docling · Issues· 906 open

AcceleratorOptions(num_threads=N) on PdfPipelineOptions never reaches the PDF parser

Threaded StandardPdfPipeline: conversion result depends on machine load (timer-based batch assembly)

[Bee] Excel backend duplicates cells of a sparse table as extra tables

harden the LibreOffice profile, flags, and timeout kill

Docx to Markdown, the EMFs and Charts are lost, and get a lot of repeat text in the complex table.

TableFormer (accurate) generates degenerate OTSL at max_steps cap → cell matcher drops table data rows (regression in 2.118.0+)

Threaded docling-parse default (2.123.0, #3764) regresses OCR on rasterised page regions vs docling_parse

Complex tables with page breaks produce hallucitating results

Bug: VLM-generated Markdown with bracketed LaTeX causes crash in document assembly

Setting settings.debug.visualize_cells=True/False causes different table output

[MarkdownTableSerializer] Chunk overflow with landscape tables in non-landscape mode

Markdown parsing

Relative hyperlinks are exported with backslashes on Windows (Markdown, HTML, LaTeX, DocLang, JSON)

Add a page break parameter to the Docling CLI when exporting markdown

How to include page numbers in markdown / Conditional OCR when text layer is missing