Commit graph

546 commits

Author SHA1 Message Date
David Schleef
ea0d666d11 oggmux: refactor how EOS is determined
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.
2011-06-10 23:54:07 -07:00
Mark Nauwelaerts
99944a5a81 oggmux: determine granulepos metadata using stream mapper whenever possible
... which unfortunately is not the case for all types, but at least so for
most common ones.
2011-06-06 13:02:49 +02:00
Mark Nauwelaerts
de1f593537 oggmux: convert incoming buffer timestamp to running time
... so all subsequent manipulation can take place in the proper timeline
without further ado.
2011-06-06 13:02:47 +02:00
Mark Nauwelaerts
2e7eddd35e oggmux: remove superfluous code
... since there is nothing in oggstream that cares (or even should)
about granulepos for what is being asked from it.
2011-06-06 13:02:43 +02:00
Vincent Penquerc'h
55e767b632 oggmux: prefer headers from caps to determine stream type
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
2011-04-16 11:55:27 +01:00
Tim-Philipp Müller
be59789664 ogg: fix unused-but-set-variable warnings with gcc 4.6
https://bugzilla.gnome.org/show_bug.cgi?id=647294
2011-04-13 23:57:56 +01:00
Tim-Philipp Müller
8f1d77bc3c oggmux: fix uninitialised variable usage and element leak
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.
2011-04-12 12:41:06 +01:00
Alessandro Decina
030f639a8e android: make it ready for androgenizer
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
2011-04-11 07:23:21 +02:00
Josep Torra
cd425d3787 oggmux: fix warning building in mac os x 2011-04-01 15:47:18 +02:00
Sebastian Dröge
1c475f10e1 oggmux: Store the segment directly inside the pad
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.
2011-04-01 11:00:38 +02:00
Sebastian Dröge
fc56c76773 oggmux: Reset the segment on flush-stop events and when going back to READY 2011-04-01 10:58:09 +02:00
Thiago Santos
d1c74779f9 oggmux: Use running time instead of timestamps
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
2011-04-01 10:56:16 +02:00
Thiago Santos
c3aae3dc17 oggmux: Keep track of pad's segments
https://bugzilla.gnome.org/show_bug.cgi?id=643775
2011-04-01 10:56:15 +02:00
Tim-Philipp Müller
5aa02968c9 oggparse: fix list iteration code
Not that it really matters, but let's fix it before someone
notices and makes fun of us.
2011-03-26 19:38:17 +00:00
Tim-Philipp Müller
a7a305a18f oggparse: make sure buffer metadata is writable before setting caps on buffers 2011-03-26 11:59:54 +00:00
Sebastian Dröge
a38fd9f9ec oggmux: Increase the seen header packets count when seeing a header packet
This fixes muxing of Speex content and possibly other formats where the
header detection works by counting the packets.

Fixes bug #644745.
2011-03-14 19:44:41 +01:00
Sebastian Dröge
461d9f2f2c oggmux: Don't handle GstCollectData as GstObject, use the pad instead 2011-02-24 16:22:53 +01:00
Edward Hervey
3fe2777b54 Revert "oggmux,adder: Check if collectpads has been freed"
This reverts commit 6d150873e8.

Depends on a core commit that was reverted.
2011-02-21 12:04:09 +01:00
David Schleef
6d150873e8 oggmux,adder: Check if collectpads has been freed
Core now calls release_pad in finalize, which is usually after
the collectpads has been unreffed.
2011-02-20 23:49:54 -08:00
David Schleef
15e23414d3 oggparse: better detection of delta unit flag 2011-02-17 18:29:56 -08:00
Vincent Penquerc'h
96a1a9dec6 oggmux: ensure serialnos are unique
We do that by checking a newly generated one is not already used in
an existing stream, and doing it again if it is.

https://bugzilla.gnome.org/show_bug.cgi?id=640211
2011-02-03 09:29:28 +00:00
Tim-Philipp Müller
54c19ba6de oggmux: free stream map caps when done 2011-02-02 17:34:42 +00:00
Tim-Philipp Müller
2eac43bd73 oggmux: keep IN_CAPS flag check for header buffers as fallback
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
2011-02-02 17:34:41 +00:00
Vincent Penquerc'h
440002a137 oggmux: use oggstream for less brittleness in recognizing headers
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
2011-02-02 17:34:41 +00:00
Vincent Penquerc'h
b7664eae71 oggmux: do not skip a pageno at start
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
2011-02-01 17:25:43 +00:00
Vincent Penquerc'h
d144f7a3a5 oggmux: cleanup
Remove a pointless string concatentation, and fix an off-by-one in
packetno in a log.

