Add H264 bitstream rewriting to limit frame reordering marker in header

webrtc/modules/rtp_rtcp/source/rtp_format_h264.cc

 /*
  *  Copyright (c) 2014 The WebRTC project authors. All Rights Reserved.
  *
  *  Use of this source code is governed by a BSD-style license
  *  that can be found in the LICENSE file in the root of the source
  *  tree. An additional intellectual property rights grant can be found
  *  in the file PATENTS.  All contributing project authors may
  *  be found in the AUTHORS file in the root of the source tree.
  */
 
+#include "webrtc/modules/rtp_rtcp/source/rtp_format_h264.h"
+
 #include <string.h>
+#include <vector>
 
+#include "webrtc/base/checks.h"
 #include "webrtc/base/logging.h"
 #include "webrtc/modules/include/module_common_types.h"
 #include "webrtc/modules/rtp_rtcp/source/byte_io.h"
-#include "webrtc/modules/rtp_rtcp/source/h264_sps_parser.h"
-#include "webrtc/modules/rtp_rtcp/source/rtp_format_h264.h"
+#include "webrtc/modules/rtp_rtcp/source/h264/sps_vui_rewriter.h"
+#include "webrtc/modules/rtp_rtcp/source/h264/h264_common.h"
+#include "webrtc/modules/rtp_rtcp/source/h264/sps_parser.h"
 
 namespace webrtc {
 namespace {
 
 enum Nalu {
   kSlice = 1,
   kIdr = 5,
   kSei = 6,
   kSps = 7,
   kPps = 8,
   kStapA = 24,
   kFuA = 28
 };
 
 static const size_t kNalHeaderSize = 1;
 static const size_t kFuAHeaderSize = 2;
 static const size_t kLengthFieldSize = 2;
 static const size_t kStapAHeaderSize = kNalHeaderSize + kLengthFieldSize;
 
 // Bit masks for FU (A and B) indicators.
-enum NalDefs { kFBit = 0x80, kNriMask = 0x60, kTypeMask = 0x1F };
+enum NalDefs : uint8_t { kFBit = 0x80, kNriMask = 0x60, kTypeMask = 0x1F };
 
 // Bit masks for FU (A and B) headers.
-enum FuDefs { kSBit = 0x80, kEBit = 0x40, kRBit = 0x20 };
+enum FuDefs : uint8_t { kSBit = 0x80, kEBit = 0x40, kRBit = 0x20 };
 
 // TODO(pbos): Avoid parsing this here as well as inside the jitter buffer.
-bool VerifyStapANaluLengths(const uint8_t* nalu_ptr, size_t length_remaining) {
+bool ParseStapAStartOffsets(const uint8_t* nalu_ptr,
+                            size_t length_remaining,
+                            std::vector<size_t>* offsets,
+                            size_t base_offset) {
+  size_t offset = 0;
   while (length_remaining > 0) {
     // Buffer doesn't contain room for additional nalu length.
     if (length_remaining < sizeof(uint16_t))
       return false;
-    uint16_t nalu_size = nalu_ptr[0] << 8 | nalu_ptr[1];
+    uint16_t nalu_size = ByteReader<uint16_t>::ReadBigEndian(nalu_ptr);
     nalu_ptr += sizeof(uint16_t);
     length_remaining -= sizeof(uint16_t);
     if (nalu_size > length_remaining)
       return false;
     nalu_ptr += nalu_size;
     length_remaining -= nalu_size;
+
+    offsets->push_back(offset + kLengthFieldSize + base_offset);

[stefan-webrtc, 2016/05/22 23:11:50]

Why do we have to pass in base_offset? It seems to always be kNalHeaderSize, so
maybe just replace it with that?

[sprang_webrtc, 2016/05/25 09:06:03]

I intended to use it in a second place, but ended up not to. Removed.

