matroskaenc.c

ffmpeg的完整源代码和作者自己写的文档。不但有在Linux的工程哦

#define FLAC_STREAMINFO_SIZE 34

static int put_flac_codecpriv(AVFormatContext *s, ByteIOContext *pb, AVCodecContext *codec)
{
    // if the extradata_size is greater than FLAC_STREAMINFO_SIZE,
    // assume that it's in Matroska's format already
    if (codec->extradata_size < FLAC_STREAMINFO_SIZE) {
        av_log(s, AV_LOG_ERROR, "Invalid FLAC extradata\n");
        return -1;
    } else if (codec->extradata_size == FLAC_STREAMINFO_SIZE) {
        // only the streaminfo packet
        put_byte(pb, 0);
        put_xiph_size(pb, codec->extradata_size);
        av_log(s, AV_LOG_ERROR, "Only one packet\n");
    }
    put_buffer(pb, codec->extradata, codec->extradata_size);
    return 0;
}

static void get_aac_sample_rates(AVFormatContext *s, AVCodecContext *codec, int *sample_rate, int *output_sample_rate)
{
    static const int aac_sample_rates[] = {
        96000, 88200, 64000, 48000, 44100, 32000,
        24000, 22050, 16000, 12000, 11025,  8000,
    };
    int sri;

    if (codec->extradata_size < 2) {
        av_log(s, AV_LOG_WARNING, "No AAC extradata, unable to determine samplerate.\n");
        return;
    }

    sri = ((codec->extradata[0] << 1) & 0xE) | (codec->extradata[1] >> 7);
    if (sri > 12) {
        av_log(s, AV_LOG_WARNING, "AAC samplerate index out of bounds\n");
        return;
    }
    *sample_rate = aac_sample_rates[sri];

    // if sbr, get output sample rate as well
    if (codec->extradata_size == 5) {
        sri = (codec->extradata[4] >> 3) & 0xF;
        if (sri > 12) {
            av_log(s, AV_LOG_WARNING, "AAC output samplerate index out of bounds\n");
            return;
        }
        *output_sample_rate = aac_sample_rates[sri];
    }
}

static int mkv_write_codecprivate(AVFormatContext *s, ByteIOContext *pb, AVCodecContext *codec, int native_id)
{
    ByteIOContext dyn_cp;
    uint8_t *codecpriv;
    int ret = 0, codecpriv_size;

    url_open_dyn_buf(&dyn_cp);

    if (native_id) {
        if (codec->codec_id == CODEC_ID_VORBIS || codec->codec_id == CODEC_ID_THEORA)
            ret = put_xiph_codecpriv(s, &dyn_cp, codec);
        else if (codec->codec_id == CODEC_ID_FLAC)
            ret = put_flac_codecpriv(s, &dyn_cp, codec);
        else if (codec->extradata_size)
            put_buffer(&dyn_cp, codec->extradata, codec->extradata_size);
    } else if (codec->codec_type == CODEC_TYPE_VIDEO) {
        if (!codec->codec_tag)
            codec->codec_tag = codec_get_tag(codec_bmp_tags, codec->codec_id);
        if (!codec->codec_tag) {
            av_log(s, AV_LOG_ERROR, "No bmp codec ID found.");
            ret = -1;
        }

        put_bmp_header(&dyn_cp, codec, codec_bmp_tags, 0);

    } else if (codec->codec_type == CODEC_TYPE_AUDIO) {
        if (!codec->codec_tag)
            codec->codec_tag = codec_get_tag(codec_wav_tags, codec->codec_id);
        if (!codec->codec_tag) {
            av_log(s, AV_LOG_ERROR, "No wav codec ID found.");
            ret = -1;
        }

        put_wav_header(&dyn_cp, codec);
    }

    codecpriv_size = url_close_dyn_buf(&dyn_cp, &codecpriv);
    if (codecpriv_size)
        put_ebml_binary(pb, MATROSKA_ID_CODECPRIVATE, codecpriv, codecpriv_size);
    av_free(codecpriv);
    return ret;
}

