Repository Issue Activity (beta)

#4049 Bug: HTML tables are missing from DoclingDocument when parsing arXiv/LaTeXML HTML

#2508 Markdown with embedded coordinates

#4030 Unlimited-OCR preset uses DeepSeek grounding prompt instead of official '<image>document parsing.'

#3583 Improve parsing of JATS documents

#3939 Document.save_as_json method does not expose json module-level options making the conversion very strict

#4058 Vector-dense pages (CAD/schematic path storms): word-cell decode runs unconditionally in preprocessing and costs GiBs per page — request config surface to bound/skip it

#4025 Complete the parsing of iWork Pages

#3495 bug: borderless table detected as both TableItem and PictureItem (dead code in _handle_cross_type_overlaps)

#4028 TableFormer ignores drawn horizontal rules, mis-pairing rows when a cell wraps over several lines

#4022 TesseractOcrCliModel lang=["auto"] silently ignores detected script on Windows (POSIX-only "script/" prefix check)

#1762 `_guess_format` method fails on multi-byte characters in UTF-8 documents

#3901 Extract avatars from pdf page scan

#4053 Invisible text reaches the exported document: `_visible_text_cells` is only wired to the OCR path, not to `get_segmented_page()`

#4043 Hyphen-minus stripped when a CLI flag (-flag) wraps across a line break during PDF text extraction

#2330 Docling misses major chunks in HTML parsing

#2620 [html] Docling produces empty Markdown file when parsing HTML document

#2528 Docling-serve Fails to Utilize GPU Despite Available CUDA Provider

#3685 [Bee] Complex Excel layouts result in duplicate tables when converted to HTML

#2584 Make RapidOcrOptions ignore `artifacts_path`

#4018 [Bee] docling parse generates text as each character separated

#4016 PDF: wrapped section_header titles split across two elements — SECTION_HEADER missing from predict_merges (companion to #3881)

#4034 DOCX textboxes are silently dropped and the result differs run to run: id() of transient lxml proxies used as node identity

#4007 pypdfium2 backend: text-cell coordinates ignore /Rotate while page size and rendering apply it

#4002 VlmPipeline (MLX): ValidationError — VlmPredictionToken.logprob receives bfloat16 array instead of float

#4027 Support the parsing of Apple's Keynotes