feat: Add PDF comparison support

[houfu]

Summary

• Adds PDFFile document class for comparing text-based PDF files
• Uses pdfplumber library for text extraction (MIT license, supports word-level bounding boxes for future chunk comparison features)
• PDF support is an optional dependency: pip install redlines[pdf]
• CLI auto-detects .pdf files and handles them automatically

Implementation Notes

This is a text dump implementation - extracts plain text only using page.extract_text():

• ✓ Works well for simple PDFs (invoices, reports, single-column documents)
• ⚠️ Limited for complex legal contracts (multi-column layouts, tables, headers/footers are flattened)
• Future enhancements tracked in Enhanced PDF support: Layout-aware extraction and structure preservation #84 for layout-aware extraction

Usage

from redlines import Redlines
from redlines.pdf import PDFFile

source = PDFFile("contract_v1.pdf")
test = PDFFile("contract_v2.pdf")

redline = Redlines(source, test)
print(redline.output_markdown)

# CLI usage - auto-detects PDF files
redlines json doc_v1.pdf doc_v2.pdf --pretty

Features

• PDFFile.text - Extracted text from all pages
• PDFFile.pages - List of PDFPage objects with page_number and text
• PDFFile.page_count - Number of pages
• preserve_pages=True option to insert [Page N] markers in text

Test plan

• Run pytest tests/test_pdf.py -v - 9 tests pass
• Run full test suite pytest tests/ - 123 pass, 12 skipped (nupunkt)
• Run mypy redlines/ - no issues
• Test CLI with PDF files manually

Closes #1

🤖 Generated with Claude Code