[HLS] Support Single File TS for HLS. #891
Conversation
| @@ -1538,6 +1538,13 @@ def testEc3AndHlsSingleSegmentMp4Encrypted(self): | |||
| ['audio', 'video'], hls=True, test_files=['bear-640x360-ec3.mp4']), | |||
| self._GetFlags(encryption=True, output_hls=True)) | |||
| self._CheckTestResults('ec3-and-hls-single-segment-mp4-encrypted') | |||
|
|
|||
| def testHlsByterange(self): | |||
There was a problem hiding this comment.
I'd prefer calling it: testHlsSingleSegmentTs, as "hls-byte-range" could be for mp4 or ts.
| @@ -33,12 +33,16 @@ | |||
| 'mp2t_media_parser.h', | |||
| 'mpeg1_header.cc', | |||
| 'mpeg1_header.h', | |||
| 'multi_segment_ts_segmenter.cc', | |||
| 'multi_segment_ts_segmenter.h', | |||
There was a problem hiding this comment.
Can you align it with the line above it?
| } | ||
| } else if (output_format == CONTAINER_WEBVTT || | ||
|
|
||
| if (output_format == CONTAINER_WEBVTT || |
There was a problem hiding this comment.
The condition should include output_format == CONTAINER_MPEG2TS || too, as TS does not have init segment. Can you also update line 232 too?
| @@ -21,7 +23,12 @@ Status TsMuxer::InitializeMuxer() { | |||
| if (streams().size() > 1u) | |||
| return Status(error::MUXER_FAILURE, "Cannot handle more than one streams."); | |||
|
|
|||
| segmenter_.reset(new TsSegmenter(options(), muxer_listener())); | |||
| if (options().segment_template.empty()) { | |||
There was a problem hiding this comment.
Prefer using the approach in PackedAudio: https://github.com/google/shaka-packager/tree/56908a83a76c85450d6b5834f3aa16e0660f4648/packager/media/formats/packed_audio.
Here is the summary:
- Having only one segmenter class instead of having one for SingleSegment and one for MultiSegment.
- The segmenter class writes the data into a temporary buffer first, which is already the case in the current TS segmenter implementation.
- Do not output the temporary buffer to the file in the segmenter, but in the muxer instead.
- In the muxer, the code can decide whether to write to a single file or a segment file based on the configuration. See https://github.com/google/shaka-packager/blob/56908a83a76c85450d6b5834f3aa16e0660f4648/packager/media/formats/packed_audio/packed_audio_writer.cc#L100 for an example.
- The muxer listener events are handled in the muxer. There is no need to handle that in the segmenter.
I think the code is cleaner and easier to maintain with this structure.
What do you think?
There was a problem hiding this comment.
Hi @kqyang, thanks for the review. Yes that makes sense. Let me try it out.
|
I had issues with this branch. Creating a new PR and branch at #934 |
|
Replaced by #934 |
This is based on comments at #891. The muxer is deciding whether to write to a single file or a segment file based on the configuration. Example: ``` ../packager 'in=TOS.ts,stream=video,output=tos_video.ts,playlist_name=tos_video.m3u8' \ 'in=TOS.ts,stream=audio,output=tos_audio.ts,playlist_name=tos_audio.m3u8' \ --hls_master_playlist_output tos.m3u8 ``` Tested the content using Exoplayer. --------- Co-authored-by: Cosmin Stejerean <cstejerean@meta.com>
github-actions
bot
added
the
status: archived
This PR is related to #291.
Example:
Tested on Exoplayer.