Skip to content

Skip empty transcript summarization using dedicated flag - #847

Open
dembrane-sam-bot wants to merge 3 commits into
mainfrom
sam/skip-empty-summarization-v2
Open

Skip empty transcript summarization using dedicated flag#847
dembrane-sam-bot wants to merge 3 commits into
mainfrom
sam/skip-empty-summarization-v2

Conversation

@dembrane-sam-bot

Copy link
Copy Markdown
Contributor

Instead of writing summary='empty' which has unintended UI consequences (such as showing 'empty' as the literal summary text to users), this PR introduces a dedicated has_empty_transcript boolean field on the conversation model.

When a finished conversation has an empty transcript, we flag has_empty_transcript = True in the DB and return early. The catch-up scheduler then excludes any conversations where has_empty_transcript == True.

@spashii

spashii commented Jul 30, 2026

Copy link
Copy Markdown
Member

Needs two things before it can land: (1) has_empty_transcript does not exist in the Directus snapshot (echo/directus/sync/snapshot/), so the write fails in every environment. It needs the field created via the migration-script + snapshot-pull flow in AGENTS.md. (2) Rebase onto main: #898 already landed the empty-string query filter fix, so this now conflicts in api/conversation.py and conversation_utils.py. The remaining value here is the dedicated flag that stops the ~5-minute catch-up scheduler from re-summarizing empty conversations forever, which is worth keeping.

@dembrane-sam-bot
dembrane-sam-bot force-pushed the sam/skip-empty-summarization-v2 branch from 4dced6a to c279535 Compare July 30, 2026 06:15
@dembrane-sam-bot

Copy link
Copy Markdown
Contributor Author

Done.

  1. Rebased onto main to resolve the conflicts from fix(summary): correct Directus empty string query filter #898.
  2. Added the has_empty_transcript field to the Directus schema snapshot (echo/directus/sync/snapshot/fields/conversation/has_empty_transcript.json) and created the idempotent migration script (echo/directus/migrations/add_has_empty_transcript_field.py).
  3. Registered has_empty_transcript in the ConversationService.update schema and updated the catch-up scheduler query in conversation_utils.py to exclude has_empty_transcript == True.

Ready for review!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants