Ffmpeg源代码解析之av_read_frame

首先网上有很多人介绍ffmpeg的源码了，其中不乏一些大神的精彩之作，在此向他们致敬。没有开源分享精神，我们程序员的学习路将会很艰难。将我研究ffmepg源码的一些经验贴出来，其中也借鉴了网上的一些内容。

av_read_frame从字面意思上来看，就是从内存中读取一帧数据，但是具体的实现如何?本文是在参考别人的代码的基础上，结合最新的源代码，进行分析研究！如果有不正确的地方，欢迎指正。

AVPacket结构体：

AVPacket结构体的作用是从内存中获取一个视频压缩帧，对于音频可能获取一个或者多个压缩帧。

av_read_frame函数：

现在的函数结构和以前的有不同，让我们来一步一步的分析，多余的代码都删了。

int av_read_frame(AVFormatContext *s, AVPacket *pkt)
{
    const int genpts = s->flags & AVFMT_FLAG_GENPTS;
    int eof = 0;
    int ret;
    AVStream *st;
    //判断缓存中是否存在Packet， s->packet_buffer，若没有存在则调用read_frame_internal()
    if (!genpts) {
        ret = s->packet_buffer
              ? read_from_packet_buffer(&s->packet_buffer,
                                        &s->packet_buffer_end, pkt)
              : read_frame_internal(s, pkt);
        if (ret < 0)
            return ret;
        goto return_packet;
    }
....
}

static int read_frame_internal(AVFormatContext *s, AVPacket *pkt)
{
    int ret = 0, i, got_packet = 0;

    av_init_packet(pkt);

    while (!got_packet && !s->parse_queue) {
        AVStream *st;
        AVPacket cur_pkt;

        /* 读取下一个包 */
        ret = ff_read_packet(s, &cur_pkt);
        if (ret < 0) {
            if (ret == AVERROR(EAGAIN))
                return ret;
            /* flush the parsers */
            for (i = 0; i < s->nb_streams; i++) {
                st = s->streams[i];
                if (st->parser && st->need_parsing)
                    parse_packet(s, NULL, st->index);
            }
            /* all remaining packets are now in parse_queue =>
             * really terminate parsing */
            break;
        }
....
}

ff_read_packet就是以前的av_read_paket。

分为3步读取Packet

1.先判断缓存s->raw_packet_buffer中是否有数据

2.如果没有数据，调用s->iformat->read_packet(s,pkt)

3.最后调用add_to_pktbuf将packet加入s->raw_packet_buffer中

int ff_read_packet(AVFormatContext *s, AVPacket *pkt)
{
    int ret, i, err;
    AVStream *st;

    for (;;) {
        AVPacketList *pktl = s->raw_packet_buffer;//首先判断s->raw_packet_buffer中是否有数据，一般开始的时候是没有的

        if (pktl) {
            *pkt = pktl->pkt;
            st   = s->streams[pkt->stream_index];
            if (s->raw_packet_buffer_remaining_size <= 0)
                if ((err = probe_codec(s, st, NULL)) < 0)
                    return err;
            if (st->request_probe <= 0) {
                s->raw_packet_buffer                 = pktl->next;
                s->raw_packet_buffer_remaining_size += pkt->size;
                av_free(pktl);
                return 0;
            }
        }

        pkt->data = NULL;
        pkt->size = 0;
        av_init_packet(pkt);
        ret = s->iformat->read_packet(s, pkt);
        if (ret < 0) {
            continue;
        }

        if ((s->flags & AVFMT_FLAG_DISCARD_CORRUPT) &&
            (pkt->flags & AV_PKT_FLAG_CORRUPT)) {
            av_log(s, AV_LOG_WARNING,
                   "Dropped corrupted packet (stream = %d)\n",
                   pkt->stream_index);
            av_free_packet(pkt);
            continue;
        }

        if (pkt->stream_index >= (unsigned)s->nb_streams) {
            av_log(s, AV_LOG_ERROR, "Invalid stream index %d\n", pkt->stream_index);
            continue;
        }

        st = s->streams[pkt->stream_index];

        if (update_wrap_reference(s, st, pkt->stream_index, pkt) && st->pts_wrap_behavior == AV_PTS_WRAP_SUB_OFFSET) {
            // correct first time stamps to negative values
            if (!is_relative(st->first_dts))
                st->first_dts = wrap_timestamp(st, st->first_dts);
            if (!is_relative(st->start_time))
                st->start_time = wrap_timestamp(st, st->start_time);
            if (!is_relative(st->cur_dts))
                st->cur_dts = wrap_timestamp(st, st->cur_dts);
        }

        pkt->dts = wrap_timestamp(st, pkt->dts);
        pkt->pts = wrap_timestamp(st, pkt->pts);

        force_codec_ids(s, st);

