PDF Stress Test
A corpus of 101 generated PDF documents for exercising document pipelines, readers, indexers and accessibility tooling. 11 are deliberately oversized; the rest vary in length. Most files are plain text-based PDFs, and the largest carries a unique high-entropy image plate on every page.
Oversized documents
Standard documents
About this corpus
Generated, not sourced
Every page is composed by a small generator in tools/stresstest/. Content is
synthetic and intentionally repetitive: it exists to give readers, parsers and accessibility
tools realistic volume, headings, lists and page furniture to process.
Simple, valid PDFs
Files use the PDF 1.4 base-14 Helvetica fonts with no images, so they are small, open quickly and are ideal for measuring per-page processing cost without file-size noise.
Regenerate any time
Run python3 tools/stresstest/build_stress_pdfs.py to rebuild all 100 files and
refresh the manifest used by this page.