Ffmpeg 4.3 add custom demuxer

1 compile ffmpeg4.3

1.1 ffmpeg 4.3 download

From abroad git Warehouse downloading ffmeg4.3 The source code is slow , It is recommended to use domestic gitee Image download .

git clone https://gitee.com/mirrors/ffmpeg.git -b release/4.3 

1.2 Compile code

./configure --enable-shared --prefix=/home/zhy/code/mypc/ffmpeg4.3/ffmpeg/install_lib
make
make install

1.3 Environment variable configuration

Build after compilation ffmpeg、ffprobe Etc , perform ffmeg See if the compilation is successful ：

./ffmpeg -version

After executing the order, it will report so A mistake we can't find ：

ffmpeg: error while loading shared libraries: libavdevice.so.58: cannot open shared object file: No such file or directory

The reason is that the environment variable is not configured , Can't find so Link path , You need to do the following ：

sudo vim  /etc/ld.so.conf

stay conf Add prefix Specified in the lib install route , Enter after adding ：

include /etc/ld.so.conf.d/*.conf
/home/zhy/code/mypc/ffmpeg4.3/ffmpeg/install_lib/lib

Finally, update the configuration ：

sudo ldconfig

Re execution ./ffmpeg -version The order will be normal .

[email protected]:~/code/mypc/ffmpeg4.3/ffmpeg$ ./ffmpeg -version
ffmpeg version n4.3.2-168-g3aba8b176f Copyright (c) 2000-2021 the FFmpeg developers
built with gcc 7 (Ubuntu 7.5.0-3ubuntu1~18.04)
configuration: --enable-shared --prefix=/home/zhy/code/mypc/ffmpeg4.3/ffmpeg/install_lib
libavutil      56. 51.100 / 56. 51.100
libavcodec     58. 91.100 / 58. 91.100
libavformat    58. 45.100 / 58. 45.100
libavdevice    58. 10.100 / 58. 10.100
libavfilter     7. 85.100 /  7. 85.100
libswscale      5.  7.100 /  5.  7.100
libswresample   3.  7.100 /  3.  7.100
[email protected]:~/code/mypc/ffmpeg4.3/ffmpeg$ 

2 Add custom demuxer

ffmpeg In a dumuxer Need to include read_header、read_packet And so on , Such as mpegts demuxer As shown below ：

AVInputFormat ff_mpegtsraw_demuxer = {
    
    .name           = "mpegtsraw",
    .long_name      = NULL_IF_CONFIG_SMALL("raw MPEG-TS (MPEG-2 Transport Stream)"),
    .priv_data_size = sizeof(MpegTSContext),
    .read_header    = mpegts_read_header,
    .read_packet    = mpegts_raw_read_packet,
    .read_close     = mpegts_read_close,
    .read_timestamp = mpegts_get_dts,
    .flags          = AVFMT_SHOW_IDS | AVFMT_TS_DISCONT,
    .priv_class     = &mpegtsraw_class,
};

Generally speaking audio and video Of demuxer More complicated , So choose to add one image Of demuxer,image The format of... Is selected webp,webp Format spec See reference article .

2.1 add to demuxer Definition

stay libavformat\rawdec.h Add ff_webp_demuxer Definition ：

#define FF_DEF_RAWIMAGE_DEMUXER(shortname, longname, probe, ext, id, flag)\ FF_RAW_DEMUXER_CLASS(shortname)\ AVInputFormat ff_ ## shortname ## _demuxer = {\ .name = #shortname,\ .long_name = NULL_IF_CONFIG_SMALL(longname),\ .read_probe = probe,\ .read_header = ff_raw_image_read_header,\ .read_packet = ff_raw_image_read_packet,\ .extensions = ext,\ .flags = flag,\ .raw_codec_id = id,\ .priv_data_size = sizeof(FFRawVideoDemuxerContext),\ .priv_class = &shortname ## _demuxer_class,\ };

