webrtc中h264的SingleNalu 解析

主要函数：这个函数解析StapA和SingleNalu两个模式，一下简单讲解一下；

 

RtpDepacketizerH264::ProcessStapAOrSingleNalu

 

 

bool RtpDepacketizerH264::ProcessStapAOrSingleNalu(
    ParsedPayload* parsed_payload,
    const uint8_t* payload_data) {
  parsed_payload->type.Video.width = 0;
  parsed_payload->type.Video.height = 0;
  parsed_payload->type.Video.codec = kRtpVideoH264;
  parsed_payload->type.Video.is_first_packet_in_frame = true;
  RTPVideoHeaderH264* h264_header =
      &parsed_payload->type.Video.codecHeader.H264;

  const uint8_t* nalu_start = payload_data + kNalHeaderSize;
  const size_t nalu_length = length_ - kNalHeaderSize;
  uint8_t nal_type = payload_data[0] & kTypeMask; //获取nal_type ；
  std::vector nalu_start_offsets;
  if (nal_type == H264::NaluType::kStapA) {  //解析StapA；
    // Skip the StapA header (StapA NAL type + length).
    if (length_ <= kStapAHeaderSize) {
      LOG(LS_ERROR) << "StapA header truncated.";
      return false;
    }

    if (!ParseStapAStartOffsets(nalu_start, nalu_length, &nalu_start_offsets)) {
      LOG(LS_ERROR) << "StapA packet with incorrect NALU packet lengths.";
      return false;
    }

    h264_header->packetization_type = kH264StapA;
    nal_type = payload_data[kStapAHeaderSize] & kTypeMask;
  } else {
    h264_header->packetization_type = kH264SingleNalu;
    nalu_start_offsets.push_back(0);
  }
  h264_header->nalu_type = nal_type;
  parsed_payload->frame_type = kVideoFrameDelta;  //在webrtc rtp H264中，仅表示P Frame；就是参考了别的Frame，不可以独立解码；

  nalu_start_offsets.push_back(length_ + kLengthFieldSize);  // End offset.
  for (size_t i = 0; i < nalu_start_offsets.size() - 1; ++i) {
    size_t start_offset = nalu_start_offsets[i];
    // End offset is actually start offset for next unit, excluding length field
    // so remove that from this units length.
    size_t end_offset = nalu_start_offsets[i + 1] - kLengthFieldSize;
    if (end_offset - start_offset < H264::kNaluTypeSize) {
      LOG(LS_ERROR) << "STAP-A packet too short";
      return false;
    }

    NaluInfo nalu;
    nalu.type = payload_data[start_offset] & kTypeMask;
    nalu.offset = start_offset;
    nalu.size = end_offset - start_offset;
    nalu.sps_id = -1;
    nalu.pps_id = -1;
    start_offset += H264::kNaluTypeSize;

    switch (nalu.type) { //判断nal类型，如果是sps，pps，sei，idr，设置parsed_payload->frame_type = kVideoFrameKey; 关于output_buffer，也可以不用；
      case H264::NaluType::kSps: {
        // Check if VUI is present in SPS and if it needs to be modified to
        // avoid
        // excessive decoder latency.

        // Copy any previous data first (likely just the first header).
        std::unique_ptr output_buffer(new rtc::Buffer());
        if (start_offset)
          output_buffer->AppendData(payload_data, start_offset);

        rtc::Optional sps;

        SpsVuiRewriter::ParseResult result = SpsVuiRewriter::ParseAndRewriteSps(
            &payload_data[start_offset], end_offset - start_offset, &sps,
            output_buffer.get());
        switch (result) {
          case SpsVuiRewriter::ParseResult::kVuiRewritten:
            if (modified_buffer_) {
              LOG(LS_WARNING)
                  << "More than one H264 SPS NAL units needing "
                     "rewriting found within a single STAP-A packet. "
                     "Keeping the first and rewriting the last.";
            }

