fluent-db-updater

Run update-db.py once at session end with a single JSON payload containing session_id, date, and optional blocks like errors, new_vocabulary, review_results, and skill_scores. The script validates the payload, backs up existing files, and applies all changes atomically.

Include a new_vocabulary array in the update-db.py payload with fully filled fields such as item_id, content, answer, difficulty, and initial_quality. The script creates the spaced-repetition records and rebuilds the review queue automatically.

Exit code 1 means payload validation failed and no files were touched; fix the payload and retry. Exit code 2 means an I/O failure such as disk or permission problems, also with no files written.

No. The review_queue is regenerated from scratch on every update-db.py run, so manual edits are overwritten. Always make changes through the payload's review_results and new_vocabulary blocks instead.

Sending the same session_id replaces the earlier session's data, which is useful for corrections but dangerous if unintentional. Always use the next_session_id value returned by read-db.py to avoid collisions.