Support MPEG1 audio in the MPEG2-TS stream parser.

media/formats/mp2t/es_parser_mpeg1audio.cc

+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "media/formats/mp2t/es_parser_mpeg1audio.h"
+
+#include <list>
+
+#include "base/basictypes.h"
+#include "base/bind.h"
+#include "base/logging.h"
+#include "base/strings/string_number_conversions.h"
+#include "media/base/audio_timestamp_helper.h"
+#include "media/base/bit_reader.h"
+#include "media/base/buffers.h"
+#include "media/base/channel_layout.h"
+#include "media/base/stream_parser_buffer.h"
+#include "media/formats/common/offset_byte_queue.h"
+#include "media/formats/mp2t/mp2t_common.h"
+#include "media/formats/mpeg/mpeg1_audio_stream_parser.h"
+
+namespace media {
+namespace mp2t {
+
+static void DummyMediaLog(const std::string& s) {

[wolenetz, 2014/09/13 00:06:38]

nit: mp2t_stream_parser has a valid log_cb passed during Init(). During
RegisterPes(), pass it to this es parser during construction? Seems better to
include than defer as TODO. Or am I missing some complexity or other breakage
here? I've no strong feeling other than this looks like possible make-work.

[damienv1, 2014/09/13 01:27:47]

Done.

+}
+
+struct EsParserMpeg1Audio::Mpeg1AudioFrame {
+  // Pointer to the ES data.
+  const uint8* data;
+
+  // Frame size.
+  int size;
+
+  // Number of samples in the frame.
+  int sample_count;
+
+  // Frame offset in the ES queue.
+  int64 queue_offset;
+};
+
+// TODO(damienv): Properly supports media logs.
+EsParserMpeg1Audio::EsParserMpeg1Audio(
+    const NewAudioConfigCB& new_audio_config_cb,
+    const EmitBufferCB& emit_buffer_cb)
+  : log_cb_(base::Bind(&DummyMediaLog)),
+    new_audio_config_cb_(new_audio_config_cb),
+    emit_buffer_cb_(emit_buffer_cb) {
+}
+
+EsParserMpeg1Audio::~EsParserMpeg1Audio() {
+}
+
+bool EsParserMpeg1Audio::ParseFromEsQueue() {
+  // Look for every MPEG1 audio frame in the ES buffer.
+  Mpeg1AudioFrame mpeg1audio_frame;
+  while (LookForMpeg1AudioFrame(&mpeg1audio_frame)) {
+    // Update the audio configuration if needed.
+    DCHECK_GE(mpeg1audio_frame.size, MPEG1AudioStreamParser::kHeaderSize);
+    if (!UpdateAudioConfiguration(mpeg1audio_frame.data))
+      return false;
+
+    // Get the PTS & the duration of this access unit.
+    TimingDesc current_timing_desc =
+        GetTimingDescriptor(mpeg1audio_frame.queue_offset);
+    if (current_timing_desc.pts != kNoTimestamp())
+      audio_timestamp_helper_->SetBaseTimestamp(current_timing_desc.pts);
+
+    if (audio_timestamp_helper_->base_timestamp() == kNoTimestamp()) {
+      DVLOG(1) << "Audio frame with unknown timestamp";
+      return false;
+    }
+    base::TimeDelta current_pts = audio_timestamp_helper_->GetTimestamp();
+    base::TimeDelta frame_duration =
+        audio_timestamp_helper_->GetFrameDuration(
+            mpeg1audio_frame.sample_count);
+
+    // Emit an audio frame.
+    bool is_key_frame = true;
+
+    // TODO(wolenetz/acolwell): Validate and use a common cross-parser TrackId
+    // type and allow multiple audio tracks. See https://crbug.com/341581.
+    scoped_refptr<StreamParserBuffer> stream_parser_buffer =
+        StreamParserBuffer::CopyFrom(
+            mpeg1audio_frame.data,
+            mpeg1audio_frame.size,
+            is_key_frame,
+            DemuxerStream::AUDIO, 0);
+    stream_parser_buffer->set_timestamp(current_pts);
+    stream_parser_buffer->set_duration(frame_duration);
+    emit_buffer_cb_.Run(stream_parser_buffer);
+
+    // Update the PTS of the next frame.
+    audio_timestamp_helper_->AddFrames(mpeg1audio_frame.sample_count);
+
+    // Skip the current frame.
+    SkipMpeg1AudioFrame(mpeg1audio_frame);
+  }
+
+  return true;
+}
+
+void EsParserMpeg1Audio::Flush() {
+}
+
+void EsParserMpeg1Audio::ResetInternal() {
+  last_audio_decoder_config_ = AudioDecoderConfig();
+}
+
+bool EsParserMpeg1Audio::LookForMpeg1AudioFrame(
+    Mpeg1AudioFrame* mpeg1audio_frame) {
+  int es_size;
+  const uint8* es;
+  es_queue_->Peek(&es, &es_size);
+
+  int max_offset = es_size - MPEG1AudioStreamParser::kHeaderSize;
+  if (max_offset <= 0)
+    return false;
+
+  for (int offset = 0; offset < max_offset; offset++) {
+    const uint8* cur_buf = &es[offset];
+    if (cur_buf[0] != 0xff)
+      continue;
+
+    int remaining_size = es_size - offset;
+    DCHECK_GE(remaining_size, MPEG1AudioStreamParser::kHeaderSize);
+    MPEG1AudioStreamParser::Header header;
+    if (!MPEG1AudioStreamParser::ParseHeader(log_cb_, cur_buf, &header))
+      continue;
+
+    if (remaining_size < header.frame_size) {

