SingleInputVideoGraph implements GraphInput now, so the asset loaders would
interface directly with SIVG, rather than VideoSampleExporter. This is to pave
way for multi-asset video processing.
PiperOrigin-RevId: 547561042
The setPlayerCommand and setSessionCommand builder methods contain assertions to ensure that at most one of these fields is set, but this left it possible to create a command button with no command by calling build on an empty builder.
PiperOrigin-RevId: 547488248
This change also marks the buttons of the custom layout as
enabled/disabled according to available commands in the controller.
Accordingly, `CommandButton.Builder.setEnabled(boolean)` is deprecated
because the value is overridden by the library.
Issue: androidx/media#38
#minor-release
PiperOrigin-RevId: 547272588
`Mp4WebvttDecoder` which used to be `SimpleSubtitleDecoder` will now be called `Mp4WebvttParser` and implement `SubtitleParser` interface. For backwards compatibility, we will have the same functionality provided by `DelegatingSubtitleDecoder` backed-up by a new `Mp4WebvttParser` instance.
PiperOrigin-RevId: 547248157
* Allow more than one input bitmap at a time.
* Allow Compositor to take in and set up an Executor. Otherwise,
Compositor resources may be created on one thread and accessed on another.
* Add a Compositor TestRunner to reuse test code more.
* Update VideoFrameProcessingTaskExecutor to use a new onError listener, so
that it's more reusable in non-DVFP contexts, like for Compositor.
PiperOrigin-RevId: 547206053
durationUs is almost always going to be a larger number than the sample
rate, so pass as the main value, rather than the multiplier.
PiperOrigin-RevId: 547193927
For sync-sample-only formats, we have an optimization to drop all buffers
with less than the start time when writing them to the queue.
For the same formats, if we set a new start time (=seek), we only seek
to the buffer at or before the start time. This means the first sample
in the queue is different depending on whether we seek to a start time
or set a start time and then write samples. This is inconsistent and
effectively means the first sample depends on a race condition between
the Loader thread (writing samples) and the playback thread (attempting
an initial seek in the already loaded samples).
The effect of this inconsistency is that we have to decode one sample
we don't need (and could have skipped) and that some tests become flaky
if the test setup runs into the mentioned race condition.
The fix is to change the SampleQueue seek method to also seek to
a sample at or after the specified time, to align the behavior to the
case where we write the same samples to an empty queue.
The change also clarifies the Javadoc of
MimeTypes.allSamplesAreSyncSamples to note that this should really only
return true if the samples have no "duration" that matters. Otherwise,
we could reasonably return true for most subtitle formats although it
would break subtitle display because we'd remove samples that start
before the seek time.
PiperOrigin-RevId: 547189941
This is needed for several subtitle formats where the `data` passed to
`SubtitleParser` doesn't contain any cue timing, such as PGS, CEA-608,
CEA-708, and DVB.
PiperOrigin-RevId: 547177127
TransformerAndroidTestRunner#run will throw any exception that occurs,
so there is no need to assert the result exception is null.
PiperOrigin-RevId: 546923823
We currently only log the input buffers to the codec, but the test
will be more complete and provide more insights into the playback
behavior if we also dump more information about these input buffers
(timestamps and flags) and also all output buffers with their
timestamps and whether they are rendererd (only relevant for video).
And also use the CapturingAudioSink to output the audio data we produce.
PiperOrigin-RevId: 546898937
Also parse the PCM encoding for lpcm in MP4, and update `MatroskaExtractor`
similarly.
Tested manually in the demo app using an MP4 with 24-bit big endian audio.
PiperOrigin-RevId: 546878505
Some adjustments to the test to make it more correct and prevent it
from becoming flaky:
- Use separate output dump files per test setup. Once we add more data
to these files, they are not guaranteed to be the same anymore.
- Use a seek position that is actually behind the midroll as described
in the test setup.
- Change ad insertion position to ensure the ad group duration doesn't
exceed the underlying media duration.
- Add a wait for isLoading to ensure the late insertion of an ad group
happens consistently at the same processing stage.
