This change pulls manifest refresh responsibility up to the
top level Dash/SS SampleSource implementations. In following
steps more of the manifest processing logic will be pulled
up (e.g. extracting track groups from the initial manifest),
which will allow ChunkSampleSource/ChunkSource instances to
be further simplified and created on demand.
I've avoided moving/renaming anything for now, so as to keep
it fairly easy to review.
Note that this change does the TODO related to releasing the
manifest fetchers.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=121001139
This is a mechanical change. The two new SampleSource classes
are forked from MultiSampleSource, with the logic that previously
was in the demo app's SourceBuilder methods copied into the
corresponding constructors. Subsequent steps will:
1. Pull the initial manifest processing from DashChunkSource and
SmoothStreamingChunkSource into the new SampleSource classes.
2. Remove the construction of ChunkSampleSource instances from
the constructors, instead instantiating children only as needed
when tracks are enabled.
3. Simplify ChunkSampleSource down into a ChunkTrackStream object.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=121001021
pageHeader.headerArray is only used in getNextSeekPosition()
in which populatePageHeader() resets them both at the beginning
anyway.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120902428
If rendererMediaClock is non-null, the rendererMediaClockSource is enabled or
started, so remove the check in updatePositionUs.
When disabling renderers for track selection, renderers with changing track
selections always transition from started -> enabled.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120819483
- You can't link to a generic type (apparently).
- Suppress unchecked conversion warning.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120817601
Plus add processOutputBuffersChanged just for consistency
with processOutputBuffer and processOutputFormat.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120811608
Functionality is moved into Eia608Parser, so that Eia608Parser
can be used with TextTrackRenderer, similar to all the other
text/subtitle formats. Modified some of the other
text/subtitle-related classes to support the new behaviour.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120755145
This CL allows near-complete support to CSS selectors (I say near because not every
CSS rule applies to WebVTT).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120717498
Now uses buffer position + limit in both new and legacy modes.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120713837
Reading the format and/or a sample needs to be done as a
single operation. Else you can have a situation where the
queue is initially empty, and this happens:
1) Read downstream format X
2) Read downstream format X (unchanged)
3) Write format Y
4) Write first sample
5) Read first sample
The first sample then appears to be format X rather than Y.
Note that readData in the SampleSource implementations always
looks roughly the same. readReset is identical in all cases.
isReady is identical in all cases now I've fixed them to be
that way. So it should be pretty easy to get DefaultTrackOutput
to implement TrackStream directly, at which point a whole load
of duplication will disappear.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120546377
- Code is simpler. We only ever reset all tracks.
- Allows the standalone media clock to be updated properly. This
allows simpler recovery for live streams in ExtractorSampleSource.
- Fixes#1285 and paves the way for a fix for #758.
Issue: #1285
Issue: #758
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120530682
I think the concept of a sparse track might need formalizing
in Format at some point. We should probably do a similar thing
with sparse tracks in ExtractorSampleSource as well. WDYT?
Issue: #551
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120530195
This is just wrong. I think there used to be an off-by-one
timestamp error in YouTube's fmp4 streams, and this code
was initially designed exclusively to play such streams. I
don't see this issue any more though, if it was ever there!
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120447694
Having moved HLS to use single sample queues per track, these
classes have become relatively similar. This CL aligns the two
to make this more obvious. It remains unclear whether it'll
ever be sensible to merge them; there are still some niggly
complications for HLS.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120429618
This CL provides the necessary infrastructure to add styling by class. This was separated
into two different CLs to ease reviewing.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120336976
This can occur in some Broadcast Wave Format (BWF) files, such as those
produced by the Zoom H2n. See the included sample for an example.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120334117
- RollingSampleBuffer -> DefaultTrackOutput
- TsChunk -> HlsMediaChunk
- Established hls.playlist package for HLS playlist things
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120325049
Notes:
- RollingSampleBuffer will be renamed DefaultTrackOutput in a
following CL, and variable naming will be sanitized.
