Revert "Reland c91b178b07b0d - Delete dead signin code (SigninGlobalError)"
[chromium-blink-merge.git] / media / formats / webm / webm_cluster_parser_unittest.cc
blobeaeb3186da3d37acacbdf6983dd14e2cfe950130
1 // Copyright 2014 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.
5 #include <algorithm>
6 #include <cstdlib>
7 #include <vector>
9 #include "base/bind.h"
10 #include "base/logging.h"
11 #include "media/base/audio_decoder_config.h"
12 #include "media/base/decrypt_config.h"
13 #include "media/formats/webm/cluster_builder.h"
14 #include "media/formats/webm/opus_packet_builder.h"
15 #include "media/formats/webm/webm_cluster_parser.h"
16 #include "media/formats/webm/webm_constants.h"
17 #include "testing/gmock/include/gmock/gmock.h"
18 #include "testing/gtest/include/gtest/gtest.h"
20 using ::testing::InSequence;
21 using ::testing::Return;
22 using ::testing::_;
24 namespace media {
26 typedef WebMTracksParser::TextTracks TextTracks;
28 namespace {
30 enum {
31 kTimecodeScale = 1000000, // Timecode scale for millisecond timestamps.
32 kAudioTrackNum = 1,
33 kVideoTrackNum = 2,
34 kTextTrackNum = 3,
35 kTestAudioFrameDefaultDurationInMs = 13,
36 kTestVideoFrameDefaultDurationInMs = 17
39 // Test duration defaults must differ from parser estimation defaults to know
40 // which durations parser used when emitting buffers.
41 static_assert(
42 static_cast<int>(kTestAudioFrameDefaultDurationInMs) !=
43 static_cast<int>(WebMClusterParser::kDefaultAudioBufferDurationInMs),
44 "test default is the same as estimation fallback audio duration");
45 static_assert(
46 static_cast<int>(kTestVideoFrameDefaultDurationInMs) !=
47 static_cast<int>(WebMClusterParser::kDefaultVideoBufferDurationInMs),
48 "test default is the same as estimation fallback video duration");
50 struct BlockInfo {
51 int track_num;
52 int timestamp;
54 // Negative value is allowed only for block groups (not simple blocks) and
55 // directs CreateCluster() to exclude BlockDuration entry from the cluster for
56 // this BlockGroup. The absolute value is used for parser verification.
57 // For simple blocks, this value must be non-negative, and is used only for
58 // parser verification.
59 double duration;
61 bool use_simple_block;
63 // Default data will be used if no data given.
64 const uint8_t* data;
65 int data_length;
68 const BlockInfo kDefaultBlockInfo[] = {
69 {kAudioTrackNum, 0, 23, true, NULL, 0},
70 {kAudioTrackNum, 23, 23, true, NULL, 0},
71 // Assumes not using DefaultDuration
72 {kVideoTrackNum, 33, 34, true, NULL, 0},
73 {kAudioTrackNum, 46, 23, true, NULL, 0},
74 {kVideoTrackNum, 67, 33, false, NULL, 0},
75 {kAudioTrackNum, 69, 23, false, NULL, 0},
76 {kVideoTrackNum, 100, 33, false, NULL, 0},
79 const uint8_t kEncryptedFrame[] = {
80 // Block is encrypted
81 0x01,
83 // IV
84 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08};
86 // Helper that hard-codes some non-varying constructor parameters.
87 WebMClusterParser* CreateParserHelper(
88 base::TimeDelta audio_default_duration,
89 base::TimeDelta video_default_duration,
90 const WebMTracksParser::TextTracks& text_tracks,
91 const std::set<int64>& ignored_tracks,
92 const std::string& audio_encryption_key_id,
93 const std::string& video_encryption_key_id,
94 const AudioCodec audio_codec) {
95 return new WebMClusterParser(
96 kTimecodeScale, kAudioTrackNum, audio_default_duration, kVideoTrackNum,
97 video_default_duration, text_tracks, ignored_tracks,
98 audio_encryption_key_id, video_encryption_key_id, audio_codec,
99 new MediaLog());
102 // Create a default version of the parser for test.
