If we already saw the keyframes that we need to find,
we do not need to bisect to find them.
This will always be the case for streams with audio only,
where each frame acts as a keyframe, but will occasionally
also happen for streams with video.
https://bugzilla.gnome.org/show_bug.cgi?id=662475
Opus streams outside of Ogg may not have headers, and oggstream
may be used by oggmux to mux an Opus stream which does not come
from Ogg - thus without headers.
Determining headerness by packet count would strip the first two
packets from such an Opus stream, leading to a very small amount
of audio being clipped at the beginning of the stream.
In push mode, we determine duration by doing a seek to the end of the
stream. However, a skeleton stream with an index will cause the duration
to be known already, and we end up never setting the push_time_duration
variable which we use to know duration has been determined.
https://bugzilla.gnome.org/show_bug.cgi?id=662049
The codec setup headers are a lot more likely to have correct information,
especially as it's easy to remux a skeleton in a file where streams don't
have the same parameters (I've even seen a file with two skeletons).
Still, this is useful in the case we have a codec we can't decode, so we
can at least (theoretically) convert granpos to time, so we discard this
information if the codec setup has already provided it.
This fixes playback on (at lesat) the original archive.org encoding of
"The Night of the Living Dead" (now replaced by another encoding).
https://bugzilla.gnome.org/show_bug.cgi?id=612443
This could happen when testing with navseek, and pressing
right and left at roughly the same time. The current chain
is temporarily moved away, and this caused the flush events
not to be sent to the source pads, which would cause the
data queues downstream to reject incoming data after the
seek, and shut down, wedging the pipeline.
Now, I can't really decide whether this is a nasty steaming
hack or a good fix, but it certainly does fix the issue, and
does not seem to break anything else so far.
https://bugzilla.gnome.org/show_bug.cgi?id=621897
This patch implements seeking in push mode (eg, over the net)
in Ogg, using the double bisection method.
As a side effect, it also fixes duration determination of network
streams, by seeking to the end to check the actual duration.
Known issues:
- Getting an EOS while seeking stops the streaming task, I can't
find a way to prevent this (eg, by issuing a seek in the event
handler).
- Seeking twice in a VERY short succession with playbin2 fails
for streams with subtitles, we end up pushing in a dataqueue
which is flushing. Rare in normal use AFAICT.
- Seeking is slow on slow links - byte ranges guesses could be
made better, decreasing the number of required requests
- If no granule position is found in the last 64 KB of a stream,
duration will be left unknown (should be pretty rare)
https://bugzilla.gnome.org/show_bug.cgi?id=621897
The first packet of a sparse stream may arrive after an initial
delay in the stream. If ogg_stream_packetout reports a discontinuity
in a sparse stream, do not propagate it to other streams in the
chain unnecessarily.
https://bugzilla.gnome.org/show_bug.cgi?id=621897
Remove the _ in front of the endianness prefix.
Remove the _3 postfix for the 24 bits formats.
Add a _32 postfix after the formats that occupy extra space beyond their
natural size.
The result is that the GST_AUDIO_NE() macro can simply append the endianness
after all formats and that we only specify a different sample width when it is
different from the natural size of the sample. This makes things more consistent
and follows the pulseaudio conventions instead of the alsa ones.
After all, we do hope to find actual data for these streams.
However, warn if we could not set up a chain when we find a
non BOS page, as that means we don't have a valid Ogg stream.
https://bugzilla.gnome.org/show_bug.cgi?id=657151
While the casual reader might end up bewildered by just why this
change might increase clarity, it just happens than, in the libogg
and associated sources, op is the canonical name for an ogg_packet
whlie og is the canonical name for an ogg_page, and reading this
code confuses me.
https://bugzilla.gnome.org/show_bug.cgi?id=657151
Headers are inherently durationless.
Instead, set duration to 0 to avoid increasing tracked granpos,
and do not warn about it, since it is totally expected.
https://bugzilla.gnome.org/show_bug.cgi?id=657151
Version written is 3.0
Base times are left empty for now.
Content-Type should be the MIME type of the stream. It is set to
the GStreamer media type for now, which is probably the same for
the streams oggmux supports.
https://bugzilla.gnome.org/show_bug.cgi?id=563251
If ints are 64 bits, 32 bits should get promoted in varargs anyway,
and we don't care about 16 bit ints.
