Skip to content

Latest commit

 

History

History
18 lines (13 loc) · 1.68 KB

File metadata and controls

18 lines (13 loc) · 1.68 KB

docproc Documentation

Guides

Document Description
CONFIGURATION.md Configuration referencedocproc.yaml schema, AI providers (OpenAI, Azure, Anthropic, Ollama, LiteLLM), ingest options (vision, LLM refinement). Used by the CLI for document extraction.
AZURE_SETUP.md Azure setup — Azure OpenAI and Azure AI Vision (Computer Vision) for PDF image extraction; credentials via env or scripts/azure_env.sh.
ARCHITECTURE.md Architecture overview — Pipeline flow, modules, Python library (Docproc). File in → markdown out; the full-stack demo lives in demo/. Examples: examples/.
USAGE.md Usage examples — CLI extract, init-config, completions.
DOCKER.md Docker — Demo infrastructure only (PostgreSQL, LocalStack, RabbitMQ). Go API and frontend run on the host.

See also CONTRIBUTING.md for development setup and running tests.

Concepts

  • docproc (CLI or library) — Document processor. Reads a file (PDF, DOCX, PPTX, XLSX), extracts text (native + optional vision for images), optionally refines with an LLM, and produces markdown (CLI writes a file; Python API returns a string or writes via Docproc.extract_to_file). No server, no database, no RAG in the core package.
  • Demo — Separate application in demo/: Go API, React frontend (demo/web/), document upload (LocalStack S3), job queue (RabbitMQ), RAG and grading (PostgreSQL + PgVector, OpenAI). Document processing is done by running the docproc CLI from the Go worker. See demo/README.md.