Validating PDFs

Basic usage

horn validate <FILES...> [OPTIONS]

Horn accepts one or more file paths or directories as arguments.

Single file

horn validate report.pdf

Multiple files

horn validate report.pdf invoice.pdf brochure.pdf

Glob patterns

horn validate *.pdf
horn validate docs/**/*.pdf

Directories

Use --recurse to scan directories recursively for PDF files:

horn validate ./output/ --recurse

Output format

Choose the output format with --format:

horn validate document.pdf --format json
horn validate document.pdf --format sarif
horn validate document.pdf --format junit
horn validate document.pdf --format text   # default

See Output Formats for details on each format.

Writing to a file

Use --output to write results to a file instead of stdout:

horn validate ./pdfs/ --recurse --format sarif -o results.sarif

Controlling failure severity

By default, Horn exits with code 1 if any error-level findings are found. Use --fail-on to change the threshold:

# Fail on warnings or errors (ignore info)
horn validate document.pdf --fail-on warning

# Fail on any finding including info
horn validate document.pdf --fail-on info

# Only fail on errors (default)
horn validate document.pdf --fail-on error

This is useful in CI/CD pipelines where you may want to gradually tighten accessibility requirements.

Performance

Horn processes PDFs in parallel using Rayon. On a modern machine, expect throughput of 500+ PDFs per second in release mode.

For best performance:

• Use release builds (cargo build --release)
• Pass directories with --recurse instead of individual files — this lets Horn's thread pool work efficiently
• Use --format json or --format sarif when piping to other tools — text output is slower for large batches