefully.

Tested with a mix of digital and scanned PDFs locally.

Closes #17

hashwnath and others added 2 commits October 17, 2026 20:35
Pages with little or no extractable text now fall back to
PyMuPDF's built-in OCR (get_textpage_ocr). Requires Tesseract
on the system but adds no new Python deps. Gracefully skips
if Tesseract isn't installed.

Closes #17

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@hashwnath
hashwnath merged commit fdd7185 into main Oct 18, 2026
7 checks passed
@hashwnath
hashwnath deleted the feat/ocr-fallback-scanned-pdfs branch October 18, 2026 03:52
hashwnath added a commit that referenced this pull request Oct 18, 2026
Pages with little or no extractable text now fall back to
PyMuPDF's built-in OCR (get_textpage_ocr). Requires Tesseract
on the system but adds no new Python deps. Gracefully skips
if Tesseract isn't installed.

Closes #17
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Add OCR fallback for scanned PDFs

1 participant