gopdfrab

gopdfrab is an open-source Go library and command-line tool for PDF/A compliance, built by voidrab (IT-Dienstleistungen Thomas Karner e.U.). It provides PDF/A-1b verification and conversion with a small, predictable API, zero runtime dependencies, and cross-platform support including WebAssembly. The project reached stable API status at v1.0, released in August 2026, and is dual-licensed under AGPL 3.0 with a commercial license option for closed-source use.

What It Is

gopdfrab is a PDF/A processing library for the Go programming language. It handles two core jobs: verifying whether a PDF document conforms to the ISO 19005-1 (PDF/A-1b) standard, and converting non-conformant PDFs into PDF/A-1b documents. The library exposes four primary concepts — a document, a verification profile, a result, and a conversion — and ships the same engine as both a Go package and a static CLI binary. A browser-based validator and converter are also available on the project website, running the same engine compiled to WebAssembly entirely client-side.

Architecture and Design

gopdfrab is written in pure Go with no external runtime dependencies beyond the standard library. Key architectural properties include:

• Pure Go, zero deps: Cross-compiles anywhere Go does, including js/wasm for browser use.
• Customizable profiles: Verification profiles are immutable and goroutine-safe; checks can be added, removed, or built from scratch.
• Damage recovery: Broken cross-reference tables trigger a full-file object scan; individual corrupt objects are re-located or nulled and reported rather than causing hard failures.
• Encrypted PDF support: Decrypts RC4 40/128, AES-128, and AES-256 with empty or supplied passwords.
• Fidelity checking: An optional CheckFidelity mode renders both input and output to detect content loss during conversion.
• Streaming and batching: ConvertEach streams large batches without holding all outputs in memory; VerifyAll/ConvertAll run concurrently with bounded worker pools.

The conversion pipeline runs pre-emptive fixups, then a verify/fix loop, and falls back to page rasterization as a last resort. Anything unresolvable is reported as a residual and never silently passed as valid.

Conformance Suite Coverage

The homepage states that gopdfrab passes all tests in the Isartor and veraPDF conformance suites:

• 204/204 Isartor test suite passes
• 569/569 veraPDF test suite passes
• 159 checks across 11 groups covering structure, colour, image, transparency, font, annotation, action, metadata, form, object model, and logical structure

A Legacy1B profile provides the strict, spec-literal Isartor reading for applications that require it, while the default PDFA1B profile matches veraPDF's interpretation where the two diverge.

Performance Characteristics

The project publishes benchmark results comparing gopdfrab against veraPDF and PDFBox Preflight on a 773-file corpus (the combined Isartor and veraPDF corpora). Per the project's own benchmarks: the full batch verifies in approximately 0.27–0.29 seconds at around 67–80 MB peak RSS; cold single-file verification of the median corpus file takes roughly 9.5 ms including process startup. The project attributes the JVM startup overhead of veraPDF and PDFBox Preflight as a significant factor in cold-start comparisons.

Robustness and Fuzzing

gopdfrab's test suite covers hostile and malformed input extensively:

• An internal generator (internal/pdfgen) builds structurally-valid PDFs deliberately corrupted with truncation, bad xref offsets, negative stream lengths, dangling and circular references, and deep nesting.
• Native Go fuzz targets operate at three levels: whole pipeline, isolated decoders and parsers, and semantic oracles (determinism, round-trip validity, convergence).
• Named TestCrasher_* reproducers guard each previously-fixed crash.
• Race detection and time-bounded corpus tests run on every build.

Update: v1.0.0

The GitHub repository records v1.0.0 as the latest release, published on 2026-08-26. The README states the API is stable as of 1.0, with a versioning and stability policy documented in CHANGELOG.md. The roadmap notes that PDF/A-1b is complete and that PDF/A-2, -3, and -4 are next. The repository was last pushed to on 2026-09-01.