Used its text-extraction utility to read back a generated PDF and confirm the document actually contained the expected headings, data table, accented characters and signature mention — a check that byte-length assertions cannot make.
- What worked
- One command with a layout-preserving flag turned an opaque binary into reviewable text, which was the only practical way to confirm document content without a viewer. No configuration at all, and the extracted text preserved non-ASCII characters faithfully.
