FFmpeg源码剖析-解码：av_read_frame()

av_read_frame()函数位于libavformat/utils.c

1. 函数概述

读取输入流的一帧数据；

2. 函数调用图

​ ​

3. 伪码分析

4. 代码分析

av_read_frame()它的声明(libavformat/avformat.h)/**
 * Return the next frame of a stream.
 * This function returns what is stored in the file, and does not validate
 * that what is there are valid frames for the decoder. It will split what is
 * stored in the file into frames and return one for each call. It will not
 * omit invalid data between valid frames so as to give the decoder the maximum
 * information possible for decoding.
 *
 * If pkt->buf is NULL, then the packet is valid until the next
 * av_read_frame() or until avformat_close_input(). Otherwise the packet
 * is valid indefinitely. In both cases the packet must be freed with
 * av_packet_unref when it is no longer needed. For video, the packet contains
 * exactly one frame. For audio, it contains an integer number of frames if each
 * frame has a known fixed size (e.g. PCM or ADPCM data). If the audio frames
 * have a variable size (e.g. MPEG audio), then it contains one frame.
 *
 * pkt->pts, pkt->dts and pkt->duration are always set to correct
 * values in AVStream.time_base units (and guessed if the format cannot
 * provide them). pkt->pts can be AV_NOPTS_VALUE if the video format
 * has B-frames, so it is better to rely on pkt->dts if you do not
 * decompress the payload.
 *
 * @return 0 if OK, < 0 on error or end of file
 */
int av_read_frame(AVFormatContext *s, AVPacket *pkt);

函数功能: 返回输入流的下一帧。
    因为它是为解码器提供尽可能多的信息，所以它返回的是存储在文件中的下一帧，
    而不会验证这一帧对解码器是否有效。
    
参数解析:
 . pkt->buf， 
   如果它的值为NULL，那这个Packet数据有效，直到下一个av_read_frame()或 avformat_close_input()。
   否则，这个packet数据不确定是否有效。
   在这两种情况下，当数据不再有效时，需要用av_packet_unref来释放。
   对于视频来说，这个packet包含一个完整的视频帧；
   对于音频来说，如果音频帧的大小是固定的(如PCM或ADPCM数据)，则包含整数帧；
                 如果音频帧的大小是可变的(如MPEG音频)，则包含一帧。
 . pkt->pts, pkt->dts和pkt->duration是以AVStream.time_base为单位表示的。
   pkt->pts，值可以是AV_NOPTS_VALUE,如果视频中包含B-frames。
   所以当不进行数据解码时，pkt->dts是更可靠的。
              
返回值:
    返回值为 0 ，则正确；
    返回值小于0，则表示出错或到了文件尾；


它的定义(libavformat/utils.c):
#define AVFMT_FLAG_GENPTS       0x0001 ///< Generate missing pts even if it requires parsing future frames.

int av_read_frame(AVFormatContext *s, AVPacket *pkt)
{
    const int genpts = s->flags & AVFMT_FLAG_GENPTS;
    int eof = 0;
    int ret;
    AVStream *st;

    if (!genpts) {
        /* 如果包的链表中有数据，则直接从链表中取一个packet; 
         * 如果没有，则要读文件获得一个packet.
         */
        ret = s->internal->packet_buffer
              ? read_from_packet_buffer(&s->internal->packet_buffer,
                                        &s->internal->packet_buffer_end, pkt)
              : read_frame_internal(s, pkt);
        if (ret < 0)
            return ret;
        goto return_packet;
    }

    for (;;) {
        AVPacketList *pktl = s->internal->packet_buffer;

        if (pktl) {
            AVPacket *next_pkt = &pktl->pkt;

            if (next_pkt->dts != AV_NOPTS_VALUE) {
                int wrap_bits = s->streams[next_pkt->stream_index]->pts_wrap_bits;

