FFmpeg的H264解码源码分析：解析器

解析器主要就是解析出NALU，以及解析一些SPS、PPS等信息，下面分析一下ff_h264_parser 

AVCodecParser ff_h264_parser = {
    .codec_ids      = { AV_CODEC_ID_H264 },
    .priv_data_size = sizeof(H264ParseContext),
    .parser_init    = init,
    .parser_parse   = h264_parse,
    .parser_close   = h264_close,
    .split          = h264_split,
};

init()

static av_cold int init(AVCodecParserContext *s)
{
    H264ParseContext *p = s->priv_data;

    p->reference_dts = AV_NOPTS_VALUE;
    p->last_frame_num = INT_MAX;
    ff_h264dsp_init(&p->h264dsp, 8, 1);
    return 0;
}

初始化比较简单，就是一些基本的参数设置

h264_close()

static void h264_close(AVCodecParserContext *s)
{
    H264ParseContext *p = s->priv_data;
    ParseContext *pc = &p->pc;

    av_freep(&pc->buffer);

    ff_h264_sei_uninit(&p->sei);
    ff_h264_ps_uninit(&p->ps);
}

结束这里就是释放AVCodecParserContext对象

h264_parse()

static int h264_parse(AVCodecParserContext *s,
                      AVCodecContext *avctx,
                      const uint8_t **poutbuf, int *poutbuf_size,
                      const uint8_t *buf, int buf_size)
{
    H264ParseContext *p = s->priv_data;
    ParseContext *pc = &p->pc;
    int next;

    // 获取extradata，一般保存SPS、PPS等信息
    if (!p->got_first) {
        p->got_first = 1;
        if (avctx->extradata_size) {
            ff_h264_decode_extradata(avctx->extradata, avctx->extradata_size,
                                     &p->ps, &p->is_avc, &p->nal_length_size,
                                     avctx->err_recognition, avctx);
        }
    }

    if (s->flags & PARSER_FLAG_COMPLETE_FRAMES) {
        next = buf_size;
    } else {
        // 查找帧结尾位置，以起始码（0x000001或0x00000001）为依据
        // SPS和PPS会被一起找到，第一个数据尾：SPS+PPS
        next = h264_find_frame_end(p, buf, buf_size, avctx);
        // next < 0 : 非一帧，数据缓存到pc->buffer
        // next >=0 : 找到一帧，将数据和之前缓存的数据打包一帧到buf
        if (ff_combine_frame(pc, next, &buf, &buf_size) < 0) {
            *poutbuf      = NULL;
            *poutbuf_size = 0;
            return buf_size;
        }

        if (next < 0 && next != END_NOT_FOUND) {
            av_assert1(pc->last_index + next >= 0);
            h264_find_frame_end(p, &pc->buffer[pc->last_index + next], -next, avctx); // update state
        }
    }

    // 解析NALU，从SPS和PPS等中获取一些信息
    parse_nal_units(s, avctx, buf, buf_size);

    if (avctx->framerate.num)
        avctx->time_base = av_inv_q(av_mul_q(avctx->framerate, (AVRational){avctx->ticks_per_frame, 1}));
    if (p->sei.picture_timing.cpb_removal_delay >= 0) {
        s->dts_sync_point    = p->sei.buffering_period.present;
        s->dts_ref_dts_delta = p->sei.picture_timing.cpb_removal_delay;
        s->pts_dts_delta     = p->sei.picture_timing.dpb_output_delay;
    } else {
        s->dts_sync_point    = INT_MIN;
        s->dts_ref_dts_delta = INT_MIN;
        s->pts_dts_delta     = INT_MIN;
    }

    if (s->flags & PARSER_FLAG_ONCE) {
        s->flags &= PARSER_FLAG_COMPLETE_FRAMES;
    }

    if (s->dts_sync_point >= 0) {
        int64_t den = avctx->time_base.den * (int64_t)avctx->pkt_timebase.num;
        if (den > 0) {
            int64_t num = avctx->time_base.num * (int64_t)avctx->pkt_timebase.den;
            if (s->dts != AV_NOPTS_VALUE) {
                // got DTS from the stream, update reference timestamp
                p->reference_dts = s->dts - av_rescale(s->dts_ref_dts_delta, num, den);
            } else if (p->reference_dts != AV_NOPTS_VALUE) {
                // compute DTS based on reference timestamp
                s->dts = p->reference_dts + av_rescale(s->dts_ref_dts_delta, num, den);
            }

            if (p->reference_dts != AV_NOPTS_VALUE && s->pts == AV_NOPTS_VALUE)
                s->pts = s->dts + av_rescale(s->pts_dts_delta, num, den);

            if (s->dts_sync_point > 0)
                p->reference_dts = s->dts; // new reference
        }
    }

