movenc: Add support for writing language codes into ISML manifests

Streaming servers appear to ignore all other language metadata.

Signed-off-by: Jan Ekström <jeebjp@gmail.com>
Signed-off-by: Josh de Kock <josh@itanimul.li>

lavf/mov.c: Make audio timestamps strictly monotonically increasing inside an edit list.

Fixes gapless decoding. Adjust skip_samples field correctly in case of DISCARDed audio frames.

Signed-off-by: Sasi Inguva <isasi@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavc/utils.c: Subtract skip_samples when frame is DISCARDed.

Signed-off-by: Sasi Inguva <isasi@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

ffmpeg_vaapi: fix choice of decoder_format

The check could previously never evaluate to true, probably due to
a typo.

Reported-By: Mihai Chindea <mihai.chindea@uti.eu.com>
Signed-off-by: Moritz Barsnick <barsnick@gmx.net>
Tested-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

doc/muxers: fix hlsenc options examples error

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Reviewed-by: Moritz Barsnick <barsnick@gmx.net>
Signed-off-by: Steven Liu <lingjiujianke@gmail.com>
Signed-off-by: Lou Logan <lou@lrcd.com>

doc/codecs.texi: fix and expand color related options

Found-by: Michael Niedermayer <michael@niedermayer.cc>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: James Almer <jamrial@gmail.com>

movenc: use similar logic to DASH when writing bit rate to ISML

This way, in case of bit rate not being set, max_bitrate will be
used instead. This enables, for example, re-using max_bitrate
information from the input or doing transcoding with a rate
control mode that is not bit rate based.

Signed-off-by: Jan Ekström <jeebjp@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/vf_colorspace: fix range for output colorspace option

Rreviewed-by: BBB
Signed-off-by: James Almer <jamrial@gmail.com>

avformat/concatdec: don't call open_file when seek position within a file

avutil/hwcontext: use CONFIG_QSV instead of CONFIG_LIBMFX for qsv

See "[FFmpeg-devel] [PATCH] hwcontext: add a QSV implementation"

Suggested-by: nablet developer <sdk@nablet.com>
Signed-off-by: James Almer <jamrial@gmail.com>

ffmpeg_cuvid: Don't unnecessarily include nvcuvid.h

Merge commit '59e7361cc791e5103be1712dc59a2055f118d0da'

* commit '59e7361cc791e5103be1712dc59a2055f118d0da':
  hwcontext: add a QSV implementation

Conflicts:
    doc/APIchanges
    libavutil/version.h

Merged-by: James Almer <jamrial@gmail.com>

Merge commit '6f19bbcf8532d018d8d6d82e000738d0ac2385c9'

* commit '6f19bbcf8532d018d8d6d82e000738d0ac2385c9':
  qsvdec: move reading the user-provided session to qsv_decode_init()

Conflicts:
    libavcodec/qsvdec.c

Merged-by: James Almer <jamrial@gmail.com>

Merge commit 'e85f6f7f8d037c0af0f294000718d9ba22753baa'

* commit 'e85f6f7f8d037c0af0f294000718d9ba22753baa':
  lavc: allow using AVCodecContext.hw_frames_ctx for decoding

Conflicts:
    doc/APIchanges
    libavcodec/version.h

Merged-by: James Almer <jamrial@gmail.com>

Merge commit '40dd5166d2ba4f9035b93748840e408cd8be40e5'

* commit '40dd5166d2ba4f9035b93748840e408cd8be40e5':
  truemotion2rt: Use ff_set_dimensions

Merged-by: James Almer <jamrial@gmail.com>

Merge commit 'a8164323374e86ce5f93759230868c98356833a2'

* commit 'a8164323374e86ce5f93759230868c98356833a2':
  pixdesc: Add new SMPTE 431, 432, and 2085 color properties

Conflicts:
    libavcodec/options_table.h
    libavcodec/version.h
    libavutil/pixdesc.c
    libavutil/pixfmt.h
    libavutil/version.h

Merged-by: James Almer <jamrial@gmail.com>

Merge commit '5d560d38deca1e4705e6d3784d737363b9c830fe'