[wolenetz, 2014/09/13 00:06:38]

|remaining_size| includes the size of the header. Is |header.frame_size|
expected to include the size of the frame *and* the header?

[damienv1, 2014/09/13 01:27:47]

Yes. An MPEG-1 audio frame includes the header.
This is the size of the entire frame.

+      // Not a full frame: will resume when we have more data.
+      // Remove all the bytes located before the frame header,
+      // these bytes will not be used anymore.
+      es_queue_->Pop(offset);
+      return false;
+    }
+
+    // Check whether there is another frame
+    // |frame_size| apart from the current one.
+    if (remaining_size >= header.frame_size + 2 &&

[wolenetz, 2014/09/13 00:06:38]

nit: +2 or +1?

[damienv1, 2014/09/13 01:27:47]

Right, that could be +1.

+        cur_buf[header.frame_size] != 0xff) {
+      continue;
+    }
+
+    es_queue_->Pop(offset);
+    es_queue_->Peek(&mpeg1audio_frame->data, &es_size);
+    mpeg1audio_frame->queue_offset = es_queue_->head();
+    mpeg1audio_frame->size = header.frame_size;
+    mpeg1audio_frame->sample_count = header.sample_count;
+    DVLOG(LOG_LEVEL_ES)
+        << "MPEG1 audio syncword @ pos=" << mpeg1audio_frame->queue_offset
+        << " frame_size=" << mpeg1audio_frame->size;
+    DVLOG(LOG_LEVEL_ES)
+        << "MPEG1 audio header: "
+        << base::HexEncode(mpeg1audio_frame->data,
+                           MPEG1AudioStreamParser::kHeaderSize);
+    return true;
+  }
+
+  es_queue_->Pop(max_offset);
+  return false;
+}
+
+bool EsParserMpeg1Audio::UpdateAudioConfiguration(
+    const uint8* mpeg1audio_header) {
+  MPEG1AudioStreamParser::Header header;
+  if (!MPEG1AudioStreamParser::ParseHeader(log_cb_,
+                                           mpeg1audio_header,
+                                           &header)) {
+    return false;
+  }
+
+  // TODO(damienv): Verify whether Android playback requires the extra data
+  // field for Mpeg1 audio. If yes, we should generate this field.
+  AudioDecoderConfig audio_decoder_config(
+      kCodecMP3,
+      kSampleFormatS16,
+      header.channel_layout,
+      header.sample_rate,
+      NULL, 0,
+      false);
+
+  if (!audio_decoder_config.Matches(last_audio_decoder_config_)) {
+    DVLOG(1) << "Sampling frequency: " << header.sample_rate;

[wolenetz, 2014/09/13 00:06:38]

nit: also log the channel layout?

[damienv1, 2014/09/13 01:27:47]

Done.

+    // Reset the timestamp helper to use a new time scale.
+    if (audio_timestamp_helper_ &&
+        audio_timestamp_helper_->base_timestamp() != kNoTimestamp()) {
+      base::TimeDelta base_timestamp = audio_timestamp_helper_->GetTimestamp();
+      audio_timestamp_helper_.reset(
+        new AudioTimestampHelper(header.sample_rate));
+      audio_timestamp_helper_->SetBaseTimestamp(base_timestamp);
+    } else {
+      audio_timestamp_helper_.reset(
+          new AudioTimestampHelper(header.sample_rate));
+    }
+    // Audio config notification.
+    last_audio_decoder_config_ = audio_decoder_config;
+    new_audio_config_cb_.Run(audio_decoder_config);
+  }
+
+  return true;
+}
+
+void EsParserMpeg1Audio::SkipMpeg1AudioFrame(
+    const Mpeg1AudioFrame& mpeg1audio_frame) {
+  DCHECK_EQ(mpeg1audio_frame.queue_offset, es_queue_->head());
+  es_queue_->Pop(mpeg1audio_frame.size);
+}
+
+}  // namespace mp2t
+}  // namespace media