1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
|
// Copyright (c) 2012 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/filters/ffmpeg_audio_decoder.h"
#include "base/bind.h"
#include "base/callback_helpers.h"
#include "base/location.h"
#include "base/message_loop_proxy.h"
#include "media/base/audio_bus.h"
#include "media/base/audio_decoder_config.h"
#include "media/base/audio_timestamp_helper.h"
#include "media/base/bind_to_loop.h"
#include "media/base/data_buffer.h"
#include "media/base/decoder_buffer.h"
#include "media/base/demuxer.h"
#include "media/base/pipeline.h"
#include "media/ffmpeg/ffmpeg_common.h"
#include "media/filters/ffmpeg_glue.h"
namespace media {
// Helper structure for managing multiple decoded audio frames per packet.
struct QueuedAudioBuffer {
AudioDecoder::Status status;
scoped_refptr<DataBuffer> buffer;
};
// Returns true if the decode result was end of stream.
static inline bool IsEndOfStream(int result, int decoded_size,
const scoped_refptr<DecoderBuffer>& input) {
// Three conditions to meet to declare end of stream for this decoder:
// 1. FFmpeg didn't read anything.
// 2. FFmpeg didn't output anything.
// 3. An end of stream buffer is received.
return result == 0 && decoded_size == 0 && input->IsEndOfStream();
}
FFmpegAudioDecoder::FFmpegAudioDecoder(
const scoped_refptr<base::MessageLoopProxy>& message_loop)
: message_loop_(message_loop),
codec_context_(NULL),
bits_per_channel_(0),
channel_layout_(CHANNEL_LAYOUT_NONE),
channels_(0),
samples_per_second_(0),
av_sample_format_(0),
bytes_per_frame_(0),
last_input_timestamp_(kNoTimestamp()),
output_bytes_to_drop_(0),
av_frame_(NULL) {
}
void FFmpegAudioDecoder::Initialize(
const scoped_refptr<DemuxerStream>& stream,
const PipelineStatusCB& status_cb,
const StatisticsCB& statistics_cb) {
DCHECK(message_loop_->BelongsToCurrentThread());
PipelineStatusCB initialize_cb = BindToCurrentLoop(status_cb);
FFmpegGlue::InitializeFFmpeg();
if (demuxer_stream_) {
// TODO(scherkus): initialization currently happens more than once in
// PipelineIntegrationTest.BasicPlayback.
LOG(ERROR) << "Initialize has already been called.";
CHECK(false);
}
demuxer_stream_ = stream;
if (!ConfigureDecoder()) {
status_cb.Run(DECODER_ERROR_NOT_SUPPORTED);
return;
}
statistics_cb_ = statistics_cb;
initialize_cb.Run(PIPELINE_OK);
}
void FFmpegAudioDecoder::Read(const ReadCB& read_cb) {
DCHECK(message_loop_->BelongsToCurrentThread());
DCHECK(!read_cb.is_null());
CHECK(read_cb_.is_null()) << "Overlapping decodes are not supported.";
read_cb_ = BindToCurrentLoop(read_cb);
// If we don't have any queued audio from the last packet we decoded, ask for
// more data from the demuxer to satisfy this read.
if (queued_audio_.empty()) {
ReadFromDemuxerStream();
return;
}
base::ResetAndReturn(&read_cb_).Run(
queued_audio_.front().status, queued_audio_.front().buffer);
queued_audio_.pop_front();
}
int FFmpegAudioDecoder::bits_per_channel() {
DCHECK(message_loop_->BelongsToCurrentThread());
return bits_per_channel_;
}
ChannelLayout FFmpegAudioDecoder::channel_layout() {
DCHECK(message_loop_->BelongsToCurrentThread());
return channel_layout_;
}
int FFmpegAudioDecoder::samples_per_second() {
DCHECK(message_loop_->BelongsToCurrentThread());
return samples_per_second_;
}
void FFmpegAudioDecoder::Reset(const base::Closure& closure) {
DCHECK(message_loop_->BelongsToCurrentThread());
base::Closure reset_cb = BindToCurrentLoop(closure);
avcodec_flush_buffers(codec_context_);
ResetTimestampState();
queued_audio_.clear();
reset_cb.Run();
}
FFmpegAudioDecoder::~FFmpegAudioDecoder() {
// TODO(scherkus): should we require Stop() to be called? this might end up
// getting called on a random thread due to refcounting.
