When the subtitles are turned off, unloadTextStream() on streaming engine is called, but currentTextStream is never set to null. When the captions are turned back on, a check in player.js sees that the current text stream is not null, so it assumes it exists, and is ready to go. Thus captions are never loaded.
Clearing (setting to null) the current stream on unloadTextStream ensures the text stream is properly initialized the next time it is called, and in a state so that captions can resume being parsed.
After further investigation, it seems that the unit tests are written in a way that assumes that the text stream is nulled when unloadTextStream is called. So this fix is definitely something that we assume already happens.
Closes#2671
During video playback, if the user switches the caption stream (e.g. CC1 to CC3 which changes languages), the first caption in the next fragment is missing.
In fragmented MP4s, the end time of a caption is determined by the start time of the next caption. Thus for the last caption in a fragment, the end time cannot be determined until the next fragment is parsed.
Before this fix: the clearing of the caption stream was being called from a chain of function calls originating from clearBuffer_() on the Media source engine. But clearing a buffer and resetting a caption stream are two independent operations. As a result, the caption parser was being reset (its buffer cleared) during video seeks, and during stream switches. This makes sense for video seeks, because the end time of the last caption in the fragment can't be determined if the entire presentation timestamp changes. However for stream switches, resetting the parser doesn't make sense. Clearing the caption parser during a stream switch would actually get rid of the last caption in that fragment (which wasn't emitted since its end time wasn't determined yet), and we would lose the data, causing the problem.
The fix is to reset (and hence clear) the caption parser during seeks, but not during stream switches.
Issue #2648
This fixes all the license headers in the main library, which corrects
the appearance of the main license in the compiled output.
It seems that the `!` in the header forces the compiler to keep it in
the output. I believe older compiler releases did this purely based
on `@license`.
Issue #2638
Change-Id: I7f0e918caad10c9af689c9d07672b7fe9be7b2f3
In getSegmentReferenceNeeded_(), we get a SegmentReference base on
(presentationTime - maxDrift) time. If no SegmentReference is avaiable,
we wait until next update.
For live edge, if the SegmentReference with the drifted time is always
out of the segment availability start/end time window, we don't append
any segments.
If we can't find a valid segment with the drifted time, we should seek
with the presentation time.
Fixes#2589
Change-Id: I3fedc002ffc5c99f9dd1d1bf9525aca141b9a2be
Various issues with the nullability of number types led to various
fixes, including:
- defaulting a nullable number to 0 to avoid propagating a null value
through calculations
- adding an assertion or runtime check that something is not null
- moving an existing null check to before the calculation
- returning early on null during an iteration
- changing a nullable number to non-nullable
- defaulting to NaN instead of null
These issues were caught by a compiler upgrade.
Issue #2528
Change-Id: I86d516c74a42ee3624c33d7513d2d4c76d3ea589
In many places, we check error codes on shaka.util.Error. But the
compiler doesn't know that what is caught in "catch" is that type, so
we add type assertions.
In some cases, we know that other types may also be thrown, so there
are also some runtime checks. Some of these had to be refactored to
allow the compiler to correctly infer types.
Change-Id: I053bd7e96213c689aae3889315052dd402124690
This adds an annotation to suppress a bogus compiler error in
StreamingEngine. In this method, we set updateTimer to null, then
await some async operation, then check to see if it's still null. The
latest Closure Compiler thinks this is a useless check, but we know
better. The updateTimer could have been created while the async
operation was in progress.
Issue #2528
Change-Id: I9a28b7dac1d7cb8bcfd836847507454030af52c6
This exception was thrown when we attempted to set up a
SegmentIterator on CEA tracks. Because they do not have their own
segments, this is not possible. The fix is to exclude CEA tracks from
this step.
Discovered while debugging #2395, but seems to be unrelated.
This bug does not affect any release versions.
Change-Id: I534b19139d88cc7ad2d6167611c04f400308a71b
This removes periods from the internal manifest structure and cleans
up code and tests accordingly. This leaves us unable to play
multi-period DASH & offline streams until the main period-flattening
algorithm is completed in shaka.util.Periods.
Three test cases have been disabled for the moment.
Multi-period playback will be restored in a smaller, more focused
follow-up commit, with disabled tests re-enabled.