                // last dts seen for this stream. if any of packets following
                // current one had no dts, we will set this to AV_NOPTS_VALUE.
                int64_t last_dts = next_pkt->dts;
                while (pktl && next_pkt->pts == AV_NOPTS_VALUE) {
                    if (pktl->pkt.stream_index == next_pkt->stream_index &&
                        (av_compare_mod(next_pkt->dts, pktl->pkt.dts, 2LL << (wrap_bits - 1)) < 0)) {
                        if (av_compare_mod(pktl->pkt.pts, pktl->pkt.dts, 2LL << (wrap_bits - 1))) {
                            // not B-frame
                            next_pkt->pts = pktl->pkt.dts;
                        }
                        if (last_dts != AV_NOPTS_VALUE) {
                            // Once last dts was set to AV_NOPTS_VALUE, we don't change it.
                            last_dts = pktl->pkt.dts;
                        }
                    }
                    pktl = pktl->next;
                }
                if (eof && next_pkt->pts == AV_NOPTS_VALUE && last_dts != AV_NOPTS_VALUE) {
                    // Fixing the last reference frame had none pts issue (For MXF etc).
                    // We only do this when
                    // 1. eof.
                    // 2. we are not able to resolve a pts value for current packet.
                    // 3. the packets for this stream at the end of the files had valid dts.
                    next_pkt->pts = last_dts + next_pkt->duration;
                }
                pktl = s->internal->packet_buffer;
            }

            /* read packet from packet buffer, if there is data */
            st = s->streams[next_pkt->stream_index];
            if (!(next_pkt->pts == AV_NOPTS_VALUE && st->discard < AVDISCARD_ALL &&
                  next_pkt->dts != AV_NOPTS_VALUE && !eof)) {
                ret = read_from_packet_buffer(&s->internal->packet_buffer,
                                               &s->internal->packet_buffer_end, pkt);
                goto return_packet;
            }
        }

        ret = read_frame_internal(s, pkt);
        if (ret < 0) {
            if (pktl && ret != AVERROR(EAGAIN)) {
                eof = 1;
                continue;
            } else
                return ret;
        }

        ret = add_to_pktbuf(&s->internal->packet_buffer, pkt,
                            &s->internal->packet_buffer_end, 1);
        av_packet_unref(pkt);
        if (ret < 0)
            return ret;
    }

return_packet:
    st = s->streams[pkt->stream_index];
    if ((s->iformat->flags & AVFMT_GENERIC_INDEX) && pkt->flags & AV_PKT_FLAG_KEY) {
        ff_reduce_index(s, st->index);
        av_add_index_entry(st, pkt->pos, pkt->dts, 0, 0, AVINDEX_KEYFRAME);
    }

    if (is_relative(pkt->dts))
        pkt->dts -= RELATIVE_TS_BASE;
    if (is_relative(pkt->pts))
        pkt->pts -= RELATIVE_TS_BASE;

    return ret;
}

4.1 read_from_packet_buffer

static int read_from_packet_buffer(AVPacketList **pkt_buffer,
                                   AVPacketList **pkt_buffer_end,
                                   AVPacket      *pkt)
{
    AVPacketList *pktl;
    av_assert0(*pkt_buffer);


    pktl        = *pkt_buffer;
    *pkt        = pktl->pkt;
    *pkt_buffer = pktl->next;
    if (!pktl->next)
        *pkt_buffer_end = NULL;
    av_freep(&pktl);
    return 0;
}

这个函数很简单，
就是从packet链表的头起取一个packet，然后将链表pkt_buffer指针向下移动一个。

4.2 read_frame_internal()

read_frame_internal()位于 libavformat/utils.c
static int read_frame_internal(AVFormatContext *s, AVPacket *pkt)
{
    int ret = 0, i, got_packet = 0;
    AVDictionary *metadata = NULL;


    /* AVPacket结构体的初始化 */
    av_init_packet(pkt);