ReleaseFFmpegResources();
}
void FFmpegAudioDecoder::ReadFromDemuxerStream() {
DCHECK(!read_cb_.is_null());
demuxer_stream_->Read(base::Bind(&FFmpegAudioDecoder::BufferReady, this));
}
void FFmpegAudioDecoder::BufferReady(
DemuxerStream::Status status,
const scoped_refptr<DecoderBuffer>& input) {
DCHECK(message_loop_->BelongsToCurrentThread());
DCHECK(!read_cb_.is_null());
DCHECK(queued_audio_.empty());
DCHECK_EQ(status != DemuxerStream::kOk, !input) << status;
if (status == DemuxerStream::kAborted) {
DCHECK(!input);
base::ResetAndReturn(&read_cb_).Run(kAborted, NULL);
return;
}
if (status == DemuxerStream::kConfigChanged) {
DCHECK(!input);
// Send a "end of stream" buffer to the decode loop
// to output any remaining data still in the decoder.
RunDecodeLoop(DecoderBuffer::CreateEOSBuffer(), true);
DVLOG(1) << "Config changed.";
if (!ConfigureDecoder()) {
base::ResetAndReturn(&read_cb_).Run(kDecodeError, NULL);
return;
}
ResetTimestampState();
if (queued_audio_.empty()) {
ReadFromDemuxerStream();
return;
}
base::ResetAndReturn(&read_cb_).Run(
queued_audio_.front().status, queued_audio_.front().buffer);
queued_audio_.pop_front();
return;
}
DCHECK_EQ(status, DemuxerStream::kOk);
DCHECK(input);
// Make sure we are notified if http://crbug.com/49709 returns. Issue also
// occurs with some damaged files.
if (!input->IsEndOfStream() && input->GetTimestamp() == kNoTimestamp() &&
output_timestamp_helper_->base_timestamp() == kNoTimestamp()) {
DVLOG(1) << "Received a buffer without timestamps!";
base::ResetAndReturn(&read_cb_).Run(kDecodeError, NULL);
return;
}
bool is_vorbis = codec_context_->codec_id == CODEC_ID_VORBIS;
if (!input->IsEndOfStream()) {
if (last_input_timestamp_ == kNoTimestamp()) {
if (is_vorbis && (input->GetTimestamp() < base::TimeDelta())) {
// Dropping frames for negative timestamps as outlined in section A.2
// in the Vorbis spec. http://xiph.org/vorbis/doc/Vorbis_I_spec.html
int frames_to_drop = floor(
0.5 + -input->GetTimestamp().InSecondsF() * samples_per_second_);
output_bytes_to_drop_ = bytes_per_frame_ * frames_to_drop;
} else {
last_input_timestamp_ = input->GetTimestamp();
}
} else if (input->GetTimestamp() != kNoTimestamp()) {
if (input->GetTimestamp() < last_input_timestamp_) {
base::TimeDelta diff = input->GetTimestamp() - last_input_timestamp_;
DVLOG(1) << "Input timestamps are not monotonically increasing! "
<< " ts " << input->GetTimestamp().InMicroseconds() << " us"
<< " diff " << diff.InMicroseconds() << " us";
base::ResetAndReturn(&read_cb_).Run(kDecodeError, NULL);
return;
}
last_input_timestamp_ = input->GetTimestamp();
}
}
RunDecodeLoop(input, false);
// We exhausted the provided packet, but it wasn't enough for a frame. Ask
// for more data in order to fulfill this read.
if (queued_audio_.empty()) {
ReadFromDemuxerStream();
return;
}
// Execute callback to return the first frame we decoded.
base::ResetAndReturn(&read_cb_).Run(
queued_audio_.front().status, queued_audio_.front().buffer);
queued_audio_.pop_front();
}
bool FFmpegAudioDecoder::ConfigureDecoder() {
const AudioDecoderConfig& config = demuxer_stream_->audio_decoder_config();
if (!config.IsValidConfig()) {
DLOG(ERROR) << "Invalid audio stream -"
<< " codec: " << config.codec()
<< " channel layout: " << config.channel_layout()
<< " bits per channel: " << config.bits_per_channel()
<< " samples per second: " << config.samples_per_second();
return false;
}
if (config.is_encrypted()) {
DLOG(ERROR) << "Encrypted audio stream not supported";
return false;
}
if (codec_context_ &&
(bits_per_channel_ != config.bits_per_channel() ||
channel_layout_ != config.channel_layout() ||
samples_per_second_ != config.samples_per_second())) {
DVLOG(1) << "Unsupported config change :";
DVLOG(1) << "\tbits_per_channel : " << bits_per_channel_
<< " -> " << config.bits_per_channel();
DVLOG(1) << "\tchannel_layout : " << channel_layout_
<< " -> " << config.channel_layout();
DVLOG(1) << "\tsample_rate : " << samples_per_second_
<< " -> " << config.samples_per_second();
return false;
}
// Release existing decoder resources if necessary.
