ARCHIVE5 SETS · ZIP · DOCUMENTED
PDF Test Collections — Grouped by Purpose
Five curated sets of PDF files engineered for specific testing workflows. Each ZIP includes documentation describing every file's purpose and expected behavior.
RAG/AI Test Sets
Papers, slides, reports, and contracts structured for retrieval-augmented generation testing. Includes header/footer variants and a README documenting expected chunking behavior per file.
OCR Test Sets
Clean sans-serif text (baseline), tables, two-column layouts, and simulated low-quality scans. Ships with ground-truth.txt so you can score OCR engine accuracy against known-correct output.
Parser Stress-Test Sets
Malformed PDFs (corrupt xref, missing trailer, invalid objects), boundary cases, font edge cases, encrypted files, and the smallest valid PDF (< 1 KB). Each file is annotated with expected parser behavior: reject, tolerate, or partial-parse.
Render-Compatibility Sets
PDFs exercising CID fonts, OpenType, Type 3, and multiple character encodings. Open them side-by-side in pdf.js, Chrome, Firefox, Adobe Acrobat, and Evince to surface rendering differences.
Security Test Sets
Files containing JavaScript, hidden text, injection payloads, and bloated constructs. For authorized security testing only — validating that your parsers, viewers, and sanitization tools handle untrusted input safely. Includes AUP and disclaimer.
All collections are free to download as ZIP archives. No account required. Files are static assets served from our CDN — no generation delay.
Need something the collections don't cover? The generator lets you set size, pages, images, and form fields to produce a custom test PDF on demand.
Each ZIP contains 4–10+ PDF files plus a documentation file (README, expected-behavior.md, ground-truth.txt, or AUP) explaining what each file tests and what the expected outcome is.
They contain structural payloads (JavaScript stubs, hidden text, oversized objects) designed to test sanitization and sandboxing — not weaponized exploits. Download and open them only in isolated test environments. The collection ships with an Acceptable Use Policy.
Use the generator (/generator) for one-off custom PDFs. Collection additions are driven by community feedback — contact us via the link on /about.