-
Notifications
You must be signed in to change notification settings - Fork 1
Open
Labels
agent-discoveredDiscovered by an AI agent during workDiscovered by an AI agent during workpriority: lowEnhancement, nice-to-haveEnhancement, nice-to-havetype: bugSomething is not workingSomething is not working
Description
Bug
calculate_transcript_metrics_from_file() in api/routers/queue.py catches all exceptions (including PermissionError, FileNotFoundError) and returns (None, None) without surfacing the error to the job record. Jobs are created with duration_minutes=None and word_count=None with no indication of why.
Impact
- Jobs missing metrics can't trigger duration-based features (e.g., timestamp phase for 30+ min content)
- Cost estimation is impossible without word count
- No way to distinguish "metrics not yet calculated" from "metrics failed to calculate"
Suggested Fix
- Log a warning with the specific error type
- Consider storing a
metrics_errorfield, or at minimum populatingerror_messageon the job - Differentiate between "file not found" (queue-time error) and "parse failure" (data quality issue)
Location
api/routers/queue.py:43-73 — calculate_transcript_metrics_from_file()
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
agent-discoveredDiscovered by an AI agent during workDiscovered by an AI agent during workpriority: lowEnhancement, nice-to-haveEnhancement, nice-to-havetype: bugSomething is not workingSomething is not working