pdf-parsinglisted

# PDF parsing Turns a PDF into data you can actually compute on — text, tables, form fields — and writes it out as a spreadsheet or a document, choosing the target from what the PDF really contains. ## The one idea that organises everything below **A PDF is not a data format. It is a page-description format**: instructions for painting glyphs at coordinates. There is no "table" inside a PDF, and often no "text" either — there are marks positioned on a canvas, which a human eye groups into rows and columns. Everything downstream follows from that. It means extraction quality is decided almost entirely by **correctly identifying what kind of PDF you have before you touch it**, because the four main kinds need four different tools and using the wrong one does not raise an error — it returns nothing, or something subtly scrambled, which is worse. The failure this skill exists to prevent: point a text extractor at a scanned document and it returns an empty string. Not an exception. The natural next thought is "the file is corrupt" or "this library is broken", and ten minutes disappear into debugging the wrong thing. The file was fine. It contained no text at all. ## Workflow ### 1. Triage first, always ```bash python3 scripts/pdf_triage.py file.pdf python3 scripts/pdf_triage.py folder/ --batch ``` Standard library only — nothing to install. It reads the PDF's own object structure, inflates the content streams, and reports what the document actually is plus the route that