Skip to content

fix: production scraper ignores structured extractor selected by qualification #644

Description

@chubes4

Problem

The same flow source produces contradictory extractor outcomes between qualification and production execution.

For Extra Chill Events flow 96 (https://www.29thstreetballroom.com/):

  • wp extrachill venues requalify-flow 96 returned qualified_structured with 55 events.
  • Immediate production job 801596 materialized one vision_flyer packet instead.
  • The AI rejected that packet as empty / placeholder / un-dated junk.
  • No structured Squarespace events reached the pipeline.

The source was originally paused for extraction_gap; DME v0.53.1 added Squarespace summary-event extraction and qualification now proves the source is extractable. The live universal_web_scraper path is not honoring the successful structured extractor/arbitration result.

Expected

Production flow execution and qualification should select the same reusable structured extractor for the same source and handler configuration. Vision fallback must not preempt a structured extractor that returns valid events.

Evidence

  • Production DME version: 0.53.1
  • Qualifier version: 0.56.0
  • Flow: 96
  • Job: 801596
  • Qualification event count: 55
  • Production source type: vision_flyer
  • Production result: agent_skipped - source-rejected

Do not add a domain-specific exception for 29th Street Ballroom; fix generic extractor arbitration/parity.

Activity

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

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions