OLD | NEW |
(Empty) | |
| 1 // Copyright (c) 2013 The Chromium Authors. All rights reserved. |
| 2 // Use of this source code is governed by a BSD-style license that can be |
| 3 // found in the LICENSE file. |
| 4 |
| 5 #include "media/mp2t/es_parser_adts.h" |
| 6 |
| 7 #include <list> |
| 8 |
| 9 #include "base/basictypes.h" |
| 10 #include "base/logging.h" |
| 11 #include "base/strings/string_number_conversions.h" |
| 12 #include "media/base/audio_decoder_config.h" |
| 13 #include "media/base/audio_timestamp_helper.h" |
| 14 #include "media/base/bit_reader.h" |
| 15 #include "media/base/buffers.h" |
| 16 #include "media/base/channel_layout.h" |
| 17 #include "media/base/stream_parser_buffer.h" |
| 18 #include "media/mp2t/mp2t_common.h" |
| 19 |
| 20 // Adts header is at least 7 bytes (can be 9 bytes). |
| 21 static const int kAdtsHeaderMinSize = 7; |
| 22 |
| 23 static const int adts_frequency_table[16] = { |
| 24 96000, |
| 25 88200, |
| 26 64000, |
| 27 48000, |
| 28 44100, |
| 29 32000, |
| 30 24000, |
| 31 22050, |
| 32 16000, |
| 33 12000, |
| 34 11025, |
| 35 8000, |
| 36 7350, |
| 37 0, |
| 38 0, |
| 39 0, |
| 40 }; |
| 41 static const int kMaxSupportedFrequencyIndex = 12; |
| 42 |
| 43 static media::ChannelLayout adts_channel_layout[8] = { |
| 44 media::CHANNEL_LAYOUT_NONE, |
| 45 media::CHANNEL_LAYOUT_MONO, |
| 46 media::CHANNEL_LAYOUT_STEREO, |
| 47 media::CHANNEL_LAYOUT_SURROUND, |
| 48 media::CHANNEL_LAYOUT_4_0, |
| 49 media::CHANNEL_LAYOUT_5_0_BACK, |
| 50 media::CHANNEL_LAYOUT_5_1_BACK, |
| 51 media::CHANNEL_LAYOUT_7_1, |
| 52 }; |
| 53 |
| 54 // Number of samples per frame. |
| 55 static const int kNumberSamplesPerAACFrame = 1024; |
| 56 |
| 57 static int ExtractAdtsFrameSize(const uint8* adts_header) { |
| 58 return ((static_cast<int>(adts_header[5]) >> 5) | |
| 59 (static_cast<int>(adts_header[4]) << 3) | |
| 60 ((static_cast<int>(adts_header[3]) & 0x3) << 11)); |
| 61 } |
| 62 |
| 63 static int ExtractAdtsFrequencyIndex(const uint8* adts_header) { |
| 64 return ((adts_header[2] >> 2) & 0xf); |
| 65 } |
| 66 |
| 67 static int ExtractAdtsChannelConfig(const uint8* adts_header) { |
| 68 return (((adts_header[3] >> 6) & 0x3) | |
| 69 ((adts_header[2] & 0x1) << 2)); |
| 70 } |
| 71 |
| 72 // Look for an ADTS syncword. |
| 73 // |new_pos| returns |
| 74 // - either the byte position of the ADTS frame (if found) |
| 75 // - or the byte position of 1st byte that was not processed (if not found). |
| 76 // |frame_sz| returns the size of the ADTS frame (if found). |
| 77 // Return whether a syncword was found. |
| 78 static bool LookForSyncWord(const uint8* raw_es, int raw_es_size, |
| 79 int pos, |
| 80 int* new_pos, int* frame_sz) { |
| 81 int max_offset = raw_es_size - kAdtsHeaderMinSize; |
| 82 if (max_offset < 0) |
| 83 max_offset = 0; |
| 84 |
| 85 for (int offset = pos; offset < max_offset; offset++) { |
| 86 const uint8* cur_buf = &raw_es[offset]; |
| 87 |
| 88 if ((cur_buf[0] != 0xff) || ((cur_buf[1] & 0xf6) != 0xf0)) |
| 89 // The first 12 bits must be 1. |
| 90 // The layer field (2 bits) must be set to 0. |
| 91 continue; |
| 92 |
| 93 int frame_size = ExtractAdtsFrameSize(cur_buf); |
| 94 if (frame_size < kAdtsHeaderMinSize) |