static int mkv_write_tracks(AVFormatContext *s)
{
    MatroskaMuxContext *mkv = s->priv_data;
    ByteIOContext *pb = &s->pb;
    ebml_master tracks;
    int i, j, ret;

    ret = mkv_add_seekhead_entry(mkv->main_seekhead, MATROSKA_ID_TRACKS, url_ftell(pb));
    if (ret < 0) return ret;

    tracks = start_ebml_master(pb, MATROSKA_ID_TRACKS, 0);
    for (i = 0; i < s->nb_streams; i++) {
        AVStream *st = s->streams[i];
        AVCodecContext *codec = st->codec;
        ebml_master subinfo, track;
        int native_id = 0;
        int bit_depth = av_get_bits_per_sample(codec->codec_id);
        int sample_rate = codec->sample_rate;
        int output_sample_rate = 0;

        if (!bit_depth)
            bit_depth = av_get_bits_per_sample_format(codec->sample_fmt);

        if (codec->codec_id == CODEC_ID_AAC)
            get_aac_sample_rates(s, codec, &sample_rate, &output_sample_rate);

        track = start_ebml_master(pb, MATROSKA_ID_TRACKENTRY, 0);
        put_ebml_uint (pb, MATROSKA_ID_TRACKNUMBER     , i + 1);
        put_ebml_uint (pb, MATROSKA_ID_TRACKUID        , i + 1);
        put_ebml_uint (pb, MATROSKA_ID_TRACKFLAGLACING , 0);    // no lacing (yet)

        if (st->language[0])
            put_ebml_string(pb, MATROSKA_ID_TRACKLANGUAGE, st->language);
        else
            put_ebml_string(pb, MATROSKA_ID_TRACKLANGUAGE, "und");

        // look for a codec ID string specific to mkv to use,
        // if none are found, use AVI codes
        for (j = 0; ff_mkv_codec_tags[j].id != CODEC_ID_NONE; j++) {
            if (ff_mkv_codec_tags[j].id == codec->codec_id) {
                put_ebml_string(pb, MATROSKA_ID_CODECID, ff_mkv_codec_tags[j].str);
                native_id = 1;
                break;
            }
        }

        switch (codec->codec_type) {
            case CODEC_TYPE_VIDEO:
                put_ebml_uint(pb, MATROSKA_ID_TRACKTYPE, MATROSKA_TRACK_TYPE_VIDEO);

                if (!native_id)
                    // if there is no mkv-specific codec ID, use VFW mode
                    put_ebml_string(pb, MATROSKA_ID_CODECID, MATROSKA_CODEC_ID_VIDEO_VFW_FOURCC);

                subinfo = start_ebml_master(pb, MATROSKA_ID_TRACKVIDEO, 0);
                // XXX: interlace flag?
                put_ebml_uint (pb, MATROSKA_ID_VIDEOPIXELWIDTH , codec->width);
                put_ebml_uint (pb, MATROSKA_ID_VIDEOPIXELHEIGHT, codec->height);
                if (codec->sample_aspect_ratio.num) {
                    AVRational dar = av_mul_q(codec->sample_aspect_ratio,
                                    (AVRational){codec->width, codec->height});
                    put_ebml_uint(pb, MATROSKA_ID_VIDEODISPLAYWIDTH , dar.num);
                    put_ebml_uint(pb, MATROSKA_ID_VIDEODISPLAYHEIGHT, dar.den);
                }
                end_ebml_master(pb, subinfo);
                break;

            case CODEC_TYPE_AUDIO:
                put_ebml_uint(pb, MATROSKA_ID_TRACKTYPE, MATROSKA_TRACK_TYPE_AUDIO);

                if (!native_id)
                    // no mkv-specific ID, use ACM mode
                    put_ebml_string(pb, MATROSKA_ID_CODECID, MATROSKA_CODEC_ID_AUDIO_ACM);