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


        /* read next packet */
        ret = ff_read_packet(s, &cur_pkt);
        if (ret < 0) { ... }
        ret = 0;
        st  = s->streams[cur_pkt.stream_index]; //从全局变量中获得当前包所在流的数据信息


        /* update context if required */
        if (st->internal->need_context_update) {
          ...
        }


        if (!st->need_parsing || !st->parser) {
            /* no parsing needed: we just output the packet as is */
            *pkt = cur_pkt;
            compute_pkt_fields(s, st, NULL, pkt, AV_NOPTS_VALUE, AV_NOPTS_VALUE);
            if ((s->iformat->flags & AVFMT_GENERIC_INDEX) &&
                (pkt->flags & AV_PKT_FLAG_KEY) && pkt->dts != AV_NOPTS_VALUE) {
                ff_reduce_index(s, st->index);
                av_add_index_entry(st, pkt->pos, pkt->dts,
                                   0, 0, AVINDEX_KEYFRAME);
            }
            got_packet = 1;
        } else if (st->discard < AVDISCARD_ALL) {
            /* 解析当前的packet，并将所有解析出的数据填充到结构体中 */
            if ((ret = parse_packet(s, &cur_pkt, cur_pkt.stream_index)) < 0)
                return ret;
            st->codecpar->sample_rate = st->internal->avctx->sample_rate;
            st->codecpar->bit_rate = st->internal->avctx->bit_rate;
            st->codecpar->channels = st->internal->avctx->channels;
            st->codecpar->channel_layout = st->internal->avctx->channel_layout;
            st->codecpar->codec_id = st->internal->avctx->codec_id;
        } else {
            /* free packet */
            av_packet_unref(&cur_pkt);
        }
        if (pkt->flags & AV_PKT_FLAG_KEY)
            st->skip_to_keyframe = 0;
        if (st->skip_to_keyframe) {
            av_packet_unref(&cur_pkt);
            if (got_packet) {
                *pkt = cur_pkt;
            }
            got_packet = 0;
        }
    }


    /* 再从上面已读取新包到包链表中取出一个包 */
    if (!got_packet && s->internal->parse_queue)
        ret = read_from_packet_buffer(&s->internal->parse_queue, &s->internal->parse_queue_end, pkt);


    if (ret >= 0) {
        AVStream *st = s->streams[pkt->stream_index];
        int discard_padding = 0;
        if (st->first_discard_sample && pkt->pts != AV_NOPTS_VALUE) {
            int64_t pts = pkt->pts - (is_relative(pkt->pts) ? RELATIVE_TS_BASE : 0);
            int64_t sample = ts_to_samples(st, pts);
            int duration = ts_to_samples(st, pkt->duration);
            int64_t end_sample = sample + duration;
            if (duration > 0 && end_sample >= st->first_discard_sample &&
                sample < st->last_discard_sample)
                discard_padding = FFMIN(end_sample - st->first_discard_sample, duration);
        }
        if (st->start_skip_samples && (pkt->pts == 0 || pkt->pts == RELATIVE_TS_BASE))
            st->skip_samples = st->start_skip_samples;
        if (st->skip_samples || discard_padding) {
            uint8_t *p = av_packet_new_side_data(pkt, AV_PKT_DATA_SKIP_SAMPLES, 10);
            if (p) {
                AV_WL32(p, st->skip_samples);
                AV_WL32(p + 4, discard_padding);
                av_log(s, AV_LOG_DEBUG, "demuxer injecting skip %d / discard %d\n", st->skip_samples, discard_padding);
            }
            st->skip_samples = 0;
        }


        if (st->inject_global_side_data) {
            for (i = 0; i < st->nb_side_data; i++) {
                AVPacketSideData *src_sd = &st->side_data[i];
                uint8_t *dst_data;


                if (av_packet_get_side_data(pkt, src_sd->type, NULL))
                    continue;


