The 'edts' write function can now generate an initial empty edit resulting in a track-specific presentation delay.
This is automatically calculated and inserted for any track where the initial DTS != 0.
Added support for long (version==1) timecodes.
Some received packets can have size 0. The return value from
av_malloc(0) may be NULL, which is ok if the size was 0. On
OS X, however, the returned pointer is non-null but leads to
crashes when trying to free it.
Signed-off-by: Martin Storsjö <martin@martin.st>
strtol could return negative values, leading to various error messages,
mainly "non-monotonically increasing dts".
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Since a private class is set for this muxer, the callers will
assume that the private data starts with an AVClass pointer.
If no such member exists, the first few bytes of the struct
will be overwritten, and the class pointer may be broken at
any later time.
Signed-off-by: Martin Storsjö <martin@martin.st>
asf_read_seek() inside the asf demuxer already does the
right thing, it tries the index and if that fails it uses
binary search. If binary search is called from outside of asfdec.c
it will fail because the asf code cannot clean up after itself.
Therefore introduce AVFMT_NOBINSEARCH that prevents the seek
code to fallback to binary search and AVFMT_NOGENSEARCH that
prevents the seek code to fallback to generic search.
The later parsing of payload data depends on the configuration
being present. If it hasn't been configured properly yet,
parsing a data packet may lead to a crash.
Signed-off-by: Martin Storsjö <martin@martin.st>
In the main loop, stream_number is incremented after checking the stream type,
so the search usually will not find the wanted stream.
This patch eliminates the useless stream_number variable and introduces a new
one, called real_stream_index to store the real stream index of the current
stream, no matter if we are looping through all the streams or only the streams
of a program.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Yet another fix for the code originally designed for use without related_stream.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
This avoids the creation of a new AVStream instead of replacing it when
a stream reset occurs (track change with some webradios for example).
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Adding _POSIX_C_SOURCE to CPPFLAGS globally produces all sorts of problems
since it causes certain system functions to be hidden on some (BSD) systems.
The solution is to only add the flag on systems that really require it, i.e.
glibc-based ones.
This change makes BSD systems compile out-of-the-box without the need for
adding specific flags manually. It also allows dropping a number of flags
set manually on a file-per-file basis, but were only present to work around
breakage introduced by the presence of _POSIX_C_SOURCE.
Also add _XOPEN_SOURCE to CPPFLAGS for glibc systems. We use XSI extensions
in several places already, so it is preferable to define it globally instead
of littering source files with individual #defines only needed for glibc.
If an MP3 file contains the string NSVs, the NSV probe will confuse it for an
NSV file. Check for 0xBEEF after a Video/Audio chunk to achieve more accuracy.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
This is more like what VLC does. If the server doesn't mention
supporting GET_PARAMETER in response to an OPTIONS request,
VLC doesn't send any keepalive requests at all. After this patch,
libavformat will still send OPTIONS keepalives if GET_PARAMETER
isn't explicitly said to be supported.
Some RTSP cameras don't support GET_PARAMETER, and will
close the connection if this is sent as keepalive request
(but support OPTIONS just fine, but probably don't need any
keepalive at all). Some other cameras don't support using
OPTIONS as keepalive, but require GET_PARAMETER instead.
Signed-off-by: Martin Storsjö <martin@martin.st>
This should fix behavior introduced by commit
96573c0d76. Av_rescale_rnd() is not
lossless so if two timestamps are equal after being rescaled they are
not always actually identical. This patch use av_compare_ts() to get
always a correct result.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
Earlier, AVIO_RDWR was handled as READ, and all the checks
for the AVIO_WRONLY flag explicitly meant the write-only case.
When changed from old AVIO/URL contants to AVIO_FLAG in
59d96941f0, these comparisons were updated incorrectly, by
mapping checks for AVIO_WRONLY to checks for AVIO_FLAG_WRITE.
When receiving RTP over UDP, the urlcontext is opened with
READ_WRITE flags.
This patch updates the flag comparisons to check for the same
conditions as the code did prior to 59d96941f0.
Signed-off-by: Martin Storsjö <martin@martin.st>
This fixes, for instance, the case that there is a track that has some samples with composition time offset and has a track run without sample-composition-time-offsets-present.