ReleaseFFmpegResources();
// Initialize AVCodecContext structure.
codec_context_ = avcodec_alloc_context3(NULL);
AudioDecoderConfigToAVCodecContext(config, codec_context_);
// MP3 decodes to S16P which we don't support, tell it to use S16 instead.
if (codec_context_->sample_fmt == AV_SAMPLE_FMT_S16P)
codec_context_->request_sample_fmt = AV_SAMPLE_FMT_S16;
AVCodec* codec = avcodec_find_decoder(codec_context_->codec_id);
if (!codec || avcodec_open2(codec_context_, codec, NULL) < 0) {
DLOG(ERROR) << "Could not initialize audio decoder: "
<< codec_context_->codec_id;
return false;
}
// Ensure avcodec_open2() respected our format request.
if (codec_context_->sample_fmt == AV_SAMPLE_FMT_S16P) {
DLOG(ERROR) << "Unable to configure a supported sample format: "
<< codec_context_->sample_fmt;
return false;
}
// Some codecs will only output float data, so we need to convert to integer
// before returning the decoded buffer.
if (codec_context_->sample_fmt == AV_SAMPLE_FMT_FLTP ||
codec_context_->sample_fmt == AV_SAMPLE_FMT_FLT) {
// Preallocate the AudioBus for float conversions. We can treat interleaved
// float data as a single planar channel since our output is expected in an
// interleaved format anyways.
int channels = codec_context_->channels;
if (codec_context_->sample_fmt == AV_SAMPLE_FMT_FLT)
channels = 1;
converter_bus_ = AudioBus::CreateWrapper(channels);
}
// Success!
av_frame_ = avcodec_alloc_frame();
bits_per_channel_ = config.bits_per_channel();
channel_layout_ = config.channel_layout();
samples_per_second_ = config.samples_per_second();
output_timestamp_helper_.reset(new AudioTimestampHelper(
config.bytes_per_frame(), config.samples_per_second()));
bytes_per_frame_ = config.bytes_per_frame();
// Store initial values to guard against midstream configuration changes.
channels_ = codec_context_->channels;
av_sample_format_ = codec_context_->sample_fmt;
return true;
}
void FFmpegAudioDecoder::ReleaseFFmpegResources() {
if (codec_context_) {
av_free(codec_context_->extradata);
avcodec_close(codec_context_);
av_free(codec_context_);
}
if (av_frame_) {
av_free(av_frame_);
av_frame_ = NULL;
}
}
void FFmpegAudioDecoder::ResetTimestampState() {
output_timestamp_helper_->SetBaseTimestamp(kNoTimestamp());
last_input_timestamp_ = kNoTimestamp();
output_bytes_to_drop_ = 0;
}
void FFmpegAudioDecoder::RunDecodeLoop(
const scoped_refptr<DecoderBuffer>& input,
bool skip_eos_append) {
AVPacket packet;
av_init_packet(&packet);
if (input->IsEndOfStream()) {
packet.data = NULL;
packet.size = 0;
} else {
packet.data = const_cast<uint8*>(input->GetData());
packet.size = input->GetDataSize();
}
// Each audio packet may contain several frames, so we must call the decoder
// until we've exhausted the packet. Regardless of the packet size we always
// want to hand it to the decoder at least once, otherwise we would end up
// skipping end of stream packets since they have a size of zero.
do {
// Reset frame to default values.
avcodec_get_frame_defaults(av_frame_);
int frame_decoded = 0;
int result = avcodec_decode_audio4(
codec_context_, av_frame_, &frame_decoded, &packet);
if (result < 0) {
DCHECK(!input->IsEndOfStream())
<< "End of stream buffer produced an error! "
<< "This is quite possibly a bug in the audio decoder not handling "
<< "end of stream AVPackets correctly.";
DLOG(ERROR)
<< "Error decoding an audio frame with timestamp: "
<< input->GetTimestamp().InMicroseconds() << " us, duration: "
<< input->GetDuration().InMicroseconds() << " us, packet size: "
<< input->GetDataSize() << " bytes";
// TODO(dalecurtis): We should return a kDecodeError here instead:
// http://crbug.com/145276
break;
}
// Update packet size and data pointer in case we need to call the decoder
// with the remaining bytes from this packet.
packet.size -= result;
packet.data += result;
if (output_timestamp_helper_->base_timestamp() == kNoTimestamp() &&
!input->IsEndOfStream()) {
DCHECK(input->GetTimestamp() != kNoTimestamp());
if (output_bytes_to_drop_ > 0) {
// Currently Vorbis is the only codec that causes us to drop samples.
