So far this wasn't possible because Robolectric's Looper and MessageQueue
implementations have multiple shortcomings:
1. The message loop of new HandlerThreads is an not an actual loop and
scheduled messages are executed on the thread the message is enqueued
(not the handler thread).
2. The scheduler used to replace the message queue is synchronizing all its
methods. Thus, when a test attempts to add messages to a Handler from
two different threads, it may easily run into a deadlock.
3. The scheduler doesn't correctly emulate the order of messages as they
would be in an actual MessageQueue:
a. If the message is enqueued on the handler thread, it gets executed
immediately (and not after all other messages at the same time).
b. The list of messages is always re-sorted by time, meaning that the
order of execution for messages at the same time is indeterminate.
4. Robolectric's SystemClock implementation returns the current scheduler
time of the main UI thread. So, unless this scheduler is used to add
messages in the future, the SystemClock time never advances.
This CL adds two helper classes which extend and replace Robolectric's
ShadowLooper and ShadowMessageQueue.
1. We intercept messages being enqueued or deleted in the message queue.
Thus Robolectric's faulty scheduler gets never used. Instead, we keep
a blocking priority queue of messages, sorted first by execution time
and then by FIFO order to correctly emulate the real MessageQueue.
2. We also keep a list of deleted messages to know which messages to ignore
when they come up in the looper.
3. When a new Looper is started, we override the dummy loop to an actual
eternal while loop which waits for new messages, checks if they haven't
been deleted, and runs the messages (similar to what Robolectric's
MessageQueue would have done at this point).
Because we don't actually use the main UI thread in our tests, we can't rely
on the SystemClock to progress in any sensible manner. To overcome this issue,
we can use the auto-advancing FakeClock also used for the simulation tests.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=182912510
This achieves two things:
1. All our tests use the same type of assertions.
2. The tests currently run as instrumentation test can be moved to
Robolectric without changing the assertions.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=182910542
These were caused by two issues:
1. The FakeMediaSource can be updated with a new timeline. The setNewSourceInfo
is called from a different thread than prepareSource and both access local
variables without synchronization.
2. For multi-window playback, the FakeRenderer claims that isReady and isEnded
are both set to false if it read the end of the stream. However isReady should be
true because it is able to "render" its data until the end of the stream.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=182785169
This fixes a very specific case where the data read has non-cached gaps
and a read-only CDS switches to read from upstream in a gap then the
cached data is deleted. When the CDS reaches the end of the gap, it
tries to open the next source. As there is no cached data, it tries to
continue with the already opened upstream data source but as it reached
end of the gap range, the code starts looping.
Also fixes infinite lock which occurs when in the previous case CDS isn't
readonly. It locks the content while filling the gap in the cache. At the
end of the gap, as the following data is deleted it tries to lock the
content for writing but the content is already locked by itself.
The last fix is preventing removal of CachedContent entry from
CachedContentIndex while associated key is locked.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=182595426
Some tests in ExoPlayerTest issue commands to the player from the test thread
while the player is actively playing media (playWhenReady=true). Due to the
indeterminate time taken to enqueue the commands on the playback thread, they
may arrive when the player already proceeded to another window or finished
playback.
To ensure the tests are always deterministic, this change pauses playback in
the tests where this may happen before issuing the commands.
Also, for tests where we need to wait for a new window before issuing the
next command, a new action is added which allows to play until a specified
position.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=182535096
As soon as the seek gets acknowledged by EPII, EPI returns the actual position
from the playback info again which is set by EPII. Thus, EPII needs to update
the position to reflect the changes expected by EPI.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=182515106
Partial reads were performed once using a partial size of 1 byte.
This was not enough to detect problems which only occur in combination
with IOExceptions. Partial reads are now only applied when no exception
is thrown.
Moreover, the tests didn't check whether the total number of sampled bytes
is what it is supposed to be. Added a field to the data dumps checking
the total number of bytes in the sampled data.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=181296545
This adds options to ExoPlayer.sendMessages which allow to specify a window index
and position at which the message should be sent. Additionally, the options can be
configured to use a custom Handler for the messages and whether the message should
be repeated when playback reaches the same position again.
The internal player converts these window positions to period index and position
at the earliest possibility. The internal player also attempts to update these
when the source info is refreshed. A sorted list of pending posts is kept and the
player triggers these posts when the playback position moves over the specified
position.
Issue:#2189
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=179683841
*** Original change description ***
Add possiblity to send messages at playback position.
This adds options to ExoPlayer.sendMessages which allow to specify a window index
and position at which the message should be sent. Additionally, the options can be
configured to use a custom Handler for the messages and whether the message should
be repeated when playback reaches the same position again.
The internal player converts these window positions to period index and position
at the earliest possibility. The internal player also at...
