This allows the AC-3 decoder to be used directly with RealMedia
decoders that unlike the libavformat one do not byte-swap automatically.
Since the new code is only used in case we would fail directly otherwise
there should be no risk for regressions.

Depending on error_recognition is not correct, low values do
certainly not mean it is ok to crash.

Based on a patch by Reimar Döffinger <Reimar.Doeffinger@gmx.de>

The Atom has SSSE3 support, which is useful in many cases, but sometimes the
SSSE3 version is slower than the SSE2 equivalent on the Atom, but is generally
faster on other processors supporting SSSE3. This flag allows for selectively
disabling certain SSSE3 functions on the Atom.

fixes issue2649.

It's pointless, since there's a seek to the end of tag later.

Conio.h is a non-standard header and may not have kbhit()
prototyped. This fixes compile on OS/2 where the EMX version (we're
using a fork) of conio.h only has getch() and getche().
Signed-off-by: NMans Rullgard <mans@mansr.com>

This updates the wmv8-drm reference after c47d3835.
Signed-off-by: NMans Rullgard <mans@mansr.com>

This is required for the generated .lib file to actually
be usable by MSVC.
Signed-off-by: NLuca Barbato <lu_zero@gentoo.org>

This makes the amr demuxer support seeking, closing roundup
issue 2593.

Previously, the field pointed to the second byte of the packet
(which is the first byte of the actual AMR payload).

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: NLuca Barbato <lu_zero@gentoo.org>

That fixes .lib creation for the win64 target.

A similar variable for the total stream duration was changed to
int64_t in b79c3df0, due to overflows in some odd
streams.
Signed-off-by: NLuca Barbato <lu_zero@gentoo.org>

This fixes crashes on exit when plaing some RealRTSP streams
(among other), e.g.
rtsp://dl.lib.brown.edu:554/areserves/1093545294660883.mp3.
These crashes have been present since 3e68b3ba.
Signed-off-by: NRonald S. Bultje <rsbultje@gmail.com>

Signed-off-by: NRonald S. Bultje <rsbultje@gmail.com>

Current code might stop in the middle of an invalid tag.

fixes issue2650
Signed-off-by: NRonald S. Bultje <rsbultje@gmail.com>

Signed-off-by: NRonald S. Bultje <rsbultje@gmail.com>

Also reset coded_block at each new slice, fixes problems in slice playback.

Introduce end: and err: labels at the end of vc1_decode_frame(), which
will clean up allocated memory consistently.

It's an evil hack that assumes an AVIOContext is always based on top of
an URLContext.
It's also not used anywhere.
Signed-off-by: NRonald S. Bultje <rsbultje@gmail.com>

Signed-off-by: NRonald S. Bultje <rsbultje@gmail.com>

Signed-off-by: NRonald S. Bultje <rsbultje@gmail.com>

This fixes a typo in the size calculation for the sample buffer
introduced in cbabccc3.
Signed-off-by: NMans Rullgard <mans@mansr.com>

Remove usage of tcgetattr and tcsetattr to modify terminal
parameters, and rely on ctrl-c to stop instead of pressing 'q'.
On systems with conio.h, keep the old behavior.

Changing the terminal settings causes problems if multiple instances
are running asynchronously on the same terminal, such as during a
parallel FATE run, or if the process crashes before restoring the
terminal.  In both cases, the terminal state is messed up requiring
a manual reset.
Signed-off-by: NMans Rullgard <mans@mansr.com>

This was missed in 0aded948.
Signed-off-by: NMans Rullgard <mans@mansr.com>

Patch by Mike Sheutzow with some additions and changes by Stefano.
Signed-off-by: NLuca Barbato <lu_zero@gentoo.org>

Change "file:stream[:syncfile:syncstream]" to
"file.stream[:syncfile.syncstream]", the latter syntax is more
consistent with the stream identifier form printed by
av_dump_format(), and more readable. Both syntaxes are currently
accepted by the opt_map() code.
Signed-off-by: NLuca Barbato <lu_zero@gentoo.org>

Signed-off-by: NMans Rullgard <mans@mansr.com>

This leaves fft.h with only the core FFT and MDCT definitions
thus making it more managable.
Signed-off-by: NMans Rullgard <mans@mansr.com>

These windows do not really belong in fft/mdct files and were
easily confused with the similarly named tables used by rdft.
Signed-off-by: NMans Rullgard <mans@mansr.com>

This function is not tightly coupled to mdct, and it's in the way
of making a fixed-point mdct implementation.
Signed-off-by: NMans Rullgard <mans@mansr.com>

This removes the rather pointless wrappers (one not even inline)
for calling the fft_calc and related function pointers.
Signed-off-by: NMans Rullgard <mans@mansr.com>

This way building ffmpeg on mingw won't require windows specific tools

This avoids the main/SDL_main declaration clash caused by SDL on
Windows and other platforms.