- TsChunk will also be renamed to HlsMediaChunk, since it can
be used for non-TS containers (e.g. MP3).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120240243
This also fixes the largest queued timestamp to be the
correct value if upstream samples are discarded.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120207054
At this point the only reason preventing the chunk package
from using RollingSampleBuffer directly, rather than
DefaultTrackOutput, is that the latter maintains the largest
parsed timestamp. This will be pushed inside the former in
the next CL. Following that, splicing logic will also be
pushed inside of RollingSampleBuffer, and HLS will be moved
over to using a single RollingSampleBuffer per track, with
the splicing done inline.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120206658
This is the first step toward merging of RollingSampleBuffer
and DefaultTrackOutput, which is a precursor to removing some
indirection for DASH/SS playbacks, and to moving HLS playbacks
to use a single output buffer per track (with inline splicing).
Where this is heading is that sample format changes will
eventually be attached to samples in the rolling buffer. This
will eliminate the need for piping sample formats around the
edges of the rolling buffer (e.g. via redirection in
ChunkExtractorWrapper and BaseMediaChunk.getSampleFormat).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120123093
There's a more nuanced version of this where we actually
disable on the loading side, but it's quite tricky to get
the threading just right. I have a change that I'll
probably manage to clean up and send out at some point.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=120106967
- Use same constant for unknown/unset microsecond times/durations.
- Change constant values to be nowhere near the "normal" range.
- Misc cleanup.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119944019
For DASH + SS, sample data is currently plumbed through:
Extractor
->ChunkExtractorWrapper
->ContainerMediaChunk
->DefaultTrackOutput
->RollingSampleNBuffer
This change bypasses the ContainerMediaChunk layer. It
should be possible to completely delete DefaultTrackOutput
in the future, but such a change may well be tied to
changes in HLS format splicing + how we buffer (or don't
buffer) disabled tracks.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119853857
prepare(SampleSource) is renamed to setSource(SampleSource). The player
immediately transitions to STATE_BUFFERING when the source is set, at which
point doSomeWork is called.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119838825
Allow styling <v Someone>Hello</v> with ::cue(v[voice="Someone"]) { ... }.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119748009
TrackRenderer and SampleSourceTrackRenderer both now use a TrackStream
so they can be merged.
(This may also be useful for adding playlist support, in case TrackStreams need
to be replaced during playback.)
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119743228
This CL allows style blocks to reference elements. For example: we could style
a cue with text "Sometimes <b>bold</b> is not enough" with the style block
::cue(b) { ... }.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119734779
This is in preparation for the player knowing about the live
window, at which point only correct playback positions will be
passed into ChunkSource implementations. Note that both
implementations bound the chunk index within range in case of,
for example, off-by-one errors caused by very recent manifest
updates.
This will temporarily cause live playbacks to always start at
the beginning of the current live window (we'll be trying to
play from t=0, which will be adjusted to the start of the live
window as a result of the bounding).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119733559
FrameworkSampleSource will still be useful for audio, where
sample interleaving isn't an issue. We could optionally add
a "don't wait for first frame" boolean to the video renderer
if we *really* need to keep some form of this workaround in
place, but I'd rather not do so for now.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119733224
In CSS, ids are references using #. The absence of # references elements.
NOTE: If the id of a cue was "1", we support its reference with ::cue(#1).
In CSS, however, this is not valid, and the number should be escaped with
\3 as in ::cue(\31). We still do not use number escaping (and I doubt
whether we should at some point).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119634708
This CL removes a warning by adding the import, and removes a qualified reference in the process.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119618343
This CL adds the support of CSS styling in Cues through id and "universal" cue selector.
The more sophisticated selectors will be left for later, because they requier a bit more
complex logic. Also narrowed a little bit the responsibilities of the WebvttCueParser to
move some to the WebvttParser.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119547731
Both of these features are being promoted to first class
citizens in V2 (multi-period support will be handled via
playlists, seeking-in-window will be handled by exposing
the window/timeline from the player and via the normal
seek API). For now, it's much easier to continue the
refactoring process with the features removed.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119518675
Note that the DataSourceInputStream read methods
are implementing a different interface (InputStream,
not DataSource), which is why -1 is still used in
that case.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119180851
This method is needed by FlacExtractor to release native resources.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=119143922
This replaces calls to unescape except for SEI unescaping.
