HLS manifest parser plugin #279

joeyparrish · 2016-01-29T17:10:45Z

Initially only for browsers which support MPEG2-TS natively (Safari, Chromecast)

avelad · 2016-04-15T06:51:53Z

Note: Microsoft Edge also supports MPEG2-TS natively

There will be support for Playready (Edge and Chromecast) and Fairplay (Safari)?

joeyparrish · 2016-04-15T07:07:30Z

There is already support for Playready on Edge & IE. Playready has not been tested on Chromecast, though.

All DRM is through EME APIs, so any interoperable DRM system in any EME-compliant browser should work. The status of Fairplay through EME is unknown at this time.

avelad · 2016-04-15T07:10:50Z

So, HLS+Playready will work on Edge(IE has not support for MPEG2-TS without transmuxing)?

joeyparrish · 2016-04-15T07:16:59Z

We expect so, yes. But we have not begun work yet, so we do not know what interoperability challenges we may find.

avelad · 2016-06-20T07:18:02Z

HLS supports Fragmented MP4 – and gets MPEG-DASH compatible! https://bitmovin.com/hls-news-wwdc-2016/

joeyparrish · 2016-06-29T03:38:08Z

Given that new development, I would expect HLS with MP4 to work on almost every browser we support. Again, we haven't started work on HLS yet.

sanbornhilland · 2016-07-13T15:15:51Z

I thought it could be useful to catalogue some of the issues with building HLS support into Shaka, as I come across them.

As reported in Support Request: sourceBuffer_ detects incorrect timestamps on content. #439 a lot of HLS VOD content seems not to start at 0 but it so far seems impossible to know what the start time is of the content until the first segment is downloaded.
HLS can include just an audio stream as it's lowest quality stream (see below), the intention being that if your bandwidth drops low enough then the player should switch down to this audio only variant. But it is not an audio stream in the sense that you can choose this audio stream vs some other stream (english vs spanish etc.). It seems to me like this doesn't fit nicely into Shaka's streaming model.

#EXT-X-STREAM-INF:BANDWIDTH=263851,CODECS="mp4a.40.2, avc1.4d400d",RESOLUTION=416x234,AUDIO="bipbop_audio",SUBTITLES="subs"
gear1/prog_index.m3u8
#EXT-X-I-FRAME-STREAM-INF:BANDWIDTH=28451,CODECS="avc1.4d400d",URI="gear1/iframe_index.m3u8"

#EXT-X-STREAM-INF:BANDWIDTH=577610,CODECS="mp4a.40.2, avc1.4d401e",RESOLUTION=640x360,AUDIO="bipbop_audio",SUBTITLES="subs"
gear2/prog_index.m3u8
#EXT-X-I-FRAME-STREAM-INF:BANDWIDTH=181534,CODECS="avc1.4d401e",URI="gear2/iframe_index.m3u8"

#EXT-X-STREAM-INF:BANDWIDTH=915905,CODECS="mp4a.40.2, avc1.4d401f",RESOLUTION=960x540,AUDIO="bipbop_audio",SUBTITLES="subs"
gear3/prog_index.m3u8
#EXT-X-I-FRAME-STREAM-INF:BANDWIDTH=297056,CODECS="avc1.4d401f",URI="gear3/iframe_index.m3u8"

#EXT-X-STREAM-INF:BANDWIDTH=1030138,CODECS="mp4a.40.2, avc1.4d401f",RESOLUTION=1280x720,AUDIO="bipbop_audio",SUBTITLES="subs"
gear4/prog_index.m3u8
#EXT-X-I-FRAME-STREAM-INF:BANDWIDTH=339492,CODECS="avc1.4d401f",URI="gear4/iframe_index.m3u8"

#EXT-X-STREAM-INF:BANDWIDTH=1924009,CODECS="mp4a.40.2, avc1.4d401f",RESOLUTION=1920x1080,AUDIO="bipbop_audio",SUBTITLES="subs"
gear5/prog_index.m3u8
#EXT-X-I-FRAME-STREAM-INF:BANDWIDTH=669554,CODECS="avc1.4d401f",URI="gear5/iframe_index.m3u8"

#EXT-X-STREAM-INF:BANDWIDTH=41457,CODECS="mp4a.40.2",AUDIO="bipbop_audio",SUBTITLES="subs"
gear0/prog_index.m3u8

raskri · 2016-10-06T06:22:25Z

Any news on the ETA for HLS support in Shaka?

joeyparrish · 2016-10-06T16:43:44Z

We have just scheduled the work for v2.1.0, which we intend to have out at the end of the year. As we make progress, you'll see commits and major milestones mentioned here.

