Unstructured
An open-source Python document parser prepares PDFs, Word files and images for LLM applications and runs locally or in Docker under Apache 2.0.
Unstructured is a document processing library for developers preparing files for LLM applications. It turns text documents and images into structured output that applications can work with.
- Document parsing: Process PDFs, HTML, Word documents and other file types through the Python library.
- Image preprocessing: Include images alongside text documents in document ingestion workflows.
- Local processing: Run the library on your own hardware or in Docker, with container images for x86_64 and Apple Silicon.
The library is free and open source under Apache 2.0. Windows use is also supported. Unstructured's hosted SaaS API and Transform API are separate offerings; the models offered through Transform run through that service rather than the local library.
The library sends telemetry by default, including platform information, processing characteristics and aggregate output counts. These events exclude document contents, filenames, paths, credentials and persistent user or machine identifiers.
Logotyper