103 WebMClusterParser* CreateDefaultParser() {
104 return CreateParserHelper(kNoTimestamp(), kNoTimestamp(), TextTracks(),
105 std::set<int64>(), std::string(), std::string(),
106 kUnknownAudioCodec);
109 // Create a parser for test with custom audio and video default durations, and
110 // optionally custom text tracks.
111 WebMClusterParser* CreateParserWithDefaultDurationsAndOptionalTextTracks(
112 base::TimeDelta audio_default_duration,
113 base::TimeDelta video_default_duration,
114 const WebMTracksParser::TextTracks& text_tracks = TextTracks()) {
115 return CreateParserHelper(audio_default_duration, video_default_duration,
116 text_tracks, std::set<int64>(), std::string(),
117 std::string(), kUnknownAudioCodec);
120 // Create a parser for test with custom ignored tracks.
121 WebMClusterParser* CreateParserWithIgnoredTracks(
122 std::set<int64>& ignored_tracks) {
123 return CreateParserHelper(kNoTimestamp(), kNoTimestamp(), TextTracks(),
124 ignored_tracks, std::string(), std::string(),
125 kUnknownAudioCodec);
128 // Create a parser for test with custom encryption key ids and audio codec.
129 WebMClusterParser* CreateParserWithKeyIdsAndAudioCodec(
130 const std::string& audio_encryption_key_id,
131 const std::string& video_encryption_key_id,
132 const AudioCodec audio_codec) {
133 return CreateParserHelper(kNoTimestamp(), kNoTimestamp(), TextTracks(),
134 std::set<int64>(), audio_encryption_key_id,
135 video_encryption_key_id, audio_codec);
138 scoped_ptr<Cluster> CreateCluster(int timecode,
139 const BlockInfo* block_info,
140 int block_count) {
141 ClusterBuilder cb;
142 cb.SetClusterTimecode(0);
144 uint8_t kDefaultBlockData[] = { 0x00 };
146 for (int i = 0; i < block_count; i++) {
147 const uint8_t* data;
148 int data_length;
149 if (block_info[i].data != NULL) {
150 data = block_info[i].data;
151 data_length = block_info[i].data_length;
152 } else {
153 data = kDefaultBlockData;
154 data_length = sizeof(kDefaultBlockData);
157 if (block_info[i].use_simple_block) {
158 CHECK_GE(block_info[i].duration, 0);
159 cb.AddSimpleBlock(block_info[i].track_num, block_info[i].timestamp, 0,
160 data, data_length);
161 continue;
164 if (block_info[i].duration < 0) {
165 cb.AddBlockGroupWithoutBlockDuration(block_info[i].track_num,
166 block_info[i].timestamp, 0, data,
167 data_length);
168 continue;
171 cb.AddBlockGroup(block_info[i].track_num, block_info[i].timestamp,
172 block_info[i].duration, 0, data, data_length);
175 return cb.Finish();
178 // Creates a Cluster with one encrypted Block. |bytes_to_write| is number of
179 // bytes of the encrypted frame to write.