avelad · 2016-10-21T09:24:36Z

There will be support for..
CEA-608?
HLS + AES?
HLS + FairPlay(#382)?
Live and VoD profiles?
What version of HLS (4,5,6,7,..)?

ismena · 2016-10-21T22:41:19Z

We are just starting investigation and design work at this point.
We'll keep this issue up to date with the decisions on what we will and won't support as we get into them.

dougdoe · 2016-12-09T15:03:07Z

Hi @joeyparrish and @ismena, I am wondering if you have an estimate for when this work will be ready for a pull request. I am attempting to do some planning, so if you have any information, it would be helpful. My team can also help perform testing when there is something to work with. Thanks.

ismena · 2016-12-09T21:38:57Z

Hi Doug,
When we did the investigation, we found out that we'll have to refactor our streaming model so it works for HLS as well as DASH (the player was initially written with DASH in mind and HLS doesn't fit nicely in our current model).
I'm working on that right now and will start looking into HLS as soon as it's done.
I plan to start with clear content support first.

My hope is to finish refactoring work this year or early January (it's a pretty big change), so you should start seeing first HLS commits early next year.

I want to be cautious in terms of speaking to the timeline until I've started the work and have full context, but I'll let you know more as soon as I can!

Thanks for the offer to help with the testing, we're always happy to get feedback.

shaharmor · 2016-12-15T08:15:54Z

@ismena Are you going to use your own HLS provider or hls.js? I highly recommend hls.js 👍

joeyparrish · 2016-12-15T17:40:02Z

We are going to write a manifest parser plugin for HLS. We do not plan on using hls.js.

Issue #279 Change-Id: Idf4241b4e2490876bdc6275685361a5c9bf132b0

These properties are no longer used since we refactored for variants. All compatible AdaptationSets are now implicitly squashed together. Issue #279 Change-Id: Idc1922c54b0c1aa62f3199369b8e819e3a374b55

Sorting by language then bandwidth led us to compare text tracks to variant tracks, provided that they had the same language. Text tracks don't have bandwidth info. t1.bandwidth - t2.bandwidth was undefined - number, which resulted in NaN. This caused the sorted order of variants to be wrong, as some video tracks got pushed to the end when compared with text. The fix is to compare type first, then language, then bandwidth. Introduced during refactors as part of #279. Change-Id: I1033b9d9252498db5fe90abe7508393fe2eddd22

Issue #279. Change-Id: I5b6f90b682d77849ce075a2a76a4202c56e5d882

birme · 2017-02-10T16:03:04Z

Ok, so no TS support on Chrome/Firefox in the library?

joeyparrish · 2017-02-10T16:11:08Z

Correct.

ismena · 2017-02-10T17:52:04Z

@avelad I'll take a look!

Issue #279. Change-Id: I860d72f7c22b2d4a8eee863dfcc4277cf80b5fc6

Issue #279. Change-Id: Ic082b268877c2b5336080113c96a73f029a1503c

ismena · 2017-02-10T21:29:19Z

@avelad Done, but there is a catch. In order to play this stream shaka.hls.HlsParser.DEFAULT_TIME_OFFSET constant will have to be changed from 10 to 0.
10 seconds seems to be a default start time offset on Apple-encoded content that's why we also set our expectation accordingly.

We're still in the process of finding a better way to detect the start time of the content if none has been specified in the manifest. If a manifest specifies EXT-X-START:TIME-OFFSET=n we will respect this value over the default.

Issue #279. Change-Id: I3b3935906e7e5989db6788648f2b2400b66369f5

Issue #279. Change-Id: Ie32ab228ab0d00f0def9f2ee8d7c1e496b346ef8

An init segment is not required for all streams, some streams are self-initializing (e.g. MPEG2-TS). Issue #279 Change-Id: I99556ee26f6568cc5e712b50d0bb4de78cb2bd9d

When appending MPEG2-TS segments, we seem to always enter a PARSING_MEDIA_SEGMENT state in the SourceBuffer. While we are in this state we can't change the timestampOffset. This means that we can't change Periods or adapt to a different resolution. To fix this, we call abort() before setting the timestampOffset to reset the state. Because abort() also resets the appendWindowEnd, this merges the two operations so we set timestampOffset and appendWindowEnd at the same time (since we do already). Issue #279 Change-Id: Icd893ded9e3137179f262ba10e55e41d06ec5a77

Issue #279 Change-Id: I491e0e7da2a814276e071fcd98deb66753c03122

Apple encoded content has a default time offset of 10 seconds. Every other packeger we've encountered so far doesn't. New config will default to 0 and allow the application to specify the offset for their content according to the packager's standards. Issue #279 Change-Id: I65b16b05a5974a2f0318cd1f9847c3a56c38b745

Currently HLS parser expects only one media tag to have a given group id. According to the spec that might not be the case. This change adds support for multiple tags with the same gruop id and insures the parser creates variants for all of them. Issue #279. Change-Id: I327e52387f7513464fc56c4b6b8d07ead689d6cc

DanielEliraz · 2017-03-16T22:01:15Z

Can we use the hls feature already?
I cloned the master branch and built it, but I had error 4011 when i injected my stream.

joeyparrish · 2017-03-16T22:09:24Z

@DanielEliraz: The implementation in master is not complete, but error 4011 indicates that the browser doesn't support the content. If it's MPEG2-TS content, Chrome & Firefox do not support that, and Shaka Player does not transmux to MP4.