            // Rewrite length field to new SPS size.
            if (h264_header->packetization_type == kH264StapA) {
              size_t length_field_offset =
                  start_offset - (H264::kNaluTypeSize + kLengthFieldSize);
              // Stap-A Length includes payload data and type header.
              size_t rewritten_size =
                  output_buffer->size() - start_offset + H264::kNaluTypeSize;
              ByteWriter::WriteBigEndian(
                  &(*output_buffer)[length_field_offset], rewritten_size);
            }

            // Append rest of packet.
            output_buffer->AppendData(
                &payload_data[end_offset],
                nalu_length + kNalHeaderSize - end_offset);

            modified_buffer_ = std::move(output_buffer);
            length_ = modified_buffer_->size();

            RTC_HISTOGRAM_ENUMERATION(kSpsValidHistogramName,
                                      SpsValidEvent::kReceivedSpsRewritten,
                                      SpsValidEvent::kSpsRewrittenMax);
            break;
          case SpsVuiRewriter::ParseResult::kPocOk:
            RTC_HISTOGRAM_ENUMERATION(kSpsValidHistogramName,
                                      SpsValidEvent::kReceivedSpsPocOk,
                                      SpsValidEvent::kSpsRewrittenMax);
            break;
          case SpsVuiRewriter::ParseResult::kVuiOk:
            RTC_HISTOGRAM_ENUMERATION(kSpsValidHistogramName,
                                      SpsValidEvent::kReceivedSpsVuiOk,
                                      SpsValidEvent::kSpsRewrittenMax);
            break;
          case SpsVuiRewriter::ParseResult::kFailure:
            RTC_HISTOGRAM_ENUMERATION(kSpsValidHistogramName,
                                      SpsValidEvent::kReceivedSpsParseFailure,
                                      SpsValidEvent::kSpsRewrittenMax);
            break;
        }

        if (sps) {
          parsed_payload->type.Video.width = sps->width;
          parsed_payload->type.Video.height = sps->height;
          nalu.sps_id = sps->id;
        } else {
          LOG(LS_WARNING) << "Failed to parse SPS id from SPS slice.";
        }
        parsed_payload->frame_type = kVideoFrameKey;
        break;
      }
      case H264::NaluType::kPps: {
        uint32_t pps_id;
        uint32_t sps_id;
        if (PpsParser::ParsePpsIds(&payload_data[start_offset],
                                    end_offset - start_offset, &pps_id,
                                    &sps_id)) {
          nalu.pps_id = pps_id;
          nalu.sps_id = sps_id;
        } else {
          LOG(LS_WARNING)
              << "Failed to parse PPS id and SPS id from PPS slice.";
        }
        break;
      }
      case H264::NaluType::kIdr:
        parsed_payload->frame_type = kVideoFrameKey;
        FALLTHROUGH();
      case H264::NaluType::kSlice: {
        rtc::Optional pps_id = PpsParser::ParsePpsIdFromSlice(
            &payload_data[start_offset], end_offset - start_offset);
        if (pps_id) {
          nalu.pps_id = *pps_id;
        } else {
          LOG(LS_WARNING) << "Failed to parse PPS id from slice of type: "
                          << static_cast(nalu.type);
        }
        break;
      }
      // Slices below don't contain SPS or PPS ids.
      case H264::NaluType::kAud:
      case H264::NaluType::kEndOfSequence:
      case H264::NaluType::kEndOfStream:
      case H264::NaluType::kFiller:
        break;
      case H264::NaluType::kSei:
        parsed_payload->frame_type = kVideoFrameKey;
        break;
      case H264::NaluType::kStapA:
      case H264::NaluType::kFuA:
        LOG(LS_WARNING) << "Unexpected STAP-A or FU-A received.";
        return false;
    }
    RTPVideoHeaderH264* h264 = &parsed_payload->type.Video.codecHeader.H264;
    if (h264->nalus_length == kMaxNalusPerPacket) {
      LOG(LS_WARNING)
          << "Received packet containing more than " << kMaxNalusPerPacket
          << " NAL units. Will not keep track sps and pps ids for all of them.";
    } else {
      h264->nalus[h264->nalus_length++] = nalu;
    }
  }

  return true;
}