* commit '5d560d38deca1e4705e6d3784d737363b9c830fe':
  pixfmt: Add ARIB STD-B76 color transfer characteristic

See 785038c92cc7fc1da437576382083246ca598fce

Conflicts:
    libavcodec/options_table.h
    libavcodec/version.h
    libavutil/pixdesc.c
    libavutil/version.h

Merged-by: James Almer <jamrial@gmail.com>

Merge commit '6ed0f70f97c882813199b3bafd724ceeb43659de'

* commit '6ed0f70f97c882813199b3bafd724ceeb43659de':
  avconv: factor out initializing stream parameters for streamcopy

Conflicts:
    ffmpeg.c

Merged-by: James Almer <jamrial@gmail.com>

Merge commit '5fa255b65c7887cc913f097aed1b581fbf1a8745'

* commit '5fa255b65c7887cc913f097aed1b581fbf1a8745':
  avconv: initialize output framerate earlier

Skipping this for now. It's not needed until several committs ahead,
and should be carefully implemented.

Merged-by: James Almer <jamrial@gmail.com>

lavc/8bps: Fix 32bit output of 24bit video.

Regression since / partial revert of ba3bb53b

lavf/mpegtsenc: Set min PID for data pkt to 0x0010.

Fixes ticket #1673.

avcodec/nvenc: nicely align AVOptions

avcodec/nvenc: Extended rate-control support as provided by SDK 7

Merged from libav commit by Yogender Gupta:
https://git.libav.org/?p=libav.git;a=commitdiff;h=70de2ea4261f860457a04e3d0c58c5543f403325

avcodec/nvenc: add HEVC REXT profile

avcodec/nvenc: Make sure that enum and array index match

Based on libav commits by Luca Barbato and Yogender Gupta:
https://git.libav.org/?p=libav.git;a=commit;h=352741b5ead1543d775ccf6040f33023e4491186
https://git.libav.org/?p=libav.git;a=commit;h=e02e2515b24bfc37ede6ca1744696230be55e50b

avformat/avidec: Check nb_streams in read_gab2_sub()

Fixes null pointer dereference
Fixes: 1/null_point.avi

Found-by: 连一汉 <lianyihan@360.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/avidec: Remove ancient assert

This assert can with crafted files fail, a warning is already printed
for this case.

Fixes assertion failure
Fixes:1/assert.avi

Found-by: 连一汉 <lianyihan@360.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavc/qdm2: increase code clarity

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Josh de Kock <josh@itanimul.li>

lavc/movtextdec.c: Avoid infinite loop on invalid data.

Signed-off-by: Sasi Inguva <isasi@google.com>

ffprobe: don't use AVStream.codec to set decoder framerate

Also don't set time_base. It's deprecated for decoding and avcodec_open2()
will overwrite it

Reviewed-by: Josh de Kock <josh@itanimul.li>
Signed-off-by: James Almer <jamrial@gmail.com>

avformat/matroskadec: set AVCodecParameters.field_order on progressive video

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: James Almer <jamrial@gmail.com>

avformat/matroskadec: retain error codes in matroska_resync() and matroska_read_packet()

Signed-off-by: Sophia Wang <skw@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

doc/filters: blend terminates on longest input

Signed-off-by: Mulvya <mulvya@gmail.com>
Signed-off-by: Lou Logan <lou@lrcd.com>

ffmpeg: stop using AVStream.codec on stream copy

This commit is based on commit 35c8580 from Anton Khirnov <anton@khirnov.net>
which was skipped in b8945c4.

The avcodec_copy_context() call in the encode path is left in place for now
as AVStream.codec is apparently still required even after porting ffmpeg to
the new bsf API.

Tested-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: James Almer <jamrial@gmail.com>

avformat: add av_stream_get_codec_timebase()

This will allow ffmpeg.c to stop using AVStream.codec in some cases

Signed-off-by: James Almer <jamrial@gmail.com>

swr: Update version & APIChanges for swr_build_matrix()

Found-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

swresample: Add swr_build_matrix()

API and Doxy documentation is taken from avresample_build_matrix()
Fixes: Ticket5780

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

swresample: Use double and float for matrixes for best quality and speed

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

doc: Mention -dn in the "Stream selection" paragraph.