Until we implement support for protected content for HLS, we reject it. We used to determine it by the presence of an 'EXT-X-KEY' tag. However, it is possible for the manifest to have the tag, but describe clear key content if the 'METHOD' attribute of the tag is 'NONE'. This CL insures we don't reject this kind of playlists. Issue #279. Change-Id: Iee853dd23116c40a43b228f0876efe373c3111af

This change enables HLS parser to get DRM information from the manifest and create DrmInfo objects. Note that until we actually add support for any of the drm systems, we will still reject encrypted content. Issue #279. Change-Id: I4d0652411a567bc75d919c5bf732f20e870a5aeb

According to the HLS spec, CODECS attribute should have a comma- separated list of codecs. The example in the spec doesn't contain spaces between the elements of the list. However, I encountered spaces in the codecs list in content encoded by Apple. It's easy to support both ways, so I decided to do it rather than fight that battle. Issue #279. Change-Id: Iecb706207365a51b5964f013e32ffe13e99604c6

We wanted to see if Media Source would accept a list of codecs for a presentation without specifying which one belongs to which stream. If it was possible, we wouldn't have to try and guess the codecs and also wouldn't be limited to a made up list of codecs that we know Media Source supports. Been there, done that, doesn't work :| Issue #279. Change-Id: Id706022887beaa576eb0f68f2f97a278a0730f30

This will allow me to use it outside of prototype methods in an upcoming change. Issue #279 Change-Id: Iaa7c9a92787d0f33b86dc1ba823f750aa268b9b1

For some reason, in some of my testing, the captured "uri" variable was undefined by the time the Promise was resolved. Instead, use "response.uri", which accounts for redirects anyway. Issue #279 Change-Id: Ibaef99a57dc01a665d095fec4765ac9b32d537ac

Widevine HLS manifests have a data URI that needs to be parsed to generate DrmInfo. Rather than write another parser, factor the data URI networking plugin to expose the parser to other classes. Issue #279 Change-Id: I71f174b86cc3b0ae0e8dbc31701ea9d36cd157ee

Issue #279 Change-Id: I7a7de1f7321210322144469188fd5d99d0040c0b

This abstracts the creation of DrmInfo to parse the EXT-X-KEY tag differently per KEYFORMAT. Issue #279 Change-Id: I2e187dcfc01f7306183c9090c54de58e81872bc4

joeyparrish · 2017-03-31T20:34:15Z

We have support for MP4 VOD HLS, in the clear and encrypted with Widevine. We have TS working on Edge and Chromecast.

I've created new issues for some of the remaining HLS features:

CEA 708 captions: #276
Safari FairPlay: #382
Live streaming: #740
Key rotation: #741
Trick mode: #742
Safari TS: #743

joeyparrish added the type: enhancement New feature or request label Jan 29, 2016

joeyparrish added this to the v2+ milestone Jan 29, 2016

sanbornhilland mentioned this issue May 17, 2016

Safari Fairplay MediaKeys Polyfill #382

Closed

joeyparrish mentioned this issue Oct 4, 2016

HLS subtitles #146

Closed

joeyparrish modified the milestones: v2+, v2.1.0 Oct 4, 2016

ismena self-assigned this Oct 18, 2016

joeyparrish mentioned this issue Nov 16, 2016

DASH trick mode tracks #538

Closed

joeyparrish mentioned this issue Dec 14, 2016

allow passing speed from http plugin #621

Closed

shaka-bot pushed a commit that referenced this issue Dec 28, 2016

Refactor shaka internals to use variants instead of stream sets.

e95d7b9

Issue #279 Change-Id: Idf4241b4e2490876bdc6275685361a5c9bf132b0

shaka-bot pushed a commit that referenced this issue Jan 9, 2017

Add HLS objects and basic routines to parse manifests into them.

8afadb4

Issue #279. Change-Id: I5b6f90b682d77849ce075a2a76a4202c56e5d882

shaka-bot pushed a commit that referenced this issue Feb 10, 2017

Add an HLS asset to the demo page.