                subinfo = start_ebml_master(pb, MATROSKA_ID_TRACKAUDIO, 0);
                put_ebml_uint  (pb, MATROSKA_ID_AUDIOCHANNELS    , codec->channels);
                put_ebml_float (pb, MATROSKA_ID_AUDIOSAMPLINGFREQ, sample_rate);
                if (output_sample_rate)
                    put_ebml_float(pb, MATROSKA_ID_AUDIOOUTSAMPLINGFREQ, output_sample_rate);
                if (bit_depth)
                    put_ebml_uint(pb, MATROSKA_ID_AUDIOBITDEPTH, bit_depth);
                end_ebml_master(pb, subinfo);
                break;

            case CODEC_TYPE_SUBTITLE:
                put_ebml_uint(pb, MATROSKA_ID_TRACKTYPE, MATROSKA_TRACK_TYPE_SUBTITLE);
                break;
            default:
                av_log(s, AV_LOG_ERROR, "Only audio, video, and subtitles are supported for Matroska.");
                break;
        }
        ret = mkv_write_codecprivate(s, pb, codec, native_id);
        if (ret < 0) return ret;

        end_ebml_master(pb, track);

        // ms precision is the de-facto standard timescale for mkv files
        av_set_pts_info(st, 64, 1, 1000);
    }
    end_ebml_master(pb, tracks);
    return 0;
}

static int mkv_write_header(AVFormatContext *s)
{
    MatroskaMuxContext *mkv = s->priv_data;
    ByteIOContext *pb = &s->pb;
    ebml_master ebml_header, segment_info;
    int ret;

    mkv->md5_ctx = av_mallocz(av_md5_size);
    av_md5_init(mkv->md5_ctx);

    ebml_header = start_ebml_master(pb, EBML_ID_HEADER, 0);
    put_ebml_uint   (pb, EBML_ID_EBMLVERSION        ,           1);
    put_ebml_uint   (pb, EBML_ID_EBMLREADVERSION    ,           1);
    put_ebml_uint   (pb, EBML_ID_EBMLMAXIDLENGTH    ,           4);
    put_ebml_uint   (pb, EBML_ID_EBMLMAXSIZELENGTH  ,           8);
    put_ebml_string (pb, EBML_ID_DOCTYPE            ,  "matroska");
    put_ebml_uint   (pb, EBML_ID_DOCTYPEVERSION     ,           2);
    put_ebml_uint   (pb, EBML_ID_DOCTYPEREADVERSION ,           2);
    end_ebml_master(pb, ebml_header);

    mkv->segment = start_ebml_master(pb, MATROSKA_ID_SEGMENT, 0);
    mkv->segment_offset = url_ftell(pb);

    // we write 2 seek heads - one at the end of the file to point to each
    // cluster, and one at the beginning to point to all other level one
    // elements (including the seek head at the end of the file), which
    // isn't more than 10 elements if we only write one of each other
    // currently defined level 1 element
    mkv->main_seekhead    = mkv_start_seekhead(pb, mkv->segment_offset, 10);
    mkv->cluster_seekhead = mkv_start_seekhead(pb, mkv->segment_offset, 0);
    if (mkv->main_seekhead == NULL || mkv->cluster_seekhead == NULL)
        return AVERROR(ENOMEM);

    ret = mkv_add_seekhead_entry(mkv->main_seekhead, MATROSKA_ID_INFO, url_ftell(pb));
    if (ret < 0) return ret;

    segment_info = start_ebml_master(pb, MATROSKA_ID_INFO, 0);
    put_ebml_uint(pb, MATROSKA_ID_TIMECODESCALE, 1000000);
    if (strlen(s->title))
        put_ebml_string(pb, MATROSKA_ID_TITLE, s->title);
    if (!(s->streams[0]->codec->flags & CODEC_FLAG_BITEXACT)) {
        put_ebml_string(pb, MATROSKA_ID_MUXINGAPP , LIBAVFORMAT_IDENT);
        put_ebml_string(pb, MATROSKA_ID_WRITINGAPP, LIBAVFORMAT_IDENT);