Use the new ParsableNalUnitBitArray for reading the slice header in HLS
access unit detection and slice_type reading.
Unescape the SPS before parsing in FLV and MP4. Before this change it was
parsed in its original (escaped) form.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=118777869
The only change that's not re-ordering is to add a Util method
for usToMs to replace the ones in the HLS and Chunk sources.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=118775359
This was made possible by the simplification of how DASH/SS
chunk replacement works. It is also a step towards eliminating
continueBuffering(), since continueBuffering() calls are no
longer relied upon to resume a backed off load.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=118774865
Step 6 of the refactor involves moving the logic that's
currently in the SourceBuilder classes in the demo app
into new SampleSource classes in the library. These classes
will construct video/audio/text pipelines on-demand (i.e.
when tracks are enabled) rather than constructing them all
up front as is currently the case in the SourceBuilder
classes. Hence we need a way to instantiate DataSource
instances (i.e. a DataSourceFactory ;)).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=118722180
DASH + SS previously had a ridiculously complicated chunk
replacement mechanism in resumeFromBackOff. It also didn't
allow replacement of the first media chunk in the queue,
even though it's possible to remove it in the case that no
corresponding samples have been consumed.
This CL moves DASH + SS to the simpler model used in the
HLS implementation, where the chunk source has a single
opportunity to cancel (and hence later replace) the chunk
when the load error occurs. With this change comes the
ability to replace the first media chunk in the queue in
all cases where it's possible to do so.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=118573418
This fixes some nuances with the initial solution. Mainly,
that the TrackStreams returned by selectTrack could not be
used safely until after endTrackSelection was invoked. It
also reduces the need for member variables to track state
between the track selection methods.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=118556651
In V2 we'll at some point start using DataSource factories
for creating DataSource instances. If there are two DataSource
interfaces this gets unnecessarily awkward.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=118470751
This removes the need for each SampleSource implementation to
implement 3x "if(condition) {noop}" tests (ChunkSampleSource
and SingleSampleSource were missing some of these checks).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=118036233
This change optimizes startup and track selection for HLS. Changes
in HlsChunkSource avoid unnecessary re-requests for media playlists.
Changes in HlsSampleSource optimize exit from the limbo state (i.e.
when endTrackSelection is first called).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=118026962
1. SampleSource now has an explicit track selection state. This state is entered
after the source is prepared, and also by calling startTrackSelection.
2. endTrackSelection commits selection changes, and is responsible for doing the
right thing w.r.t starting/stopping/restarting load operations.
3. All sources now start or restart a load in the case of a new track selection.
This fixes a problem where a source could be advanced by repeatedly disabling
and re-enabling whilst paused. Some sources didn't restart a load in this case,
since the position was unchanged, however the downstream renderer would then
consume media up to the first keyframe in order to render something. Hence
each disable/re-enable would advance by a keyframe.
4. This change will enable a subsequent change where we'll discard media for
non-selected tracks earlier than we do currently (i.e. we'll hook the extractor
to a dummy track output, so the samples will never be written to a rolling
buffer). This will enable a further subsequent change where buffer contributions
are per-renderer rather than per-source.
Issue: #1041
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=118024436
This fixes an issue where the PsExtractor would start reading
unsynchronized if sniff was called.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117958077
Expose the input buffer for Exoplayer V2. This allows subclasses to
parse the input buffer before it is decoded. One particular usage
of this is to allow parsing user data stored in the tracks
(e.g. SEI in H264), and incorporate the user data into the rendering.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117865971
Also use MediaCodec buffer flag constants instead of those on MediaExtractor.
