PDF Extract

Fast, structured PDF text extraction. Superior to summarize.sh for PDFs because it:

• Supports page selection (--pages 1,3-5)
• Uses pdfplumber for structured layouts and tables (default)
• Uses PyMuPDF for fast raw text extraction
• Outputs clean, page-delimited text suitable for analysis

When to use

Use this skill immediately when the user asks any of:

• "extract text from this PDF"
• "read/analyze/summarize this PDF file"
• "what's in this PDF?"
• "get content from these PDF pages"
• Any task involving local PDF file analysis

Quick start

cd ~/.openclaw/workspace/pdf-tools
source bin/activate 2>/dev/null || true

# Extract all text (pdfplumber - best for tables/structure)
./pdf_extract.py /path/to/file.pdf

# Extract specific pages
./pdf_extract.py /path/to/file.pdf --pages 1,3-5

# Fast extraction (PyMuPDF - best for speed/simple text)
./pdf_extract.py /path/to/file.pdf --method pymupdf

# Combine: fast extract of specific pages
./pdf_extract.py /path/to/file.pdf --method pymupdf --pages 1-10

Methods

Page selection syntax

• --pages 1 — page 1 only
• --pages 1,3,5 — specific pages
• --pages 1-5 — range (inclusive)
• --pages 1,3-5,10 — mixed

Integration with LLM analysis

Extract first, then analyze:

# Extract to file for further processing
./pdf_extract.py paper.pdf --pages 1-3 > /tmp/extracted.txt
# Then read the file and ask questions about it

Or pipe directly for quick checks:

./pdf_extract.py paper.pdf --pages 1 | head -50

Error handling

• File not found: Check path, try absolute path
• Empty output: PDF may be scanned image (no text layer) — use OCR tools instead
• Import errors: Ensure virtualenv is activated: source ~/.openclaw/workspace/pdf-tools/bin/activate

Dependencies

Located in ~/.openclaw/workspace/pdf-tools/:

• pdfplumber — Python PDF parsing
• pymupdf (fitz) — Fast PDF rendering
• Virtual environment pre-configured

Compared to summarize.sh

Workflow: Use pdf_extract.py to extract precisely what you need, then use the model directly for analysis/summarization.