180 scoped_ptr<Cluster> CreateEncryptedCluster(int bytes_to_write) {
181 CHECK_GT(bytes_to_write, 0);
182 CHECK_LE(bytes_to_write, static_cast<int>(sizeof(kEncryptedFrame)));
184 ClusterBuilder cb;
185 cb.SetClusterTimecode(0);
186 cb.AddSimpleBlock(kVideoTrackNum, 0, 0, kEncryptedFrame, bytes_to_write);
187 return cb.Finish();
190 bool VerifyBuffers(const WebMClusterParser::BufferQueue& audio_buffers,
191 const WebMClusterParser::BufferQueue& video_buffers,
192 const WebMClusterParser::BufferQueue& text_buffers,
193 const BlockInfo* block_info,
194 int block_count) {
195 int buffer_count = audio_buffers.size() + video_buffers.size() +
196 text_buffers.size();
197 if (block_count != buffer_count) {
198 DVLOG(1) << __FUNCTION__ << " : block_count (" << block_count
199 << ") mismatches buffer_count (" << buffer_count << ")";
200 return false;
203 size_t audio_offset = 0;
204 size_t video_offset = 0;
205 size_t text_offset = 0;
206 for (int i = 0; i < block_count; i++) {
207 const WebMClusterParser::BufferQueue* buffers = NULL;
208 size_t* offset;
209 StreamParserBuffer::Type expected_type = DemuxerStream::UNKNOWN;
211 if (block_info[i].track_num == kAudioTrackNum) {
212 buffers = &audio_buffers;
213 offset = &audio_offset;
214 expected_type = DemuxerStream::AUDIO;
215 } else if (block_info[i].track_num == kVideoTrackNum) {
216 buffers = &video_buffers;
217 offset = &video_offset;
218 expected_type = DemuxerStream::VIDEO;
219 } else if (block_info[i].track_num == kTextTrackNum) {
220 buffers = &text_buffers;
221 offset = &text_offset;
222 expected_type = DemuxerStream::TEXT;
223 } else {
224 LOG(ERROR) << "Unexpected track number " << block_info[i].track_num;
225 return false;
228 if (*offset >= buffers->size()) {
229 DVLOG(1) << __FUNCTION__ << " : Too few buffers (" << buffers->size()
230 << ") for track_num (" << block_info[i].track_num
231 << "), expected at least " << *offset + 1 << " buffers";
232 return false;
235 scoped_refptr<StreamParserBuffer> buffer = (*buffers)[(*offset)++];
237 EXPECT_EQ(block_info[i].timestamp, buffer->timestamp().InMilliseconds());
238 EXPECT_EQ(std::abs(block_info[i].duration),
239 buffer->duration().InMillisecondsF());
240 EXPECT_EQ(expected_type, buffer->type());
241 EXPECT_EQ(block_info[i].track_num, buffer->track_id());
244 return true;
247 bool VerifyBuffers(const scoped_ptr<WebMClusterParser>& parser,
248 const BlockInfo* block_info,
249 int block_count) {
250 const WebMClusterParser::TextBufferQueueMap& text_map =
251 parser->GetTextBuffers();
252 const WebMClusterParser::BufferQueue* text_buffers;
253 const WebMClusterParser::BufferQueue no_text_buffers;
254 if (!text_map.empty())
255 text_buffers = &(text_map.rbegin()->second);
256 else
257 text_buffers = &no_text_buffers;
259 return VerifyBuffers(parser->GetAudioBuffers(),
260 parser->GetVideoBuffers(),
261 *text_buffers,
262 block_info,
263 block_count);
266 bool VerifyTextBuffers(const scoped_ptr<WebMClusterParser>& parser,
267 const BlockInfo* block_info_ptr,
268 int block_count,
269 int text_track_num,
270 const WebMClusterParser::BufferQueue& text_buffers) {
271 const BlockInfo* const block_info_end = block_info_ptr + block_count;
273 typedef WebMClusterParser::BufferQueue::const_iterator TextBufferIter;
274 TextBufferIter buffer_iter = text_buffers.begin();
275 const TextBufferIter buffer_end = text_buffers.end();
277 while (block_info_ptr != block_info_end) {
278 const BlockInfo& block_info = *block_info_ptr++;