Add an extra size validity check in asf_read_frame_header(). Without
this asf->packet_size_left may become negative, which triggers an
assertion failure later.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
As per issue2629, most 23.976fps matroska H.264 files are incorrectly
detected as 24fps, as the matroska timestamps usually have only
millisecond precision.
Fix that by doubling the amount of timestamps inspected for frame rate
for streams that have coarse time base. This also fixes 29.970 detection
in matroska.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
(cherry picked from commit 78431098f9)
Tested with mplayer based on this report
http://thread.gmane.org/gmane.comp.video.mplayer.user/66043/focus=66063
Signed-off-by: Reinhard Tartler <siretart@tauware.de>
While deprecated, they're totally useless as long as their values
are different from the AVIO_FLAG values that are used internally.
Currently, this leads to old libav applications still compiling
correctly (since we haven't removed the fallback wrappers), but
failing since the functions internally compare to the new AVIO_FLAG
values.
These should be removed at some point, but they aren't removed yet.
The intent is to be able to recompile an old application against
the new ABI without modifying the code, and this doesn't work
currently.
Signed-off-by: Martin Storsjö <martin@martin.st>
Previously, the function would lead to an infinite wait (by
returning AVERROR(EAGAIN)) on sockets indicating an error
via either of these poll flags.
Signed-off-by: Martin Storsjö <martin@martin.st>
Official AVI specification says that stream header in case of video contains
BITMAPINFO, which is equal to BITMAPINFOHEADER and optional palette. Currently
lavf AVI demuxer thinks otherwise which produces garbage on codecs that have
both palette and extradata (luckily, there are not so many such codecs).
An example of such file is:
http://samples.multimedia.cx/V-codecs/KMVC/baseball1.avi
(IIRC, MSS1 or MSS2 also had such situation but they are still not supported
by lavc).
As a side note, passing palette in extradata as it's been done previously is
not quite correct since proper _extra_ data is surplus bytes in
BITMAPINFOHEADER, not including palette.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
If there are no variants, the total bitrate of the single
stream isn't known, and exporting variant_bitrate = 0 does
look weird, since there really aren't any variants.
Signed-off-by: Martin Storsjö <martin@martin.st>
All file names should be in UTF-8 within libavformat.
This is handled by mapping the open() function to an internal one
in os_support.h for windows.
fopen() could be overridden in the same way, but if that would be
used from ffmpeg.c, it would add a dependency on an ff prefixed
internal lavf function.
Signed-off-by: Martin Storsjö <martin@martin.st>
This should hopefully fix roundup issue 2586.
This commit only implements it in the demuxer, not in the
protocol handler. If desired, some of the code could be
refactored to be shared by both implementations.
Signed-off-by: Martin Storsjö <martin@martin.st>
This can later be extended to support other AES bit sizes,
encryption, other crypto algorithms, reading the key from a URL, etc.
In order to use it, the key and initialization vector has to be
passed via AVOptions. Since such options can't be passed to
protocols from the command line, the protocol is currently
only for libavformat internal use.
Signed-off-by: Martin Storsjö <martin@martin.st>
AVERROR_NUMEXPECTED is used only in the image muxer and demuxer, and
has a too much specific meaning, which is better explained through a
log message. Thus it can be replaced by AVERROR(EINVAL).
This breaks API.
Signed-off-by: Stefano Sabatini <stefano.sabatini-lala@poste.it>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
After switching this from a statically allocated array to a
dynamically allocated one in the major bump, this needs explicit
freeing.
Signed-off-by: Martin Storsjö <martin@martin.st>
The problem with url_exist() is that it tries to open a resource in
RDONLY mode. If the file is a FIFO and there is already a reading
client, the open() call will hang.
By using avio_check() with access mode of 0, the second reading
process will check if the file exists without attempting to open it,
thus avoiding the lock.
Fix issue #1663.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Make AVIO_FLAG_ access constants work as flags, and in particular fix
the behavior of functions (such as avio_check()) which expect them to
be flags rather than modes.
This breaks API.
Also add inttypes.h #include for PRId64 macros.
