ractogateway.pipelines.video_processor._summarizer
Comprehensive summary generation for VideoProcessorPipeline.
Combines visual frame analyses + audio transcripts → one structured summary covering whiteboard equations, screen content, and spoken explanations.
- ractogateway.pipelines.video_processor._summarizer.generate_summary_sync(sections, transcript, kit, usage)[source]
Generate summary synchronously using kit.
Updates usage with summary token counts. Returns the summary string.
- Return type:
- async ractogateway.pipelines.video_processor._summarizer.generate_summary_async(sections, transcript, kit, usage)[source]
Async variant of
generate_summary_sync().- Return type:
- ractogateway.pipelines.video_processor._summarizer.build_sections(frames, transcript)[source]
Merge visual frame analyses and transcript segments into VideoSections.
Each transcript segment becomes one section; frames are matched by timestamp overlap. If there are no transcript segments, each kept frame becomes its own section.
- Return type:
list[VideoSection]