+    offset += kLengthFieldSize + nalu_size;
   }
   return true;
 }
 
-bool ParseSingleNalu(RtpDepacketizer::ParsedPayload* parsed_payload,
-                     const uint8_t* payload_data,
-                     size_t payload_data_length) {
-  parsed_payload->type.Video.width = 0;
-  parsed_payload->type.Video.height = 0;
-  parsed_payload->type.Video.codec = kRtpVideoH264;
-  parsed_payload->type.Video.isFirstPacket = true;
-  RTPVideoHeaderH264* h264_header =
-      &parsed_payload->type.Video.codecHeader.H264;
-
-  const uint8_t* nalu_start = payload_data + kNalHeaderSize;
-  size_t nalu_length = payload_data_length - kNalHeaderSize;
-  uint8_t nal_type = payload_data[0] & kTypeMask;
-  if (nal_type == kStapA) {
-    // Skip the StapA header (StapA nal type + length).
-    if (payload_data_length <= kStapAHeaderSize) {
-      LOG(LS_ERROR) << "StapA header truncated.";
-      return false;
-    }
-    if (!VerifyStapANaluLengths(nalu_start, nalu_length)) {
-      LOG(LS_ERROR) << "StapA packet with incorrect NALU packet lengths.";
-      return false;
-    }
-
-    nal_type = payload_data[kStapAHeaderSize] & kTypeMask;
-    nalu_start += kStapAHeaderSize;
-    nalu_length -= kStapAHeaderSize;
-    h264_header->packetization_type = kH264StapA;
-  } else {
-    h264_header->packetization_type = kH264SingleNalu;
-  }
-  h264_header->nalu_type = nal_type;
-
-  // We can read resolution out of sps packets.
-  if (nal_type == kSps) {
-    H264SpsParser parser(nalu_start, nalu_length);
-    if (parser.Parse()) {
-      parsed_payload->type.Video.width = parser.width();
-      parsed_payload->type.Video.height = parser.height();
-    }
-  }
-  switch (nal_type) {
-    case kSps:
-    case kPps:
-    case kSei:
-    case kIdr:
-      parsed_payload->frame_type = kVideoFrameKey;
-      break;
-    default:
-      parsed_payload->frame_type = kVideoFrameDelta;
-      break;
-  }
-  return true;
-}
-
-bool ParseFuaNalu(RtpDepacketizer::ParsedPayload* parsed_payload,
-                  const uint8_t* payload_data,
-                  size_t payload_data_length,
-                  size_t* offset) {
-  if (payload_data_length < kFuAHeaderSize) {
-    LOG(LS_ERROR) << "FU-A NAL units truncated.";
-    return false;
-  }
-  uint8_t fnri = payload_data[0] & (kFBit | kNriMask);
-  uint8_t original_nal_type = payload_data[1] & kTypeMask;
-  bool first_fragment = (payload_data[1] & kSBit) > 0;
-
-  uint8_t original_nal_header = fnri | original_nal_type;
-  if (first_fragment) {
-    *offset = kNalHeaderSize;
-    uint8_t* payload = const_cast<uint8_t*>(payload_data + *offset);
-    payload[0] = original_nal_header;
-  } else {
-    *offset = kFuAHeaderSize;
-  }
-
-  if (original_nal_type == kIdr) {
-    parsed_payload->frame_type = kVideoFrameKey;
-  } else {
-    parsed_payload->frame_type = kVideoFrameDelta;
-  }
-  parsed_payload->type.Video.width = 0;
-  parsed_payload->type.Video.height = 0;
-  parsed_payload->type.Video.codec = kRtpVideoH264;
-  parsed_payload->type.Video.isFirstPacket = first_fragment;
-  RTPVideoHeaderH264* h264_header =
-      &parsed_payload->type.Video.codecHeader.H264;
-  h264_header->packetization_type = kH264FuA;
-  h264_header->nalu_type = original_nal_type;
-  return true;
-}
 }  // namespace
 
 RtpPacketizerH264::RtpPacketizerH264(FrameType frame_type,
                                      size_t max_payload_len)
-    : payload_data_(NULL),
-      payload_size_(0),
-      max_payload_len_(max_payload_len) {
-}
+    : max_payload_len_(max_payload_len) {}
 
