A toolset for cleaning LaTeX projects for arXiv/camera-ready submissions and verifying the resulting PDFs.
This repository contains two main tools:
-
LaTeX Paper Cleaner (
latex_cleaner.py): Creates a clean version of your LaTeX project by:- Only keeping necessary TeX files that are actually included in the main document
- Only keeping figures that are actually used in the paper
- Removing all comments from TeX files
- Only keeping bibliography entries that are actually cited in the paper
-
PDF Comparison Tool (
compare_pdfs.py): Verifies that your cleaned project compiles to the same PDF as the original by:- Automatically compiling both the original and cleaned versions
- Comparing PDF content using multiple methods (hash, text, visual)
- Identifying specific differences when they exist
- Python 3.6+
- LaTeX distribution (TeX Live, MiKTeX, etc.)
- Python dependencies:
PyPDF2,pytesseract,pdf2image,Pillow,numpy
pip install PyPDF2 pytesseract pdf2image Pillow numpyFor PDF to image conversion, you also need to install Poppler:
- Ubuntu/Debian:
apt-get install poppler-utils - macOS:
brew install poppler - Windows: Download from here
For OCR capabilities, you need Tesseract:
- Ubuntu/Debian:
apt-get install tesseract-ocr - macOS:
brew install tesseract - Windows: Download from here
python latex_cleaner.py --source_dir /path/to/original/project --output_dir /path/to/cleaned/project --main_tex main.texOptions:
--source_diror-s: Source LaTeX project directory (required)--output_diror-o: Output directory for cleaned project (default: "./cleaned_project")--main_texor-m: Main TeX file (relative to source_dir, will try to detect automatically if not specified)--verboseor-v: Enable verbose output (print all dependencies)
python compare_pdfs.py --original /path/to/original/project --cleaned /path/to/cleaned/project --main-tex main.texOptions:
--originalor-o: Original LaTeX project directory (required)--cleanedor-c: Cleaned LaTeX project directory (required)--main-texor-m: Main TeX file (relative to project directories, required)--visual: Perform visual comparison (slower but more accurate)--verboseor-v: Show detailed differences
- Recursively tracks dependencies from the main TeX file
- Identifies and keeps only necessary files
- Removes all comments while preserving document structure
- Creates a clean dependency tree visualization
- Preserves both image and bibliography references
- Automatic compilation of both original and cleaned projects
- Multiple verification methods:
- Hash comparison (fastest, exact match required)
- Text content comparison (robust to metadata changes)
- Visual comparison (catches rendering differences)
- Detailed reporting of differences when found
# Clean a LaTeX project
python latex_cleaner.py -s ~/projects/my_paper -o ./cleaned_paper -m main.tex
# Verify the cleaned project produces the same PDF
python compare_pdfs.py -o ~/projects/my_paper -c ./cleaned_paper -m main.tex --visualThis project is licensed under the MIT License - see the LICENSE file for details.