https://bugzilla.gnome.org/show_bug.cgi?id=640189
2011-01-24 20:01:24 +01:00
Sreerenj Balachandran
7cfa18545c oggdemux: Remove dead code 2011-01-24 19:40:00 +01:00
Yang Xichuan
cb9607632f oggparse: Make gst_ogg_parse_submit_buffer() safe
By not passing zero-sized buffers to ogg_sync_buffer()
and checking the return values of libogg functions.

Fixes bug #639136.
2011-01-24 19:39:59 +01:00
Yang Xichuan
43553b4dcb oggdemux: remove outdated comment
https://bugzilla.gnome.org/show_bug.cgi?id=639121
2011-01-10 12:34:54 +00:00
David Schleef
3c4466b816 oggdemux: ignore header pages when looking for keyframe
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.
2011-01-06 12:34:32 -08:00
Vincent Penquerc'h
b03b223fb1 oggstream: when the last keyframe position is not known, do not use -1
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
2011-01-06 12:34:32 -08:00
Tim-Philipp Müller
542256fd55 ogg: fix typo in comment 2010-12-29 11:51:42 +00:00
Vincent Penquerc'h
a5533c21d9 oggstream: fix interpretation of Theora granule position
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.
2010-12-29 12:40:25 +01:00
Wim Taymans
68e64770a0 oggdemux: handle pads that are not added yet
Don't try to stream data on pads that are not added yet. This happens while we
discover the different streams.
2010-12-28 19:39:18 +01:00
Vincent Penquerc'h
85cafac6af oggstream: implement tag extraction for Kate streams
This will mainly allow Totem to know the language of those streams,
so the subtitle selection menu gets properly filled out.

https://bugzilla.gnome.org/show_bug.cgi?id=638005
2010-12-27 10:53:24 +00:00
Vincent Penquerc'h
dd135119d9 oggdemux: set headers on caps
This will allow switching from one stream to another without having to send
the headers for the new stream again.

https://bugzilla.gnome.org/show_bug.cgi?id=637927
2010-12-25 16:55:15 +01:00
David Schleef
0c4820914c oggstream: Fix parsing of theora size 2010-12-22 15:30:36 -08:00
Vincent Penquerc'h
e443ae6000 oggdemux: Don't use gst_pad_alloc_buffer()
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
2010-12-22 20:19:15 +01:00
Vincent Penquerc'h
22aa87e98f ogg: implement packet duration query for kate streams
https://bugzilla.gnome.org/show_bug.cgi?id=637519
2010-12-18 20:33:09 +01:00
Mark Nauwelaerts
524b7c60f2 oggstream: additional tag extraction
... supporting theora, flac, speex, celt.

Fixes #629349.
2010-12-03 13:54:08 +01:00
Mark Nauwelaerts
d5055a9f23 oggstream: use separate tag extraction vfunction 2010-12-03 13:54:06 +01:00
Mark Nauwelaerts
cc3dfd31d0 oggstream: refactor vorbis comment tag extraction 2010-12-03 13:54:04 +01:00
Mark Nauwelaerts
a4aacb9d2c oggdemux: plug some oggstream leaks 2010-12-03 13:54:01 +01:00
Mark Nauwelaerts
515f33bd3b oggstream: streamline tag extraction and prevent some leaks 2010-12-03 13:53:59 +01:00
Mark Nauwelaerts
63ba9eafb1 oggdemux: send stream tags after newsegment and global tags 2010-12-03 13:53:56 +01:00
Sreerenj Balachandran
2319c85f46 oggdemux: perform more (vorbis comment header) tag extractions
In particular, move comment header parsing to gstoggstrem.c.
Thanks to Felipe Contreras.

