Repository Issue Activity (beta)

#267 Expose configurable confidence threshold for OCR routing and document threshold behavior

#297 Raised numeric glyphs are never re-attached: footnote markers lead their cell, exponents corrupt the next one

#296 Rotated column headers are scrambled across cells when the page is not mostly rotated

#266 Document confidence scores and make OCR routing thresholds configurable

#293 Adopt/state policy/licensing for externally-sourced PDFs included as test fixtures

#291 ERROR: arxiv paper title and people name format process error

#286 Valid PDF rejected as `malformed document: invalid PDF structure` when more than 506 trailing bytes follow `%%EOF`

#288 Table row is dropped into prose when its label starts with a footnote marker (is_footnote_row only checks the text before the first ")")

#290 Making V2RAY Proxies

#251 process_pdf_bytes().markdown silently drops content that extract_text_bytes() finds (confidence stays 1.0)

#208 Subset GID remap corrupts CIDFontType0 (CFF) text — Japanese Adobe-Japan1 PDFs extract as unrelated glyphs

#283 Page-number stripping deletes body content on A4 pages (PAGE_NUMBER_TOP_Y is US-Letter-calibrated)

#272 pdf-inspector fails on Chinese z-library PDFs (crash + misclassification + silent empty extraction)

#281 Electricity and magnetism

#270 Ruled table splits into multiple tables when a single cell wraps to more than one line

#247 Document and expose configurable classification thresholds and sampling options for OCR routing

#269 Markdown is always None

#276 Scanned/ImageBased page_signals empty case untested

#271 pdf-inspector is near the top of the star-velocity board at 1,717 stars a day

#212 Arabic (RTL) text is returned in visual order, not logical order

#228 No xref recovery: files that pypdf and pdfium repair raise Invalid PDF structure

#275 test-probe-permission

#227 needs_ocr from extract_pages_markdown() contradicts classify_pdf() on the same page

#220 Feature: distinct-n self-check to catch duplicated output (well-encoded but repeated text)

#219 Two-column pages: columns are detected but markdown is emitted in raster order (lines interleaved)