| 95 // Too short to be an ADTS frame. |
| 96 continue; |
| 97 |
| 98 // Check whether there is another frame |
| 99 // |size| apart from the current one. |
| 100 int remaining_size = raw_es_size - offset; |
| 101 if (remaining_size >= frame_size + 2) { |
| 102 if ((cur_buf[frame_size] != 0xff) || |
| 103 (cur_buf[frame_size + 1] & 0xf6) != 0xf0) |
| 104 continue; |
| 105 } |
| 106 |
| 107 *new_pos = offset; |
| 108 *frame_sz = frame_size; |
| 109 return true; |
| 110 } |
| 111 |
| 112 *new_pos = max_offset; |
| 113 return false; |
| 114 } |
| 115 |
| 116 namespace media { |
| 117 namespace mp2t { |
| 118 |
| 119 EsParserAdts::EsParserAdts( |
| 120 NewAudioConfigCB new_audio_config_cb, |
| 121 EmitBufferCB emit_buffer_cb) |
| 122 : new_audio_config_cb_(new_audio_config_cb), |
| 123 emit_buffer_cb_(emit_buffer_cb) { |
| 124 } |
| 125 |
| 126 EsParserAdts::~EsParserAdts() { |
| 127 } |
| 128 |
| 129 bool EsParserAdts::Parse(const uint8* buf, int size, |
| 130 base::TimeDelta pts, |
| 131 base::TimeDelta dts) { |
| 132 int raw_es_size = 0; |
| 133 const uint8* raw_es = NULL; |
| 134 |
| 135 // The incoming PTS applies to the access unit that comes just after |
| 136 // the beginning of |buf|. |
| 137 if (pts != kNoTimestamp()) { |
| 138 es_byte_queue_.Peek(&raw_es, &raw_es_size); |
| 139 pts_list_.push_back(EsPts(raw_es_size, pts)); |
| 140 } |
| 141 |
| 142 // Copy the input data to the ES buffer. |
| 143 es_byte_queue_.Push(buf, size); |
| 144 es_byte_queue_.Peek(&raw_es, &raw_es_size); |
| 145 |
| 146 // Look for every ADTS frame in the ES buffer starting at offset = 0 |
| 147 int es_position = 0; |
| 148 int frame_size; |
| 149 while (LookForSyncWord(raw_es, raw_es_size, es_position, |
| 150 &es_position, &frame_size)) { |
| 151 DVLOG(LOG_LEVEL_ES) << "ADTS syncword @ pos=" << es_position |
| 152 << " frame_size=" << frame_size; |
| 153 DVLOG(LOG_LEVEL_ES) << "ADTS header: " |
| 154 << base::HexEncode(&raw_es[es_position], 7); |
| 155 |
| 156 // Do not process the frame if this one is a partial frame. |
| 157 int remaining_size = raw_es_size - es_position; |
| 158 if (frame_size > remaining_size) |
| 159 break; |
| 160 |
| 161 // Update the audio configuration if needed. |
| 162 DCHECK_GE(frame_size, kAdtsHeaderMinSize); |
| 163 if (!UpdateAudioConfiguration(&raw_es[es_position])) |
| 164 return false; |
| 165 |
| 166 // Get the PTS & the duration of this access unit. |
| 167 DCHECK(audio_timestamp_helper_); |
| 168 while (!pts_list_.empty() && |
| 169 pts_list_.front().first <= es_position) { |
| 170 audio_timestamp_helper_->SetBaseTimestamp(pts_list_.front().second); |
| 171 pts_list_.pop_front(); |
| 172 } |
| 173 |
| 174 base::TimeDelta current_pts = audio_timestamp_helper_->GetTimestamp(); |
| 175 base::TimeDelta frame_duration = |
| 176 audio_timestamp_helper_->GetFrameDuration(kNumberSamplesPerAACFrame); |
| 177 |
| 178 // Emit an audio frame. |
| 179 bool is_key_frame = true; |
| 180 scoped_refptr<StreamParserBuffer> stream_parser_buffer = |
| 181 StreamParserBuffer::CopyFrom( |
| 182 &raw_es[es_position], |
| 183 frame_size, |
| 184 is_key_frame); |
| 185 stream_parser_buffer->SetDecodeTimestamp(current_pts); |
| 186 stream_parser_buffer->set_timestamp(current_pts); |
