LaTeX to Word Conversion Tool

简体中文

In daily work, colleagues or supervisors unfamiliar with LaTeX may request Word documents for review and collaboration. This project provides a Python script that uses Pandoc and Pandoc-Crossref to automatically convert LaTeX files into Word documents following a specified format. Although there's no perfect method for converting LaTeX to Word, the output generated by this project meets informal review needs. However, around 5% of the content (such as author information) may require manual corrections post-conversion.

Project GitHub

Features

• Supports formula conversion
• Automatically numbers and cross-references images, tables, formulas, and citations
• Converts multi-figure LaTeX files
• Outputs Word files in a specified format
• Supports Chinese

Examples are shown below; more results are in tests:

Quick Start

Ensure Pandoc and Pandoc-Crossref are correctly installed (see Install Dependencies). Execute the following command in your terminal:

tex2docx convert --input-texfile <your_texfile> --output-docxfile <your_docxfile> 

Replace <...> in the command with the corresponding file path and name.

Install Dependencies

Ensure you have installed Pandoc, Pandoc-Crossref, and related Python libraries.

Pandoc

Install Pandoc as described in the official documentation. It is recommended to download the latest release from Pandoc Releases.

Pandoc-Crossref

Install Pandoc-Crossref by following the official documentation. Ensure compatibility between Pandoc and Pandoc-Crossref and configure the path correctly.

Python Libraries

Install from PyPI:

pip install tex2docx

Usage and Examples

This tool supports both command-line and script-based usage. Ensure the required dependencies are installed.

Command-Line Usage

Run the following command in your terminal:

tex2docx convert --input-texfile <your_texfile> --output-docxfile <your_docxfile> --reference-docfile <your_reference_docfile> --bibfile <your_bibfile> --cslfile <your_cslfile>

Use convert --help to see details on these parameters.

For example, using tests/en:

convert --input-texfile ./tests/en/main.tex --output-docxfile ./tests/en/main_cli.docx --reference-docfile ./my_temp.docx --bibfile ./tests/ref.bib --cslfile ./ieee.csl

This will generate the Word file main_cli.docx in the tests/en directory.

Script Usage

from tex2docx import LatexToWordConverter

config = {
    'input_texfile': '<your_texfile>',
    'output_docxfile': '<your_docxfile>',
    'reference_docfile': '<your_reference_docfile>',
    'cslfile': '<your_cslfile>',
    'bibfile': '<your_bibfile>',
    'fix_table': True,
    'debug': False
}

converter = LatexToWordConverter(**config)
converter.convert()

For more examples, refer to tests/test_integration.py.

FAQ

1. Inconsistent Multi-Figure Layout
   The relative positions of sub-figures may differ between LaTeX compilation and Word conversion, as shown below:

   

   This may result from redefined page size or parameters in the LaTeX file. To address this, adjust the MULTIFIG_TEXFILE_TEMPLATE variable. Below is an example for reference:

   import tex2docx
   
   my_multifig_texfile_template = r"""
   \documentclass[preview,convert,convert={outext=.png,command=\unexpanded{pdftocairo -r 600 -png \infile}}]{standalone}
   \usepackage{graphicx}
   \usepackage{subfig}
   \usepackage{xeCJK}
   \usepackage{geometry}
   \newgeometry{
       top=25.4mm, bottom=33.3mm, left=20mm, right=20mm,
       headsep=10.4mm, headheight=5mm, footskip=7.9mm,
   }
   \graphicspath{{%s}}
   
   \begin{document}
   \thispagestyle{empty}
   %s
   \end{document}
   """
   
   config = {
       'input_texfile': 'tests/en/main.tex',
       'output_docxfile': 'tests/en/main.docx',
       'reference_docfile': 'my_temp.docx',
       'cslfile': 'ieee.csl',
       'bibfile': 'tests/ref.bib',
       'multifig_texfile_template': my_multifig_texfile_template,
   }
   
   converter = tex2docx.LatexToWordConverter(**config)
   converter.convert()
   

2. The Word Output Doesn't Meet Formatting Requirements
   Use Word's style management tools to adjust the styles in my_temp.docx.