280 if (block_info.track_num != text_track_num)
281 continue;
283 EXPECT_FALSE(block_info.use_simple_block);
284 EXPECT_FALSE(buffer_iter == buffer_end);
286 const scoped_refptr<StreamParserBuffer> buffer = *buffer_iter++;
287 EXPECT_EQ(block_info.timestamp, buffer->timestamp().InMilliseconds());
288 EXPECT_EQ(std::abs(block_info.duration),
289 buffer->duration().InMillisecondsF());
290 EXPECT_EQ(DemuxerStream::TEXT, buffer->type());
291 EXPECT_EQ(text_track_num, buffer->track_id());
294 EXPECT_TRUE(buffer_iter == buffer_end);
295 return true;
298 void VerifyEncryptedBuffer(scoped_refptr<StreamParserBuffer> buffer) {
299 EXPECT_TRUE(buffer->decrypt_config());
300 EXPECT_EQ(static_cast<unsigned long>(DecryptConfig::kDecryptionKeySize),
301 buffer->decrypt_config()->iv().length());
304 void AppendToEnd(const WebMClusterParser::BufferQueue& src,
305 WebMClusterParser::BufferQueue* dest) {
306 for (WebMClusterParser::BufferQueue::const_iterator itr = src.begin();
307 itr != src.end(); ++itr) {
308 dest->push_back(*itr);
312 } // namespace
314 class WebMClusterParserTest : public testing::Test {
315 public:
316 WebMClusterParserTest() : parser_(CreateDefaultParser()) {}
318 protected:
319 void ResetParserToHaveDefaultDurations() {
320 base::TimeDelta default_audio_duration = base::TimeDelta::FromMilliseconds(
321 kTestAudioFrameDefaultDurationInMs);
322 base::TimeDelta default_video_duration = base::TimeDelta::FromMilliseconds(
323 kTestVideoFrameDefaultDurationInMs);
324 ASSERT_GE(default_audio_duration, base::TimeDelta());
325 ASSERT_GE(default_video_duration, base::TimeDelta());
326 ASSERT_NE(kNoTimestamp(), default_audio_duration);
327 ASSERT_NE(kNoTimestamp(), default_video_duration);
329 parser_.reset(CreateParserWithDefaultDurationsAndOptionalTextTracks(
330 default_audio_duration, default_video_duration));
333 scoped_ptr<WebMClusterParser> parser_;
335 private:
336 DISALLOW_COPY_AND_ASSIGN(WebMClusterParserTest);
339 TEST_F(WebMClusterParserTest, HeldBackBufferHoldsBackAllTracks) {
340 // If a buffer is missing duration and is being held back, then all other
341 // tracks' buffers that have same or higher (decode) timestamp should be held
342 // back too to keep the timestamps emitted for a cluster monotonically
343 // non-decreasing and in same order as parsed.
344 InSequence s;
346 // Reset the parser to have 3 tracks: text, video (no default frame duration),
347 // and audio (with a default frame duration).
348 TextTracks text_tracks;
349 text_tracks.insert(std::make_pair(TextTracks::key_type(kTextTrackNum),
350 TextTrackConfig(kTextSubtitles, "", "",
351 "")));
352 base::TimeDelta default_audio_duration =
353 base::TimeDelta::FromMilliseconds(kTestAudioFrameDefaultDurationInMs);
354 ASSERT_GE(default_audio_duration, base::TimeDelta());
355 ASSERT_NE(kNoTimestamp(), default_audio_duration);
356 parser_.reset(CreateParserWithDefaultDurationsAndOptionalTextTracks(
357 default_audio_duration, kNoTimestamp(), text_tracks));
359 const BlockInfo kBlockInfo[] = {
360 {kVideoTrackNum, 0, 33, true, NULL, 0},
361 {kAudioTrackNum, 0, 23, false, NULL, 0},
362 {kTextTrackNum, 10, 42, false, NULL, 0},
363 {kAudioTrackNum, 23, kTestAudioFrameDefaultDurationInMs, true, NULL, 0},
364 {kVideoTrackNum, 33, 33, true, NULL, 0},
365 {kAudioTrackNum, 36, kTestAudioFrameDefaultDurationInMs, true, NULL, 0},
366 {kVideoTrackNum, 66, 33, true, NULL, 0},
367 {kAudioTrackNum, 70, kTestAudioFrameDefaultDurationInMs, true, NULL, 0},
368 {kAudioTrackNum, 83, kTestAudioFrameDefaultDurationInMs, true, NULL, 0},