PiperOrigin-RevId: 546825183
This changes all MediaSources in our library to allow updates to
their MediaItems (if supported).
Issue: google/ExoPlayer#9978
Issue: androidx/media#33
PiperOrigin-RevId: 546808812
This removes some stack traces in unit tests when a
`RemoveMediaController` was created with `waitForConnection=false`
in which case the missing controller in the
`MediaControllerProviderService` caused a `NullPointerException`.
PiperOrigin-RevId: 546795181
`SubripDecoder` which used to be `SimpleSubtitleDecoder` will now be called `SubripParser` and implement `SubtitleParser` interface. For backwards compatibility, we will have the same functionality provided by `DelegatingSubtitleDecoder` backed-up by a new `SubripParser` instance.
PiperOrigin-RevId: 546538113
`SsaDecoder` which used to be `SimpleSubtitleDecoder` will now be called `SsaParser` and implement `SubtitleParser` interface. For backwards compatibility, we will have the same functionality provided by `DelegatingSubtitleDecoder` backed-up by a new `SsaParser` instance.
PiperOrigin-RevId: 546336035
`DelegatingSubtitleDecoder` is a `SimpleSubtitleDecoder` that delegates to a `SubtitleParser` instance (e.g. future implementations like `WebvttParser`, `SubripParser`, `SsaParser`, `DvbParser` etc)
It will be in the `exoplayer` package, rather than `extractor` and only be used in the decoding flow which happens after SampleQueue (and to be deprecated). When this decoding flow is deprecated and removed, the class can be later deleted.
PiperOrigin-RevId: 546317528
Frame buffers are focused, not current. EGL Contexts may be current but this
is a prerequisite for focusing a framebuffer, and this alone isn't
sufficient to make a bitmap.
Clear this up as it can make some complex tests harder to understand, if the
methods needed don't match up in naming.
PiperOrigin-RevId: 546305118
Add documentation for threading requirements at the class level (in
addition to existing documentation on the methods) to improve
discoverablility. Also fix a couple of nits in the javadoc (US English
spelling, avoid passive voice) and in `OnInputFrameProcessedListener`.
PiperOrigin-RevId: 546303732
Otherwise, errors like `GL_INVALID_FRAMEBUFFER_OPERATION` will only
report a `null` error string, instead of the proper error string.
PiperOrigin-RevId: 546273328
This makes tests more realistic because the returned value matches
the rest of the simulated test time.
It also prevents test flakiness in (yet to be written) tests that
may not correctly advance the playback time in the position tracker.
PiperOrigin-RevId: 546011842
When selecting progressive and HLS tracks, we currently check if
we can either seek in the existing samples or are starting from the
beginning of the stream. In both cases, we don't need to reload the
stream and can continue reading.
Seeking to the beginning isn't side-effect free though because it
also sets the startTimeUs to zero. Whether a seek to the beginning
is successful also depends on whether we already loaded a sample or
not. This mean that the startTimeUs value is set (or not set) based
on the speed of the Loader thread even for the same input starting
from the beginning. This race condition means the actual samples
we write to the stream may differ if the stream has initial samples
with negative timestamps.
We can avoid this race condition by checking if the queue is empty
first, so that we only call seekTo if really needed.
PiperOrigin-RevId: 546010951
Robolectric writes warnings to the log output because the Surfaces
we create are not released and pollute the test environment. This
causes performance issues if tests are run repeatedly.
PiperOrigin-RevId: 545906757
We have the optimization to skip buffers if the release time is
exactly the same as the one for the previous buffer. This makes
sense under the assumption that these buffers get released to
a visible Surface and will be ignored anyway.
However, it's also helpful to provide a customization option to not
do this for cases where the outputting MediaCodecAdapter is not
directly talking to a visible Surface or for tests where we can't
fully control the vsync timing of the Surface and want to ensure
we output all samples that are meant to be shown.