There is no "Data options" paragraph.
Requested-by: ilker tezcan

ffprobe: don't access AVCodecContext.pkt_timebase directly

Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/ansi: Check dimensions

Fixes: 1.avi

Found-by: 连一汉 <lianyihan@360.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/mpegvideo_enc: fix memory leak

When the input frames contain side data, it will accumulate endlessly in
the coded frame, as av_frame_copy_props will append any new side data.

Fixes ticket #5799.

lavf/movenc: Allow to disable writing the timecode track.

Fixes ticket #5492.

avcodec/hevc: Add YUV420P10 to vaapi hardware decode - permits hardware decoding of HEVC Main 10 on AMD RX 480

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

ffmpeg: fix memleak of bitstream filter context on failure

Increase the nb_bitstream_filters value as soon as the context is allocated, so
if option parsing fails the last context is actually freed.

Reviewed-by: Josh de Kock <josh@itanimul.li>
Signed-off-by: James Almer <jamrial@gmail.com>

ffmpeg: fix memleak of encoder options AVDictionary on failure

Reviewed-by: Josh de Kock <josh@itanimul.li>
Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/Makefile: Fix mlpenc dependencies

Signed-off-by: James Almer <jamrial@gmail.com>

avformat/avidec: remove warning about deprecated declarations

Signed-off-by: James Almer <jamrial@gmail.com>

doc/libav-merge: change gmane link to a ffmpeg.org one

doc: move out merge script to tools

doc: remove codecpar mention in libav-merge.txt

Merge commit '1e93c1e30ff0e8bf6094a426ca60f005e9cdaed3'

* commit '1e93c1e30ff0e8bf6094a426ca60f005e9cdaed3':
  avconv: do not set encoder options when streamcopy is used

This chunk was removed in 955b818cf947473ec94a3fe8aa7f408b119fbbc9
already.

Merged-by: Clément Bœsch <u@pkh.me>

lavd/sdl2: add sdl alias

This commit also adds an sdl alias for the configure script.

lavd/sdl2: remove unused code

avformat/avidec: Fix memleak with dv in avi

Found-by: 连一汉 <lianyihan@360.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

ffmpeg: Fix bistream typos

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec: fix vc1dsp dependencies

lavf/utils: Avoid an overflow for huge negative durations.

Fixes ticket #5135.

avcodec/nvenc: use AVERROR_BUFFER_TOO_SMALL instead of ENOBUFS

Should fix compilation with mingw32

Reviewed-by: Timo Rothenpieler <timo@rothenpieler.org>
Signed-off-by: James Almer <jamrial@gmail.com>

avfilter/tests/integral: Remove unused variables

Reviewed-by: Clément Bœsch <u@pkh.me>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

ffmpeg: switch to the new BSF API

This commit is initially largely based on commit 4426540 from Anton
Khirnov <anton@khirnov.net> and two following fixes (80fb19b and
fe7b21c) which were previously skipped respectively in 98e3153, c9ee36e,
and 7fe7cdc.

mpeg4-bsf-unpack-bframes FATE reference is updated because the bsf
filter now actually fixes the extradata (mpeg4_unpack_bframes_init()
changing one byte is now honored on the output extradata).

The FATE references for remove_extra change because the packet flags
were wrong and the keyframes weren't marked, causing the bsf relying on
these proprieties to not actually work as intended.

The following was fixed by James Almer:

The filter option arguments are now also parsed correctly.

A hack to propagate extradata changed by bitstream filters after the
first av_bsf_receive_packet() call is added to maintain the current
behavior. This was previously done by av_bitstream_filter_filter() and
is needed for the aac_adtstoasc bsf.

The exit_on_error was not being checked anymore, and led to an exit
error in the last frame of h264_mp4toannexb test. Restoring this
behaviour prevents erroring out. The test is still changed as a result
due to the badly filtered frame now not being written after the failure.