        // reserve space to write the segment UID later
        mkv->segment_uid = url_ftell(pb);
        put_ebml_void(pb, 19);
    }

    // reserve space for the duration
    mkv->duration = 0;
    mkv->duration_offset = url_ftell(pb);
    put_ebml_void(pb, 11);                  // assumes double-precision float to be written
    end_ebml_master(pb, segment_info);

    ret = mkv_write_tracks(s);
    if (ret < 0) return ret;

    ret = mkv_add_seekhead_entry(mkv->cluster_seekhead, MATROSKA_ID_CLUSTER, url_ftell(pb));
    if (ret < 0) return ret;

    mkv->cluster_pos = url_ftell(pb);
    mkv->cluster = start_ebml_master(pb, MATROSKA_ID_CLUSTER, 0);
    put_ebml_uint(pb, MATROSKA_ID_CLUSTERTIMECODE, 0);
    mkv->cluster_pts = 0;

    mkv->cues = mkv_start_cues(mkv->segment_offset);
    if (mkv->cues == NULL)
        return AVERROR(ENOMEM);

    return 0;
}

static int mkv_block_size(AVPacket *pkt)
{
    int size = 4;           // track num + timecode + flags
    return size + pkt->size;
}

static int mkv_blockgroup_size(AVPacket *pkt)
{
    int size = mkv_block_size(pkt);
    size += ebml_num_size(size);
    size += 2;              // EBML ID for block and block duration
    size += 8;              // max size of block duration
    size += ebml_num_size(size);
    size += 1;              // blockgroup EBML ID
    return size;
}

static void mkv_write_block(AVFormatContext *s, unsigned int blockid, AVPacket *pkt, int flags)
{
    MatroskaMuxContext *mkv = s->priv_data;
    ByteIOContext *pb = &s->pb;

    av_log(s, AV_LOG_DEBUG, "Writing block at offset %" PRIu64 ", size %d, "
           "pts %" PRId64 ", dts %" PRId64 ", duration %d, flags %d\n",
           url_ftell(pb), pkt->size, pkt->pts, pkt->dts, pkt->duration, flags);
    put_ebml_id(pb, blockid);
    put_ebml_num(pb, mkv_block_size(pkt), 0);
    put_byte(pb, 0x80 | (pkt->stream_index + 1));     // this assumes stream_index is less than 126
    put_be16(pb, pkt->pts - mkv->cluster_pts);
    put_byte(pb, flags);
    put_buffer(pb, pkt->data, pkt->size);
}

static int mkv_write_packet(AVFormatContext *s, AVPacket *pkt)
{
    MatroskaMuxContext *mkv = s->priv_data;
    ByteIOContext *pb = &s->pb;
    AVCodecContext *codec = s->streams[pkt->stream_index]->codec;
    int keyframe = !!(pkt->flags & PKT_FLAG_KEY);
    int ret;

    // start a new cluster every 5 MB or 5 sec
    if (url_ftell(pb) > mkv->cluster_pos + 5*1024*1024 || pkt->pts > mkv->cluster_pts + 5000) {
        av_log(s, AV_LOG_DEBUG, "Starting new cluster at offset %" PRIu64
               " bytes, pts %" PRIu64 "\n", url_ftell(pb), pkt->pts);
        end_ebml_master(pb, mkv->cluster);

        ret = mkv_add_seekhead_entry(mkv->cluster_seekhead, MATROSKA_ID_CLUSTER, url_ftell(pb));
        if (ret < 0) return ret;

        mkv->cluster_pos = url_ftell(pb);
        mkv->cluster = start_ebml_master(pb, MATROSKA_ID_CLUSTER, 0);
        put_ebml_uint(pb, MATROSKA_ID_CLUSTERTIMECODE, pkt->pts);
        mkv->cluster_pts = pkt->pts;
        av_md5_update(mkv->md5_ctx, pkt->data, FFMIN(200, pkt->size));
    }