8687584

Issue #279. Change-Id: I860d72f7c22b2d4a8eee863dfcc4277cf80b5fc6

shaka-bot pushed a commit that referenced this issue Feb 10, 2017

Add m4s to supported file extensions.

5c5f313

Issue #279. Change-Id: Ic082b268877c2b5336080113c96a73f029a1503c

shaka-bot pushed a commit that referenced this issue Feb 10, 2017

Update the comment about default time offset for HLS.

933f5f9

Issue #279. Change-Id: I3b3935906e7e5989db6788648f2b2400b66369f5

shaka-bot pushed a commit that referenced this issue Feb 15, 2017

Add support for subtitles to HLS parser.

109065b

Issue #279. Change-Id: Ie32ab228ab0d00f0def9f2ee8d7c1e496b346ef8

shaka-bot pushed a commit that referenced this issue Feb 21, 2017

Remove HLS requirement for init segment.

33c277a

An init segment is not required for all streams, some streams are self-initializing (e.g. MPEG2-TS). Issue #279 Change-Id: I99556ee26f6568cc5e712b50d0bb4de78cb2bd9d

shaka-bot pushed a commit that referenced this issue Feb 28, 2017

Comment on the lack of HLS CLOSED-CAPTIONS support

6632897

Issue #279 Change-Id: I491e0e7da2a814276e071fcd98deb66753c03122

shaka-bot pushed a commit that referenced this issue Mar 30, 2017

Make HlsParser.getRequiredAttributeValue_ static

3dfc61a

This will allow me to use it outside of prototype methods in an upcoming change. Issue #279 Change-Id: Iaa7c9a92787d0f33b86dc1ba823f750aa268b9b1

shaka-bot pushed a commit that referenced this issue Mar 30, 2017

Add more mp4 extensions to HLS parser

b549c3d

Issue #279 Change-Id: I7a7de1f7321210322144469188fd5d99d0040c0b

shaka-bot pushed a commit that referenced this issue Mar 31, 2017

Add support for Widevine-encrypted HLS content

db909b0

This abstracts the creation of DrmInfo to parse the EXT-X-KEY tag differently per KEYFORMAT. Issue #279 Change-Id: I2e187dcfc01f7306183c9090c54de58e81872bc4

joeyparrish added the component: HLS The issue involves Apple's HLS manifest format label Mar 31, 2017

joeyparrish closed this as completed Mar 31, 2017

shaka-bot added the archived label Mar 22, 2018

shaka-project locked and limited conversation to collaborators Mar 22, 2018

shaka-bot added the status: archived Archived and locked; will not be updated label Apr 15, 2021

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

HLS manifest parser plugin #279

HLS manifest parser plugin #279

joeyparrish commented Jan 29, 2016

avelad commented Apr 15, 2016

joeyparrish commented Apr 15, 2016

avelad commented Apr 15, 2016

joeyparrish commented Apr 15, 2016

avelad commented Jun 20, 2016

joeyparrish commented Jun 29, 2016

sanbornhilland commented Jul 13, 2016

raskri commented Oct 6, 2016

joeyparrish commented Oct 6, 2016

avelad commented Oct 21, 2016

ismena commented Oct 21, 2016

dougdoe commented Dec 9, 2016

ismena commented Dec 9, 2016

shaharmor commented Dec 15, 2016

joeyparrish commented Dec 15, 2016

birme commented Feb 10, 2017

joeyparrish commented Feb 10, 2017

ismena commented Feb 10, 2017

ismena commented Feb 10, 2017

DanielEliraz commented Mar 16, 2017

joeyparrish commented Mar 16, 2017

joeyparrish commented Mar 31, 2017

HLS manifest parser plugin #279

HLS manifest parser plugin #279

Comments

joeyparrish commented Jan 29, 2016

avelad commented Apr 15, 2016

joeyparrish commented Apr 15, 2016

avelad commented Apr 15, 2016

joeyparrish commented Apr 15, 2016

avelad commented Jun 20, 2016

joeyparrish commented Jun 29, 2016

sanbornhilland commented Jul 13, 2016

raskri commented Oct 6, 2016

joeyparrish commented Oct 6, 2016

avelad commented Oct 21, 2016

ismena commented Oct 21, 2016

dougdoe commented Dec 9, 2016

ismena commented Dec 9, 2016

shaharmor commented Dec 15, 2016

joeyparrish commented Dec 15, 2016

birme commented Feb 10, 2017

joeyparrish commented Feb 10, 2017

ismena commented Feb 10, 2017

ismena commented Feb 10, 2017

DanielEliraz commented Mar 16, 2017

joeyparrish commented Mar 16, 2017

joeyparrish commented Mar 31, 2017