PiperOrigin-RevId: 545906113
BaseGlShaderProgram and SingleFrameGlShaderProgram are technically color space agnostic. useHdr is passed to the TexturePool, which only uses the value to choose between high/regular precision. useHdr is therefore a bit misleading and should be generalized to "useHighPrecisionColorComponents."
PiperOrigin-RevId: 545726629
This makes tests more realistic because the returned value matches
the rest of the simulated test time.
It also prevents test flakiness in (yet to be written) tests that
may accidentally drop output buffers or calculate the wrong values.
PiperOrigin-RevId: 545690008
Based on the spec, ETSI TS 102 366 V1.4.1 Annex F, 6 bits should have skipped instead of 6 bytes.
This correction was pointed out in Issue: androidx/media#474.
PiperOrigin-RevId: 545658365
This allows us to simplify and reduce thread initialization, hops and the
synchronization cost associated with them, especially when more and more VFPs
are used.
PiperOrigin-RevId: 545622169
Upon track transition of offloaded playback of gapless tracks, the framework will reset the playback head position. The AudioTrackPositionTracker must be made to expect the reset and cache accumulated sum of rawPlaybackHeadPosition.
#minor-release
PiperOrigin-RevId: 545602979
This allows MediaSources to accept MediaItem updates after creation.
This CL adds the handling and plumbing logic in `ExoPlayerImpl`,
`ExoPlayerImplInternal`, `MediaSourceList` and `MaskingMediaSource`.
It also updates all forwarding/wrapping sources to forward these calls
to their wrapped instance.
The actual functionality is only added to `FakeMediaSource` instances in
tests so far.
PiperOrigin-RevId: 545450210
This plumbs a duration through `SampleQueue` which will make it easier
to handle transcoding muxed subtitles from e.g. SubRip to
`text/x-exoplayer-cues`.
This change is a no-op to the end-to-end behaviour of ExoPlayer because
currently we only support parsing sideloaded subtitles before
`SampleQueue`, and by adding the duration we don't affect the cues that
are ultimately output by `Player.Listener.onCues` (as shown by no change
to the golden files for `WebvttPlaybackTest` in this commit).
I considered making `CuesWithStartTimeAndDuration` implement
`Bundleable` (and deleting `CueEncoder/Decoder`) but decided against
it because we are deliberately not encoding `startTimeUs` (since that's
encoded as the sample time in `SampleQueue`). I also considered
introducing another type that only has `List<Cue>` and `durationUs`
fields, but it didn't seem necessary, since we want `startTimeUs`
everywhere else (except inside `SampleQueue`).
PiperOrigin-RevId: 545226847
Also make some casting that is needed a bit safer.
Error Prone was complaining about casting `long` to `int` before passing
it to `Extractor.seek(long, long)`:
> Converting a `long` or `Long` to an `int` to pass as a `long`
> parameter is usually not necessary. If this conversion is intentional,
> consider `Longs.constrainToRange()` instead.
PiperOrigin-RevId: 545226063
This is a corollary of `SubtitleDecoderFactory`. We may add a
`boolean supportsFormat(Format)` method later if it turns out to be
useful.
PiperOrigin-RevId: 545225489
The timeout time of 5000ms is quite long, and means that for tests with >1
VFPTestRunner, like texture output and compositor tests, we have to wait 10s.
The latch allows us to continue with the test once we're done processing input,
instead of needing to wait longer, until the timeout is reached.
PiperOrigin-RevId: 545159839
The actual errors are all positive hex values. Without this CL, we must first
convert decimal errors to hex ones before figuring out what went wrong.
PiperOrigin-RevId: 544695961
We introduced truncation to 32 chars in <unknown commit>
and included indent and offset in the calculation. I think this is
technically correct, but it causes problems with the content in
Issue: google/ExoPlayer#11019 and it doesn't seem a problem to only truncate actual
cue text (i.e. ignore offset and indent).
#minor-release
PiperOrigin-RevId: 544677965
This avoids issues with potential number clashes with ID_UNSET.
Also fixes some further parsing where adaptation set ids are referenced
from manifest properties.
