Sibling case to #308: when a fragmented MP4 segment begins with a sidx
box (no ftyp/styp/moof at the file root), Mp4InputFormat._canReadInput
returned false, causing Input.getTracks() to throw
UnsupportedInputFormatError.
This shape is standard in CMAF when the DASH on-demand profile is used
or when HLS playlists are derived from one (Vimeo's vod-adaptive-ak CDN,
for example). The segment is still a valid ISOBMFF fragment — the sidx
just sits before the moof.
Adds 'sidx' alongside the existing 'moof' check so the format probe
accepts these segments, and a focused regression test that fails on
the previous behaviour.
* fix: release targets from Output._targets on finalize
Long-running HLS transcodes leak memory. Every finalized BufferTarget
stays in _targets until the outer Output closes, pinning its buffer.
Writer.finalize() already does this cleanup for writer-based flows;
extend it to buffer-finalize paths via the public 'finalized' event.
* fix: ReadOrchestrator LRU eviction picks only drained workers
assert(pendingSlices.length === 0) fires under heavy concurrent reads
(e.g. multi-rendition HLS decode from BlobSource). LRU filter only
checked !running; workers with queued slices could be evicted.
Add pendingSlices.length === 0 to the filter.
* fix: export AppendOnlyStreamTarget from index
Missing from the re-export; public docs import it by name.
* fix: join HLS init segment path with root + playlist path
Init path went through _getTarget bare; segments got joined with rootPath + playlist.path. Playlist-relative URI then can't resolve when the playlist lives in a subdirectory.
* Fix targets not being cleaned up, fix paused workers with remaining pending slices, modify doc block, fixed isRoot not being changed on proxied requests
---------
Co-authored-by: Vanilagy <[email protected]>
* Add BufferTarget.onFinalize option for awaitable async action on finalize
The `finalized` event fires synchronously and its return value is ignored,
making it unsuitable for use cases where the muxer should wait (e.g.
uploading the buffer to S3, R2, or other object stores that require a
known Content-Length and therefore can't stream via StreamTarget).
Adds a new `BufferTargetOptions` type with an `onFinalize` callback that
the muxer awaits before resolving. Matches the existing callback pattern
used by `HlsOutputFormatOptions.onSegment`, `onMaster`, etc.
When used with PathedTarget, this provides proper backpressure: the next
segment won't start being produced until the previous one has finished
uploading, keeping memory bounded regardless of video length.
- Adds `BufferTargetOptions` type, exported from the package root
- `BufferTarget` constructor now accepts optional options (backward compatible)
- `_finalize()` awaits `onFinalize` before emitting the `finalized` event
- Adds runtime validation for non-function callbacks
- Updates "Upload to a server" docs in writing-hls.md with S3 pattern
- Documents `onFinalize` in writing-media-files.md BufferTarget section
- Adds tests for callback invocation, async awaiting, backward compat, and validation
* Add ConcurrentRunner, add OutputOptions.onFinalize, adjust docs accordingly, clean up tests
* give me more control baby
* forgot to tell about it
* polish guide and js docs
* vanilagy says remove this, docs will reveal it
---------
Co-authored-by: Vanilagy <[email protected]>
Vanilagy