// If we have to drop samples it always means the timeline starts at 0.
DCHECK_EQ(codec_context_->codec_id, CODEC_ID_VORBIS);
output_timestamp_helper_->SetBaseTimestamp(base::TimeDelta());
} else {
output_timestamp_helper_->SetBaseTimestamp(input->GetTimestamp());
}
}
int decoded_audio_size = 0;
if (frame_decoded) {
if (av_frame_->sample_rate != samples_per_second_ ||
av_frame_->channels != channels_ ||
av_frame_->format != av_sample_format_) {
DLOG(ERROR) << "Unsupported midstream configuration change!"
<< " Sample Rate: " << av_frame_->sample_rate << " vs "
<< samples_per_second_
<< ", Channels: " << av_frame_->channels << " vs "
<< channels_
<< ", Sample Format: " << av_frame_->format << " vs "
<< av_sample_format_;
// This is an unrecoverable error, so bail out.
QueuedAudioBuffer queue_entry = { kDecodeError, NULL };
queued_audio_.push_back(queue_entry);
break;
}
decoded_audio_size = av_samples_get_buffer_size(
NULL, codec_context_->channels, av_frame_->nb_samples,
codec_context_->sample_fmt, 1);
// If we're decoding into float, adjust audio size.
if (converter_bus_ && bits_per_channel_ / 8 != sizeof(float)) {
DCHECK(codec_context_->sample_fmt == AV_SAMPLE_FMT_FLT ||
codec_context_->sample_fmt == AV_SAMPLE_FMT_FLTP);
decoded_audio_size *=
static_cast<float>(bits_per_channel_ / 8) / sizeof(float);
}
}
int start_sample = 0;
if (decoded_audio_size > 0 && output_bytes_to_drop_ > 0) {
DCHECK_EQ(decoded_audio_size % bytes_per_frame_, 0)
<< "Decoder didn't output full frames";
int dropped_size = std::min(decoded_audio_size, output_bytes_to_drop_);
start_sample = dropped_size / bytes_per_frame_;
decoded_audio_size -= dropped_size;
output_bytes_to_drop_ -= dropped_size;
}
scoped_refptr<DataBuffer> output;
if (decoded_audio_size > 0) {
DCHECK_EQ(decoded_audio_size % bytes_per_frame_, 0)
<< "Decoder didn't output full frames";
// Convert float data using an AudioBus.
if (converter_bus_) {
// Setup the AudioBus as a wrapper of the AVFrame data and then use
// AudioBus::ToInterleaved() to convert the data as necessary.
int skip_frames = start_sample;
int total_frames = av_frame_->nb_samples;
if (codec_context_->sample_fmt == AV_SAMPLE_FMT_FLT) {
DCHECK_EQ(converter_bus_->channels(), 1);
total_frames *= codec_context_->channels;
skip_frames *= codec_context_->channels;
}
converter_bus_->set_frames(total_frames);
DCHECK_EQ(decoded_audio_size,
(converter_bus_->frames() - skip_frames) * bytes_per_frame_);
for (int i = 0; i < converter_bus_->channels(); ++i) {
converter_bus_->SetChannelData(i, reinterpret_cast<float*>(
av_frame_->extended_data[i]));
}
output = new DataBuffer(decoded_audio_size);
output->SetDataSize(decoded_audio_size);
converter_bus_->ToInterleavedPartial(
skip_frames, converter_bus_->frames() - skip_frames,
bits_per_channel_ / 8, output->GetWritableData());
} else {
output = DataBuffer::CopyFrom(
av_frame_->extended_data[0] + start_sample * bytes_per_frame_,
decoded_audio_size);
}
output->SetTimestamp(output_timestamp_helper_->GetTimestamp());
output->SetDuration(
output_timestamp_helper_->GetDuration(decoded_audio_size));
output_timestamp_helper_->AddBytes(decoded_audio_size);
} else if (IsEndOfStream(result, decoded_audio_size, input) &&
!skip_eos_append) {
DCHECK_EQ(packet.size, 0);
output = DataBuffer::CreateEOSBuffer();
}
if (output) {
QueuedAudioBuffer queue_entry = { kOk, output };
queued_audio_.push_back(queue_entry);
}
// Decoding finished successfully, update statistics.
if (result > 0) {
PipelineStatistics statistics;
statistics.audio_bytes_decoded = result;
statistics_cb_.Run(statistics);
}
} while (packet.size > 0);
}
} // namespace media
|