Auto-resume playback when the removal of playback suppression due to unsuitable output is conveyed via change in playback suppression to Player.PLAYBACK_SUPPRESSION_REASON_NONE within a configurable timeout defaulting to 5 minutes.
PiperOrigin-RevId: 544411987
Instead of playing or pausing itself, the ExoPlayer implementation should only update the playback suppression reason as and when audio outputs are added or removed dynamically.
PiperOrigin-RevId: 544379033
On a MediaItem change, the input Format (and Effects to apply) may be
different. Therefore the AudioProcessingPipeline must be reconfigured
to determine what processing is active, and what the AudioFormat of the
data output is. In the event that it is different, additional
AudioProcessor instances must be used to ensure the encoder will still
be able to accept the audio buffers.
PiperOrigin-RevId: 544338451
Add a fail-fast check in `ExoPlayerImpl` to ensure the equality of the lengths of `ShuffleOrder` and the current playlist. Also improve the documentation of `setShuffleOrder(ShuffleOrder)` with explicit instruction on this.
Issue: androidx/media#480
#minor-release
PiperOrigin-RevId: 544009359
Providing the sync token in the api allows the client to decide which waiting method they would like to use depending on the use case, allowing them to optimise if possible.
PiperOrigin-RevId: 543997311
This fixes a broken javadoc link and makes it possible to customize media
source creation via the asset loader factory without also having to
implementing a decoder factory from scratch.
PiperOrigin-RevId: 543995538
It currently wrongly documents that it is only called before reading
streams (that has never been the case and all MediaPeriods already need
to handle calls after reading samples from the streams).
It was also a bit unclear what a discontinuity implies and the new
Javadoc calls out the main use case for discontinuties and the intended
meaning of returning a discontinuity.
#minor-release
PiperOrigin-RevId: 543989124
We use `@Nullable` for IntDefs in other places and don't use this suffix
elsewhere, so I don't think we need it here either.
PiperOrigin-RevId: 543687285
Goal of tests (SequenceExportTest) that use this media is for the
silence and the media to match exactly with audio format, however
`sample_with_increasing_timestamps.mp4` had a different sample rate.
testvid_1022ms.mp4: channel count = 2, sample rate = 44100.
PiperOrigin-RevId: 543458948
MP4 edit lists sometimes ask to start playback between two samples.
If this happens, we currently change the timestamp of the first
sample to zero to trim it (e.g. to display the first frame for a
slightly shorter period of time). However, we can't do this to audio
samples are they have an inherent duration and trimming them this
way is not possible.
#minor-release
PiperOrigin-RevId: 543420218
*** Original commit ***
Mark output sample as decode-only based on start time
We currently do the same check on the input timestamps and
expect the output timestamps to match. Some codecs produce
samples with modified timestamps and the logic is a lot safer
when the comparison with the start time is done on the output
side of the codec.
Issue: google/ExoPlayer#11000
***
PiperOrigin-RevId: 543379665
FallbackListener.onTransformationRequestFinalized() is called from the
AssetLoader thread for audio, and from the GL thread for video.
PiperOrigin-RevId: 542851284
By passing this class where it's needed, implementations don't need to store it
in a field (reducing boilerplate) and it's clearer that it can't be unset when
needed.
PiperOrigin-RevId: 542823522
`MediaControllerImplBase` has 2 methods for updating listeners about `PlayerInfo` changes - `updatePlayerInfo` (for masking the state) and `onPlayerInfoChanged` (when communicating with the session). There is a set number of listener callbacks related to `PlayerInfo` updates and both methods should go through the same control flow (whether we know that masking will ignore most of them or not).
A unified method `notifyPlayerInfoListenersWithReasons` encapsulates only the shared logic of 2 methods - listeners' callbacks. This ensures that both methods call them in the same order and none are missed out.
PiperOrigin-RevId: 542587879
With the upcoming "handle format changes" CL, stereo -> mono audio
would add an AudioProcessor. Robolectric decodes output encoded data,
which crashes some AudioProcessors because the number of frames may not
be an integer.
