It turns out that, on some browsers, incomplete XML does not properly
parse. This means the parseFirstCue tests did not work on those platforms.
This modifies those tests on platforms that do not support partial XML;
when no support is detected, it instead tries to get the first cue
from a non-partial segment.
This also modifies the HLS parser to only load partial text segments if
the text parser supports it; otherwise, the hls parser will load full
captions in order to extract start times.
b/137945635
Issue #2037
Change-Id: I58112864a818baf155b365a65fba1c8a56eb9520
Previously, when the text engine tried to load the start time of
a segment, it would parse every cue in that segment, then check the
time of the first cue. This was judged to not be a significant
performance issue, as parsing cues is a fast operation. However, it
did have an unintended side-effect: in some situations, this method
was being passed partial segments; notably, the HLS parser would load
the first 2048kb of the stream's texts to extract timing data.
If the caption parsers tried to actually parse an incomplete caption,
they would error.
This gives the text parsers "parseFirstCue" methods, and uses those
methods when it only needs the first cue anyway.
Fixes#2037
Change-Id: I2a1fb2f1a96d98967f0c6e6a5c277914a28b42ad
This ensures that we get the expected types and that type coercion
doesn't convert between types. This also ensures we are consistent
in how we check for equality of primitives in tests.
Change-Id: I9f3aacdf25ab1afe5e8d6e4b895b5299ee687d54
Instead of checking each property separately, just use
jasmine.objectContaining. This also adds a utility for matching a
number that is close to an expected value. This is required when
matching objects with toEqual().
Change-Id: I4de49cfaf254e0752beffc97c578a77f9ec8ee95
Creating them in beforeAll will usually cause test contamination, so
they should be created in beforeEach instead so they are reset for
each test.
Change-Id: I51f11de85555cbf5b8c18969eabeecbe8617b8f3
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
This is an automated change to convert use of "function" functions
to arrow functions. This doesn't change all uses of bind() that
could be converted. This also doesn't remove all "function" functions.
Change-Id: I40ac7d086bcef947a1be083359c8fd1d4499a9c3
A coming update to the Google eslint config will require using "const"
over "let". This makes that one change to isolate the big changes.
Change-Id: I7d0974c3ae15c53cc45a6b07bf9f6586e2d34aca
When CEA captions are extracted from video segments, the
presentationTimeOffset of the video segment should be applied to the
extracted captions, as well.
Fixes b/70902665
Change-Id: I446333a14b8b6374786ab594a579b6e18bc73ac1
There we a couple different fake text displayers in our tests. This CL
take them all and consolidates them into one fake text displayer.
Change-Id: I889aa4f11660d641920b0628e7e91549cb715c6a
CSS style 'direction' supports only horizontal directions, (left to
right, right to left), and writing-mode supports horizontal top to
bottom, vertical left to right, vertical left to right.
Separating the field 'writingDirection' to two fields 'direction'
and 'writingMode', so that they're the same as the css style names.
Issue #1708
Change-Id: Ie730bd7d13e5483d7425261854c268cd7f095400
In our tests we regularly have checks that repeatedly check if a
function/spy has been called once more. To check this we would check
the call and then reset the calls each time.
To abstract this out, this CL defines two custom jasmine matchers that
allow us to check if a function/spy has been called once more.
To validate the matchers, TextEngine tests were update to make use of
it.
Change-Id: Ia6bf8d0a585648126881e1713b35be674bd47ba7
Browser support is currently buggy and limited, so until they fix it,
this removes VTTRegion support.
Closes#1584
Change-Id: Ia5bb34dd3ea94737adcdd1d441201bb97b303f81
This reverts commit c38d4dd8d3, which
actually broke text range calculations in v2.3.10, and v2.4.2-v2.4.4.
The original commit was meant to account for the period start, but
resulted in a double-accounting of presentationTimeOffset.
The start and ends times passed into TextEngine's appendBuffer were
period-relative, so timestampOffset had already been applied. To
avoid further confusion and to fix the original issue the reverted
commit tried to address, these have been changed to
presentation-relative timestamps. Now the period start and all
offsets have been accounted for before the metadata reaches
MediaSourceEngine and TextEngine.
The tests added in the bad commit have been modified to test for the
opposite: that we do not erroneously account for timestamp offset when
calculating the buffered ranges for text.
Closes#1562
Change-Id: I9fa7a3f59906c4f3e623f411e48551f86f5c2ff7
https://goo.gl is being turned-down, so we can't use it for new URLS.
So we have consistent short links, this converts them to be
https://bit.ly.
Change-Id: I07a86cba807b67157664893341f648023918d0de
For multi-Period content, we need to account for the timestamp offset
in the buffered range calculation since the start/end times are
given as Period-relative.
Change-Id: I6d32de2abc316918333b1c58ba73c71d028284c4
Previously, we would read each individual VTTC and VTTE box in the
MDAT and associate them with durations in the TRUN. This worked for a
lot of content, but if sampleSize is defined in the TRUN, a duration
might refer to multiple VTTC boxes.
This changes the VTT in mp4 reader to parse the MDAT based on the TRUN,
to account for such situations.