This makes the code a lot more readable, and still gets us nice
hexadecimal 32 bit serialnos.
https://bugzilla.gnome.org/show_bug.cgi?id=656775
This decreases the number of buffers held on each pad by one,
eliminating next_buffer. Simplifies the logic by relying solely
on CollectPads to let us know when a pad is in EOS. As a side
benefit, the collect pads related code is structured more like
other CollectPad users.
The previous code would occasionally mark the wrong pad as EOS,
causing the code to get in a state where all the streams were
finished, but EOS hadn't been sent to the source pad.
Ogg mandates the first header packet must determine a stream's type.
However, some streams (such as VP8) do not include such a header
when muxed in other containers, and thus do not include this header
as a buffer, but only in caps. We thus use headers from caps when
available to determine a new stream's type.
https://bugzilla.gnome.org/show_bug.cgi?id=647856
gcc on OSX complains about ret being used uninitialized in
this function, and it is right. Don't leak element ref
when returning early because newsegment event is not in
TIME format.
Remove the android/ top dir
Fixe the Makefile.am to be androgenized
To build gstreamer for android we are now using androgenizer which generates the
needed Android.mk files.
Androgenizer can be found here:
http://git.collabora.co.uk/?p=user/derek/androgenizer.git
Also initialize it always in TIME format. We require TIME segments
in oggmux anyway and drop newsegment events in other formats and
assume an open-ended segment starting at 0.
Theora and vorbis use running time (which is correct) for calculating
the granulepos for their ogg packets. Oggmux, however, used
timestamps to order the received buffers.
This patch makes it use the running time to compare buffer times
and also to timestamp pushed buffers.
Some bits of the code still use timestamps, but they are only
used to calculate durations, so it should be fine.
https://bugzilla.gnome.org/show_bug.cgi?id=643775
In case the ogg mapper doesn't handle all the accepted input formats
(although it really should). Saves us error handling for that case
though. Also log caps properly.
https://bugzilla.gnome.org/show_bug.cgi?id=629196
Using the IN_CAPS flag for this is brittle, and will fail if either
vorbisparse or vorbistag (which is itself based on vorbisparse) is
inserted between oggdemux and oggmux. Possibly other elements too
(eg, theoraparse, etc).
Using oggstream ensures we Get It Right More Often Than Not.
https://bugzilla.gnome.org/show_bug.cgi?id=629196
Discontinuities are automatically signalled by oggdemux at the start
of a new stream. When oggmux is yet to output actual data pages,
do not signal these discontinuities in the ogg stream.
This patch may miss some actual discontinuities at the very start of
a stream, but avoids the spurious missing pages when encoding happens
normally.
A better fix might involve finding a way to distinguish between actual
data discontinuities and discontinuities merely marking the start of
a new stream.
Fixes an issue with ogg page numbering (would skip a number for no
reason, which then looks like a packet was lost somewhere) when
re-muxing an ogg stream, e.g. when re-tagging in rhythmbox.
https://bugzilla.gnome.org/show_bug.cgi?id=629196
This was causing keyframe_granule to be set to 0 for all streams
when seeking to the beginning of the stream, i.e., at the
beginning of playback. Fixes#619778.
Instead, use either 0 or 1, depending on bitstream version, which give
the correct result for streams which aren't cut off at start.
This allows that function to not return negative granpos.
https://bugzilla.gnome.org/show_bug.cgi?id=638276
The offset part of the granpos is not a sign of the newer encoding.
Use the version number instead.
This fixes the criticals thrown by theoraparse, and (at last) the
remaining part of #553244.
allocate buffers using gst_buffer_new_and_alloc() instead of
gst_pad_alloc_buffer_and_set_caps(), as the first one will
cause the pad to block, and we don't want that since that will
prevent subsequent pads from being fed if a block occurs at
start, when all pads must be fed for playback to start.
This fixes autoplugging of the tiger element and other things.
https://bugzilla.gnome.org/show_bug.cgi?id=637822
Oggdemux will currently try to pad alloc a buffer from the peer when it is
reading the header files. This is a relic from the time where we had an internal
parser and needs to be removed at some point in time.
The problem is that when there is no peer pad yet (which is normal when
collecting headers) we should still continue to parse all the packets of a
page instead of erroring out on NOT_LINKED.
