# Core packages for crawl.py and build_sheet.py. All small.
requests>=2.28
beautifulsoup4>=4.11
openpyxl>=3.1

# Optional. Each line improves one step. Uncomment what you want.
# trafilatura>=1.6            # cleaner main-content extraction in crawl.py
# lxml>=4.9                   # faster HTML parsing in crawl.py
# scikit-learn>=1.2           # TF-IDF similarity backend in match.py
# google-auth>=2.20           # needed by gsc_pages.py
# sentence-transformers>=2.2  # bge embeddings in match.py. Pulls in PyTorch, which is large.

# match.py runs with no packages at all. It falls back to pure-Python token overlap.