For the moment, this only accounts for a single MDAT, and assumes that
the MDAT contains nothing but vttc or vtte boxes, listed in order.
Issue #1266
Change-Id: I56e310d085abdda16e968761ed3b4fd0cc5e24d0
This is an automated fix to be inline with Google style guides, this
was created with eslint's --fix option.
Change-Id: I860eecbc8152603e730aa17a1393f16d26b3b6fc
Google style guide requires adding curly braces to all block statements
even if it is only has one line. This fixes it by using eslint's
--fix flag followed by running clang-format to reformat the change.
Change-Id: Idc086c2aa8c02df5ef8b2140a11bfb9128eeb4bd
This is part of a change to convert all usages of 'var' with either
'let' or 'const'. This takes a conservative approach for 'const' where
it will only be used for aliases and storing the "original" values in
tests.
Change-Id: I6a329d28e13a81c9f7136737518c6bb8fa18402e
We incorrectly added the presentationTimeOffset to the segment times as
a fix for #1164. The correct fix is to use the include the PTO in the
time structure passed to the text parsers. This is a partial revert
of 207505.
Issue #1164Closes#1232
Change-Id: I1f2805e0dbdc44be71e2160b3d37a73732c97a4f
When the MPEGTS field rolls over in HLS live WebVTT, we need to avoid
our parsed cue timestamps rolling over as well. To achieve this, we
simply ignore the X-TIMESTAMP-MAP tag once the segment index is built.
Rollover in TS video and audio segments is handled by mux.js already.
Closes#1191
Change-Id: Ie52734509921973ff47517fab34367ec413a6ca6
Instead of taking Uint8Array for media segments and ArrayBuffer for
init segments, take Uint8Array for everything.
Clean-up after #1022, discovered while preparing the upgrade guide for
v2.3.
Change-Id: I1f284f6f51f345e663c06d96d788bd9cfb941c52
We do not support CSS embedded in WebVTT at this time. Instead of
failing to parse the cues, skip the style block.
Closes#1104
Change-Id: I3d500ba11afe43e81bbdef9924e4dd9e05db2b85
This change enables an additional conformance check in the compiler
which prevents us from committing code where the compiler fails to
infer what "this" means.
None of these changes constituted actual bugs, just inference failures
in the compiler.
Issue #1130
Change-Id: If9b18203768e197258042a9b4339530f371ed831
This avoids having media from one period replaced by media from the
next period. Instead, media that comes before the period start will
be chopped off by MediaSource.
Closes#1098
Change-Id: Idf6dc2ffafe78214e94bc75aca63920e153f1a2c
This converts from using ArrayBuffer in the text parsers to using
Uint8Array. This avoids doing extra buffer copies when using
ArrayBuffer.slice().
Closes#1022
Change-Id: I34a789ec4eb76c95169ba81f27a8433ef25f27f0
The tts:origin and tts:extent attributes specify the sub-region to
render cues onto. We parsed them as positioning info for the cues
themselves. This converts to using VTTRegion for this. However, this
is ignored on browsers since none expose VTTRegion as of now.
Closes#1020
Change-Id: I0f922868690461cdd8847c3eba21d944d854f537
Beforehand, if we were parsing a ttml subtitles stored in an mp4, we did
not account for the possibility that there would be multiple mdat boxes,
each with its own ttml data. In such a case, we would only use the
payload from the final mdat.
I don't know if having multiple mdat like that is really in the spec, but
at least one user has had this problem, and it's a straightforward change
to make that doesn't really slow down the text parser at all (since we
parse every mdat anyway).
Closes#1028
Change-Id: Id782695548cb92f7aa7278843695b670d8c15ff0
The fields in shaka.text.Cue were getting renamed in compiled builds,
which would cause problems with custom text parsers or text
displayers.
Change-Id: I5a1c05ad373aca722268417c55a84f195e53ee20
When a text track's mode is set to "disabled" the cues exposed cue
list is set to null. However, if you save a reference to the list,
the reference is valid. If the track is disabled, the add and remove
calls still work.
This change caches the reference to the cue's list when the track is
first created so that we can always read it - regardless of what mode
the track is set to.
Fix#990
Change-Id: I8274ea8450e664eeaa359bf9e78d13405fd2e503
IE/Edge don't accept cues that are out of time order. So when we
reverse the order of cues for #848, the cues would be rejected. Now
we sort the cues and only reverse the order of cues with equal times.
Change-Id: I860e1ea9694eb95ff2e74d9545c92373eb371686
If line and position are null, the corresponding VttCue should have
default values. Before this change, we treated 0 as null, which would
have prevented setting line or position to 0.
Change-Id: I63c3ce301016eb96bf3e6c1299433262c998cdc2
Since the browser displays identical ranges from bottom up, we must
reverse the order before feeding them to the browser. This way, the
first one parsed shows up on top.
Issue #848
Change-Id: Id2e6582e610808f7061bd0f8281c0705ecf1d6dc
Theodore Abshire
Vincent Valot
Jacob Trimble
Sandra Lokshina
Joey Parrish
Aaron Vaage
Michelle Zhuo