 RtpPacketizerH264::~RtpPacketizerH264() {
 }
 
+RtpPacketizerH264::Fragment::Fragment(const uint8_t* buffer, size_t length)
+    : buffer(buffer), length(length) {}
+RtpPacketizerH264::Fragment::Fragment(
+    std::unique_ptr<rtc::Buffer> buffer_writer)
+    : buffer(buffer_writer->data()),
+      length(buffer_writer->size()),
+      tmp_buffer(std::move(buffer_writer)) {}
+RtpPacketizerH264::Fragment::Fragment(const Fragment& fragment)
+    : buffer(fragment.buffer), length(fragment.length) {}
+
 void RtpPacketizerH264::SetPayloadData(
     const uint8_t* payload_data,
     size_t payload_size,
     const RTPFragmentationHeader* fragmentation) {
-  assert(packets_.empty());
-  assert(fragmentation);
-  payload_data_ = payload_data;
-  payload_size_ = payload_size;
-  fragmentation_.CopyFrom(*fragmentation);
+  RTC_DCHECK(packets_.empty());
+  RTC_DCHECK(input_fragments_.empty());
+  RTC_DCHECK(fragmentation);
+  for (int i = 0; i < fragmentation->fragmentationVectorSize; ++i) {
+    const uint8_t* buffer =
+        &payload_data[fragmentation->fragmentationOffset[i]];
+    size_t length = fragmentation->fragmentationLength[i];
+
+    bool updated_sps = false;
+    H264Common::NaluType nalu_type = H264Common::ParseNaluType(buffer[0]);
+    if (nalu_type == H264Common::NaluType::kSps) {
+      // Check if stream uses picture order count type 0, and if so rewrite it
+      // to enable faster decoding. Streams in that format incur additional
+      // delay because it allows decode order to differ from render order.
+      // The mechanism used is to rewrite (edit or add) the SPS's VUI to contain
+      // restrictions on the maximum number of reordered pictures. This reduces
+      // latency significantly, though it still adds about a frame of latency to
+      // decoding.

[stefan-webrtc, 2016/05/22 23:11:50]

Please comment on the possible side effects, if any.

[sprang_webrtc, 2016/05/25 09:06:03]

Consequences for rewriting the VUI, in general? I'll add a comment and an assert
that don't get any b-frames as well.

[stefan-webrtc, 2016/05/27 20:56:59]

Acknowledged.

+      std::unique_ptr<rtc::Buffer> rbsp_buffer =
+          H264Common::ParseRbsp(buffer + H264Common::kNaluTypeSize,
+                                length - H264Common::kNaluTypeSize);
+      rtc::Optional<SpsParser::SpsState> sps;
+      std::unique_ptr<rtc::Buffer> output_buffer(new rtc::Buffer());
+      output_buffer->AppendData((*rbsp_buffer)[0]);

[stefan-webrtc, 2016/05/26 18:08:13]

Was this wrong? Covered by a test?

[noahric, 2016/05/26 18:56:52]

And either way, it definitely needs a comment, since the next line appears to
use rbsp_buffer[0] again.

Also, slightly prefer rbsp_buffer->data()[0], to be a bit more obvious (the
parens and explicit unique_ptr deref are sorta confusing).

[sprang_webrtc, 2016/05/27 13:12:21]

Acknowledged.

[sprang_webrtc, 2016/05/27 13:12:21]

Yes. Uncovered and cover by a test I added.

+      SpsVuiRewriter::ParseResult result = SpsVuiRewriter::ParseAndRewriteSps(
+          rbsp_buffer->data(), rbsp_buffer->size(), &sps, output_buffer.get());
+      if (result == SpsVuiRewriter::ParseResult::kParsedAndModified) {
+        input_fragments_.push_back(Fragment(std::move(output_buffer)));
+        updated_sps = true;
+      }
+    }
+
+    if (!updated_sps)
+      input_fragments_.push_back(Fragment(buffer, length));
+  }
   GeneratePackets();
 }
 
 void RtpPacketizerH264::GeneratePackets() {
-  for (size_t i = 0; i < fragmentation_.fragmentationVectorSize;) {
-    size_t fragment_offset = fragmentation_.fragmentationOffset[i];
-    size_t fragment_length = fragmentation_.fragmentationLength[i];
-    if (fragment_length > max_payload_len_) {
-      PacketizeFuA(fragment_offset, fragment_length);
+  for (size_t i = 0; i < input_fragments_.size();) {
+    if (input_fragments_[i].length > max_payload_len_) {
+      PacketizeFuA(i);
       ++i;
     } else {
-      i = PacketizeStapA(i, fragment_offset, fragment_length);
+      i = PacketizeStapA(i);
     }
   }
 }
 
