Exercises: Build Parser Quality Gates
Use starter.py to classify parsed documents, detect cleaning risk, and decide whether content can move to chunking.
1 Detect parser readiness
Return ready, ready_with_warnings, needs_review, or rejected based on quality signals.
2 Preserve required metadata
Validate source id, format, version, access label, owner, and source positions.
3 Explain the first production risk
Identify whether the top risk is OCR, table extraction, boilerplate noise, missing metadata, or security metadata loss.
4 Recommend the next action
Decide whether to publish to chunking, route to review, fix parser config, or reject the source.