Fixes #629349 (partially).
2010-12-03 13:53:54 +01:00
David Schleef
d421ccb440 oggstream: fix incorrect warning on skeleton headers 2010-12-02 17:36:33 -08:00
David Schleef
2b91bd1bad oggparse: Set DELTA_UNIT on buffers 2010-12-02 17:36:26 -08:00
Wim Taymans
e0d683f3ed oggdemux: fix wrong flowreturn handling
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
2010-10-29 11:48:18 +01:00
Wim Taymans
6c14f6c970 ogg: add some more debug statements 2010-10-29 11:47:53 +01:00
Wim Taymans
fea50233bb oggdemux: only keep last valid granulepos
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
2010-10-12 16:06:57 +02:00
David Schleef
9dc59cff15 oggdemux: Fix check for last page 2010-10-12 16:06:57 +02:00
David Schleef
64dfad46f2 oggdemux: change checks from is_skeleton to is_sparse 2010-10-12 16:06:57 +02:00
David Schleef
5ae2f04856 oggdemux: move is_sparse into stream map 2010-10-12 16:06:56 +02:00
Guillaume Emont
b41cd04289 oggdemux: fix seeking with negative rate with skeleton
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.
2010-10-10 18:26:13 +02:00
Sebastian Dröge
6357bdef63 oggdemux: Don't reset the pad when pushing resulted in NOT_LINKED
The pad might be linked later and after resetting it it will
only work after resetting all of oggdemux.
2010-08-30 15:50:26 +02:00
Sebastian Dröge
b899bca94d oggdemux: Don't use GST_FLOW_IS_FATAL()
And while we're at it, handle WRONG_STATE as error too
in oggdemux and WRONG_STATE and NOT_LINKED in oggaviparse.
2010-08-27 17:23:46 +02:00
Sebastian Dröge
5fd9fc13d5 oggstream: static forward declarations are forbidden by the C standard
...and actually cause compiler errors on VC++. Change it to an extern
forward declaration and non-static definition.
2010-08-08 17:42:04 +02:00
Fredrik Söderquist
3c06cabb0b oggdemux: Handle errors from _get_next_page in _do_seek.
If the source element failed here, oggdemux would crash.

