Removes duplicated code and starts cleaning up handling of media clocks.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=174829840
This prevents users from having to check sideloaded subtitles URLs before
preparing a SingleSampleMediaSource with it.
Issue:#3140
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=174475274
This is the first CL in a series to add chunkless preparation support.
Also did a bit a tidying up in HlsSampleStreamWrappen and
HlsMasterPlaylistParserTest.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=174461737
This also allows exposing multiple CC channels to any fMP4 extractor client.
Issue:#1661
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=174458725
First fix, prevents forced rewriting when cipher is set but encrypt is
false.
Second, removes the store() call in SimpleCache.initialize() so
initialization doesn't fail because of CachedContentIndex write issues.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=174450586
The mediaChunks.size() > 1 check was supposed to ensure this, and
did roughly the right thing when there was only a single stream
(although it was unnecessarily restrictive in preventing chunk
cancelation for the first chunk, where bytesLoaded != 0 and none
of the samples had been consumed).
Now we have multiple streams the check doesn't do the right thing,
and adding a back-buffer feature will make even more incorrect.
This change switches to checking the condition we actually want
to check directly :).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=174449398
*** Reason for rollback ***
Breaks setting PlaybackParameters before start of playback
*** Original change description ***
Add support for float output in DefaultAudioSink
Also switch from using MIME types to C.ENCODING_* encodings in DefaultAudioSink.
***
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=174445506
Client can add this functionality by extending DownloadService.
Also made DownloadManager accept multiple listeners. So instead of
broadcast event, client can listen to DownloadManager directly.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=174335820
Instead of using an Executor to run DownloadTasks creates and manages
threads internally.
Also added DownloadThread internal class to better separate the code
that doesn't run on the main thread.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=174036872
1. Move Timeline/Manifest into PlaybackInfo
2. Don't update externally visible Timeline/Manifest during preparation
3. Ignore MSG_POSITION_DISCONTINUITY during preparation
4. Correctly set masking variables at start of preparation, and use them
Once this change goes in, PlaybackInfo will contain timeline, manifest
and position, which should always be self-consistent with one another.
The next step would then be to move a bunch of logic in ExoPlayerImpl
that derives state from timeline and position into PlaybackInfo, and
split that into its own top level class that can be easily tested to make
sure it never IndexOutOfBounds.
I think we could also replace the masking variables and instead just assign
a new PlaybackInfo to the playbackInfo variable whenever we're doing
something that requires masking. This should be possible because we no
longer update playbackInfo whenever we have pending acks. It would
require allowing PlaybackInfo to mask the window position internally when
the timeline is empty, but I think this is ok, and again is something we
could test pretty easily.
Issue: #3362
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=173909791
This makes it a bit more obvious what's going on during
preparation. In particular, it makes it clear that
MSG_SOURCE_INFO_REFRESHED arrives before MSG_TRACKS_CHANGED.
Issue: #3362
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=173392080
Also switch from using MIME types to C.ENCODING_* encodings in DefaultAudioSink.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=173379623
We don't expect this case to occur, since track selection is
normally expected to check canAcquireSession before selecting
a track. Nevertheless, if an attempt is made to acquire a
session when the media doesn't support the manager's UUID, we
should fail in a more graceful way.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=173124170
Other catch blocks in this class catch everything. This one should
too.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=173118891
Tasks conflict if both of them work on the same media and at least one
of them is remove action.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=172741795
We have been using USE_CHORD_PITCH == false for a while and the quality of
pitch changes seems fine. It's now possible to set the sample rate too, but
this only works if USE_CHORD_PITCH is false, so remove the constant.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=172736631
- When transitioning to a new period, the value of bufferedDurationUs
passed to TrackSelection.updateSelectedTrack was incorrectly set to
0. It should have been set to correctly reflect buffered media in
previous periods still being played out.
- This change fixes the issue described above, and also propagates the
playback position through to this method. The position of the next
load within the period can be calculated by adding the position and
bufferedDurationUs.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=172736101
If connecting a Bluetooth audio device fails, the AudioTrack may be left in a
bad state, where it is not actually playing and its position has jumped back to
zero. Detect and work around this case by resetting the track.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=172600912
prepare and selectTracks receive the position from which any
loading should start, where-as continueLoading receives the
actual playback position. These are different in the case that
a previous period is still being played out.
Also removed "relative to the start of the period" from prepare
documentation because it couldn't really be relative to anything
else.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=172592769
newPlayingPeriodHolder could be set then updated if seeking to a repeated period
that was loaded more than once. This led to MediaPeriodHolders leaking.
Only set newPlayingPeriodHolder once so that any later holders with the same
period identifier get released.
Also add a regression test. FakeMediaSource checks that all created
MediaPeriods were released when it is released.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=172591937
This is not really useful with the DefaultAudioSink, but could be used in a
custom AudioSink when mixing audio from sources that have different sample
rates.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=172434482
1. Ignore edit list where the sequence doesn't contain a sync
sample, rather than failing.
2. Make Mp4Extractor.readAtomPayload so it doesn't try and read
the same payload twice if a failure occurs parsing it.
