LaTeX to ePub Converter

Convert LaTeX academic papers to high-quality ePub format so you can read them on a mobile device. Uses LaTeXML processing with AI-enhanced bibliography formatting.

🚀 Quick Start

# Install dependencies
pip install -r requirements.txt

# Convert LaTeX to ePub (two-step process)
python3 latex_to_xml.py path/to/paper.tex
python3 xml_to_epub.py output/paper.xml

# Output: epub_books/paper.epub

📋 Requirements

System Dependencies

# LaTeXML (Ubuntu/Debian)
sudo apt-get install latexml

# LaTeXML (macOS)
brew install latexml

# AWS CLI for cognitive processing
aws configure

Python Dependencies

pip install -r requirements.txt

🎯 Production Architecture

LaTeX Source → LaTeX Expansion → LaTeXML → Cognitive Enhancement → XSLT → Professional ePub

Core Files

• latex_to_xml.py - LaTeXML + Bedrock cognitive converter
• xml_to_epub.py - XSLT-based ePub generator
• xml_to_epub.xsl - High quality LaTeXML stylesheet
• review_structure.py - Quality assessment module

✨ Features

Professional-Quality LaTeX Processing

• ✅ Multi-file support - Automatic \input{}/\include{} expansion
• ✅ Industry-standard conversion - LaTeXML engine with MathML
• ✅ Excellent table rendering - Complex academic tables preserved
• ✅ Image handling - Automatic figure processing and embedding

AI-Enhanced Bibliography

• ✅ Cognitive processing - Amazon Bedrock Claude Sonnet handles complex, variables formats
• ✅ Smart citation formatting - Converts to [AuthorYear] format
• ✅ Full author preservation - Complete citation details
• ✅ Clean text flow - Flattened citations, no line breaks

Quality Assurance

• ✅ LaTeX cleanup - Removes artifacts like \xspace
• ✅ Comprehensive assessment - LaTeXML-adapted quality scoring
• ✅ Professional output - Mobile-compatible ePub format
• ✅ Validation - Schema-compliant XML generation

🔧 Advanced Usage

Directory Processing

# Automatically detects main LaTeX file
python3 latex_to_xml.py path/to/latex/directory/

Quality Assessment

The converter provides detailed quality metrics:

• Metadata preservation (title, authors, abstract)
• Structure conversion (sections, equations, tables, figures)
• Bibliography processing (citation formatting, reference completeness)
• LaTeXML quality indicators (XML structure, namespacing, content preservation)

Caching

• LaTeX expansion - Cached with timestamp validation
• Bedrock calls - Automatic response caching for efficiency

📁 Project Structure

~/workspace/
├── latex_to_xml.py              # Main LaTeX converter
├── xml_to_epub.py               # ePub generator
├── xml_to_epub.xsl              # XSLT stylesheet
├── review_structure.py          # Quality assessment
├── requirements.txt             # Dependencies
├── output/                      # Generated XML files
├── epub_books/                  # Generated ePub files
└── approaches/                  # Historical development (see below)

📚 Historical Development

The approaches/ directory contains the research and development history of this converter:

approaches/xml_to_epub/

• latex_to_xml_rules_based.py - Pure Python regex-based parsing
• latex_to_xml_cognitive.py - Amazon Q Developer CLI approach
• latex_to_xml_hybrid.py - TexSoup + pylatexenc + LLM combination
• latex_to_xml_latexml.py - Original LaTeXML implementation
• xml_to_epub_latexml.py - Original ePub generator
• structural_review/ - Quality assessment framework
• tools/ - Development utilities and testing tools

Evolution Summary

1. Rules-Based (99.8% quality) - Reliable but limited LaTeX cleaning
2. Cognitive (experimental) - Q CLI integration challenges
3. Hybrid (100% when working) - Best architecture but XML namespace issues
4. LaTeXML (96-100% quality) - Production choice - Professional processing with cognitive enhancements

The production version represents the culmination of this research, combining LaTeXML's professional LaTeX processing with Bedrock's cognitive bibliography enhancement.

🛠 Troubleshooting

Common Issues

# Check LaTeXML installation
latexml --version

# Verify AWS credentials
aws sts get-caller-identity

# Test dependencies
python3 -c "import lxml.etree, boto3; print('Dependencies OK')"

File Organization

• Generated XML files: output/
• Generated ePub files: epub_books/
• Expanded LaTeX files: *_expanded.tex (cached)

🎉 Production Ready

This converter produces publication-quality ePub files suitable for:

• ✅ Academic publishing - Professional formatting and citations
• ✅ Mobile reading - Compatible with all ePub readers
• ✅ Archival storage - Standards-compliant format
• ✅ Distribution - Optimized file sizes with complete content

The LaTeXML + Bedrock cognitive approach delivers consistent, high-quality results for academic document conversion.