Tables out of PDFs, typed like a spreadsheet — not pasted like a picture. Load a text-layer PDF and this extracts its tables, types numeric columns as real values (currency and counts, thousands separators removed, so totals and sorts work), and runs a totals spot-check: every numeric column is re-summed from the raw extracted text and compared against the typed sheet. A page with no text layer — a scan — is flagged OCR REQUIRED and excluded. It is never guessed at.
The sample is the studio's published 14-row price-list PDF — extraction reproduces all 14 rows and the totals check passes (Price 289.55 · Stock 5,470). The PDF reader library (PDF.js, by Mozilla) loads from a public CDN; your PDF itself is read on your device and never uploaded. If the library can't load, the tool says so and stops — reload on a connection. Results always appear on this page first; the CSV download is a convenience.