About this project

pdf-inspector is a Rust library built by Firecrawl for fast PDF classification and text extraction. It detects whether a PDF is text-based, scanned, image-based, or mixed, and extracts text with position awareness, font information, and multi-column reading order. The library converts PDFs to clean Markdown, including headings, lists, code blocks, tables, bold/italic formatting, URLs, and page breaks. It supports CID fonts, RTL text, encoding issue detection, and selective OCR routing for pages that need it. Bindings are available for Python, Node.js, and browser WebAssembly, along with CLI tools (pdf2md and detect-pdf). The default builds remain lightweight, with OCR dependencies loaded only when needed. Benchmarks on the opendataloader-bench corpus show competitive speed and quality for native-text PDFs such as reports, research papers, invoices, and legal documents.