Extractous

Extractous is an open-source data extraction engine for unstructured documents, published on GitHub by Yobix AI and written in Rust. Built on the proven Apache Tika parsing toolkit, it extracts clean text and metadata from PDFs, Word documents and many other formats at high speed, with OCR support — exactly the preprocessing step that RAG pipelines and LLM ingestion workflows depend on. The project has around 1,800 stars and was last updated in 2024.

Overview

Most document parsers are either slow JVM services or fragile scripts. Extractous wraps Tika’s mature format support in a native Rust core and exposes it through bindings for multiple languages, so the same extraction engine can sit inside Python data pipelines, ETL jobs or LLM preprocessing services without running a separate Java service. It focuses on throughput and predictable output for machine consumption rather than pretty rendering.

Key Features

• Fast text and metadata extraction from PDFs, DOCX and many other document formats
• Built on Apache Tika for broad, battle-tested format coverage
• Native Rust core with bindings for multiple programming languages
• OCR support for scanned documents and images
• Designed for RAG and LLM ingestion pipelines and ETL workloads
• Apache-2.0 licensed and self-hostable with no per-page fees

💳 Pricing

Free. Extractous is open source under the Apache-2.0 license; you run it in your own infrastructure at no cost.

Who Is It For

Data engineers building document pipelines, RAG developers who need reliable PDF and Office parsing, and teams replacing slow or paid extraction services with a self-hosted engine.