Quickstart

python generate_docx_submission.py main.tex \
  --outdir submission_build \
  --bibliography-style apa-numbered

./run_submission_build.sh main.tex submission_build --bibliography-style apa-numbered

--bibliography-style apa
APA-like author-date style.

--bibliography-style apa-numbered
APA-like reference formatting, but citations are numbered and references are ordered by first appearance.

--bibliography-style vancouver
Vancouver-like numbered references.

--bibliography-style ama
AMA-like numbered references.

--csl /path/to/style.csl
Overrides the named styles with a custom CSL file.

LaTeX-to-DOCX submission builder

This package creates journal-submission DOCX files from a complete LaTeX source while preserving manuscript text.

It performs four steps:

1. Expands \input{...} and \include{...} files.
2. Extracts table and figure environments.
3. Creates a clean LaTeX manuscript with \ref{...} values hard-coded, floats removed, and figure captions moved to the back.
4. Runs Pandoc or a supplied todocx.sh to create the main Word file, then separately creates a Word document containing parsed Word tables.

Files

• generate_docx_submission.py: main generator.
• run_submission_build.sh: convenience wrapper.
• todocx.sh: optional Pandoc converter supplied by the user.

Basic usage

./run_submission_build.sh main.tex submission_build

or directly:

python generate_docx_submission.py main.tex --outdir submission_build --todocx todocx.sh

Outputs

Inside submission_build/:

• main_expanded_inputs.tex: input-expanded LaTeX source.
• main_clean_refs_hardcoded.tex: main source with floats removed and references hard-coded.
• tables_extracted.tex: raw table environments only, no preamble.
• figure_captions.tex: figure captions only, no preamble.
• figures/: figure files copied from \includegraphics{...} when found.
• manuscript.docx: main Word document if Pandoc succeeds.
• tables.docx: separate Word document with parsed tables.
• manifest.json: labels, captions, figure paths, and conversion status.

Notes

The script does not rewrite, shorten, or JAMA-structure the manuscript text. It only removes floats, moves figure legends to the back, hard-codes cross-references, and converts the resulting document.

The table parser is deliberately conservative. It handles ordinary tabular tables, \hline, \rowcolor, basic formatting commands, \newline, and common math symbols. If a table is too complex, the table block is preserved as raw LaTeX in tables.docx for manual cleanup.

If todocx.sh fails because a custom reference document is not available, the generator falls back to direct Pandoc conversion.