                dst_data = av_packet_new_side_data(pkt, src_sd->type, src_sd->size);
                if (!dst_data) {
                    av_log(s, AV_LOG_WARNING, "Could not inject global side data\n");
                    continue;
                }


                memcpy(dst_data, src_sd->data, src_sd->size);
            }
            st->inject_global_side_data = 0;
        }


        if (!(s->flags & AVFMT_FLAG_KEEP_SIDE_DATA))
            av_packet_merge_side_data(pkt);
    }


    av_opt_get_dict_val(s, "metadata", AV_OPT_SEARCH_CHILDREN, &metadata);
    if (metadata) {
        s->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
        av_dict_copy(&s->metadata, metadata, 0);
        av_dict_free(&metadata);
        av_opt_set_dict_val(s, "metadata", NULL, AV_OPT_SEARCH_CHILDREN);
    }


#if FF_API_LAVF_AVCTX
    update_stream_avctx(s);
#endif


    if (s->debug & FF_FDEBUG_TS)
        av_log(s, AV_LOG_DEBUG,
               "read_frame_internal stream=%d, pts=%s, dts=%s, "
               "size=%d, duration=%"PRId64", flags=%d\n",
               pkt->stream_index,
               av_ts2str(pkt->pts),
               av_ts2str(pkt->dts),
               pkt->size, pkt->duration, pkt->flags);


    return ret;
}

4.3 parse_packet()

libavformat/utils.c
/**
 * Parse a packet, add all split parts to parse_queue.
 *
 * @param pkt Packet to parse, NULL when flushing the parser at end of stream.
 */
static int parse_packet(AVFormatContext *s, AVPacket *pkt, int stream_index)
{
    AVPacket out_pkt = { 0 }, flush_pkt = { 0 };
    AVStream *st = s->streams[stream_index];
    uint8_t *data = pkt ? pkt->data : NULL;
    int size      = pkt ? pkt->size : 0;
    int ret = 0, got_output = 0;


    if (!pkt) {
        av_init_packet(&flush_pkt);
        pkt        = &flush_pkt;
        got_output = 1;
    } else if (!size && st->parser->flags & PARSER_FLAG_COMPLETE_FRAMES) {
        // preserve 0-size sync packets
        compute_pkt_fields(s, st, st->parser, pkt, AV_NOPTS_VALUE, AV_NOPTS_VALUE);
    }


    while (size > 0 || (pkt == &flush_pkt && got_output)) {
        int len;
        int64_t next_pts = pkt->pts;
        int64_t next_dts = pkt->dts;


        av_init_packet(&out_pkt);
        /* 调用视频或音频解析器进行对应的解析 */
        len = av_parser_parse2(st->parser, st->internal->avctx,
                               &out_pkt.data, &out_pkt.size, data, size,
                               pkt->pts, pkt->dts, pkt->pos);


        pkt->pts = pkt->dts = AV_NOPTS_VALUE;
        pkt->pos = -1;
        /* increment read pointer */
        data += len;
        size -= len;


        got_output = !!out_pkt.size;


        if (!out_pkt.size)
            continue;


        if (pkt->side_data) {
            out_pkt.side_data       = pkt->side_data;
            out_pkt.side_data_elems = pkt->side_data_elems;
            pkt->side_data          = NULL;
            pkt->side_data_elems    = 0;
        }


        /* set the duration */
        out_pkt.duration = (st->parser->flags & PARSER_FLAG_COMPLETE_FRAMES) ? pkt->duration : 0;
        if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
            if (st->internal->avctx->sample_rate > 0) {
                out_pkt.duration =
                    av_rescale_q_rnd(st->parser->duration,
                                     (AVRational) { 1, st->internal->avctx->sample_rate },
                                     st->time_base,
                                     AV_ROUND_DOWN);
            }
        }


        out_pkt.stream_index = st->index;
        out_pkt.pts          = st->parser->pts;
        out_pkt.dts          = st->parser->dts;
        out_pkt.pos          = st->parser->pos;