    if (codec->codec_type != CODEC_TYPE_SUBTITLE) {
        mkv_write_block(s, MATROSKA_ID_SIMPLEBLOCK, pkt, keyframe << 7);
    } else {
        ebml_master blockgroup = start_ebml_master(pb, MATROSKA_ID_BLOCKGROUP, mkv_blockgroup_size(pkt));
        mkv_write_block(s, MATROSKA_ID_BLOCK, pkt, 0);
        put_ebml_uint(pb, MATROSKA_ID_DURATION, pkt->duration);
        end_ebml_master(pb, blockgroup);
    }

    if (codec->codec_type == CODEC_TYPE_VIDEO && keyframe) {
        ret = mkv_add_cuepoint(mkv->cues, pkt, mkv->cluster_pos);
        if (ret < 0) return ret;
    }

    mkv->duration = FFMAX(mkv->duration, pkt->pts + pkt->duration);
    return 0;
}

static int mkv_write_trailer(AVFormatContext *s)
{
    MatroskaMuxContext *mkv = s->priv_data;
    ByteIOContext *pb = &s->pb;
    offset_t currentpos, second_seekhead, cuespos;
    int ret;

    end_ebml_master(pb, mkv->cluster);

    if (!url_is_streamed(pb)) {
        cuespos = mkv_write_cues(pb, mkv->cues, s->nb_streams);
        second_seekhead = mkv_write_seekhead(pb, mkv->cluster_seekhead);

        ret = mkv_add_seekhead_entry(mkv->main_seekhead, MATROSKA_ID_CUES    , cuespos);
        if (ret < 0) return ret;
        ret = mkv_add_seekhead_entry(mkv->main_seekhead, MATROSKA_ID_SEEKHEAD, second_seekhead);
        if (ret < 0) return ret;
        mkv_write_seekhead(pb, mkv->main_seekhead);

        // update the duration
        av_log(s, AV_LOG_DEBUG, "end duration = %" PRIu64 "\n", mkv->duration);
        currentpos = url_ftell(pb);
        url_fseek(pb, mkv->duration_offset, SEEK_SET);
        put_ebml_float(pb, MATROSKA_ID_DURATION, mkv->duration);

        // write the md5sum of some frames as the segment UID
        if (!(s->streams[0]->codec->flags & CODEC_FLAG_BITEXACT)) {
            uint8_t segment_uid[16];
            av_md5_final(mkv->md5_ctx, segment_uid);
            url_fseek(pb, mkv->segment_uid, SEEK_SET);
            put_ebml_binary(pb, MATROSKA_ID_SEGMENTUID, segment_uid, 16);
        }
        url_fseek(pb, currentpos, SEEK_SET);
    }

    end_ebml_master(pb, mkv->segment);
    av_free(mkv->md5_ctx);
    return 0;
}

AVOutputFormat matroska_muxer = {
    "matroska",
    "Matroska File Format",
    "video/x-matroska",
    "mkv",
    sizeof(MatroskaMuxContext),
    CODEC_ID_MP2,
    CODEC_ID_MPEG4,
    mkv_write_header,
    mkv_write_packet,
    mkv_write_trailer,
    .codec_tag = (const AVCodecTag*[]){codec_bmp_tags, codec_wav_tags, 0},
    .subtitle_codec = CODEC_ID_TEXT,
};

AVOutputFormat matroska_audio_muxer = {
    "matroska",
    "Matroska File Format",
    "audio/x-matroska",
    "mka",
    sizeof(MatroskaMuxContext),
    CODEC_ID_MP2,
    CODEC_ID_NONE,
    mkv_write_header,
    mkv_write_packet,
    mkv_write_trailer,
    .codec_tag = (const AVCodecTag*[]){codec_wav_tags, 0},
};