This commit implements the functionality to play back audio through
PlaybackManager.
To decode the audio data, AudioDataProviders are created for each track
in the provided media data. These providers will fill their audio block
queue, then sit idle until their corresponding tracks are enabled.
In order to output the audio, one AudioMixingSink is created which
manages a PlaybackStream which requests audio blocks from multiple
AudioDataProviders and mixes them into one buffer with sample-perfect
precision.
With this commit, all PlaybackManager can do is autoplay a file from
start to finish, with no pausing or seeking functionality.
All audio playback functionality has been removed from HTMLMediaElement
and HTMLAudioElement in anticipation of PlaybackManager taking that
over, for both audio-only and audio/video.
Very minor change, which doesn't actually affect our output, since we
were already inputting and outputting microseconds, but it can't hurt
to give FFmpeg's decoder this information as well.
We should be fine to just fall back to zero or the last returned value
if we encounter an error in PlaybackStream::total_time_played(), and
this also simplifies the using code.
There was a warning for the Optional initializer having no effect, but
removing the initializer caused the call to add a track to the HashMap
to complain. A constructor looks a little nicer here anyway.
Most users will only care about the total file duration, and shouldn't
be required to determine the file duration from multiple track
durations. To facilitate that, add a total_duration() function that
returns the demuxer's duration not associated to any particular track.
The existing conversion was rounding to the nearest millisecond, which
is much less precision than most videos will want. Instead, use only
integer math to directly convert the presentation time to seconds and
nanoseconds for our AK::Duration to represent accurately.
This first pass only applies to the following two cases:
- Public functions returning a view type into an object they own
- Public ctors storing a view type
This catches a grand total of one (1) issue, which is fixed in
the previous commit.
DecoderError::formatted() made a ByteString, took a view into it,
dropped the ByteString, then propagated the (now invalid) view back to
the caller as an error.
Since the object has to keep the ByteString alive, keep it as a variant
to make sure the "happy" path with no alloc remains alloc-free.
The SharedSingleProducerCircularQueue used here has dubious value, This
queue is used to pass commands to the audio thread, such as play/pause/
seek/volume change/etc. We can make do with a simple locked vector, as
we were blocking to enqueue tasks anyways. We can also use an atomic
bool to tell the audio thread when it needs to take a lock on the task
queue, to keep the thread lock-free most of the time.
This commit removes the -Wno-unusued-private-field flag, thus
reenabling the warning. Unused field were either removed or marked
[[maybe_unused]] when unsure.
