extractous

Why we included this project

Most RAG pipelines hit the same wall early: the files you feed them don't come as clean text. Extractous removes that step from the critical path by parsing PDFs, Word docs, HTML, and other formats in-process with a Rust core, so there is no separate extraction server to run and the bindings for Rust and Python let you call it straight from your own code. It returns both content and metadata through one simple API, and can OCR scanned pages when you need it. The README backs up the performance claims with benchmarks showing it roughly 25 times faster than the popular unstructured-io library, which matters for teams pushing large document volumes through tight memory budgets. If document ingestion is the slow part of your pipeline, this is a solid, lightweight building block.