Fixes #623218.
2010-06-30 17:20:55 +01:00
Sebastian Dröge
9a971980b7 oggdemux: activate_chain must not be called with a NULL chain
It will crash later and shouldn't really happen anyway unless
something is really wrong.
2010-06-14 11:11:56 +02:00
Tim-Philipp Müller
0f3ea721e4 ogg: fix debug message printf format some more
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'".
2010-06-14 00:33:04 +01:00
Philippe Normand
c25bf9777a oggdemux: Fix format string compiler warning on OS X 2010-06-12 13:04:43 +02:00
Sebastian Dröge
4caee01a06 oggdemux: Handle SEEKING query in push mode too 2010-06-09 17:02:59 +02:00
Sebastian Dröge
bb7773b40a oggdemux: Update the total time from the Skeleton 4 indexes
Fixes bug #620939, see bug #607945.
2010-06-09 16:38:34 +02:00
Sebastian Dröge
7bade30f5e oggstream: Implement latest version of the Skeleton 4.0 spec
Fixes bug #620939.
2010-06-09 16:34:05 +02:00
Sebastian Dröge
3f3691085a oggdemux: Implement correct parsing of Skeleton 4.0 index packets 2010-06-08 13:11:05 +02:00
Sebastian Dröge
cba98e05cf oggdemux: Add parsing of Skeleton 4.0 indexes 2010-06-08 12:01:15 +02:00
Sebastian Dröge
f213c8ebe3 oggdemux: Parse segment length and content offset from fishead
And print them for debugging purposes. Not sure if we can do anything useful
with this information.
2010-06-08 11:40:32 +02:00
Sebastian Dröge
aa8daa862e oggdemux: Parse Skeleton stream major/minor version 2010-06-08 11:31:35 +02:00
Sebastian Dröge
7eea3c25c9 oggdemux: Use binary search for searching in the index 2010-06-08 11:26:26 +02:00
Sebastian Dröge
981af36b71 oggmux: Start a new page for every CMML buffer 2010-06-08 07:34:53 +02:00
Sebastian Dröge
8a678da1fe oggdemux: Fix sizes again, this time for real 2010-05-20 14:21:19 +02:00
Sebastian Dröge
bf9d573cd1 oggdemux: Fix size checks 2010-05-20 13:58:14 +02:00
Sebastian Dröge
8fd69f2cb7 oggdemux: Drop all other Ogg VP8 header packets and make VP8 mapping check a bit more strict 2010-05-20 10:19:29 +02:00
Sebastian Dröge
563a96ef1a ogg: Some more minor adjustments for the VP8 Ogg mapping 2010-05-20 08:52:49 +02:00
Sebastian Dröge
838c96fe8f ogg: Update to the latest VP8 mapping 2010-05-19 21:35:19 +02:00
Sebastian Dröge
ac2d0170a8 oggdemux: Fix granulepos->key granule calculation for Dirac 2010-05-19 19:23:07 +02:00
Sebastian Dröge
d6b6d0cdad oggdemux: Add support for mapping specific granulepos to key granule mapping 2010-05-19 19:23:07 +02:00
Sebastian Dröge
5fc1309703 ogg: Implement Ogg VP8 mapping 2010-05-19 19:23:07 +02:00
Wim Taymans
02a78f528c oggdemux: don't seek when no current chain
Avoid a crash when we try to seek when there is no current chain.
2010-05-06 13:10:54 +02:00
Wim Taymans
e33425e4d0 oggdemux: ignore the skeleton start time
Ignore the skeleton start time as it is usually wrong for live streams
and we have the needed logic to calculate it anyway.
2010-05-06 12:21:38 +02:00
Wim Taymans
27de725038 oggdemux: wait for headers before exposing chains
Wait until we have all the stream headers before we start exposing the streams
of a chain.
2010-05-06 12:06:09 +02:00
Wim Taymans
549bc3c80e oggdemux: use index to estimate bitrate
When we have an index, use it to much more accurately estimate the total stream
bitrate.
2010-05-06 10:58:01 +02:00
Tim-Philipp Müller
f22b115804 oggdemux: printf format fixes 2010-05-05 13:25:02 +01:00
Wim Taymans
7b3f9aa504 oggdemux: handle corrupt indexes
Make sure we handle and receover from corrupt indexes.
2010-05-04 17:38:50 +02:00
Wim Taymans
e84b3c838c oggdemux: fix EOS check 2010-05-04 15:47:29 +02:00
Wim Taymans
50e556cc84 oggstream: avoild division by 0 2010-05-04 13:51:07 +02:00
Wim Taymans
12929944a7 oggdemux: cleanup unused defines 2010-05-04 13:50:46 +02:00
Wim Taymans
a57b461af8 oggdemux: use the index in push mode when we can
When seeking in push mode, try to use the index first before we use the bitrate
estimation.
2010-05-04 13:36:58 +02:00
Wim Taymans
03c3aa6377 oggdemux: use skeleton duration when possible 2010-05-04 13:06:24 +02:00
Wim Taymans
5c770a17db oggstream: parse duration from 3.3 skeleton 2010-05-04 13:02:19 +02:00
Wim Taymans
ff4479f00a oggdemux: more index parsing work 2010-05-04 12:11:18 +02:00
Wim Taymans
97319a6276 oggdemux: clean up fishead/fisbone parsing
Remove some redundant code for parsing fishead streams. Actually use the data we
parsed (mostly start_time).
2010-05-04 12:05:15 +02:00
Wim Taymans
f96caa17b3 oggdemux: implement seek in push mode
Refactor start time collection code.
When we receive a flush_stop, resync to the new start time and push out a new
segment event.
2010-05-04 11:26:14 +02:00
Wim Taymans
9be4e53001 oggdemux: make event handling better
Explicitly handle FLUSH events and resync on FLUSH_STOP.
Make send_event return a boolean.
Use more performant send_event function to forward events.
2010-05-04 11:26:14 +02:00
Wim Taymans
c4ec1c4c62 oggdemux: implement seeking in pushmode
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.
2010-05-04 11:26:14 +02:00
Wim Taymans
d9b7003926 oggdemux: refactor for seeking in pushmode
refactor the code a little to prepare for seeking in push mode
2010-05-04 11:26:14 +02:00
Wim Taymans
f9ca4f6097 oggdemux: use bitrate to estimate length in pushmode
Parse the bitrate from the various streams.
Use the bitrate and the upstream length in bytes to estimate the total stream
duration in push mode.
2010-04-30 17:43:06 +02:00
Mark Nauwelaerts
e23d6bbda7 oggdemux: only EOS when all streams are EOS 2010-04-30 12:51:54 +02:00
Mark Nauwelaerts
e8ae2adef1 oggdemux: fix debug message 2010-04-30 12:51:51 +02:00