Add H264 bitstream rewriting to limit frame reordering marker in header

webrtc/modules/rtp_rtcp/source/h264/bitstream_parser.cc

Index: webrtc/modules/rtp_rtcp/source/h264/bitstream_parser.cc
diff --git a/webrtc/modules/rtp_rtcp/source/h264/bitstream_parser.cc b/webrtc/modules/rtp_rtcp/source/h264/bitstream_parser.cc
new file mode 100644
index 0000000000000000000000000000000000000000..b4537275348d00675415e777e9ca291a58e9d344
--- /dev/null
+++ b/webrtc/modules/rtp_rtcp/source/h264/bitstream_parser.cc
@@ -0,0 +1,305 @@
+/*
+ *  Copyright (c) 2015 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/h264/bitstream_parser.h"
+
+#include <memory>
+#include <vector>
+
+#include "webrtc/base/bitbuffer.h"
+#include "webrtc/base/bytebuffer.h"
+#include "webrtc/base/checks.h"
+
+#include "webrtc/modules/rtp_rtcp/source/h264/h264_common.h"
+#include "webrtc/base/logging.h"
+
+namespace webrtc {
+
+#define RETURN_FALSE_ON_FAIL(x)       \
+  if (!(x)) {                         \
+    LOG_F(LS_ERROR) << "FAILED: " #x; \
+    return false;                     \
+  }
+
+H264BitstreamParser::H264BitstreamParser() {}
+H264BitstreamParser::~H264BitstreamParser() {}
+
+bool H264BitstreamParser::ParseNonParameterSetNalu(const uint8_t* source,
+                                                   size_t source_length,
+                                                   uint8_t nalu_type) {
+  RTC_CHECK(sps_);
+  RTC_CHECK(pps_);
+  last_slice_qp_delta_ = rtc::Optional<int32_t>();
+  std::unique_ptr<rtc::ByteBufferWriter> slice_rbsp(H264Common::ParseRbsp(
+      source + H264Common::kNaluHeaderAndTypeSize,
+      source_length - H264Common::kNaluHeaderAndTypeSize));
+  rtc::BitBuffer slice_reader(
+      reinterpret_cast<const uint8_t*>(slice_rbsp->Data()),
+      slice_rbsp->Length());
+  // Check to see if this is an IDR slice, which has an extra field to parse
+  // out.
+  bool is_idr = (source[H264Common::kNaluHeaderSize] & 0x0F) ==
+                H264Common::NaluType::kIdr;
+  uint8_t nal_ref_idc = (source[H264Common::kNaluHeaderSize] & 0x60) >> 5;
+  uint32_t golomb_tmp;
+  uint32_t bits_tmp;
+
+  // first_mb_in_slice: ue(v)
+  RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(&golomb_tmp));
+  // slice_type: ue(v)
+  uint32_t slice_type;
+  RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(&slice_type));
+  // slice_type's 5..9 range is used to indicate that all slices of a picture
+  // have the same value of slice_type % 5, we don't care about that, so we map
+  // to the corresponding 0..4 range.
+  slice_type %= 5;
+  // pic_parameter_set_id: ue(v)
+  RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(&golomb_tmp));
+  if (sps_->separate_colour_plane_flag == 1) {
+    // colour_plane_id
+    RETURN_FALSE_ON_FAIL(slice_reader.ReadBits(&bits_tmp, 2));
+  }
+  // frame_num: u(v)
+  // Represented by log2_max_frame_num_minus4 + 4 bits.
+  RETURN_FALSE_ON_FAIL(
+      slice_reader.ReadBits(&bits_tmp, sps_->log2_max_frame_num_minus4 + 4));
+  uint32_t field_pic_flag = 0;
+  if (sps_->frame_mbs_only_flag == 0) {
+    // field_pic_flag: u(1)
+    RETURN_FALSE_ON_FAIL(slice_reader.ReadBits(&field_pic_flag, 1));
+    if (field_pic_flag != 0) {
+      // bottom_field_flag: u(1)
+      RETURN_FALSE_ON_FAIL(slice_reader.ReadBits(&bits_tmp, 1));
+    }
+  }
+  if (is_idr) {
+    // idr_pic_id: ue(v)
+    RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(&golomb_tmp));
+  }
+  // pic_order_cnt_lsb: u(v)
+  // Represented by sps_.log2_max_pic_order_cnt_lsb_minus4 + 4 bits.