Signed-off-by: Clément Bœsch <u@pkh.me>
Signed-off-by: James Almer <jamrial@gmail.com>

lavd/sdl2: Fix 32bit rgb formats on little-endian hardware.

lavd/sdl2: Move unsupported formats SDL_PIXELFORMAT_xxx888 updwards.

There is a paragraph for unsupported SDL formats.

lavf/aacdec: Do not autodetect a single frame inside the file.

lavd: drop SDL1 device and SDL1 support

Signed-off-by: Josh de Kock <josh@itanimul.li>

ffplay: add SDL2 support

This commit also drops SDL1 support for ffplay.

Tested-by: James Almer <jamrial@gmail.com> (Windows, mingw-w64)
Signed-off-by: Josh de Kock <josh@itanimul.li>

lavd/opengl: use SDL2

Signed-off-by: Lukasz Marek <lukasz.m.luki2@gmail.com>
Signed-off-by: Josh de Kock <josh@itanimul.li>

MAINTAINERS: update my entries

Acked-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Josh de Kock <josh@itanimul.li>

lavd: Add SDL2 output device

Acked-by: Michael Niedermayer
Signed-off-by: Josh de Kock <josh@itanimul.li>

lavfi: add nlmeans filter

Fixes Ticket #4910

avformat/hls: Fix handling of EXT-X-BYTERANGE streams over 2GB

Replace uses of atoi() with strtoll() when trying to read values into
int64_t variables.

Fixes Kodi trac #16926:
http://trac.kodi.tv/ticket/16926

avformat/hlsenc: refine EXT-X-BYTERANGE support for segments

refine EXT-X-BYTERANGE tag,
the spec link:
https://tools.ietf.org/html/draft-pantos-http-live-streaming-19#section-4.3.2.2
the apple doc:
https://developer.apple.com/library/ios/technotes/tn2288/_index.html#
//apple_ref/doc/uid/DTS40012238-CH1-BYTE_RANGE_SUPPORT_FOR_SEGMENTS

command line:
./ffmpeg -i ~/Movies/objectC/a.mp4 -c copy -f hls -hls_time 7
 -hls_list_size 0 -hls_segment_size 2500000 -t 40 output-test.m3u8
output:
localhost:ffmpeg liuqi$ ll *.ts ;cat output-test.m3u8
 -rw-r--r--  1 liuqi  staff  2792176  9 12 14:44 output-test0.ts
 -rw-r--r--  1 liuqi  staff  3112528  9 12 14:44 output-test3.ts
 -rw-r--r--  1 liuqi  staff  3377420  9 12 14:44 output-test6.ts
 -rw-r--r--  1 liuqi  staff  1228016  9 12 14:44 output-test7.ts
 #EXTM3U
 #EXT-X-VERSION:4
 #EXT-X-TARGETDURATION:10
 #EXT-X-MEDIA-SEQUENCE:0
 #EXTINF:9.021000,
 #EXT-X-BYTERANGE:1334988@0
 output-test0.ts
 #EXTINF:3.000000,
 #EXT-X-BYTERANGE:721356@1334988
 output-test0.ts
 #EXTINF:3.000000,
 #EXT-X-BYTERANGE:735832@2056344
 output-test0.ts
 #EXTINF:6.000000,
 #EXT-X-BYTERANGE:1645940@0
 output-test3.ts
 #EXTINF:3.000000,
 #EXT-X-BYTERANGE:715152@1645940
 output-test3.ts
 #EXTINF:3.000000,
 #EXT-X-BYTERANGE:751436@2361092
 output-test3.ts
 #EXTINF:9.000000,
 #EXT-X-BYTERANGE:3377420@0
 output-test6.ts
 #EXTINF:3.960000,
 #EXT-X-BYTERANGE:1228016@0
 output-test7.ts
 #EXT-X-ENDLIST
 localhost:ffmpeg liuqi$

 ticket-id: #5839

Signed-off-by: Steven Liu <lingjiujianke@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

ffprobe.c: Indicate decode-but-discard packets when doing -show_packets.