Implementation Details

This project relies on Pandoc and Pandoc-Crossref to convert LaTeX files to Word documents. The core command used is:

pandoc texfile -o docxfile \
    --lua-filter resolve_equation_labels.lua \
    --filter pandoc-crossref \
    --reference-doc=temp.docx \
    --number-sections \
    -M autoEqnLabels \
    -M tableEqns \
    -M reference-section-title=Reference \
    --bibliography=ref.bib \
    --citeproc --csl ieee.csl \
    -t docx+native_numbering

1. --lua-filter resolve_equation_labels.lua handles equation numbering and cross-references, inspired by Constantin Ahlmann-Eltze's script.
2. --filter pandoc-crossref handles cross-references for other elements.
3. --reference-doc=my_temp.docx applies the styles from my_temp.docx to the generated Word file. Two template files are included: TIE-temp.docx (for TIE journal submission, double-column format) and my_temp.docx (single-column, designed for easier annotation).
4. --number-sections adds numbering to section headings.
5. -M autoEqnLabels, -M tableEqns enable automatic numbering of equations and tables.
6. -M reference-section-title=Reference adds a section title for references.
7. --bibliography=my_ref.bib generates the bibliography from ref.bib.
8. --citeproc --csl ieee.csl formats citations and the bibliography using the IEEE citation style.
9. -t docx+native_numbering improves captions for images and tables.

The conversion for multi-figure LaTeX content may not be perfect. This project extracts multi-figure code from the LaTeX file and uses the convert and pdftocairo tools to compile the figures into a single large PNG file, replacing the original LaTeX image code and updating references to ensure smooth import into Word.

Known Issues

1. Captions for figures and tables in Chinese still start with "Figure" and "Table".
2. Author information is not fully converted.

Changelog

v1.3.0

1. Major code refactoring: Complete modular restructuring of the codebase

   • Split monolithic tex2docx.py (1139 lines) into 8 specialized modules for better maintainability
   • Introduced clear separation of concerns with dedicated modules for configuration, parsing, conversion, etc.
   • Enhanced type annotations and error handling throughout the codebase

2. Improved testing infrastructure:

   • Renamed and reorganized test files for better clarity:
     • test_tex2docx_refactored.py → test_unit.py (unit tests for individual components)
     • test_tex2docx.py → test_integration.py (end-to-end integration tests)
   • Added comprehensive test documentation in tests/README.md
   • Enhanced pytest configuration with proper markers and test discovery

3. Critical bug fixes:

   • Fixed LaTeX reference line break issue where \ref{} commands were incorrectly split as \nef{}
   • Resolved CLI import errors with proper module structure
   • Enhanced reference numbering accuracy for tables, figures, and equations

4. Developer experience improvements:

   • Better project structure with clear module boundaries
   • Comprehensive documentation updates
   • Cleaner development workflow with organized test suites
   • Preserved all existing functionality while improving code quality

v1.2.4

1. Add support for \include in LaTeX texfiles. (#3)

2. Enhanced the display of figures and tables for better formatting and presentation.

3. Fixed conflicts between cm and varwidth in tables.

4. Resolved conflict issues between subfig and varwidth.

v1.2.3

1. Add feature and option to fix table (issue #2).

v1.2.2

1. fix comments bug #1.

v1.2.1

1. Improved default value settings, including built-in Word style templates and ieee.csl (used as default values).

v1.2.0

1. Fixed module import issues, improving stability.
2. Enhanced the command-line tool for a more intuitive and efficient user experience.
3. Switched to pyproject.toml for dependency management, replacing setup.py.
4. Released on PyPI; users can install via pip install tex2docx.

Miscellaneous

There are two kinds of people: those who use LaTeX and those who don't. The latter often ask the former for a Word version. Hence, the following command:

pandoc input.tex -o output.docx\
  --filter pandoc-crossref \
  --reference-doc=my_temp.docx \
  --number-sections \
  -M autoEqnLabels -M tableEqns \
  -M reference-section-title=Reference \
  --bibliography=my_ref.bib \
  --citeproc --csl ieee.csl \
  -t docx+native_numbering