2.2 add to demuxer Definition

stay libavformat\rawdec.c Add implementation related functions in ff_raw_image_read_header、ff_raw_image_read_packet Function and webp Of probe function .
probe Function is mainly used to speculate container Of fmt,webp The format of pictures is mainly through "RIFF" and ”WEBP“ Two conjectures ,probe The function is shown below ：

static int webp_probe(const AVProbeData *p)
{
    
    uint8_t *b = p->buf;
	
    printf("[debug] webp_probe\n");

    if (AV_RB32(b)  == MKBETAG('R', 'I', 'F', 'F')
        && AV_RB32(b + 8)  == MKBETAG('W', 'E', 'B', 'P'))
    {
    
         printf("[debug] is webp image \n");
	 return AVPROBE_SCORE_MAX - 1;
    }
    return 0;
}

ff_raw_image_read_packet Functions do not need special treatment , Every read must size Of buffer That's all right. .

int ff_raw_image_read_packet(AVFormatContext *s, AVPacket *pkt)
{
    
    int ret, size = 1024;

    ret = av_get_packet(s->pb, pkt, size);

     if (ret < 0) {
    
          return ret;
     }

    pkt->size = ret;
    pkt->stream_index = 0;
    pkt->flags = AV_PKT_FLAG_KEY;

    return ret;
}

read_header Function usually reads pictures width、height As well as some flag Information ,ff_raw_image_read_header The function is implemented as follows ：

int ff_raw_image_read_header(AVFormatContext *s)
{
    
    AVStream *st;
    FFRawVideoDemuxerContext *s1 = s->priv_data;
    int ret = 0;

   printf("[debug] %s: %d\n",  __func__, __LINE__);

    st = avformat_new_stream(s, NULL);
    if (!st) {
    
        ret = AVERROR(ENOMEM);
        goto fail;
    }

    st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
    st->codecpar->codec_id = s->iformat->raw_codec_id;
    st->need_parsing = AVSTREAM_PARSE_FULL_RAW;

    if (AV_CODEC_ID_WEBP ==  st->codecpar->codec_id)
    {
    
        webp_read_info(s, st);
    }

fail:
    return ret;
}

among webp_read_info Mainly based on webp container Extracted some information ,webp container The most commonly used structures are shown below , See the reference article for details webp Official documents ：

 0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                   WebP file header (12 bytes)                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                      ChunkHeader('VP8X')                      |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|Rsv|I|L|E|X|A|R|                   Reserved                    |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|          Canvas Width Minus One               |             ...
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
...  Canvas Height Minus One    |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

among , WebP file header as follows ：

 0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|      'R'      |      'I'      |      'F'      |      'F'      |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                           File Size                           |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|      'W'      |      'E'      |      'B'      |      'P'      |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

According to the above structure ,webp_read_info The content in is as follows ：

static int webp_read_info (AVFormatContext *s ,AVStream *st)
{
    

#define VP8X_FLAG_ANIMATION 0x02
#define VP8X_FLAG_XMP_METADATA 0x04
#define VP8X_FLAG_EXIF_METADATA 0x08
#define VP8X_FLAG_ALPHA 0x10
#define VP8X_FLAG_ICC 0x20

    int ret = 0;
    int len = 0;
    int64_t pos = 0;
    int64_t cur_pos = 0;
    AVIOContext *pb = s->pb;
    int flag = 0;

    printf("[debug] %s: %d\n",  __func__, __LINE__);

    if (NULL == pb)
    {
    
       printf("[error] %s: %d invalid input \n",  __func__, __LINE__);
        goto END;
    }

    pos = avio_tell(pb);
    avio_seek(pb, 0, SEEK_SET);

     /*read RIFF tag*/
    ret = avio_rb32(pb);
    if (ret == MKBETAG('R', 'I', 'F', 'F'))
    {
    
         printf("[debug] read RIFF tag \n");
    }
    else
    {
    
        goto END;
    }
    
    /*read size*/
    ret = avio_rb32(pb);
    
    /*read WEBP tag*/
    ret = avio_rb32(pb);
    if (ret == MKBETAG('W', 'E', 'B', 'P'))
    {
    
         printf("[debug] read WEBP tag \n");
    }

    /*read VP8X tag*/
    ret = avio_rb32(pb);
    if (ret == MKBETAG('V', 'P', '8', 'X'))
    {
    
         printf("[debug] read VP8X tag \n");
    }
    
    /*read size*/
    ret = avio_rb32(pb);
    /*read flag*/
    flag = avio_r8(pb);
    
    printf(" ICCP: %d\n Alpha: %d\n EXIF: %d\n XMP: %d\n Animation: %d\n",
           (flag & VP8X_FLAG_ICC) != 0,
           (flag & VP8X_FLAG_ALPHA) != 0,
           (flag & VP8X_FLAG_EXIF_METADATA) != 0,
           (flag & VP8X_FLAG_XMP_METADATA) != 0,
           (flag & VP8X_FLAG_ANIMATION) != 0);

    avio_seek(pb, pos, SEEK_SET);
    
END:

    return ret;
}

Mainly read VP8X Chunk Some of flag, More detailed webp demux You can refer to libwebp Realization .