***
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=179666357
This adds options to ExoPlayer.sendMessages which allow to specify a window index
and position at which the message should be sent. Additionally, the options can be
configured to use a custom Handler for the messages and whether the message should
be repeated when playback reaches the same position again.
The internal player converts these window positions to period index and position
at the earliest possibility. The internal player also attempts to update these
when the source info is refreshed. A sorted list of pending posts is kept and the
player triggers these posts when the playback position moves over the specified
position.
Issue:#2189
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=179563355
Skipping short periods in a while loop is conceptually a new operation
and thus we need to send out the updated playback info in between for
the listeners to receive multiple period transition discontinuities.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=179027334
This is a no-op change replacing the local variables in ExoPlayerImplInternal
with the new ones in PlaybackInfo.
***
Use playbackState, isLoading and trackSelectorResult from playbackInfo in ExoPlayerImpl.
***
Move duplicated listener notification in ExoPlayerImpl to new method.
Also split reset method in one parts which creates the new playback info
and one part which notifies the listeners. The increment of the pending
operation counter needs to happen in between.
***
Use only one pending operation counter in ExoPlayerImpl.
This also allows to move onSeekProcessed into the notification chain.
***
Replace playback info changing messages to ExoPlayerImpl by single message type.
As they are all handled in the same way, they can be summarized to one message.
***
Only send playback info change notifications once per playback thread message.
This ensures that all concurrent changes actually reach ExoPlayerImpl concurrently.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=178907165
1. The player doesn't acknowledge phantom stops when an exception is thrown anymore.
2. It also makes sure it doesn't reset the pendingPrepareCount unless it's actually
immediately acknowledging these prepares.
3. It ensures a seek is acknowledged even though an exception is thrown during seeking.
Added tests (which previously failed) for all three cases.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=178876362
Waiting for the timeline change didn't work correctly because the timeline was
already equal to Timeline.EMPTY (due to the masking). Now waiting explicitly
for the empty Timeline exposed by the source.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=177749292
- Remove skipping of the VBRI/XING frame before calculating position
offsets. This was incorrect. Instead, a constraint is used to ensure
we don't return positions within these frames, the difference being
that the constraint adjusts only positions that would fall within
the frames, where-as the previous approach shifted positions through
the whole stream.
- Excluded last entry in the VBRI table because it has an invalid
position (the length of the stream).
- Give variables in XingSeeker descriptive names.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=177451295
- This change snaps the seek position for constant bitrate MP3s
to the nearest frame boundary, avoiding the need to skip one
byte at a time to re-synchronize (this may still happen if the
MP3 does not really have fixed size frames).
- Tweaked both ConstantBitrateSeeker and WavHeader to ensure the
returned positions are valid.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=177441798
This allows to keep the state synced with ExoPlayerImpl after stopping the player,
but still releases the media source immediately as it needs to be reprepared.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=177167980
We can acknoledge seeks before preparation finished immediately now,
because ExoPlayerImpl won't leave the masking state until the first prepare
operation is processed.
As a side effect, it also cleans up the responsibility of the callbacks.
Prepares are always acknowledged with a SOURCE_INFO_REFRESHED, while seeks
are always acknowledged with a SEEK_ACK.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=177144089
The ExoPlayerImpl implementation forwards the stop request with this optional
parameter. To ensure correct masking (e.g. when timeline updates arrive after
calling reset in ExoPlayerImpl but before resetInternal in
ExoPlayerImplInternal), we use the existing prepareAck counter and extend it
also count stop operations. For this to work, we also return the updated
empty timeline after finishing the reset.
The CastPlayer doesn't support the two reset options so far.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=177132107
Unconditionally waiting for the action schedule to finish in ExoPlayerTestRunner
doesn't work if the action schedule is not intended to be finished.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=177024139
Fixed by explicitly waiting for the timeline update. This shouldn't be
necessary and will be removed as soon as the correct order of events
can be guaranteed (timeline change -> state change -> onSeekProcessed).
The waiting for the timeline update is implemented by introducing the
feature that the test runner also waits until the action schedule has
finished before stopping the test.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=176848540
This is mostly useful for suppressing the initial position
discontinuity reported by ClippingMediaPeriod.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=176758972
This brings ClippingMediaSource clip failures in line with
what MergingMediaSource does when it cannot merge.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=176660123
We (eventually - albeit possibly infinitely far in the future)
expect a timeline update with a window of known duration. This
also stops live radio stream playbacks transitioning to ended
state when their tracks are disabled.
As part of this fix, I found an issue where getPeriodPosition
could return null even when defaultPositionProjectionUs is 0,
which is not as documented.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=176492024
The ExoPlayerImplInternal.reset method now takes the same set of options
as the ExoPlayer.prepare method. This also allows to
- Remove some code duplication within ExoPlayerImplInternal
- Fix calls to prepare(sameSource, resetPosition=true, resetState=false)
with enabled shuffle mode where the position was not correctly reset to the
first period index.