        /* TODO: audio: time filter; video: frame reordering (pts != dts) */
        if (s->use_wallclock_as_timestamps)
            pkt->dts = pkt->pts = av_rescale_q(av_gettime(), AV_TIME_BASE_Q, st->time_base);

        if (!pktl && st->request_probe <= 0)
            return ret;

        add_to_pktbuf(&s->raw_packet_buffer, pkt, &s->raw_packet_buffer_end);//将packet加入到s->raw_packet_buffer
        s->raw_packet_buffer_remaining_size -= pkt->size;

        if ((err = probe_codec(s, st, pkt)) < 0)
            return err;
    }
}

对于ts流来讲：

s->iformat->read_packet调用的是mpegts_read_packet 具体的看我的博文ffmpeg对于ts流的解析

static int mpegts_read_packet(AVFormatContext *s, AVPacket *pkt)
{
    MpegTSContext *ts = s->priv_data;
    int ret, i;

    pkt->size = -1;
    ts->pkt = pkt;
    ret = handle_packets(ts, 0);
    if (ret < 0) {
        av_free_packet(ts->pkt);
        /* flush pes data left */
        for (i = 0; i < NB_PID_MAX; i++)
            if (ts->pids[i] && ts->pids[i]->type == MPEGTS_PES) {
                PESContext *pes = ts->pids[i]->u.pes_filter.opaque;
                if (pes->state == MPEGTS_PAYLOAD && pes->data_index > 0) {
                    new_pes_packet(pes, pkt);
                    pes->state = MPEGTS_SKIP;
                    ret = 0;
                    break;
                }
            }
    }

    if (!ret && pkt->size < 0)
        ret = AVERROR(EINTR);
    return ret;
}

包的解析：

int av_parser_parse2(AVCodecParserContext *s, AVCodecContext *avctx,
                     uint8_t **poutbuf, int *poutbuf_size,
                     const uint8_t *buf, int buf_size,
                     int64_t pts, int64_t dts, int64_t pos)
{
    int index, i;
    uint8_t dummy_buf[FF_INPUT_BUFFER_PADDING_SIZE];

    if (!(s->flags & PARSER_FLAG_FETCHED_OFFSET)) {
        s->next_frame_offset =
        s->cur_offset        = pos;
        s->flags            |= PARSER_FLAG_FETCHED_OFFSET;
    }

    if (buf_size == 0) {
        /* padding is always necessary even if EOF, so we add it here */
        memset(dummy_buf, 0, sizeof(dummy_buf));
        buf = dummy_buf;
    } else if (s->cur_offset + buf_size != s->cur_frame_end[s->cur_frame_start_index]) { /* skip remainder packets */
        /* add a new packet descriptor */
        i = (s->cur_frame_start_index + 1) & (AV_PARSER_PTS_NB - 1);
        s->cur_frame_start_index = i;
        s->cur_frame_offset[i]   = s->cur_offset;
        s->cur_frame_end[i]      = s->cur_offset + buf_size;
        s->cur_frame_pts[i]      = pts;
        s->cur_frame_dts[i]      = dts;
        s->cur_frame_pos[i]      = pos;
    }

    if (s->fetch_timestamp) {
        s->fetch_timestamp = 0;
        s->last_pts        = s->pts;
        s->last_dts        = s->dts;
        s->last_pos        = s->pos;
        ff_fetch_timestamp(s, 0, 0);
    }
    /* 真正的解析函数，从avctx可以看出应该是调用了响应的解码器的*/
    index = s->parser->parser_parse(s, avctx, (const uint8_t **) poutbuf,
                                    poutbuf_size, buf, buf_size);
    /* update the file pointer */
    if (*poutbuf_size) {
        /* fill the data for the current frame */
        s->frame_offset = s->next_frame_offset;

        /* offset of the next frame */
        s->next_frame_offset = s->cur_offset + index;
        s->fetch_timestamp   = 1;
    }
    if (index < 0)
        index = 0;
    s->cur_offset += index;
    return index;
}