Issue #1339 (flatten periods)
Issue #1698 (rapid period transitions issue)
Issue #856 (audio change causes bitrate change)
Closes#892 (refactor StreamingEngine)
Change-Id: I0cbf3b56bfdb51add15229df323b902f0b2e643a
Playing content in our demo with my WiFi acting CRAZY, I found an
exception in StreamingEngine. The bad connection seems to have
exacerbated and exposed a race condition that can be triggered when
the active stream is switched during an async appendBuffer() call.
The fix is to be careful not to assume that the mediaState object is
the same as it was after the async call. If mediaState.stream has
changed, we should not manipulate mediaState.segmentIterator, which
may refer to a different Stream or may be null.
This bug does not affect any release versions.
Change-Id: I3d00009e78978cae95b47bb9f862fca4294b7b20
As part of Period-flattening, I'm trying to remove our dependence on
the "position" field of SegmentReference. With that eliminated, we
can more easily concatenate Arrays of SegmentReferences without
modifying them.
- Make SegmentIndex iterable
- Add specialized seek() and current() methods to SegmentIterator
- Remove position from SegmentReference
- Make positions in SegmentIndex API stable without field in
reference
- Remove brittle hard-coded positions in tests (except SegmentIndex
tests, where they would be hard to avoid in testing methods
separately)
- Use SegmentIterator in StreamingEngine to track the next segment
between switches
Issue #892 (refactor StreamingEngine)
Issue #1339 (period flattening)
Change-Id: I666cc21249c34ee6cbc138a59109d9f1159fa127
This reverts commit 235e4e11ad.
The effort to remove SegmentReference's position field will be handled
in a different way.
Issue #892 (refactor StreamingEngine)
Issue #1339 (period flattening)
Change-Id: I62b115137abc89f498b30467e574b0401dcad05d
As part of Period-flattening, I'm trying to reduce our dependence on
the "position" field of SegmentReference. If it can be eliminated, we
can more easily concatenate Arrays of SegmentReferences without
modifying them.
SegmentIndex can now track the last reference you asked for and
iterate through the list of references. This means we don't need the
"position" field of SegmentReference, which means we don't need to
know positions in advance or globally. StreamingEngine will no longer
use position to request segments.
The old methods find(time):position and get(position):SegmentReference
have been replaced with seek(time), current(), and next(), all of
which return a SegmentReference and maintain an internal pointer to
the "current" reference. Care has been taken to maintain that pointer
during the evict() and fit() operations. Recent changes to merge()
made sure that the pointer does not need to change during that
operation.
All test updates are related to the SegmentIndex API change, not
changing expectations or behavior.
Issue #892 (refactor StreamingEngine)
Issue #1339 (period flattening)
Change-Id: I1682dcc2dd625c6e390711538e46d31e6eb6cea8
We fetch the previous segment in Live streams to combat differences
between the manfiest times and the segment times. Now this can be
configured so apps with correct manifests can avoid the extra segment
request.
Also, we'll update the buffering state when a segment appends so we
leave the buffering state sooner.
Issue #2291
Change-Id: Id12c8132dc11739e4c8d42cb1f08e6ae7da1a966
By having an explicit enum, we can make sure there are no typos in
FakeEvent creation calls. This change also allows the cast proxy and
receiver to proxy events based entirely on the enum; that way, they
do not need to maintain their own independent list of events, that
can potentially fall out of sync (see #2318).
Change-Id: I08fa6653f35d86e87efda7fc822c8988688cfcff
This changes the eslint rule to enforce a strict pattern for the
argument comments. The comment must appear before the argument and
must be /* foo= */. This still ignores line comments.
Change-Id: I3afb01c65e1088eda13facb3aeeaa7595a2f5aee
This is a step along the way to flattening periods, and also
simplifies a few small things in StreamingEngine and tests.
Issue #1339 (flatten periods)
Issue #892 (refactor StreamingEngine)
Change-Id: Ie17cd5e15ed6ec9290a918c3a69c05c74581e0fc
SegmentReference used to have presentationTimeOffset, which,
subtracted from the period start time, was then _added_ to the
timestamps in the segment by MediaSource.
Now, SegmentReference has a timestampOffset field, which is exactly
what MediaSource's timestampOffset field is set to on the SourceBuffer
before this segment is appended. For DASH, this is periodStart minus
presentationTimeOffset.