-void RtpPacketizerH264::PacketizeFuA(size_t fragment_offset,
-                                     size_t fragment_length) {
+void RtpPacketizerH264::PacketizeFuA(size_t fragment_index) {
   // Fragment payload into packets (FU-A).
   // Strip out the original header and leave room for the FU-A header.
-  fragment_length -= kNalHeaderSize;
-  size_t offset = fragment_offset + kNalHeaderSize;
+  const Fragment& fragment = input_fragments_[fragment_index];
+
+  size_t fragment_length = fragment.length - kNalHeaderSize;
+  size_t offset = kNalHeaderSize;
   size_t bytes_available = max_payload_len_ - kFuAHeaderSize;
-  size_t fragments =
+  const size_t num_fragments =
       (fragment_length + (bytes_available - 1)) / bytes_available;
-  size_t avg_size = (fragment_length + fragments - 1) / fragments;
+
+  const size_t avg_size = (fragment_length + num_fragments - 1) / num_fragments;
   while (fragment_length > 0) {
     size_t packet_length = avg_size;
     if (fragment_length < avg_size)
       packet_length = fragment_length;
-    uint8_t header = payload_data_[fragment_offset];
-    packets_.push(Packet(offset,
-                         packet_length,
-                         offset - kNalHeaderSize == fragment_offset,
-                         fragment_length == packet_length,
-                         false,
-                         header));
+    packets_.push(PacketUnit(Fragment(fragment.buffer + offset, packet_length),

[stefan-webrtc, 2016/05/22 23:11:50]

What happens here if fragment is a rewritten sps? Are we sure that fragment will
outlive this new Fragment?

[sprang_webrtc, 2016/05/25 09:06:03]

Yes. In the case of FU-A, the input fragment won't be popped from the queue
until the last PacketUnit for that fragment has been consumed (from
RtpPacketizerH264::NextFragmentPacket below):

if (packet->last_fragment)
   input_fragments_.pop_front();

[stefan-webrtc, 2016/05/26 18:08:13]

Acknowledged.

+                             offset - kNalHeaderSize == 0,
+                             fragment_length == packet_length, false,
+                             fragment.buffer[0]));
     offset += packet_length;
     fragment_length -= packet_length;
   }
+  RTC_CHECK_EQ(0u, fragment_length);
 }
 
-int RtpPacketizerH264::PacketizeStapA(size_t fragment_index,
-                                      size_t fragment_offset,
-                                      size_t fragment_length) {
+size_t RtpPacketizerH264::PacketizeStapA(size_t fragment_index) {
   // Aggregate fragments into one packet (STAP-A).
   size_t payload_size_left = max_payload_len_;
   int aggregated_fragments = 0;
   size_t fragment_headers_length = 0;
-  assert(payload_size_left >= fragment_length);
-  while (payload_size_left >= fragment_length + fragment_headers_length) {
-    assert(fragment_length > 0);
-    uint8_t header = payload_data_[fragment_offset];
-    packets_.push(Packet(fragment_offset,
-                         fragment_length,
-                         aggregated_fragments == 0,
-                         false,
-                         true,
-                         header));
-    payload_size_left -= fragment_length;
+  const Fragment* fragment = &input_fragments_[fragment_index];
+  RTC_CHECK_GE(payload_size_left, fragment->length);
+  while (payload_size_left >= fragment->length + fragment_headers_length) {
+    RTC_CHECK_GT(fragment->length, 0u);
+    packets_.push(PacketUnit(*fragment, aggregated_fragments == 0, false, true,
+                             fragment->buffer[0]));
+    payload_size_left -= fragment->length;
     payload_size_left -= fragment_headers_length;
 