This is in preparation for merging InputBuffer and SampleHolder.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117810136
This version only supports 16-bit uncompressed PCM. A follow-up CL will
add support for other sample bit depths.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117809475
If this situation is encountered, we assume that the encoder has a good reason to
do this and use the last pts + frameDuration as new pts.
Issue: #1295
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117808961
- Remove special DefaultTrackOutput.sampleData method, and have
SingleSampleMediaChunk use the regular one instead.
- Make DummyTrackOutput behave correctly is allowEndOfInput is
false.
- Simplify progress tracking in ExtractorSampleSource.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117808659
This is the first version and is still not linked to the WebVTT parser nor
does it support all the intended features, but it was left this way to
ease the review a little bit.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117722492
- Not doing any renaming for now. It'll be easier to wait
until after the extensions themselves are brought across.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117438338
When the ctts contained an entry that had a 0-valued entry count, the
parser would miss every other entry, failing the final assertion.
The standard does not seem to prevent the value 0 in the sample_count
field, so we need to allow it.
Issue: #1326
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117241945
When the edited sample sequence does not contain any sync sample Exoplayer does not provide
support. This CL changes the ArrayIndexOutOfBoundsException for a more explicative
ParserException.
Issue: #1336
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117241805
The MP4 standard considers the udta box as a regular container box. Quicktime, however,
considers it as a container (of only leaf atoms) that can have a terminating 32 bit
integer with value 0. Since this breaks the principle of not having content in container
boxes, this CL considers the udta box as a leaf box that contains other boxes and does
the parsing manually.
Issue: #1315
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117237255
This fix derives from issue #1308, which came up in unfragmented mp4 files.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117236416
The parsing of multiple moov boxes for a single ExtractorOutput incurred in
an assertion failure due to repeated track declarations. This CL makes each
new moov box replace any previous one. This change is transparent to the
client, no flags are provided to allow this feature.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117236246
The minimum compression ratio matches the Nexus 5X MPEG-4 video decoder.
Issue: #1290
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117234657
If a container box is empty, it is never removed
from the container box stack, breaking the extractor.
Issue: #1308
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117228211
- Made enabledRenderers an array to avoid loads of method calls.
- Made if so that enabled renderers are always called in a consistent
order, rather than their order changing if they're enabled/disabled
over time. This is likely to make performance more predictable.
- Split out reading of resets into a separate method. This method is
now called directly after seeking on the source, so as to ensure
instant propagation of the new position from source->renderers in
the common case.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117225639
Some devices fail to decode an avc3 stream that doesn't start with an SPS (for
example, if an access unit delimiter appears first). Workaround the issue by
discarding input sample data up to the first SPS on those devices.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117224602
When reading unknown duration files with CBR seeking, the Mp3Extractor could
try to read sample data from 0. This happened because synchronization did not
skip over the ID3 data when it immediately found valid frames. When the invalid
sample data is read, the extractor tries to resynchronize from the next byte
(at offset 1), and this fails because the ID3 data at the start of the file is
longer than the synchronization search distance.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117224270
This is needed to support fully demuxed audio in HLS. For the
sample I have the video (only) variant still declares an AAC
stream. I suspect there's at least one toolchain out there that
hardcodes H264 and AAC streams in TS output.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117224224
The upstream source (or its stream) should always provide data
starting from a sync frame, so this logic shouldn't be necessary.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117220759
allow "center".
This value (and not "middle") is listed in
https://w3c.github.io/webvtt/ ( WebVTT: The Web Video Text Tracks Format, Draft
Community Group Report, 21 December 2015).
Leaving the behavior for "middle" unchanged.
It was the value listed in older drafts, e.g.
https://www.w3.org/TR/2014/WD-webvtt1-20141113/ ( W3C First Public Working
Draft 13 November 2014 )
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117220612
Failing to parse a UUID from a ContentProtection should only
result in filtering if there was actually a cenc:pssh element
there for us to get it from.
Issue: #1256
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117220417
This avoids accessing PtsTimestampAdjuster through a thunk method,
and allows the PesReader classes to be static, which is nice in
general.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=117220139