- Keep the current timeline when calling stop (in line with ExoPlayerImpl).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=176481878
Currently onTimelineChanged doesn't allow to distinguish easily between the
different reasons why it's being called. Especially, finding
out whether a new media source has been prepared or the current source
refreshed dynamically was impossible without tightly coupling the player
operations with the listener.
The new reasons provide this disdinction by either indicating a newly
initialized media source, a dynamic update to an existing timeline
or manifest, or a reset of the player (which usually results in an
empty timeline).
The original onTimelineChanged method without reason is kept in the
DefaultEventListener as deprecated to prevent the need to update all
existing listeners in one go.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=176478701
- Properly report internal discontinuities
- Add DISCONTINUITY_REASON_SEEK_ADJUSTMENT to distinguish
seek adjustments from other internal discontinuity events
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=176367365
- MediaSourceTestRunner aims to encapsulate some of the logic
currently used in DynamicConcatenatingMediaSourceTest, so it
can be re-used for testing other MediaSource implementations.
- The change also fixes DynamicConcatenatingMediaSourceTest to
execute calls on the correct threads, and to release handler
threads at the end of each test.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=176117535
This test seems to obtain a timeline from a prepared
FakeMediaSource, but that's identical to the timeline
passed into that source to start with :).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=176117133
This is helpful for tests which don't care about detailled timeline set-ups
besides the number of windows.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=176097369
The purpose of this change isn't to fix anything. It's just to
simplify things a little bit. There will be following CLs that
make some changes to get things onto correct threads.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=175800354
This allows to remove the LazyMediaSource used within
DynamicConcatenatingMediaSourceTest and also allows to write test which
simulates dynamic timeline or manifest updates.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=175680371
Also added tests which verify the intended behaviour.
GitHub:#3452
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=175656478
Add onEnable() and onDisable() call-backs to TrackSelection. This allows
TrackSelection to perform interesting operations (like subscribe to
NetworkStatus) and clean up after itself.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=175558485
These callbacks are executed on the app thread after the corresponding
timeline update was triggered. This ensures that seek operations see the
updated timelines and are therefore valid, even if the seek is performed into a
window which didn't exist before.
GitHub:#3407
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=175136187
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
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
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
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 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
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
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
Both cases were not supported so far. Added tests which all failed in the
previous code version and adapted the concatenated media sources to cope with
empty timelines and empty concatenations.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=166480344
This is mostly connecting the already stored shuffleMode with the timeline queries
for the playback order.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=166199330
The media source is initialized with a DefaultShuffleOrder which can be changed at
any time. Whenever the list of media source is changed, the shuffle order is adapted
accordingly (either on the app thread if the player is not prepared yet, or on the
player thread). The shuffle order is then used to construct the timeline.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=166198488
The media source is initialized with a DefaultShuffleOrder which can be changed at
any time. This shuffle order is then used within the corresponding timeline.
The isRepeatOneAtomic flag is extended to also suppress shuffling (now called
isAtomic only).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=166197184
In addition, let unit test assert window indices for both shuffle modes.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=166191069
This functionality is most likely needed by UI modules which currently need
to obtain the timeline, the current repeat and shuffle modes and are only then
able to query the next/previous window index using this information.
Adding these methods simplifies these cumbersome requests.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=166181202
This allows to test the expected behaviour of timeline with different shuffle modes.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=166181091
And implement a basic version of the methods in all implementations.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=166041342
These classes maintain a shuffled order of indices allowing to query the
next, previous, first, and last indices. And also support inserting and
removing elements without changing the shuffled order of the rest.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=164116287
- Fix Ogg extractor to work without sniffing.
- Fix extractors to handle seek() before init().
- Add tests for both issues.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=163992343
This is for consistency with what we do elsewhere;
specifically in FragmentedMp4Extractor.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=163974960
Currently the ExoPlayerWrapper is used to run tests using an ExoPlayer
implementation. Some properties of the test are set in the constructor, some
are set by overloading the class, others are hard-coded in the
ExoPlayerWrapper class, and a mechanism similar to the existing ActionSchedule
is missing.
This change does the following:
1. Renames ExoPlayerWrapper to ExoPlayerTestRunner as it better reflects
its purpose.
2. Adds an internal Builder to easily set-up the test in a coherent way. This
allows to only set necessary test components while using defaults for the rest.
3. Integrate ActionSchedule.
4. Apply the new structure to the existing tests currently using ExoPlayerWrapper.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=163186578
This is the first step towards facilitating Cast integration to ExoPlayer.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=162366198
This enables adaptive media test cases using TrackGroups with multiple
Formats.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=162182005
This allows to create extensions of FakeDataSet and also simplifies the
FakeDataSource class.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=162180952