3. Make processAtomEnded so that it doesn't pop the moov if
parsing it fails.
Issue: #3351
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=172106244
This change fixes various issues:
- MobileHarness sometimes allocated devices with SDK < 16. As we have no tests running
on these SDKs, a new dimension filter for the mobile_test target ensures that only
devices with SDK >= 16 are selected. A similar filter for SDK version is also added
to the ABR playback tests to ensure no old devices are selected.
- DRM specific tests are skipped for Api < 18, but were not able to run because the
DashTestRunner class tried to link to the MediaDrm constructor. Moved the
constructor to a seperate Builder class to allow execution on Api levels 16 and 17.
- DashWidevineOfflineTest also tried to access code for Api >= 18 without checking
the current level.
- Action implementations which are waiting for events did not ensure that they have a
nextAction to wait for. This caused NullPointerExceptions when this next action was
scheduled.
- DefaultDrmSession always restored the offline keys when a new license was requested,
even if the keys were already restored. These repeated slow calls to restoreKeys
resulted in high numbers of dropped buffers.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=171974859
Update the audio track selection logic in DefaultTrackSelector:
- When forcing lowest bitrate, use bitrate as tie-breaker when track scores are
the same, prefer the lower bitrate.
- Otherwise, use one of the following values as tie-breaker in order:
- ChannelCount
- SampleRate
- BitRate
If the format being checked is within renderer's capabilities, select it if it
has higher tie-break value, else, select it if it has lower tie-break value.
If all tie-break values are the same, prefer the already selected track.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=171803092
MediaCodecRenderer implementations require DrmSessionManager<FrameworkMediaCrypto>,
but it's currently not possible for an app to provide a custom implementation due
to FrameworkMediaCrypto having a package private constructor. This change exposes
public FrameworkMediaCrypto constructors, hence removing this restriction.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=171718853
Calling HandlerThread.quit() or .quitSafely() doesn't immediately terminate
the thread. It just instructs the Looper not to accept any new messages and
to terminate at the next opportunity. Added a HandlerThread.join() everywhere
where the intention is to close and release all resources and to stop all
threads.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=171525241
This change also replaces individual DownloadAction versions with a
single master version.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=171273880
For initial DRM provisioning and key request, we allow the requests to be
retried (with increasing delay for each successive retry) before failing.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=171271384
Controls are still hidden while playing ads, but if the app pauses the player,
controls will be shown. During ads, the player is not seekable.
When the player enters the background then returns to the foreground, the
content period may not be prepared, so also cache the content window duration.
This means that if the app reenters the foreground while an ad is paused the
time bar can be populated.
Issue: #3303
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=171123428
Now this counter includes input buffers too, which are dropped as part of
skipping to keyframes for catch up.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=171119930
If the current output buffer is very late and the playback position is in a
later group of pictures, drop all buffers to the keyframe preceding the
playback position.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=170975259
This is useful to determine when a seek request was processed by the player
and all playback state changes (mostly to BUFFERING) have been performed.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=170826793
Updates DefaultDrmSessionManager to use the prefered Widevine version (v1
on >= 23 and v0 for < 23).
For other DRM schemes, uses the first scheme found.
* Always assume a renderer is ready if it's read to the end of
its current stream and there's a subsequent period already
prepared. This prevents getting stuck when a non-clock renderer
has a short stream.
* Switch to the standalone clock if the renderer providing the
media clock has read to the end of its current stream, is no
longer ready, and there's a subsequent period already prepared.
This prevents getting stuck when a clock renderer has a short
stream.
* Remove unnecessary clock synchronization logic (since it would
need to be made more complicated as a result of this change).
* Don't update the playing period holder when playWhenReady is
false. This avoids the position jumping to the start of the
next period when seeking to the very end of the current period
whilst paused (we still end up showing the first frame of video
from the next period, but fixing that will have to wait).
Github: #1874
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=170717481
This avoids spurious position reports following an underrun.
Github: #1874
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=170344399
This change allows applications to provide custom AudioSinks, which could be
based on android.media.AudioTrack like AudioTrackAudioSink, or could be
completely custom.
The refactoring is mostly mechanical and shouldn't result in any functionality
changes.
Some android.media.AudioTrack-specific details have to appear in the AudioSink
interface so this change modifies the javadoc on the AudioTrack (now AudioSink)
to note that some methods will have no effect.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=170311083
This allows simplified listener implementations as most listeners
will not listen to all possible notifications.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=170177821
This includes both cbcs and cenc. Will only work for streams that require a single
pssh.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=169382884
There's no reason to perform the discontinuity check or skip
the adaptation field if we don't have a payload reader for
the packet.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=169374609
This fixes infinite loop in LeastRecentlyUsedCacheEvictor.evictCache when index store fails.
Also made CachedContentIndex not final so it can be mocked and added a package protected SimpleCache
constructor so mock index can be injected.
Issue: #3260
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=169249517
This case may happen if the max span size is more than the max size the
evictor is configured.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=169101093
1- Avoid dropped buffers by using a PriorityQueue instead of a set.
2- Process the end of stream after non-EOS buffers.
