SDK / exoplayer

This fixes a bug that renderer errors are currently falsely associated
with the playing period.

PiperOrigin-RevId: 321381705

This method has been replaced by onPlaybackSpeedChanged

PiperOrigin-RevId: 321369921

ISSUE: #7565
PiperOrigin-RevId: 321367089

Also remove some duplicate logging from ExoHostedTest - EventLogger
logs the same info already.

PiperOrigin-RevId: 321366052

Float values are allowed to be > 0dbfs,
it is just not nominal as it will might
distort the signal when played without
attenuation.

This is also consistent with
[AudioTrack.write(FloatBuffer)](https://developer.android.com/reference/android/media/AudioTrack#write(float[],%20int,%20int,%20int)) that explicitly
allows it up to 3dbfs.

PiperOrigin-RevId: 321345077

PiperOrigin-RevId: 321340777

The method has been called from two call sites in EPII triggered by EPII.updatePeriods(). The first call site was calling it when the MediaSourceList is empty or not yet prepared. This can be removed because if empty or not prepared no source ever could have thrown yet.

The second call site was checking for potential source refresh exceptions when queue.getNextMediaPeriodInfo() returns null when trying to getting the next loading period. Looking into all reasons for why the method returns null, none of them is caused by an exception of a media source. The reasons are:

- if we are at the last period of the timeline
- if the defaultPosition of the next period in the timeline is null (if the window.durationUs == C.TIME_UNSET or defaultPositionProjectionUs is projected beyond the duration of the window)
- if we are waiting for an ad uri to arrive (period.isAdAvailable(...) == false)
- if we are waiting for the ad group count to be updated (adCountInCurrentAdGroup == C.LENGTH_UNSET)

The above reasons are not caused by a source error and may be resolved when doSomeWork is called the next time. Hence it is save to remove the calls to maybeThrowSourceInfoRefreshError().

Beside this, an actual sourceInfoRefreshError will be reported by maskingMediaSource.maybeThrowPrepareError(), which is called each time doSomeWork() is called and the playing period is not yet prepared (EPII:L836). So the player is notified by source errors that way, which confirms removing the above calls is fine.

PiperOrigin-RevId: 321331777

PiperOrigin-RevId: 321280295

PiperOrigin-RevId: 321218451

PiperOrigin-RevId: 321181453

It's only used in BandwidthMeter so inline it there.

PiperOrigin-RevId: 321177126

Plumb this down into BaseRenderer.onStreamChanged and use it when
deciding whether to render the first frame of a new period.

PiperOrigin-RevId: 321175627

PiperOrigin-RevId: 321174738

This doesn't affect the nullness checker or Kotlin, but it does make
weird warnings appear in Android Studio. It seems mildly preferable to
have the same spurious warnings in these files that we have in other
tests, rather than different spurious warnings.

PiperOrigin-RevId: 321173760

PiperOrigin-RevId: 321169585

PiperOrigin-RevId: 321168965

PiperOrigin-RevId: 321168125

PiperOrigin-RevId: 321166822

PiperOrigin-RevId: 321163229

This change masks playbackInfo.periodId and playbackInfo.loadingPeriodId for operations which change these periods (set/add/remove sources and seeks).

Because this masking is reflected in the playbackInfo object, player attributes can be retrieved without the maskingXyz variables in EPI. This has the advantage that the playbackInfo object always reflects the public state of the player even when operations are still pending. The maskingXyz variables in EPI are only required for the deprecated use case of an initial seek in an empty timeline.

PiperOrigin-RevId: 321160092

PiperOrigin-RevId: 321158149

PiperOrigin-RevId: 321157794

PiperOrigin-RevId: 321157115

PiperOrigin-RevId: 321156463

PiperOrigin-RevId: 321155415

In order to use DeviceInfo class in media2, this CL moves the class to
common module. It didn't move the other file in the same package,
DeviceListener, as it's for DeviceComponent but media2 SessionPlayer
doesn't have components as it is already flattened. PlayerCallback will
have equivalent methods of DeviceListener.

PiperOrigin-RevId: 321154997

PiperOrigin-RevId: 321153963

The constructor that takes a Uri is deprecated

PiperOrigin-RevId: 321148326

PiperOrigin-RevId: 321147910

All the public Cue constructors are deprecated

PiperOrigin-RevId: 321122512

PiperOrigin-RevId: 321121988

PiperOrigin-RevId: 321121735

createMediaSource(Uri) is deprecated.

PiperOrigin-RevId: 321121383

PiperOrigin-RevId: 321109232

ISSUE: #7565
PiperOrigin-RevId: 321108417

PiperOrigin-RevId: 320970814

PiperOrigin-RevId: 320960833

The method currently advances the read position and returns the number
of skipped samples. This prevents checking how many samples are skipped
before the operation is executed.

Instead, we have a new method that returns the number of to be skipped
samples and a skip method that executes the skipping.

PiperOrigin-RevId: 320953439

Remove occurrences in comments and private fields.

ISSUE: #7565
PiperOrigin-RevId: 320948364

- DefaultAudioSink always supports floating point input. Make it
  advertise this fact.
- Remove the ability to enable/disable floating point output in
  FfmpegAudioRenderer, since this ability is now also provided on
  DefaultAudioSink.
- Let FfmpegAudioRenderer query the sink to determine whether it
  will output floating point PCM directly or resample it to 16-bit
  PCM.

PiperOrigin-RevId: 320945360