     // Next fragment.
     ++fragment_index;
-    if (fragment_index == fragmentation_.fragmentationVectorSize)
+    if (fragment_index == input_fragments_.size())
       break;
-    fragment_offset = fragmentation_.fragmentationOffset[fragment_index];
-    fragment_length = fragmentation_.fragmentationLength[fragment_index];
+    fragment = &input_fragments_[fragment_index];
 
     fragment_headers_length = kLengthFieldSize;
     // If we are going to try to aggregate more fragments into this packet
     // we need to add the STAP-A NALU header and a length field for the first
     // NALU of this packet.
     if (aggregated_fragments == 0)
       fragment_headers_length += kNalHeaderSize + kLengthFieldSize;
     ++aggregated_fragments;
   }
   packets_.back().last_fragment = true;
   return fragment_index;
 }
 
 bool RtpPacketizerH264::NextPacket(uint8_t* buffer,
                                    size_t* bytes_to_send,
                                    bool* last_packet) {
   *bytes_to_send = 0;
   if (packets_.empty()) {
     *bytes_to_send = 0;
     *last_packet = true;
     return false;
   }
 
-  Packet packet = packets_.front();
+  PacketUnit packet = packets_.front();
 
   if (packet.first_fragment && packet.last_fragment) {
     // Single NAL unit packet.
-    *bytes_to_send = packet.size;
-    memcpy(buffer, &payload_data_[packet.offset], packet.size);
+    *bytes_to_send = packet.source_fragment.length;
+    memcpy(buffer, packet.source_fragment.buffer, *bytes_to_send);
     packets_.pop();
-    assert(*bytes_to_send <= max_payload_len_);
+    input_fragments_.pop_front();
+    RTC_CHECK_LE(*bytes_to_send, max_payload_len_);
   } else if (packet.aggregated) {
     NextAggregatePacket(buffer, bytes_to_send);
-    assert(*bytes_to_send <= max_payload_len_);
+    RTC_CHECK_LE(*bytes_to_send, max_payload_len_);
   } else {
     NextFragmentPacket(buffer, bytes_to_send);
-    assert(*bytes_to_send <= max_payload_len_);
+    RTC_CHECK_LE(*bytes_to_send, max_payload_len_);
   }
   *last_packet = packets_.empty();
   return true;
 }
 
 void RtpPacketizerH264::NextAggregatePacket(uint8_t* buffer,
                                             size_t* bytes_to_send) {
-  Packet packet = packets_.front();
-  assert(packet.first_fragment);
+  PacketUnit* packet = &packets_.front();
+  RTC_CHECK(packet->first_fragment);
   // STAP-A NALU header.
-  buffer[0] = (packet.header & (kFBit | kNriMask)) | kStapA;
+  buffer[0] = (packet->header & (kFBit | kNriMask)) | kStapA;
   int index = kNalHeaderSize;
   *bytes_to_send += kNalHeaderSize;
-  while (packet.aggregated) {
+  while (packet->aggregated) {
+    const Fragment& fragment = packet->source_fragment;
     // Add NAL unit length field.
-    ByteWriter<uint16_t>::WriteBigEndian(&buffer[index], packet.size);
+    ByteWriter<uint16_t>::WriteBigEndian(&buffer[index], fragment.length);
     index += kLengthFieldSize;
     *bytes_to_send += kLengthFieldSize;
     // Add NAL unit.
-    memcpy(&buffer[index], &payload_data_[packet.offset], packet.size);
-    index += packet.size;
-    *bytes_to_send += packet.size;
+    memcpy(&buffer[index], fragment.buffer, fragment.length);
+    index += fragment.length;
+    *bytes_to_send += fragment.length;
     packets_.pop();
-    if (packet.last_fragment)
+    input_fragments_.pop_front();
+    if (packet->last_fragment)
       break;
-    packet = packets_.front();
+    packet = &packets_.front();
   }
-  assert(packet.last_fragment);
+  RTC_CHECK(packet->last_fragment);
 }
 
 void RtpPacketizerH264::NextFragmentPacket(uint8_t* buffer,
                                            size_t* bytes_to_send) {
-  Packet packet = packets_.front();
+  PacketUnit* packet = &packets_.front();
   // NAL unit fragmented over multiple packets (FU-A).
   // We do not send original NALU header, so it will be replaced by the
   // FU indicator header of the first packet.
-  uint8_t fu_indicator = (packet.header & (kFBit | kNriMask)) | kFuA;
+  uint8_t fu_indicator = (packet->header & (kFBit | kNriMask)) | kFuA;
   uint8_t fu_header = 0;
 
