PDF Date Parser Benchmark.
Twenty synthetic, text-based PDF test files include bills, invitations, renewals, ambiguous numeric dates and multiple-date documents. Test results are local program output and must not be interpreted as production accuracy on real-world PDFs.
How to reproduce
Run npm run generate and npm test in the source folder. The fixtures are built from synthetic English text, encoded as PDFs via pdf-lib, then the test runner checks every expected normalized ISO date. The parser uses chrono-node with additional numeric date-order handling.
Coverage boundaries
These deterministic fixtures are not a statistically representative validation dataset. The test runner also checks IANA timezone conversion (New York summer/winter), detects invalid daylight-saving times, and validates ICS output. OCR accuracy, photographed invoices, multilingual date formats and layout-heavy scanned documents require separate testing.