DOCX math: degHide is ignored and a hidden radical degree is emitted

Aperta
  • Qualcosa non va?
#2,806 0 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 1 giorno

@feiiiiii5 ci sta già lavorando.

Dal 1/11/2026.

  • #2807 di @feiiiiii5 — aperta
  • #2814 di @phanipaladugula — aperta

Valutazione

Difficoltà
2/5
Tempo stimato
1-3 ore
Idoneità per principianti
30/100
Tipo di issue
Bug
Chiarezza
Specificata chiaramente
Stato di attività
Ferma
Stack tecnologico
python
Ambito
backend

Direzione di ricerca

Start with do_rad in packages/markitdown/src/markitdown/converter_utils/docx/math/omml.py, which picks the indexed root only from whether the degree text is nonempty, and with the radical properties parser that ignores degHide. Run the reproduction snippet from the issue to confirm $\sqrt[3]{x}$ is emitted. Linked PR #2807 is already open for this, so review it before starting. Done means a hidden degree gives \sqrt{x}, the val and bare-element defaults are honored, and visible degrees are unchanged.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

Problem

When an OMML radical retains a nonempty m:deg but marks it hidden with m:radPr/m:degHide, DOCX conversion still emits the degree. For example, a radical whose degree 3 is hidden becomes $\sqrt[3]{x}$ instead of $\sqrt{x}$.

This changes the displayed mathematical expression in the Markdown consumed by downstream readers or LLMs. Conversion succeeds without a warning.

Reproduced on current main at 7482c092, reporting version 0.1.9. I searched existing issues and PRs for degHide, hidden degree, radical, and square root and did not find this case.

Reproduction

From a checkout of that commit:

pip install -e "./packages/markitdown[docx]" python-docx

The following creates a synthetic DOCX in memory and uses the public conversion API. No LLM or external service is involved.

import io

from docx import Document
from docx.oxml import parse_xml
from markitdown import MarkItDown, StreamInfo

document = Document()
document.add_paragraph()._p.append(parse_xml("""
<m:oMath xmlns:m="http://schemas.openxmlformats.org/officeDocument/2006/math">
  <m:rad>
    <m:radPr><m:degHide m:val="1"/></m:radPr>
    <m:deg><m:r><m:t>3</m:t></m:r></m:deg>
    <m:e><m:r><m:t>x</m:t></m:r></m:e>
  </m:rad>
</m:oMath>
"""))
stream = io.BytesIO()
document.save(stream)
stream.seek(0)

result = MarkItDown(enable_plugins=False).convert_stream(
    stream, stream_info=StreamInfo(extension=".docx")
)
print(result.markdown)
# Actual:   $\sqrt[3]{x}$
# Expected: $\sqrt{x}$

Expected behavior and controls

Microsoft's HideDegree documentation specifies that degHide=1/true hides the radical degree. A present degHide with no val also means enabled; an absent element means disabled.

I checked the following through both oMath2Latex and serialized DOCX conversion, keeping the stored degree 3 and radicand x constant:

Radical properties Expected LaTeX Actual
<m:degHide m:val="1"/> \sqrt{x} \sqrt[3]{x}
<m:degHide m:val="true"/> \sqrt{x} \sqrt[3]{x}
<m:degHide/> \sqrt{x} \sqrt[3]{x}
<m:degHide m:val="0"/> \sqrt[3]{x} correct
<m:degHide m:val="false"/> \sqrt[3]{x} correct
<m:radPr/> \sqrt[3]{x} correct
no m:radPr \sqrt[3]{x} correct

The first five rows place the listed element inside m:radPr.

Likely cause

oMath2Latex.do_rad chooses an indexed root solely from whether the converted degree text is nonempty. It never checks degHide; the properties parser does not read that flag either.

The radical handler should honor the visibility flag and its on/off defaults while preserving the existing visible-degree behavior.

Environment

  • macOS 26.4, Apple Silicon; Python 3.13.3
  • markitdown 0.1.9 from the commit above
  • python-docx 1.2.0, mammoth 1.13.0, lxml 6.1.3
  • beautifulsoup4 4.15.0, markdownify 1.2.3
Lingua principale
Python
Stelle
194k
Fork
14.5k
Merge medio
24m
PR unite (30g)
7

Preparare l'ambiente

Apri in Codespaces

Avvia il container di sviluppo del progetto nel browser, con il tuo account GitHub.

  • Include un Dockerfile o un file Docker Compose
  • Nessun modello di pull request
  • Nessuna guida per i contributori

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di microsoft/markitdown

Tutte le issue di microsoft/markitdown

Issue simili

Altre issue su Python

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.