Issue:#3250
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=169077365
1. Only tell sessions that want provisioning when provisioning occurs.
2. Also propagate failure to provision to these sessions.
3. If a session responsible for provisioning is released, start
provisioning using another session instead.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=168713918
- MediaDrmEventListener.onEvent is typically called on the
app's main thread. mediaDrmHandler is instantiated on the
playback thread. Hence mediaDrmHandler should be volatile
to ensure visibility.
- Nulling mediaDrmHandler could result in a NPE in onEvent.
Instantiate mediaDrmHandler (and don't null it again) to
avoid this. MediaDrmHandler.handleMessage will correctly
discard any events for sessions that are now closed.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=168707938
AdsMediaSource lives in the core library so only ImaAdsLoader remains in
the ima extension. AdsMediaSource takes an AdsLoader implementation.
ImaAdsMediaSource is deprecated rather than removed for now.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=168707921
Currently, to make transition to next media period seamless, after the renderer
has read until the end of the current SampleStream, we may send it the next
SampleStream so the renderer may read from the next SampleStream ahead of the
transition.
For NoSampleRenderer, we should avoid doing this: skipping ahead for such
renderer doesn't have any benefit (the renderer does not consume data from
SampleStream), and it will change the provided rendererOffsetUs while the
renderer is still rendering from the playing media period.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=168669800
This allows listeners to easily determine the source of the discontinuity.
Reasons can be period transitions, seeks, and internal reasons.
Listeners still using the deprecated ExoPlayer.EventListener interface were
updated to Player.EventListener.
GitHub: #3252
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=168549612
Remove gapless functionality that relies on MediaCodec, and implement this in
an AudioProcessor instead.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=168547487
Currently our Renderer is always associated with and consume data from
some SampleStreams, which were constructed from the provided MediaSource.
There are use-cases, in which the users want to have simple Renderer
implementation that does not consume data from SampleStream at all, but
render using their custom logic at each rendering position - they mostly just
need ExoPlayer to keep track of the playback position and enable/disable the
renderer.
This CL adds support for such Renderer by adding a TRACK_TYPE_NONE.
Renderer of such type will be:
- Associated with null TrackSelection as the result of track-selection
operation.
- Associated with EmptySampleStream.
GitHub: #3212
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=168545749
In live streaming, if the playback position is very close to live edge,
the buffered duration will never reach minDurationForQualityIncreaseMs,
which prevents switching from ever happening. So we will provide the
durationToLiveEdgeUs to AdaptiveTrackSelection in live streaming case,
so it can handle this edge case.
GitHub: #3017
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=168535969
The release callback handler in Loader might not be alive
anymore. Catch this case to prevent warnings about sending
messages on dead threads.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=167996538
Remaining instrumentation tests either use android.os.Handler or rely on assets.
In the latter case, the tests are difficult to migrate due to differences
between the internal and external build systems, and configuration needed in
Android Studio. In addition, SimpleCacheSpanTest remains as an instrumentation
test because it fails due to a problem with string encoding on the internal
build (and two other tests in its package are kept with it because they depend
on it).
This test removes a dependency from testutils on Mockito, as a different
version of Mockito needs to be used for instrumentation tests vs Robolectric
tests, yet both sets of tests need to rely on testutils. Mockito setup is now
done directly in the tests that need it.
Move OggTestData to testutils so it can be used from both instrumentation and
Robolectric tests.
It may be possible to simplify assertions further using Truth but this is left
for possible later changes.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=167831435
Everything should go through the ExoMediaDrm layer. We still
need to abstract away the android.media exception classes, but
this is left as future work.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=167710213
Reporting incorrect positions for ad playbacks was causing IMA to
think the ad wasn't playing, when in fact it was.
Issue: #3180
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=167702032
Also pass an unresolved end point to ClippingMediaPeriod. This removes some
assertions checking timestamps in the ClippingMediaPeriod, but makes it
possible to identify when the end point is at the end of the media.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=167683358
The default implementation is just calling through to handler.postDelayed,
while the fake clock uses its internal time value to trigger the handler
calls at the correct time.
This is useful to apply a fake clock in situations where a handler is used to
post delayed messages.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=167567914
Currently FlacDecoder/FlacExtractor always perform resampling to 16bit. In some
case (with 24bit audio), this might lower the audio quality if the system
supports 24bit audio.
Since AudioTrack implementation supports resampling, we will remove the
resampling step, and return an output with the same bits-per-sample as the original stream.
User can choose to re-sample to 16bit in AudioTrack if necessary.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=167494350
The bytesRemaining didn't always take into account any skipped bytes, which
meant that reaching the end of the file was not correctly detected in read().
Issue: #3216
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=167016672
UTF-8 is the default charset on Android so this should be a no-op change, but
makes the code portable (in case it runs on another platform).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=167011583
*** Reason for rollback ***
Doesn't work because trackOutputProvider can be null when extracting init data.
*** Original change description ***
Don't copy primary-track format to non-primary tracks
Copying non-primary-track formats to non-primary tracks
looks non-trivial (I tried; went down a dead-end), so
leaving that for now.
***
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=166883654
