create-annotated-pdf

Generate annotated PDFs showing what the extraction pipeline detected. Color-coded bounding boxes for text blocks, tables, figures, equations overlaid on the original PDF.

Quick Start

Single PDF (from run directory)

cd /path/to/extractor
uv run python -m extractor.pipeline.tools.render_annotated_pdf from-run \
  --pdf /path/to/original.pdf \
  --run-dir /path/to/run_directory \
  --out /tmp/annotated.pdf \
  --export-pages --png-dpi 144

Single PDF (by stem — auto-discovers run dir)