The official source code lives at github.com/firecrawl/pdf-inspector, maintained by Firecrawl under the MIT License. Whether you want to read the source, file a bug, or contribute, this is the entry point — this page rounds up the key links and local-build steps.

Key Links

What	Where
Source repository	github.com/firecrawl/pdf-inspector
Issues	github.com/firecrawl/pdf-inspector/issues
Releases	github.com/firecrawl/pdf-inspector/releases
OCR runtime guide	docs/ocr-runtime.md
README	github.com/firecrawl/pdf-inspector#readme

The one true repo is firecrawl/pdf-inspector; @firecrawl/pdf-inspector on npm, pdf-inspector on PyPI, and the pdf-inspector crate on crates.io are its official release channels.

Clone the Repository

git clone https://github.com/firecrawl/pdf-inspector.git
cd pdf-inspector

What's Inside

pdf-inspector is a Rust-first, multi-language workspace:

• Rust core — five modules: detection (detector), extraction (extractor), layout (layout), tables (tables), and Markdown rendering (markdown/); the only PDF parsing dependency is lopdf;
• Node.js binding (napi) — the @firecrawl/pdf-inspector npm package with prebuilt binaries for Linux / macOS / Windows, TypeScript definitions, and a CLI;
• Browser binding (wasm) — the @firecrawl/pdf-inspector-wasm npm package, single-threaded build with embedded CMaps; it powers the online demo on this site's homepage;
• Python binding — the pdf-inspector PyPI package, built with maturin;
• CLI — the pdf2md and detect-pdf binaries shipped with the Rust crate;
• Test assets — snapshot tests, mutation tests, and cargo-fuzz targets.

Architecture details are in the project introduction.

Building from Source

Rust (core + CLI)

cargo build --release          # build the core library and CLI
cargo install pdf-inspector    # or install pdf2md / detect-pdf onto your PATH
cargo test                     # run the test suite

Node.js binding

npm install @firecrawl/pdf-inspector   # prebuilt binaries — no Rust toolchain needed

To modify the binding itself you'll need a Rust toolchain and the napi CLI; see the development section of the repo README.

Python binding

pip install pdf-inspector      # prebuilt wheel — no Rust toolchain needed

For in-repo development:

maturin develop --release

Browser WASM

npm install @firecrawl/pdf-inspector-wasm

The WASM build embeds CMap data, so browsers need no extra font-mapping files.

Don't Want to Compile?

Most users never need a source build — the prebuilt packages are the everyday path; see the install page. Just want to try it? Drop a PDF into the homepage demo.

Filing a Good Issue

Three things make issues dramatically easier to triage:

1. Classify first: paste the output of detect-pdf file.pdf --analyze --json — many "garbled output" reports turn out to be scanned or encoding-corrupted pages;
2. Minimal reproduction: the smallest PDF that triggers the problem (redacted!) or repro steps;
3. Environment: platform, package and version (@firecrawl/pdf-inspector / pdf-inspector / the pdf-inspector crate), and how you call it.

Contributing Code

1. Fork the repo and branch off main;
2. Include tests with your change — snapshot cases against real-world dirty PDFs are especially welcome;
3. Make sure cargo test and formatting pass, then open a PR;
4. For larger changes (new dependencies, new APIs, new formats), open an issue first.

Note

pdf-inspector is MIT licensed. Submitting a PR means you agree to license your contribution under the same terms.

Sister Repository: AnyDoc

Around the same time, Firecrawl also open-sourced anydoc (official site anydoc.wiki): it converts 14 office formats — Word, PPT, Excel, OpenDocument, RTF, EPUB, CSV, and more — into Markdown, with the PDF path powered by pdf-inspector itself. PDF-only? Stay here. Mixed office files? Use AnyDoc — details on the AnyDoc page.