   // S | E | R | 5 bit type.
-  fu_header |= (packet.first_fragment ? kSBit : 0);
-  fu_header |= (packet.last_fragment ? kEBit : 0);
-  uint8_t type = packet.header & kTypeMask;
+  fu_header |= (packet->first_fragment ? kSBit : 0);
+  fu_header |= (packet->last_fragment ? kEBit : 0);
+  uint8_t type = packet->header & kTypeMask;
   fu_header |= type;
   buffer[0] = fu_indicator;
   buffer[1] = fu_header;
 
-  if (packet.last_fragment) {
-    *bytes_to_send = packet.size + kFuAHeaderSize;
-    memcpy(buffer + kFuAHeaderSize, &payload_data_[packet.offset], packet.size);
-  } else {
-    *bytes_to_send = packet.size + kFuAHeaderSize;
-    memcpy(buffer + kFuAHeaderSize, &payload_data_[packet.offset], packet.size);
-  }
+  const Fragment& fragment = packet->source_fragment;
+  *bytes_to_send = fragment.length + kFuAHeaderSize;
+  memcpy(buffer + kFuAHeaderSize, fragment.buffer, fragment.length);
+  if (packet->last_fragment)
+    input_fragments_.pop_front();
   packets_.pop();
 }
 
 ProtectionType RtpPacketizerH264::GetProtectionType() {
   return kProtectedPacket;
 }
 
 StorageType RtpPacketizerH264::GetStorageType(
     uint32_t retransmission_settings) {
   return kAllowRetransmission;
 }
 
 std::string RtpPacketizerH264::ToString() {
   return "RtpPacketizerH264";
 }
 
+RtpDepacketizerH264::RtpDepacketizerH264() : offset_(0), length_(0) {}
+RtpDepacketizerH264::~RtpDepacketizerH264() {}
+
 bool RtpDepacketizerH264::Parse(ParsedPayload* parsed_payload,
                                 const uint8_t* payload_data,
                                 size_t payload_data_length) {
-  assert(parsed_payload != NULL);
+  RTC_CHECK(parsed_payload != nullptr);
   if (payload_data_length == 0) {
     LOG(LS_ERROR) << "Empty payload.";
     return false;
   }
 
+  offset_ = 0;
+  length_ = payload_data_length;
+  modified_buffer_.reset();
+
   uint8_t nal_type = payload_data[0] & kTypeMask;
-  size_t offset = 0;
   if (nal_type == kFuA) {
     // Fragmented NAL units (FU-A).
-    if (!ParseFuaNalu(
-            parsed_payload, payload_data, payload_data_length, &offset)) {
+    if (!ParseFuaNalu(parsed_payload, payload_data))
       return false;
-    }
   } else {
     // We handle STAP-A and single NALU's the same way here. The jitter buffer
     // will depacketize the STAP-A into NAL units later.
-    if (!ParseSingleNalu(parsed_payload, payload_data, payload_data_length))
+    // TODO(sprang): Parse STAP-A offsets here and store in fragmentation vec.
+    if (!ParseSingleNalu(parsed_payload, payload_data))
       return false;
   }
 
-  parsed_payload->payload = payload_data + offset;
-  parsed_payload->payload_length = payload_data_length - offset;
+  const uint8_t* payload =
+      modified_buffer_ ? modified_buffer_->data() : payload_data;
+
+  parsed_payload->payload = payload + offset_;
+  parsed_payload->payload_length = length_;
   return true;
 }
+
+bool RtpDepacketizerH264::ParseSingleNalu(ParsedPayload* parsed_payload,

[stefan-webrtc, 2016/05/22 23:11:50]

Could you point out which parts of these functions actually changed?

[sprang_webrtc, 2016/05/25 09:06:03]

391-417 and 442-444, essentially. Otherwise maybe some formatting changes
(moving enums to h264_common, etc)