    // 将完整的一帧输出到poutbuf
    *poutbuf      = buf;
    *poutbuf_size = buf_size;
    return next;
}

这里主要做了几件事：

1. 解析extradata
2. 查找一帧结束位置
3. 组装一帧NALU
4. 解析NALU，获取一些SPS、PPS等信息
5. 输出buf

ff_h264_decode_extradata()

int ff_h264_decode_extradata(const uint8_t *data, int size, H264ParamSets *ps,
                             int *is_avc, int *nal_length_size,
                             int err_recognition, void *logctx)
{
    int ret;

    if (!data || size <= 0)
        return -1;

    if (data[0] == 1) {
        int i, cnt, nalsize;
        const uint8_t *p = data;

        *is_avc = 1;

        if (size < 7) {
            av_log(logctx, AV_LOG_ERROR, "avcC %d too short
", size);
            return AVERROR_INVALIDDATA;
        }

        // Decode sps from avcC
        // 解码SPS
        cnt = *(p + 5) & 0x1f; // Number of sps
        p  += 6;
        for (i = 0; i < cnt; i++) {
            nalsize = AV_RB16(p) + 2;
            if (nalsize > size - (p - data))
                return AVERROR_INVALIDDATA;
            ret = decode_extradata_ps_mp4(p, nalsize, ps, err_recognition, logctx);
            if (ret < 0) {
                av_log(logctx, AV_LOG_ERROR,
                       "Decoding sps %d from avcC failed
", i);
                return ret;
            }
            p += nalsize;
        }
        // Decode pps from avcC
        // 解码PPS
        cnt = *(p++); // Number of pps
        for (i = 0; i < cnt; i++) {
            nalsize = AV_RB16(p) + 2;
            if (nalsize > size - (p - data))
                return AVERROR_INVALIDDATA;
            ret = decode_extradata_ps_mp4(p, nalsize, ps, err_recognition, logctx);
            if (ret < 0) {
                av_log(logctx, AV_LOG_ERROR,
                       "Decoding pps %d from avcC failed
", i);
                return ret;
            }
            p += nalsize;
        }
        // Store right nal length size that will be used to parse all other nals
        *nal_length_size = (data[4] & 0x03) + 1;
    } else {
        *is_avc = 0;
        ret = decode_extradata_ps(data, size, ps, 0, logctx);
        if (ret < 0)
            return ret;
    }
    return size;
}

这里extradata存储的其实是SPS和PPS，解析可以获取SPS和PPS，像FLV这些格式SPS和PPS是存储在extradata里的

h264_find_frame_end()

static int h264_find_frame_end(H264ParseContext *p, const uint8_t *buf,
                               int buf_size, void *logctx)
{
    int i, j;
    uint32_t state;
    ParseContext *pc = &p->pc;

    int next_avc = p->is_avc ? 0 : buf_size;
//    mb_addr= pc->mb_addr - 1;
    state = pc->state;
    if (state > 13)
        state = 7;

    if (p->is_avc && !p->nal_length_size)
        av_log(logctx, AV_LOG_ERROR, "AVC-parser: nal length size invalid
");

    for (i = 0; i < buf_size; i++) {
        if (i >= next_avc) {
            int nalsize = 0;
            i = next_avc;
            for (j = 0; j < p->nal_length_size; j++)
                nalsize = (nalsize << 8) | buf[i++];
            if (nalsize <= 0 || nalsize > buf_size - i) {
                av_log(logctx, AV_LOG_ERROR, "AVC-parser: nal size %d remaining %d
", nalsize, buf_size - i);
                return buf_size;
            }
            next_avc = i + nalsize;
            state    = 5;
        }

        // 这里有个状态机，通过状态机查找start_code来找一帧
        // 7 - 初始化状态
        // 2 - 找到1个0
        // 1 - 找到2个0
        // 0 - 找到大于等于3个0
        // 4 - 找到2个0和1个1，即001（即找到了起始码）
        // 5 - 找到至少3个0和1个1，即0001等等（即找到了起始码）
        // >=8 - 找到2个Slice Header
        if (state == 7) {
            // 查找0的位置
            i += p->h264dsp.startcode_find_candidate(buf + i, next_avc - i);
            if (i < next_avc)
                state = 2;
        } else if (state <= 2) {
            if (buf[i] == 1)
                state ^= 5;            // 2->7, 1->4, 0->5
            else if (buf[i])
                state = 7;
            else
                state >>= 1;           // 2->1, 1->0, 0->0
        } else if (state <= 5) {
            int nalu_type = buf[i] & 0x1F;
            if (nalu_type == H264_NAL_SEI || nalu_type == H264_NAL_SPS ||
                nalu_type == H264_NAL_PPS || nalu_type == H264_NAL_AUD) {
                if (pc->frame_start_found) {
                    i++;
                    goto found;
                }
            } else if (nalu_type == H264_NAL_SLICE || nalu_type == H264_NAL_DPA ||
                       nalu_type == H264_NAL_IDR_SLICE) {
                state += 8;
                continue;
            }
            state = 7;
        } else {
            unsigned int mb, last_mb = p->parse_last_mb;
            GetBitContext gb;
            p->parse_history[p->parse_history_count++] = buf[i];