2.3 Replace ffmpeg Medium webp demuxer

ffmpeg The Central Plains originally had webp Of demuxer , So you need to libavformat\allformats.c Before the note ff_image_webp_pipe_demuxer And add custom ff_webp_demuxer

/* image demuxers */
//extern AVInputFormat ff_image_webp_pipe_demuxer;
extern AVInputFormat  ff_image_xpm_pipe_demuxer;
extern AVInputFormat  ff_image_xwd_pipe_demuxer;
extern AVInputFormat  ff_webp_demuxer;/*new add*/

Comment out ff_image_webp_pipe_demuxer Compilation only requires ff_image_webp_pipe_demuxer Just kill the place you call , Finally, it needs to be in config.h Close your own webp demuxer, Set to 0.

#define CONFIG_IMAGE_WEBP_PIPE_DEMUXER 0

2.4 Running results

Recompile and install, test ffprobe:

./ffprobe samples/webp/animation/animated-webp-supported.webp

Running results ：

[email protected]:~/code/mypc/ffmpeg4.3/ffmpeg$ ./ffprobe samples/webp/animation/animated-webp-supported.webp
ffprobe version n4.3.2-168-g3aba8b176f Copyright (c) 2007-2021 the FFmpeg developers
  built with gcc 7 (Ubuntu 7.5.0-3ubuntu1~18.04)
  configuration: --enable-shared --prefix=/home/zhy/code/mypc/ffmpeg4.3/ffmpeg/install_lib
  libavutil      56. 51.100 / 56. 51.100
  libavcodec     58. 91.100 / 58. 91.100
  libavformat    58. 45.100 / 58. 45.100
  libavdevice    58. 10.100 / 58. 10.100
  libavfilter     7. 85.100 /  7. 85.100
  libswscale      5.  7.100 /  5.  7.100
  libswresample   3.  7.100 /  3.  7.100
[debug]  webp_probe
[debug]  is webp image 
[debug]  ff_raw_image_read_header: 338
[debug]  webp_read_info: 251
[debug]  read RIFF tag 
[debug]  read WEBP tag 
[debug]  read VP8X tag 
  ICCP: 0
  Alpha: 1
  EXIF: 0
  XMP: 0
  Animation: 1
[webp @ 0x5618e1c420c0] skipping unsupported chunk: ANIM
[webp @ 0x5618e1c420c0] skipping unsupported chunk: ANMF
    Last message repeated 11 times
[webp @ 0x5618e1c420c0] image data not found
[webp @ 0x5618e1c40d80] decoding for stream 0 failed
[webp @ 0x5618e1c40d80] Could not find codec parameters for stream 0 (Video: webp, none): unspecified size
Consider increasing the value for the 'analyzeduration' and 'probesize' options
Input #0, webp, from 'samples/webp/animation/animated-webp-supported.webp':
  Duration: N/A, bitrate: N/A
    Stream #0:0: Video: webp, none, 90k tbr, 90k tbn, 90k tbc
[email protected]:~/code/mypc/ffmpeg4.3/ffmpeg$ 

libwebp in webpinfo Read the results ：

[email protected]:~/code/mypc/mutil_media/libwebp/examples$ ./webpinfo samples/webp/animation/animated-webp-supported.webp 
File: samples/webp/animation/animated-webp-supported.webp
RIFF HEADER:
  File size:  37342
Chunk VP8X at offset     12, length     18
[leif] webp_info->feature_flags_  = 18
  ICCP: 0
  Alpha: 1
  EXIF: 0
  XMP: 0
  Animation: 1
  Canvas size 400 x 400

It can be seen that , Read the flag And use libwebp Consistent reading , Prove the added customization webp Of demuxer function OK.

Reference article

1 https://developers.google.com/speed/webp/docs/riff_container