This also adds append window start & end times to the
SegmentReference. Now segments can be appended to SourceBuffers
without reference to the period.
Note that start & end times of the SegmentReference in each segment
index are still relative to the period. This will change in a
follow-up.
Issue #1339 (flatten periods)
Issue #892 (refactor StreamingEngine)
Change-Id: I9d54eb2b529ec643c9475b8e9d218c3e2e826a26
This reflects changes in Google's policy on JavaScript license
headers, which should be smaller to avoid increasing the size of the
binary unnecessarily.
This also updates the company name from "Google, Inc" to "Google LLC".
Change-Id: I3f8b9ed3700b6351f43173d50c94d35c333e82b4
Because aborting requires knowledge of the new stream, this process
must be done asynchronously. This makes the abort logic async, and
checks carefully for any stream or operation changes during the
process.
Issue #1339 (flatten periods)
Issue #892 (refactor StreamingEngine)
Change-Id: Ic187676eeca907603efeb0ffa11855b9af2fc5ca
Instead of making many internal methods async to accomodate
createSegmentIndex being called lazily, just call createSegmentIndex
during the update cycle instead. This greatly simplifies things and
allows me to revert some of the changes I made in the earlier commit.
Issue #1339 (flatten periods)
Issue #892 (refactor StreamingEngine)
Change-Id: I72be8e88f0cf8b04b63d3cda129fa38cef727c0f
In StreamingEngine, rather than wait to enable ABR after indexes have
been created for all streams, create each stream's SegmentIndex
on-demand as needed during playback. This means ABR can be enabled
much more quickly, and also eliminates some complexity from
StreamingEngine's startup sequence.
This required several test changes, since many of our tests were
accidentally structured to depend on certain operations either being
synchronous or happening early during startup.
Issue #1339 (flatten periods)
Issue #892 (refactor StreamingEngine)
Change-Id: I4bc1d0cdf9022aad14a008accf0aac37c870a83f
We would incorrectly initialize the embedded captions multiple times
during a Period transition, which caused duplicate cues to be given to
the displayer. We also wouldn't handle the case when switching between
embedded captions and external text during a Period transition. This
fixes both cases and adds tests for them.
This also avoids passing an empty cue list to the displayer.
Fixes#2076
Change-Id: I89add3eb86ad8d93644bba14eabd11f98d57bc5e
These are intermittently used and cannot be verified. Most of our
methods can throw a shaka.util.Error, so having it doesn't really add
anything. Plus, if we change a function to throw, we'd need to update
all their callers to ensure they have an accurate description of what
they throw; otherwise we can't trust the directive.
Change-Id: I520bd0fc4c33443e967bf5b103ca5aa9e3274884
Since embedded text tracks are not "real" tracks, we need to ignore them
when handling Period transitions. This ensures that the other streams
will handle the transition instead of waiting for the text stream to get
updated.
Fixes#2075Fixes#2094
Change-Id: I2c7c92cb04795edb3f7695677ee6745bad3f4471
This changes the network API to use BufferSource instead of ArrayBuffer,
which allows plugins to return a "view" on a buffer instead of the
whole buffer. This also adds some utilities for changing between
views and buffers.
Lastly this forbids the use of the "buffer" property of TypedArrays
since it doesn't work with partial "views". This audits and fixes the
usages of the "buffer" property to ensure correct usage.
It should be noted that both MSE and EME accept a BufferSource as input,
so we don't need to convert a "view" into an ArrayBuffer before passing
to it.
Change-Id: Iaa417773f8ce5304424e43c7372ce10ebf540d2a
To prepare for flattening out the manifest structure to remove
periods, this change moves initSegmentReference and
presentationTimeOffset fields into the SegmentReference object. This
way, the segments on either side of a period transition or HLS
discontinuity can have different offsets or init segments, eventually
allowing us to create a single array of SegmentReferences for
multi-period content.
Issue #1339
Change-Id: Ic7eff0483789644881247ecf8044c5fb6a48f0e6
This replaces find/get callbacks in Stream with a SegmentIndex. With
the exception of DASH's SegmentTemplate+duration, all manifests were
already backed by SegmentIndex. Now, all manifests are backed by
SegmentIndex. This will simplify Period-flattening, in which all
tracks will be represented by a list of segments, some of which come
from different Periods.