This fixes the following warnings:
libavformat/xwma.c:147: warning: too many arguments for format
libavformat/xwma.c:151: warning: too many arguments for format
ff_get_wav_header is reading data from a WAVE file and then uses it
(without validation) to malloc a buffer. It then proceeded to read
data into the buffer, without verifying that the allocation succeeded.
To address this, change ff_get_wav_header to return an error if
allocation failed, and adapted all calling code to handle that error.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
The mpegts muxer does not set the stream time base using
av_set_pts_info, but expects it to have the default value of 1/90000.
If the calling code changes stream pts before writing the header,
other muxers override the time base at that point (like mpegenc.c).
This code could be executed if the demuxer reads more than one
segment before returning from av_open_input_stream.
Signed-off-by: Martin Storsjö <martin@martin.st>
The new function is more flexible than url_exist(), as it allows to
specify which access flags to check, and does not require an explicit
open of the checked resource.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Stop the avio input at a point where amf_parse_object can
continue parsing the end of the object seamlessly, when all
data is available.
If unsupported data is encountered within the keyframes object,
try seeking to the start of the keyframes object - if the seek
back was successful, the caller can continue parsing the rest
of the AMF data.
Signed-off-by: Martin Storsjö <martin@martin.st>
Current keyframes data parser unconditionally rewind metadata to
the end at the end of function. As result ALL metadata located
after keyframes index not parsed, and as metadata object can have
ANY placement inside metadata it can lead to unpredictable result
(bitrate can not be found, etc.). As result FLV movie will not
play at all in such situation.
Signed-off-by: Martin Storsjö <martin@martin.st>
The printf function is forbidden in library code, but not in test programs,
so #undefine the printf macro that disables the system printf function.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Support the URL_FLAG_NONBLOCK semantic and uniform the protocol.
The quick retry loop is already part of retry_transfer_wrapper.
The polling routine is common to the network protocols:
udp, tcp and, once merged, sctp.
This avoids issues where EOF at the end of the segment is given
the variant demuxer. Now the demuxers only see one single data
stream (as when using the applehttp protocol handler).
It doesn't look fit to be a part of the public API.
Adding a temporary hack to ffserver to be able to use it, should be
cleaned up when somebody is up for it.
In particular, now it assumes that
a) chapters are chronologically ordered
b) chapters have the same timebases
c) duration of the stream is known
and asserts if any of these is not met.
Make it properly deal with harsher conditions.
fixes issue2320
Allows distinguishing between EOF and IO error in read_packet return code.
Signed-off-by: Jindrich Makovicka <makovick@gmail.com>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
'keyframes' metatag is not part of the standard, it is just
convention to use such kind of metatag information for indexing.
Structure is following, it allows to have it inconsistent:
keyframes:
times (array):
time0 (num)
time1 (num)
time2 (num)
filepositions (array)
position0 (num)
position1 (num)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
A zero sized stsc run doesn't make a lot of sense but the spec does not
prohibit them and MPlayer VLC demuxers support them.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
According to the RFC, GET_PARAMETER should be used for
this, and according to a report from Tim Ouellette,
OPTIONS doesn't work for keeping the connection alive for some
servers. Also, live555 uses GET_PARAMETER for this purpose.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
FFmpeg did not seek back to the original position, but to "0", making
reading a VBR tag impossible.
(issue 2645)
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
Amazon S3 sends header field names all lowercase.
This is actually acceptable according to the HTTP standard.
http://tools.ietf.org/html/rfc2616#section-4.2
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
When a normal Block is parsed, duration is initialized to
AV_NOPTS_VALUE. If it is not changed, then the track's default
duration is used. But for SimpleBlock, duration is initialized to
0 instead of AV_NOPTS_VALUE. This is due to the difference in how
EBML_NEST vs EBML_PASS are processed. Setting duration to 0 leads
eventually to wrongly estimate the frame duration in util.c
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
This allows libavformat to guess an estimated duration for
amr files.
For streams with varying bit rates (or with silence descriptors
or "no frame" blocks) the guess is, of course, inaccurate.
According to the RFC, the default is multicast if nothing is
specified, which doesn't make sense for TCP.
According to a bug report, some Axis camera models give a
"400 Bad Request" error if this is omitted.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
A similar variable for the total stream duration was changed to
int64_t in b79c3df088, due to overflows in some odd
streams.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