            init_get_bits(&gb, p->parse_history, 8*p->parse_history_count);
            mb= get_ue_golomb_long(&gb);
            if (get_bits_left(&gb) > 0 || p->parse_history_count > 5) {
                p->parse_last_mb = mb;
                if (pc->frame_start_found) {
                    if (mb <= last_mb) {
                        i -= p->parse_history_count - 1;
                        p->parse_history_count = 0;
                        goto found;
                    }
                } else
                    pc->frame_start_found = 1;
                p->parse_history_count = 0;
                state = 7;
            }
        }
    }
    pc->state = state;
    if (p->is_avc)
        return next_avc;
    return END_NOT_FOUND;

found:
    pc->state             = 7;
    pc->frame_start_found = 0;
    if (p->is_avc)
        return next_avc;
    return i - (state & 5);
}

这里主要是通过startcode去查找一帧NALU的位置。

ff_combine_frame()

int ff_combine_frame(ParseContext *pc, int next,
                     const uint8_t **buf, int *buf_size)
{
    if (pc->overread) {
        ff_dlog(NULL, "overread %d, state:%"PRIX32" next:%d index:%d o_index:%d
",
                pc->overread, pc->state, next, pc->index, pc->overread_index);
        ff_dlog(NULL, "%X %X %X %X
",
                (*buf)[0], (*buf)[1], (*buf)[2], (*buf)[3]);
    }

    /* Copy overread bytes from last frame into buffer. */
    for (; pc->overread > 0; pc->overread--)
        pc->buffer[pc->index++] = pc->buffer[pc->overread_index++];

    if (next > *buf_size)
        return AVERROR(EINVAL);

    /* flush remaining if EOF */
    if (!*buf_size && next == END_NOT_FOUND)
        next = 0;

    pc->last_index = pc->index;

    /* copy into buffer end return */
    // 没找到数据缓存到pc->buffer
    if (next == END_NOT_FOUND) {
        void *new_buffer = av_fast_realloc(pc->buffer, &pc->buffer_size,
                                           *buf_size + pc->index +
                                           AV_INPUT_BUFFER_PADDING_SIZE);

        if (!new_buffer) {
            av_log(NULL, AV_LOG_ERROR, "Failed to reallocate parser buffer to %d
", *buf_size + pc->index + AV_INPUT_BUFFER_PADDING_SIZE);
            pc->index = 0;
            return AVERROR(ENOMEM);
        }
        pc->buffer = new_buffer;
        memcpy(&pc->buffer[pc->index], *buf, *buf_size);
        pc->index += *buf_size;
        return -1;
    }

    av_assert0(next >= 0 || pc->buffer);

    *buf_size          =
    pc->overread_index = pc->index + next;

    /* append to buffer */
    // 找到就把数据组装buf里
    if (pc->index) {
        void *new_buffer = av_fast_realloc(pc->buffer, &pc->buffer_size,
                                           next + pc->index +
                                           AV_INPUT_BUFFER_PADDING_SIZE);
        if (!new_buffer) {
            av_log(NULL, AV_LOG_ERROR, "Failed to reallocate parser buffer to %d
", next + pc->index + AV_INPUT_BUFFER_PADDING_SIZE);
            pc->overread_index =
            pc->index = 0;
            return AVERROR(ENOMEM);
        }
        pc->buffer = new_buffer;
        if (next > -AV_INPUT_BUFFER_PADDING_SIZE)
            memcpy(&pc->buffer[pc->index], *buf,
                   next + AV_INPUT_BUFFER_PADDING_SIZE);
        pc->index = 0;
        *buf      = pc->buffer;
    }

    /* store overread bytes */
    for (; next < 0; next++) {
        pc->state   = pc->state   << 8 | pc->buffer[pc->last_index + next];
        pc->state64 = pc->state64 << 8 | pc->buffer[pc->last_index + next];
        pc->overread++;
    }

    if (pc->overread) {
        ff_dlog(NULL, "overread %d, state:%"PRIX32" next:%d index:%d o_index:%d
",
                pc->overread, pc->state, next, pc->index, pc->overread_index);
        ff_dlog(NULL, "%X %X %X %X
",
                (*buf)[0], (*buf)[1], (*buf)[2], (*buf)[3]);
    }

    return 0;
}

这里主要做了两件事：

1. 要是没找到就缓存到buffer
2. 找到就组装一帧NALU到buffer