+  if (sps_->pic_order_cnt_type == 0) {
+    RETURN_FALSE_ON_FAIL(slice_reader.ReadBits(
+        &bits_tmp, sps_->log2_max_pic_order_cnt_lsb_minus4 + 4));
+    if (pps_->bottom_field_pic_order_in_frame_present_flag &&
+        field_pic_flag == 0) {
+      // delta_pic_order_cnt_bottom: se(v)
+      RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(&golomb_tmp));
+    }
+  }
+  if (sps_->pic_order_cnt_type == 1 &&
+      !sps_->delta_pic_order_always_zero_flag) {
+    // delta_pic_order_cnt[0]: se(v)
+    RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(&golomb_tmp));
+    if (pps_->bottom_field_pic_order_in_frame_present_flag && !field_pic_flag) {
+      // delta_pic_order_cnt[1]: se(v)
+      RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(&golomb_tmp));
+    }
+  }
+  if (pps_->redundant_pic_cnt_present_flag) {
+    // redundant_pic_cnt: ue(v)
+    RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(&golomb_tmp));
+  }
+  if (slice_type == H264Common::SliceType::B) {
+    // direct_spatial_mv_pred_flag: u(1)
+    RETURN_FALSE_ON_FAIL(slice_reader.ReadBits(&bits_tmp, 1));
+  }
+  switch (slice_type) {
+    case H264Common::SliceType::P:
+    case H264Common::SliceType::B:
+    case H264Common::SliceType::Sp:
+      uint32_t num_ref_idx_active_override_flag;
+      // num_ref_idx_active_override_flag: u(1)
+      RETURN_FALSE_ON_FAIL(
+          slice_reader.ReadBits(&num_ref_idx_active_override_flag, 1));
+      if (num_ref_idx_active_override_flag != 0) {
+        // num_ref_idx_l0_active_minus1: ue(v)
+        RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(&golomb_tmp));
+        if (slice_type == H264Common::SliceType::B) {
+          // num_ref_idx_l1_active_minus1: ue(v)
+          RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(&golomb_tmp));
+        }
+      }
+      break;
+    default:
+      break;
+  }
+  // assume nal_unit_type != 20 && nal_unit_type != 21:
+  RTC_CHECK_NE(nalu_type, 20);
+  RTC_CHECK_NE(nalu_type, 21);
+  // if (nal_unit_type == 20 || nal_unit_type == 21)
+  //   ref_pic_list_mvc_modification()
+  // else
+  {
+    // ref_pic_list_modification():
+    // |slice_type| checks here don't use named constants as they aren't named
+    // in the spec for this segment. Keeping them consistent makes it easier to
+    // verify that they are both the same.
+    if (slice_type % 5 != 2 && slice_type % 5 != 4) {
+      // ref_pic_list_modification_flag_l0: u(1)
+      uint32_t ref_pic_list_modification_flag_l0;
+      RETURN_FALSE_ON_FAIL(
+          slice_reader.ReadBits(&ref_pic_list_modification_flag_l0, 1));
+      if (ref_pic_list_modification_flag_l0) {
+        uint32_t modification_of_pic_nums_idc;
+        do {
+          // modification_of_pic_nums_idc: ue(v)
+          RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(
+              &modification_of_pic_nums_idc));
+          if (modification_of_pic_nums_idc == 0 ||
+              modification_of_pic_nums_idc == 1) {
+            // abs_diff_pic_num_minus1: ue(v)
+            RETURN_FALSE_ON_FAIL(
+                slice_reader.ReadExponentialGolomb(&golomb_tmp));
+          } else if (modification_of_pic_nums_idc == 2) {
+            // long_term_pic_num: ue(v)
+            RETURN_FALSE_ON_FAIL(
+                slice_reader.ReadExponentialGolomb(&golomb_tmp));
+          }
+        } while (modification_of_pic_nums_idc != 3);
+      }
+    }
+    if (slice_type % 5 == 1) {
+      // ref_pic_list_modification_flag_l1: u(1)
+      uint32_t ref_pic_list_modification_flag_l1;
+      RETURN_FALSE_ON_FAIL(
+          slice_reader.ReadBits(&ref_pic_list_modification_flag_l1, 1));
+      if (ref_pic_list_modification_flag_l1) {
+        uint32_t modification_of_pic_nums_idc;
+        do {
+          // modification_of_pic_nums_idc: ue(v)
+          RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(
+              &modification_of_pic_nums_idc));
+          if (modification_of_pic_nums_idc == 0 ||
+              modification_of_pic_nums_idc == 1) {
+            // abs_diff_pic_num_minus1: ue(v)
+            RETURN_FALSE_ON_FAIL(
+                slice_reader.ReadExponentialGolomb(&golomb_tmp));
+          } else if (modification_of_pic_nums_idc == 2) {
+            // long_term_pic_num: ue(v)
+            RETURN_FALSE_ON_FAIL(
+                slice_reader.ReadExponentialGolomb(&golomb_tmp));
+          }
+        } while (modification_of_pic_nums_idc != 3);
+      }
+    }
+  }
+  // TODO(pbos): Do we need support for pred_weight_table()?