+                                          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.isFirstPacket = true;
+  RTPVideoHeaderH264* h264_header =
+      &parsed_payload->type.Video.codecHeader.H264;
+
+  const uint8_t* nalu_start = payload_data + kNalHeaderSize;
+  size_t nalu_length = length_ - kNalHeaderSize;
+  uint8_t nal_type = payload_data[0] & kTypeMask;
+  std::vector<size_t> nalu_start_offsets;
+  if (nal_type == kStapA) {
+    // 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,
+                                kNalHeaderSize)) {
+      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;
+
+  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;
+    nal_type = payload_data[start_offset] & kTypeMask;
+    start_offset += H264Common::kNaluTypeSize;
+
+    if (nal_type == kSps) {
+      // Check if VUI is present in SPS and if it needs to be modified to avoid
+      // excessive decoder latency.
+      std::unique_ptr<rtc::Buffer> rbsp_buffer = H264Common::ParseRbsp(
+          &payload_data[start_offset], end_offset - start_offset);
+      rtc::Optional<SpsParser::SpsState> sps;
+      std::unique_ptr<rtc::Buffer> output_buffer(new rtc::Buffer());
+      // If StapA, copy any previous data first.
+      size_t prefix_size =
+          h264_header->packetization_type == kH264StapA ? start_offset : 0;
+      if (prefix_size)
+        output_buffer->AppendData(payload_data, prefix_size);
+
+      SpsVuiRewriter::ParseResult result = SpsVuiRewriter::ParseAndRewriteSps(
+          rbsp_buffer->data(), rbsp_buffer->size(), &sps, output_buffer.get());
+      if (result == SpsVuiRewriter::ParseResult::kParsedAndModified) {
+        // No support for TWO modified SPS in one RTP packet. Who DOES that?!
+        RTC_CHECK(!modified_buffer_);
+        size_t rewritten_size = output_buffer->size() - prefix_size;
+
+        // Rewrite length field to new SPS size.
+        if (h264_header->packetization_type == kH264StapA) {
+          size_t length_field_offset =
+              start_offset - (H264Common::kNaluTypeSize + kLengthFieldSize);
+          ByteWriter<uint16_t>::WriteBigEndian(
+              &(*output_buffer)[length_field_offset], rewritten_size);
+        }
+
+        // Append rest of packet.
+        output_buffer->AppendData(&payload_data[end_offset],
+                                  length_ - end_offset);
+
+        modified_buffer_ = std::move(output_buffer);
+        length_ = modified_buffer_->size();
+      }
+
+      if (sps) {
+        parsed_payload->type.Video.width = sps->width;
+        parsed_payload->type.Video.height = sps->height;
+      }
+      parsed_payload->frame_type = kVideoFrameKey;
+    } else if (nal_type == kPps || nal_type == kSei || nal_type == kIdr) {
+      parsed_payload->frame_type = kVideoFrameKey;
+    }
+  }
+
+  return true;
+}
+
+bool RtpDepacketizerH264::ParseFuaNalu(
+    RtpDepacketizer::ParsedPayload* parsed_payload,
+    const uint8_t* payload_data) {
+  if (length_ < kFuAHeaderSize) {
+    LOG(LS_ERROR) << "FU-A NAL units truncated.";
+    return false;
+  }
+  uint8_t fnri = payload_data[0] & (kFBit | kNriMask);
+  uint8_t original_nal_type = payload_data[1] & kTypeMask;
+  bool first_fragment = (payload_data[1] & kSBit) > 0;
+
+  if (first_fragment) {
+    offset_ = 0;
+    length_ -= kNalHeaderSize;
+    uint8_t original_nal_header = fnri | original_nal_type;
+    modified_buffer_.reset(new rtc::Buffer());
+    modified_buffer_->AppendData(payload_data + kNalHeaderSize, length_);
+    (*modified_buffer_)[0] = original_nal_header;
+  } else {
+    offset_ = kFuAHeaderSize;
+    length_ -= kFuAHeaderSize;
+  }
+
+  if (original_nal_type == kIdr) {
+    parsed_payload->frame_type = kVideoFrameKey;
+  } else {
+    parsed_payload->frame_type = kVideoFrameDelta;
+  }
+  parsed_payload->type.Video.width = 0;
+  parsed_payload->type.Video.height = 0;
+  parsed_payload->type.Video.codec = kRtpVideoH264;
+  parsed_payload->type.Video.isFirstPacket = first_fragment;
+  RTPVideoHeaderH264* h264_header =
+      &parsed_payload->type.Video.codecHeader.H264;
+  h264_header->packetization_type = kH264FuA;
+  h264_header->nalu_type = original_nal_type;
+  return true;
+}
+
 }  // namespace webrtc