Command line and Python API

How to install and drive the package is documented in the user guide, which is built from the released code at each release, so it describes the version on PyPI and cannot drift from it. The reference pages here are the things the code cannot document about itself: the corpus, the engines as measured, routing and licensing.

In brief, installing the package provides one command:

pip install 'tetrak[qa]'                # plus: brew install tesseract poppler

tetrak-ocr backends                     # what this machine can run
tetrak-ocr ocr scan.jpg --backend tesseract-auto
tetrak-ocr batch --backend auto-local   # workspace/scans/ → processed/, failures to triage/
tetrak-ocr evaluate --all --save        # the benchmark, from a checkout

The guide covers the rest:

• Installation — the package, the optional backend extras and what each pulls in, the system dependencies, model weights and credentials.
• Quick start — one file, then a folder, and what comes back.
• Using the command line — every subcommand, the output formats and searchable PDFs, the triage queue, the run log, dry runs and the evaluation harness.
• Choosing a backend — what each engine is for, what auto-local does, and what a run costs.
• Using the Python API — the registry, the errors, backend options, fan-out and the quality gate, outputs, multi-page documents, the batch pipeline and the run log from code.
• Tutorials — batch an archive, transcribe Armenian material and build your own pipeline.
• Command line reference and Python API reference — every flag and every public function, generated from the parser and the docstrings.