Signed-off-by: Sasi Inguva <isasi@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

fate: add Test for h264_mp4toannexb (ticket2991)

avcodec/cavsdsp: use av_clip_uint8() for idct

Fixes out of array read
Fixes: 1.swf

Found-by: 连一汉 <lianyihan@360.cn>
Tested-by: 连一汉 <lianyihan@360.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/remove_extradata_bsf: Fix AVoption parameter max value

tests/fate/ffmpeg: Remove dead automatic remove extradata test update the keyframe test

Found-by: jamrial
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

x86/h264_weight: use appropriate register size for weight parameters

Fixes trac 5579

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Acked-by: Michael Niedermayer <michael@niedermayer.cc>

tests/fate: Add fate-ffmpeg-bsf-remove-* tests

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

compat/cuda: convert to unix line endings

avcodec/cuvid: mark as avoid for probing

avformat/utils: avoid using marked decoders for probing

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec: add new AVOID_PROBING capability

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>

fate: update fate-source reference file

Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/h264: Use ptrdiff_t for (bi)weight functions

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

cuvid: Pass bit depth information to decoder

Although cuvid can only output 8bit, it can consume HEVC Main10 if
the bit depth is set properly. In cases where >8bit is not supported,
this change is still beneficial as the decoder will fail to be
created instead of plowing throw and decoding as 8bit.

cuvid: Use bundled headers

We need to remove the dynlink fanciness and replace it with normal
function prototypes and update the include paths and configure logic.

We don't need to explicitly check for PICPARMS now - they're going
to be there.

cuvid: Add MIT licenced nvcuid headers from Video SDK 7.0

For unknown reasons, the only accurately descriptive version of
cuviddec.h is in the Video SDK - the one in CUDA 7.5 lacks vp8
PICPARAMS and the vp9 struct definition is inaccurate. The CUDA 8 RC
includes an ancient version of this file from many many years go.

However, the one in the Video SDK is modified to work through a
dynamic link mechanism which we don't really want to use, so the
next change will modify the files to just declare functions in
the normal way.

I've split the changes so it's clear to see what changed between
the original files and ones that work for us.

avformat/movenc: write pasp atom even if sar.num == sar.den

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avformat/utils: force native h264 decoder for probing

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>

lavc/avpacket: Fix undefined behaviour, do not pass a null pointer to memcpy().

Fixes ticket #5857.

crystalhd: Use up-to-date bsf API

Although the old API is supposed to be functional, the crystalhd
decoder is currently not working for non-annex.b h.264 content.

So, let's update to the modern API and make it work again.

Signed-off-by: Philip Langdale <philipl@overt.org>

avcodec/cuvid: add support for hardware deinterlacing

Currently does not work with the ffmpeg cli tool, due do it using the
old one in one out API.
Anything using the new API, like mpv, can make use of it, provided it is
prepared for a decoder modifying the framerate and outputing multiple
frames per input. FFmpeg itself is not.

avcodec/cuvid: implement new send_packet/receive_frame api

avcodec/mlz: Check output chars before using it

Fixes hypothetical integer overflow

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/mlz: Remove 'l' postfixes from numbers

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

ffmpeg: switch to codecpar

This commit is largely based on commit 15e84ed3 from Anton Khirnov
<anton@khirnov.net> which was previously skipped in bbf5ef9d.

There are still a bunch of things raising codecpar related warnings that
need fixing, such as:
- the use of codec->debug in the interactive debug mode
- read_ffserver_streams(): it's probably broken now but there is no test
- lowres stuff
- codec copy apparently required by bitstream filters

The matroska references are updated because they now properly forward
the field_order (previously unknown, now progressive).

Thanks to James Almer for fixing a bunch of FATE issues in this commit.

Signed-off-by: Clément Bœsch <clement@stupeflix.com>
Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/on2avc: add 0x500 stereo support and improve 0x500 mono support

0x500 can be stereo.
0x500 mono can use extended window types.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/dds: add support for 4bpp format

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/drawutils: allow drawing opaque text on transparent background

Signed-off-by: Paul B Mahol <onemda@gmail.com>