PiperOrigin-RevId: 542568875
This change uses this new method everywhere we currently `instanceof`
check an `Extractor` directly. This allows us to introduce
wrapping/delegating `Extractor` instances - because the `instanceof`
checks will continue to operate on the underlying instance.
HLS is a slightly different case, because it directly re-instantiates
`Extractor` instances, which is not compatible with an arbitrary
wrapping structure. Luckily the only `Extractor` instances that HLS
re-instantiates do not support muxed subtitles, so won't be wrapped
in the first place (although future changes might use the
delegating-`Extractor` pattern for other purposes, which might affect
HLS).
PiperOrigin-RevId: 542550928
The callbacks for `PlayerInfo` changes are currently in both `MediaControllerImplBase.updatePlayerInfo` (masking) and `MediaControllerImplBase.onPlayerInfoChanged`. But the order was different between them both and `ExoPlayerImpl.updatePlaybackInfo` which they are trying to mimic.
#minor-release
PiperOrigin-RevId: 542519070
Devices pre-API 33 are not able to comprehend the position reset that occurs by the HAL in offloaded gapless track transitions.
PiperOrigin-RevId: 542503662
The effects pipeline must receive images in the sRGB colorspace due to the color transfers applied in the shaders. Currently the burden to making sure images are in the right colorspaces falls onto apps. This CL ensures that this is not the case anymore.
PiperOrigin-RevId: 542323613
When an app tried to re-prepare a live streeam with server side inserted
ad after a playback exception, the player tried to find the ad group by
its index in the ad playback state of the next timeline when creating
the first period.
If a source that supports server side ad, has removed the ad playback
state when the source has been removed, this causes a crash. For live
streams this is a reasonable thing to do given the exception could be
caused by an invalid ad playback state.
This change removes the ad metadata from the current period for live
streams and the timeline. In case the ad playback state is not reset
by the source, the first timeline refresh would ad the metadata again.
PiperOrigin-RevId: 541959628
Some events may arrive after the playlist is cleared (e.g. load
cancellation). In this case, the DefaultPlaybackSessionManager may
create a new session for the already removed item.
We already have checks in place that ignore events with old
windowSequenceNumbers, but these checks only work if the current
session is set (i.e. the playlist is non-empty). The fix is to add
the same check for empty playlists by keeping note of the last
removed window sequence number.
PiperOrigin-RevId: 541870812
FakeClock currently doesn't work well with Espresso and Compose UI
tests because view interactions in both frameworks intentionally idle
the main looper to handle pending UI effects. However, this also
advances playback progress even though we want to deterministically
trigger progress from the test itself.
To solve this problem, we can detect the idling Robolectric call and
postpone any further updates until we leave this state.
PiperOrigin-RevId: 541831050
Add a Wear OS specific implementation of 'Player.Listener' to help resolving the playback suppression due to unsuitable output by launching a system media output switcher dialog.
PiperOrigin-RevId: 541698125
Audio only tests are now using RAW audio where possible, which is
passed through the Robolectric decoders/encoders, and can be handled by
the AudioProcessor instances accurately.
PiperOrigin-RevId: 541648853
To more accurately describe what they do, especially as Compositor will
starts to use more contexts or threads, and it's important to know what
needs to be reset/recreated/focused before what methods.
PiperOrigin-RevId: 541010135
Issue: When running the Transformer related test cases, the tests are flaky
because the order in which audio and video samples are interleaved seems to
differ in few instances.
Root cause: When running a transformation the sample producer (Asset loader)
and sample consumer (Sample pipeline) both runs on different thread and
theoretically there is no reason for behaviour to be deterministic because
the number of samples produced/written depends on how fast individual thread
works. So it is indeed surprising that test somehow worked deterministically in
majority of instances (may be something to do with Robolectric environment).
Solution: Since we don't expect the order of sample interleaving to be deterministic, make the dumping logic deterministic where all the video
samples will be collected and then dumped together (similarly for audio). This would mean we won't be able to see the interleaving so for that we need to
add separate test case verifying the interleaving logic only.
Pending: Test case for interleaving.
PiperOrigin-RevId: 540930871