+  RTC_CHECK(!(
+      (pps_->weighted_pred_flag && (slice_type == H264Common::SliceType::P ||
+                                    slice_type == H264Common::SliceType::Sp)) ||
+      (pps_->weighted_bipred_idc != 0 &&
+       slice_type == H264Common::SliceType::B)))
+      << "Missing support for pred_weight_table().";
+  // if ((weighted_pred_flag && (slice_type == P || slice_type == SP)) ||
+  //    (weighted_bipred_idc == 1 && slice_type == B)) {
+  //  pred_weight_table()
+  // }
+  if (nal_ref_idc != 0) {
+    // dec_ref_pic_marking():
+    if (is_idr) {
+      // no_output_of_prior_pics_flag: u(1)
+      // long_term_reference_flag: u(1)
+      RETURN_FALSE_ON_FAIL(slice_reader.ReadBits(&bits_tmp, 2));
+    } else {
+      // adaptive_ref_pic_marking_mode_flag: u(1)
+      uint32_t adaptive_ref_pic_marking_mode_flag;
+      RETURN_FALSE_ON_FAIL(
+          slice_reader.ReadBits(&adaptive_ref_pic_marking_mode_flag, 1));
+      if (adaptive_ref_pic_marking_mode_flag) {
+        uint32_t memory_management_control_operation;
+        do {
+          // memory_management_control_operation: ue(v)
+          RETURN_FALSE_ON_FAIL(slice_reader.ReadExponentialGolomb(
+              &memory_management_control_operation));
+          if (memory_management_control_operation == 1 ||
+              memory_management_control_operation == 3) {
+            // difference_of_pic_nums_minus1: ue(v)
+            RETURN_FALSE_ON_FAIL(
+                slice_reader.ReadExponentialGolomb(&golomb_tmp));
+          }
+          if (memory_management_control_operation == 2) {
+            // long_term_pic_num: ue(v)
+            RETURN_FALSE_ON_FAIL(
+                slice_reader.ReadExponentialGolomb(&golomb_tmp));
+          }
+          if (memory_management_control_operation == 3 ||
+              memory_management_control_operation == 6) {
+            // long_term_frame_idx: ue(v)
+            RETURN_FALSE_ON_FAIL(
+                slice_reader.ReadExponentialGolomb(&golomb_tmp));
+          }
+          if (memory_management_control_operation == 4) {
+            // max_long_term_frame_idx_plus1: ue(v)
+            RETURN_FALSE_ON_FAIL(
+                slice_reader.ReadExponentialGolomb(&golomb_tmp));
+          }
+        } while (memory_management_control_operation != 0);
+      }
+    }
+  }
+  // cabac not supported: entropy_coding_mode_flag == 0 asserted above.
+  // if (entropy_coding_mode_flag && slice_type != I && slice_type != SI)
+  //   cabac_init_idc
+  int32_t last_slice_qp_delta;
+  RETURN_FALSE_ON_FAIL(
+      slice_reader.ReadSignedExponentialGolomb(&last_slice_qp_delta));
+  last_slice_qp_delta_ = rtc::Optional<int32_t>(last_slice_qp_delta);
+  return true;
+}
+
+void H264BitstreamParser::ParseSlice(const uint8_t* slice, size_t length) {
+  // Parse out the RBSP. It should be small, so it's OK that we create a copy.
+  // We'll eventually write this back.
+  std::unique_ptr<rtc::ByteBufferWriter> rbsp(
+      H264Common::ParseRbsp(slice + H264Common::kNaluHeaderAndTypeSize,
+                            length - H264Common::kNaluHeaderAndTypeSize));
+  const uint8_t* parsed_data = reinterpret_cast<const uint8_t*>(rbsp->Data());
+  const size_t parsed_length = rbsp->Length();
+  H264Common::NaluType nalu_type = H264Common::ParseNaluType(slice[4]);
+  switch (nalu_type) {
+    case H264Common::NaluType::kSps: {
+      SpsParser sps_parser(parsed_data, parsed_length);
+      if (!sps_parser.Parse())
+        FATAL() << "Unable to parse SPS from H264 bitstream.";
+      sps_ = rtc::Optional<SpsParser::SpsState>(sps_parser.GetState());
+      break;
+    }
+    case H264Common::NaluType::kPps: {
+      PpsParser pps_parser(parsed_data, parsed_length);
+      if (!pps_parser.Parse())
+        FATAL() << "Unable to parse PPS from H264 bitstream.";
+      pps_ = rtc::Optional<PpsParser::PpsState>(pps_parser.GetState());
+      break;
+    }
+    default:
+      RTC_CHECK(ParseNonParameterSetNalu(slice, length, nalu_type))
+          << "Failed to parse picture slice.";
+      break;
+  }
+}
+
+void H264BitstreamParser::ParseBitstream(const uint8_t* bitstream,
+                                         size_t length) {
+  std::vector<size_t> offsets =
+      H264Common::FindNaluStartOffsets(bitstream, length);
+  RTC_CHECK(!offsets.empty());
+  offsets.push_back(length);  // End for the last slice.
+  for (size_t i = 0; i < offsets.size() - 1; ++i)
+    ParseSlice(bitstream + offsets[i], offsets[i + 1] - offsets[i]);
+}
+
+bool H264BitstreamParser::GetLastSliceQp(int* qp) const {
+  if (!last_slice_qp_delta_ || !pps_)
+    return false;
+  *qp = 26 + pps_->pic_init_qp_minus26 + *last_slice_qp_delta_;
+  return true;
+}
+
+}  // namespace webrtc