Extract Overleaf Comments

Extract visible comments from a Chrome-saved Overleaf page and optionally map Overleaf character offsets (data-pos) back to line numbers in a .tex file.

This is useful because Overleaf review-panel comments are not included in the downloaded LaTeX source archive.

Install

The extractor uses only the Python standard library.

From PyPI, once released:

pip install extract-overleaf-comments

From a local checkout:

python3 -m pip install .

Then run:

extract-overleaf-comments --help

Development

python3 -m pip install -e ".[test]"
python3 -m pytest tests
python3 -m build
python3 -m twine check dist/*

Usage

1. Open the Overleaf project.
2. Open the review/comments panel.
3. Use the browser's "Save page as..." feature and save the complete page.
4. Zip the saved .html file and its companion _files/ directory, or pass a zip archive created by the browser.
5. Run:

python3 src/overleaf_comment_extractor.py Archive.zip --tex main.tex --out-prefix comments

The command writes:

• comments.csv
• comments.md

To also create a compilable annotated LaTeX copy:

python3 src/overleaf_comment_extractor.py Archive.zip \
  --tex main.tex \
  --out-prefix comments \
  --comment-tex

This writes main_comments.tex by default. The annotated file inserts red thread markers such as [T001] near the Overleaf character offsets and adds a final Extracted Overleaf Comments section with the full comment text.

To put comments directly in the PDF margin instead:

python3 src/overleaf_comment_extractor.py Archive.zip \
  --tex main.tex \
  --out-prefix comments \
  --comment-tex \
  --comment-placement margin

The margin mode groups all replies from one thread into one \marginpar note in \footnotesize text. The default is --comment-placement appendix.

By default, this annotated copy is made standalone: the original document class and package list are replaced by a minimal article setup, and missing figures are rendered as placeholders. This makes the review PDF compile even when the full private Overleaf project is not available locally. Use --preserve-comment-tex-preamble if you want to keep the original LaTeX class and package setup.

Example

The repository contains a fake HTML snapshot and fake .tex file only. It does not contain private Overleaf projects or real manuscripts.

cd examples/fake_overleaf_save
zip -r ../fake_overleaf_save.zip .
cd ../..
python3 src/overleaf_comment_extractor.py examples/fake_overleaf_save.zip \
  --tex examples/fake_project.tex \
  --out-prefix examples/fake_comments \
  --comment-tex examples/fake_project_comments.tex

Release

PyPI publishing is configured through GitHub Actions Trusted Publishing. To publish a release:

1. Create a pending publisher on PyPI for:
   • PyPI project: extract-overleaf-comments
   • Owner: adakite
   • Repository: extract-overleaf-comments
   • Workflow: publish-pypi.yml
   • Environment: pypi
2. Tag a version and push the tag:

git tag v0.1.0
git push origin v0.1.0

Limitations

This tool parses comments that are present in the saved browser DOM. If Overleaf has not loaded a thread, or if a collapsed/truncated comment is not present in the DOM, the extractor cannot recover it from the HTML snapshot. For a more complete capture, make sure the review panel is open and the relevant comments are loaded before saving the page.