docling-project/docling

Markdown serialization of headings in rich table cells

Open

#2,722 建立於 2025年12月4日

在 GitHub 查看
 (3 留言) (0 反應) (1 負責人)Python (59,751 star) (4,140 fork)batch import
buggood first issuemarkdown

描述

Bug

In docling-core, the MarkdownTableSerializer transforms DoclingDocument headings and titles into markdown headings (i.e., text with preceding # symbols). According to the markdown specs, you can’t use headings, blockquotes, lists, horizontal rules, images, or most HTML tags when formatting text within tables and most applications will not render those headings properly.

Steps to reproduce

Convert Docling's test file table_with_heading_02.html to markdown or check its ground truth file table_with_heading_02.html.md

Docling version

Docling version: 2.64.0 Docling Core version: 2.51.1 Docling IBM Models version: 3.10.2 Docling Parse version: 4.7.1 Python: cpython-313 (3.13.5) Platform: macOS-14.7.1-arm64-arm-64bit-Mach-O

Python version

Python 3.13.5

貢獻者指南

技術棧
python
領域
backenddata
議題類型
bug
難度面向新貢獻者的預計實作難度，1 表示很小改動，5 表示專家級工作。
2
預計時間有經驗貢獻者完成調查、實作、測試並準備 pull request 的粗略時間範圍。
1-3 hours
活動狀態議題目前的可參與程度：新鮮、活躍、陳舊、阻塞或等待維護者輸入。
blocked
清晰度議題是否清楚說明預期改動、驗收標準和下一步。
clear
前置要求
PythonMarkdown
新手友善度1-100 的估計分數，表示該議題對首次貢獻者的友善程度。
30
研究方向
The bug is in `docling core/transforms/serializer/markdown.py` where headings are serialized as `#` prefixed lines even within table cells, violating markdown specs. The issue is assigned and has recent comments; check if the assignee has a branch or plan. The fix likely involves detecting table context and replacing heading markup with plain text or bold inside tables. Reference the markdown spec link provided and the ground truth file to verify correct output.