trac3r00 Doc Reader
Office files are ZIPs of XML. This reads them with the stdlib, so an agent can work with documents on any machine with zero installs.
Commands
python3 scripts/doc_reader.py report.docx
python3 scripts/doc_reader.py slides.pptx --json
python3 scripts/doc_reader.py data.xlsx | head -50
python3 scripts/doc_reader.py page.html notes.md book.pdf
Supported formats
| Format | How | Notes |
|---|---|---|
.docx |
ZIP + word/document.xml paragraphs | headings preserved as paragraphs |
.pptx |
ZIP + ppt/slides/slideN.xml | per-slide text, slide count |
.xlsx |
ZIP + shared strings + sheet rows | tab-separated rows |
.html/.htm |
html.parser, scripts/styles stripped | clean readable text |
.md/.txt |
raw text | passthrough |
.pdf |
delegates to pdftotext |
honest error if not installed |
Exit codes
0 = all files read, 1 = some failed, 2 = all failed or unsupported format. Corrupt files (bad ZIP, malformed XML) are reported as corrupt, not silently skipped.
Pairs with
resume-audit (convert a PDF resume with doc-reader first, then audit),
seo-audit (read the HTML doc-reader extracts, then score it).