| 187 stream_parser_buffer->set_duration(frame_duration); |
| 188 emit_buffer_cb_.Run(stream_parser_buffer); |
| 189 |
| 190 // Update the PTS of the next frame. |
| 191 audio_timestamp_helper_->AddFrames(kNumberSamplesPerAACFrame); |
| 192 |
| 193 // Skip the current frame. |
| 194 es_position += frame_size; |
| 195 } |
| 196 |
| 197 // Discard all the bytes that have been processed. |
| 198 DiscardEs(es_position); |
| 199 |
| 200 return true; |
| 201 } |
| 202 |
| 203 void EsParserAdts::Flush() { |
| 204 // All the complete frames have been emitted, |
| 205 // so just clear the ES buffer. |
| 206 es_byte_queue_.Reset(); |
| 207 pts_list_.clear(); |
| 208 } |
| 209 |
| 210 void EsParserAdts::Reset() { |
| 211 es_byte_queue_.Reset(); |
| 212 pts_list_.clear(); |
| 213 last_audio_decoder_config_ = AudioDecoderConfig(); |
| 214 } |
| 215 |
| 216 bool EsParserAdts::UpdateAudioConfiguration(const uint8* adts_header) { |
| 217 int frequency_index = ExtractAdtsFrequencyIndex(adts_header); |
| 218 if (frequency_index > kMaxSupportedFrequencyIndex) { |
| 219 // Frequency index 13 & 14 are reserved |
| 220 // while 15 means that the frequency is explicitly written |
| 221 // (not supported). |
| 222 return false; |
| 223 } |
| 224 |
| 225 // TODO(damienv): support HE-AAC frequency doubling (SBR) |
| 226 // based on the incoming ADTS profile. |
| 227 int samples_per_second = adts_frequency_table[frequency_index]; |
| 228 int channel_configuration = ExtractAdtsChannelConfig(adts_header); |
| 229 int adts_profile = (adts_header[2] >> 6) & 0x3; |
| 230 |
| 231 AudioDecoderConfig audio_decoder_config( |
| 232 kCodecAAC, |
| 233 kSampleFormatS16, |
| 234 adts_channel_layout[channel_configuration], |
| 235 samples_per_second, |
| 236 NULL, 0, |
| 237 false); |
| 238 |
| 239 if (!audio_decoder_config.Matches(last_audio_decoder_config_)) { |
| 240 DVLOG(1) << "Sampling frequency: " << samples_per_second; |
| 241 DVLOG(1) << "Channel config: " << channel_configuration; |
| 242 DVLOG(1) << "Adts profile: " << adts_profile; |
| 243 // Reset the timestamp helper to use a new time scale. |
| 244 if (audio_timestamp_helper_) { |
| 245 base::TimeDelta base_timestamp = audio_timestamp_helper_->GetTimestamp(); |
| 246 audio_timestamp_helper_.reset( |
| 247 new AudioTimestampHelper(samples_per_second)); |
| 248 audio_timestamp_helper_->SetBaseTimestamp(base_timestamp); |
| 249 } else { |
| 250 audio_timestamp_helper_.reset( |
| 251 new AudioTimestampHelper(samples_per_second)); |
| 252 } |
| 253 // Audio config notification. |
| 254 last_audio_decoder_config_ = audio_decoder_config; |
| 255 new_audio_config_cb_.Run(audio_decoder_config); |
| 256 } |
| 257 |
| 258 return true; |
| 259 } |
| 260 |
| 261 void EsParserAdts::DiscardEs(int nbytes) { |
| 262 DCHECK_GE(nbytes, 0); |
| 263 if (nbytes <= 0) |
| 264 return; |
| 265 |
| 266 // Adjust the ES position of each PTS. |
| 267 for (EsPtsList::iterator it = pts_list_.begin(); it != pts_list_.end(); ++it) |
| 268 it->first -= nbytes; |
| 269 |
| 270 // Discard |nbytes| of ES. |
| 271 es_byte_queue_.Pop(nbytes); |
| 272 } |
| 273 |
| 274 } // namespace mp2t |
| 275 } // namespace media |
| 276 |
OLD | NEW |