events
Pipeline event callbacks. PipelineEvents is a dataclass of optional callables that any stage can fire.
Container for all pipeline event callbacks.
Every field is None by default — assign only the callbacks you need.
Example::
events = PipelineEvents(
on_frame_written=lambda idx, total: print(f" frame {idx + 1}/{total}"),
)
PipelineRunner(manifest, config, events=events).run()
on_clip_found = field(default=None)
class-attribute
instance-attribute
on_clip_found(clip_name, clip_root) — fired as each valid clip is discovered.
on_clip_skipped = field(default=None)
class-attribute
instance-attribute
on_clip_skipped(reason, path) — fired when a path is skipped during scanning.
on_queue_depth = field(default=None)
class-attribute
instance-attribute
on_queue_depth(preprocess_queue_depth, postwrite_queue_depth)
Both values are point-in-time snapshots; treat them as approximate since worker threads may be draining concurrently.