        if (st->need_parsing == AVSTREAM_PARSE_FULL_RAW)
            out_pkt.pos = st->parser->frame_offset;


        if (st->parser->key_frame == 1 ||
            (st->parser->key_frame == -1 &&
             st->parser->pict_type == AV_PICTURE_TYPE_I))
            out_pkt.flags |= AV_PKT_FLAG_KEY;


        if (st->parser->key_frame == -1 && st->parser->pict_type ==AV_PICTURE_TYPE_NONE && (pkt->flags&AV_PKT_FLAG_KEY))
            out_pkt.flags |= AV_PKT_FLAG_KEY;


        /* 计算当前包的pts,dts,duration三个值 */
        compute_pkt_fields(s, st, st->parser, &out_pkt, next_dts, next_pts);


        /* 将当前包添加到包链表中 */
        ret = add_to_pktbuf(&s->internal->parse_queue, &out_pkt,
                            &s->internal->parse_queue_end, 1);
        av_packet_unref(&out_pkt);
        if (ret < 0)
            goto fail;
    }


    /* end of the stream => close and free the parser */
    if (pkt == &flush_pkt) {
        av_parser_close(st->parser);
        st->parser = NULL;
    }


fail:
    av_packet_unref(pkt);
    return ret;
}

4.4 av_parser_parse2()

libavformat/avcodec.h
/**
 * Parse a packet.
 *
 * @param s             parser context.
 * @param avctx         codec context.
 * @param poutbuf       set to pointer to parsed buffer or NULL if not yet finished.
 * @param poutbuf_size  set to size of parsed buffer or zero if not yet finished.
 * @param buf           input buffer.
 * @param buf_size      buffer size in bytes without the padding. I.e. the full buffer
                        size is assumed to be buf_size + AV_INPUT_BUFFER_PADDING_SIZE.
                        To signal EOF, this should be 0 (so that the last frame
                        can be output).
 * @param pts           input presentation timestamp.
 * @param dts           input decoding timestamp.
 * @param pos           input byte position in stream.
 * @return the number of bytes of the input bitstream used.
 *
 * Example:
 * @code
 *   while(in_len){
 *       len = av_parser_parse2(myparser, AVCodecContext, &data, &size,
 *                                        in_data, in_len,
 *                                        pts, dts, pos);
 *       in_data += len;
 *       in_len  -= len;
 *
 *       if(size)
 *          decode_frame(data, size);
 *   }
 * @endcode
 */
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);
解析一个packet。


libavcodec/parser.c
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[AV_INPUT_BUFFER_PADDING_SIZE];


    av_assert1(avctx->codec_id != AV_CODEC_ID_NONE);


    /* Parsers only work for the specified codec ids. */
    av_assert1(avctx->codec_id == s->parser->codec_ids[0] ||
               avctx->codec_id == s->parser->codec_ids[1] ||
               avctx->codec_id == s->parser->codec_ids[2] ||
               avctx->codec_id == s->parser->codec_ids[3] ||
               avctx->codec_id == s->parser->codec_ids[4]);


    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, 0);
    }
    /* WARNING: the returned index can be negative */
    /* 此处函数指针将指向当前包所对应流的解析器，
     * 如H.264的视频流，则调用libavcodec/h264_parser.c:h264_parse()函数，
     * 它将分解析H.264的NAL单元， 从而获得这个包的完整信息 
     */
    index = s->parser->parser_parse(s, avctx, (const uint8_t **) poutbuf,
                                    poutbuf_size, buf, buf_size);
    av_assert0(index > -0x20000000); // The API does not allow returning AVERROR codes
#define FILL(name) if(s->name > 0 && avctx->name <= 0) avctx->name = s->name
    if (avctx->codec_type == AVMEDIA_TYPE_VIDEO) {
        FILL(field_order);
    }


    /* 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;
}




​