The invalid test did not cause any actual problem since
the first branch is only possible with bogus filters.

Fix coverity issue CID 733850.

Fixes: CID733804
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

The code would crash before if it was NULL

Fixes CID733804
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

This fixes a regression with the scale filters input changing.
In the long run filters should get a flag to indicate support of this
and then this flag be used here.
But the regression should not be left standing until thats done.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Several filters support this already.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Fixes CID717571
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Make the filter accept named options for the first argument, and update
documentation accordingly.

Also make sure we use the actual width of the image and not the whole
linesize.

Fixes CID718988
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Fixes: CID718989
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Fixes CID733727.

Fixes CID733798
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Fixes CID733799
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Fixes: CID733841
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Fixes CID733842
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

A number of compilers, for example those from TI and IBM, choke on
these initialisers.  The current style is also quite ugly.

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

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

This avoids a memleak with noise_str.

Does not make a difference in this case since sizeof(uint8_t*) ==
sizeof(uint8_t**). Fixes CID703814.

Fixes CID732273.

Fixes CID732272.

Prevents writes to freed memory and the leak of the old extended data.
Fixes CID732303.

When there is only 1 channel, the planar and interleaved formats of the same
data type should be treated as identical.

Signed-off-by: Bobby Bingham <uhmmmm@gmail.com>

Anonymous structs cannot be forward declared and have no benefit.

Fixes Bug 203.

CC:libav-stable@libav.org

This is more consistent with what the rest of Libav does.

This breaks API.

Right now it incorrectly assumes that the frames are sent in the proper
order, which worked with old ffmpeg and avconv versions by accident.

Output frames correspond 1:1 to input frames on the main input.
So use the main input timebase for output.

Allow extensibility.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Implement av_buffersink_read() and av_buffersink_read_samples()
for ffmpeg's version of buffersink.

With this change, avconv linked against ffmpeg's libraries passes
the same number of tests whether it uses ffbuffersink or
buffersink_old.

They are part of the public API. Their libav compatibility
counterpart are already enabled unconditionally.

Parse expression only when a new value is explicitly specified.

In particular, avoid double free in case an old expression value is
cached, it is set in the context, it is freed as the old value, and
finally the pointer stored in the context is freed again when the filter
is released.

Fix programmatic selection of accepted pixel formats.

Signed-off-by: Stefano Sabatini <stefasab@gmail.com>

Fixes ticket1752

Commit message by commiter
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>