pdf

Open the PDF with the office tool to read its form fields from the snapshot, then use the fill_form batch operation with values keyed by field name. Unknown names or options fail with a list of what exists, and oversized values are reported as clipped.

Run a pdf-tables query on the target pages; ruled tables are read exactly from cell borders while alignment-based ones are guessed from text positions. Providing an output directory writes one CSV per table for downstream spreadsheet processing.

Yes, the ocr_pages batch operation adds an invisible searchable text layer to scanned pages in place. Check available cached languages with the detect action first, since uncached languages require a network download.

Use the secure action with encrypt or decrypt, which writes an AES-256 copy of the file. This requires qpdf installed on the machine; verify availability with the detect action before promising a decrypt.

The page is likely a scanned image, reported in the snapshot's likelyScannedPages field. Run OCR on those pages or render them as images; never report a picture page as empty.

Digital signatures, redaction, and PDF/A conformance are unsupported. Marks like highlights or opaque text do not redact content since the underlying text remains extractable from the file.