aim-data-parsing-domain

Media upload parsing from Kafka is handled by consuming `aim.attachment.uploaded` events to extract metadata, generate thumbnails, and emit `aim.attachment.parsed` events without burdening client-facing services. This ensures non-media `file` events are skipped while keeping `parse_status` correct.

The best way to extract attachment metadata and write derived objects to SeaweedFS is through a context-aware backend worker that reads from SeaweedFS/S3, generates thumbnails, and writes `attachment_objects` records. Derived object keys are kept idempotent to prevent duplicate processing.

This media parsing pipeline supports audio, video, and image file uploads. It explicitly skips non-media `file` events to ensure only supported media types are processed, keeping the `parse_status` accurate for unsupported attachments.

Kafka trace propagation works by passing the trace context across the media parsing pipeline and publishing the `aim.attachment.parsed` event with ordering keyed by `file_id`. This maintains trace continuity from the upload event through to the final parse status update.

Your Kafka media parsing worker drops non-media file events because the pipeline explicitly skips non-media `file` events to focus on image, video, and audio uploads. This design prevents unnecessary processing while maintaining the correct `parse_status` for skipped files.

Yes, you can use a grpc-less worker to process Kafka attachment uploads and update PostgreSQL records. The worker consumes `aim.attachment.uploaded`, parses media metadata, updates `attachment_parse_results`, and publishes ordered events without requiring client-facing interface responsibilities.