The SegmentIndex in Stream will not be created until
createSegmentIndex is called. Prior to this change, the find/get
callbacks could be invoked without createSegmentIndex() in some cases
(excluding DASH's SegmentBase), which some lazy tests took advantage
of. Now that find/get are methods on SegmentIndex, createSegmentIndex
must be called in all cases. The tests have been updated accordingly.
Making SegmentIndex generation async in all cases exposed some issues
with the parser context being modified in-place between one
Representation and the next. So the parser now makes a shallow copy
of the context before it is bound into an async callback.
To facilitate updating the SegmentIndex for SegmentTemplate+duration
content, SegmentIndex now has a method to update its list on a timer.
Once per segment duration, the index will be updated to add and remove
SegmentReferences.
The initial expansion of SegmentTemplate+duration will be limited to a
relatively small number of segments, to avoid excessive CPU or memory
consumption. This defaults to 1000 segments, but is configurable.
Issue #1339
Change-Id: I99c007b1096c3b396d04a729750cd7b743cb899d
We start streaming text once subtitles have been enabled.
With external text, we used to create a media state for it
as soon as text was added even if captions were disabled
and we did not actually stream, although no text was
being parsed and shown.
This lead to the following problem:
If external text was added with subtitles disabled, we
created the media state with no actual cues in it.
Once the subtitles were enabled, we saw that a media
state existed for text and assumed it had content.
So, we started streaming, but nothing was shown.
This change makes us only create the media state if
subtitles are enabled and defer creation otherwise.
Closes#1938.
Change-Id: Iee39c30fbd6b7f0abe7772dfb75cba561fcc9998
This changes several classes to use the Destroyer class to handle
destroy(). This also changes the behavior to not ignore destruction
and instead throw a new error. This is more clear for callers and
ensures we propagate errors.
Change-Id: I756c085639558509c22e5c43d69ddf4acd28d46f
When switch() is called, it checks for what segment/period the media
state will need to fetch next. As text is much smaller than audio/
video, it might be the case that text has all the segments in
period i while audio and video are still in the process of fetching
them.
Thus, the next period needed for text will be i+1, while for other
media states it will be i.
Switch() assumes that if the period needed is not the same as
current period, a perios transition is about to happen and there's
no point in switching streams now since they're about to change on
the next update anyway.
However, the period transition only happenes if all the media
states require it. In our edge case, only text is ready for the
next period, so transition will not happen.
This change corrects the assumption "if a media state is ready
for the new period, don't switch" to assumption "if ALL media
states are ready for the new period, don't switch."
Issue #1774
Change-Id: I35f1b7ae10704922fb5692e02fc5f2edc6982575
The new support for replacing the text displayer mid-playback had an
uncaught problem. Specifically, loading a new asset while an old asset
was playing caused an error.
This was because it was trying to reload the existing text stream when
there was no text stream.
This changes the reloadTextStream method on streaming engine to do
nothing if there is no text stream, fixing that problem.
Change-Id: I53f25fb14a54670697ad014bc6e6f72f6cc28577
Rather than only aborting a request if the new request is smaller
(effective when adapting down), abort if the new request will be done
"on time" (effective when adapting up) without causing buffering.
Change-Id: I7572b8bfc566a4f264a5080e4c1bc983f1363ce0
This sets a TextDisplayer to the native text displayer if native
controls were desired, or a custom UI text displayer otherwise.
This required adding methods to the player to switch out the current
text displayer if a new factory is provided, and a method to streaming
engine to reload the current text stream.
Change-Id: I85595e9ac9db0b60464ee7f20fa35855efb26424
This is a fully automated change. The linter will fail because the
extra indentation caused line-length errors. These won't be fixed
automatically. They are fixed in a follow-up to make this one fully
automated.
Change-Id: I4d8cf9c998985add2bcd24a81c8d65495668c4f3
With the new style rule, we cannot have two statements on the same line.
So we can no longer have an "if" on a single line and we cannot have
an arrow function with a body on the same line as when it is used.
This is mostly a manual change.
Change-Id: I2285202dd5ecbad764308bc725e6d317ff2ee7f0
Muhammad Haris
Joey Parrish
michellezhuo
Jacob Trimble
Álvaro Velad Galván
Theodore Abshire
Sandra Lokshina