Fixes#632167
Only keep the last valid granulepos we see when scanning the last
pages. It is possible that the last page that we inspect has a -1 granulepos, in
which case we want to keep the previous valid time instead.
Fixes#631703
Files with a skeleton, or other files with a stream that ends before the end of
the chain would start playing from the end of the chain when trying to seek with
a negative rate at a position between the end of any stream and the end of the
chain.
This is due to the loop in _do_seek() assuming that pages will be encountered
for all streams shortly after the place where we want to seek, as found by
do_binary_search().
In the first iteration of the loop, stream ends are now checked against the
time of the current page.
Just cast the pointer diff, so it works everywhere without
warnings. Can't use %tu, because that modifier is C99. Warning
was: "format '%li' expects type 'long int', but argument 8 has
type 'int'".
Convert seek requests to bytes using the bitrate and forward them upstream. Does
not quite work because the flushing and resyncing is not implemented yet.
We update the passed begintime argument to narrow our search region in the
binary search. This means that it does not always contain the chain begin time
after a couple of bisects. Use the real chain->begin_time to bring the
granuletime to the time in the chain instead.
Fixes#610005
This can happen if the combined flow return is not OK although the
allocation succeeded or if the packet in question is a BOS and we're
not going to push headers.
Fixes bug #608699.
Use _ogg_packet_free() instead of gst_mini_object_unref in one more
place now that the header list contains ogg packets and not buffers.
file: Stephen_Fry-Happy_Birthday_GNU-nq_600px_425kbit.ogv
Last stop is already based on the chain start and there is no need
to subtract the chain start as it may lead to a negative overflow.
This was causing seeking issues when the target chain was not
the first one (that has chain start = 0)
Fixes#606382
When performing seeks, the granulepos should be offset by
its chain start time to avoid using wrong values to
update segment's last_stop. A sample file is indicated on
bug #606382
Keep a list of headers for each stream of a chain. When a chain is activated,
push the headers before pushing the data so that decoders can sync.
Fix seeking in chains, take the chain start time into account when comparing
timestamps.
See #606382
Implement keyframe seeking in oggdemux by doing the double seek trick. First
seek to the required position, then read pages for all streams to grab the
granulepos (to know the timing of the keyframe) of each stream, then seek back
to the first keyframe.
After a seek, discard all packets before the packet with the granulepos on it so
that the output buffers contain valid timestamps.
Reorder some code so that we check the timestamps before allocating and pushing
an output buffer.
Do more checks on valid packets in ogm mode.
OGM demuxing no longer requires helper elements. It's done internally
in oggdemux. Vorbis comments are still not handled because I don't
have anything to test with.
gstoggstream.c:419: error: format ‘%lld’ expects type ‘long long int’, but argument 8 has type ‘gint64’
gstoggdemux.c:2253: error: format ‘%lld’ expects type ‘long long int’, but argument 8 has type ‘GstClockTime’
gstoggdemux.c:2333: error: format ‘%lld’ expects type ‘long long int’, but argument 8 has type ‘GstClockTime’
Timestamp generation was broken by the last commit for formats
with a non-zero granule shift. Also keep track of the last keyframe
so that we can regenerate granulepos for theora.
Add a granule to granulepos conversion function. Fix the duration
function for vorbis. Handle timestamps on header packets differently
and be more careful about calculating OFFSET and OFFSET_END. After
this change, timestamps for vorbis don't exactly match up with the
timestamps that vorbisparse outputs, but it's unclear if vorbisparse
is actually correct and it would add a lot more code to make oggdemux
match vorbisparse. Fixes#602790.
Adds code that parses headers of various formats encapsulated in
Ogg in order to calculate timestamps and durations of each buffer.
Removes the creation of helper decoder elements to do this calculation
via conversion queries.
Fixes: #344013, #568014.
Revert previous 'fix' for bug #588717 and fix it properly, whilst
maintaining the streamheader field on the output caps. Also make
sure we don't leak header buffers we couldn't push when downstream
is unlinked. Add unit test for the presence of the streamheader
field on the output caps and for the issue from bug #588717.
Fix caps warning when there's no element linked downstream, and pass
not-linked flow return value correctly up the chain, so we error out
correctly. Fixes#588717.
For this to work properly, theoradec and vorbisdec need to put
tag events received from upstream into the pending_events list
so they get pushed out after any newsegment event, not before.
