firecrawl/pdf-inspector

Fast Rust library for PDF inspection, classification, and text extraction. Intelligently detects scanned vs text-based PDFs to enable smart routing decisions.

What it solves

pdf-inspector solves the problem of expensive and slow PDF processing by providing a fast way to classify PDFs and extract their content. Many pipelines waste resources by running Optical Character Recognition (OCR) on every document, even when the PDF is already text-based. This library allows developers to quickly determine if a PDF is text-based or scanned and extract structured Markdown from text-based PDFs locally and rapidly.

How it works

The library uses a two-stage process:

  1. Smart Classification: It samples content streams to detect if a PDF is TextBased, Scanned, ImageBased, or Mixed. This happens in milliseconds by looking for text and image operators without loading the full document.
  2. Text Extraction: For text-based PDFs, it performs position-aware extraction. It analyzes font sizes, X/Y coordinates, and drawing operations to identify headings, tables, lists, and multi-column layouts, which it then converts into clean Markdown.

For documents that are scanned or mixed, the library offers selective OCR, allowing users to run OCR only on the specific pages that actually need it using PP-OCRv6 Small locally.

Who it’s for

It is designed for developers building large-scale document processing pipelines, specifically those handling reports, research papers, financial documents, invoices, and legal PDFs where speed, reading order, and table structure are critical.

Highlights

  • High Performance: Processes text-based PDFs locally in under 200ms.
  • Multi-Language Support: Provides bindings for Rust, Python, Node.js, and Browser WebAssembly.
  • Structured Output: Converts PDFs to Markdown with support for H1-H4 headings, tables (rectangle and heuristic detection), code blocks, and multi-column layouts.
  • Selective OCR: Routes only necessary pages to OCR to minimize latency and cost.
  • Position Awareness: Extracts text with font information and coordinates to maintain reading order.

Related

  • Project
    firecrawl/anydoc

    A fast Rust library that converts various office documents (Word, PowerPoint, Excel, PDF, etc.) into clean, LLM-ready GitHub-Flavored Markdown.

  • Project
    run-llama/liteparse

    LiteParse is a fast, locally‑run PDF (and office‑format) parser that extracts text with bounding boxes, optional OCR, screenshots, and rich JSON/Markdown output. It ships a Rust core and bindings for Python, Node.js/TypeScript, and WebAssembly, making it easy to feed structured document data into LLM and RAG pipelines without cloud dependencies.

  • Project
    yfedoseev/pdf_oxide

    A high-performance, multi-language PDF toolkit for text and image extraction and markdown conversion, optimized for RAG pipelines and AI assistants.

  • Project
    zibo-chen/rust-paddle-ocr

    A lightweight Rust OCR library based on PaddleOCR models and the MNN inference runtime, providing text detection and recognition across multiple languages.

  • Project
    Dicklesworthstone/llm_aided_ocr

    An advanced OCR pipeline that uses LLMs to correct errors in raw Tesseract OCR output and transform it into high-quality, formatted Markdown documents.