371 const int kExpectedBuffersOnPartialCluster[] = {
372 0, // Video simple block without DefaultDuration should be held back
373 0, // Audio buffer ready, but not emitted because its TS >= held back video
374 0, // Text buffer ready, but not emitted because its TS >= held back video
375 0, // 2nd audio buffer ready, also not emitted for same reason as first
376 4, // All previous buffers emitted, 2nd video held back with no duration
377 4, // 2nd video still has no duration, 3rd audio ready but not emitted
378 6, // All previous buffers emitted, 3rd video held back with no duration
379 6, // 3rd video still has no duration, 4th audio ready but not emitted
380 9, // Cluster end emits all buffers and 3rd video's duration is estimated
383 ASSERT_EQ(arraysize(kBlockInfo), arraysize(kExpectedBuffersOnPartialCluster));
384 int block_count = arraysize(kBlockInfo);
386 // Iteratively create a cluster containing the first N+1 blocks and parse all
387 // but the last byte of the cluster (except when N==|block_count|, just parse
388 // the whole cluster). Verify that the corresponding entry in
389 // |kExpectedBuffersOnPartialCluster| identifies the exact subset of
390 // |kBlockInfo| returned by the parser.
391 for (int i = 0; i < block_count; ++i) {
392 if (i > 0)
393 parser_->Reset();
394 // Since we don't know exactly the offsets of each block in the full
395 // cluster, build a cluster with exactly one additional block so that
396 // parse of all but one byte should deterministically parse all but the
397 // last full block. Don't |exceed block_count| blocks though.
398 int blocks_in_cluster = std::min(i + 2, block_count);
399 scoped_ptr<Cluster> cluster(CreateCluster(0, kBlockInfo,
400 blocks_in_cluster));
401 // Parse all but the last byte unless we need to parse the full cluster.
402 bool parse_full_cluster = i == (block_count - 1);
403 int result = parser_->Parse(cluster->data(), parse_full_cluster ?
404 cluster->size() : cluster->size() - 1);
405 if (parse_full_cluster) {
406 DVLOG(1) << "Verifying parse result of full cluster of "
407 << blocks_in_cluster << " blocks";
408 EXPECT_EQ(cluster->size(), result);
409 } else {
410 DVLOG(1) << "Verifying parse result of cluster of "
411 << blocks_in_cluster << " blocks with last block incomplete";
412 EXPECT_GT(cluster->size(), result);
413 EXPECT_LT(0, result);
416 EXPECT_TRUE(VerifyBuffers(parser_, kBlockInfo,
417 kExpectedBuffersOnPartialCluster[i]));
421 TEST_F(WebMClusterParserTest, Reset) {
422 InSequence s;
424 int block_count = arraysize(kDefaultBlockInfo);
425 scoped_ptr<Cluster> cluster(CreateCluster(0, kDefaultBlockInfo, block_count));
427 // Send slightly less than the full cluster so all but the last block is
428 // parsed.
429 int result = parser_->Parse(cluster->data(), cluster->size() - 1);
430 EXPECT_GT(result, 0);
431 EXPECT_LT(result, cluster->size());
433 ASSERT_TRUE(VerifyBuffers(parser_, kDefaultBlockInfo, block_count - 1));
434 parser_->Reset();
436 // Now parse a whole cluster to verify that all the blocks will get parsed.
437 result = parser_->Parse(cluster->data(), cluster->size());
438 EXPECT_EQ(cluster->size(), result);
439 ASSERT_TRUE(VerifyBuffers(parser_, kDefaultBlockInfo, block_count));
442 TEST_F(WebMClusterParserTest, ParseClusterWithSingleCall) {
443 int block_count = arraysize(kDefaultBlockInfo);
444 scoped_ptr<Cluster> cluster(CreateCluster(0, kDefaultBlockInfo, block_count));
446 int result = parser_->Parse(cluster->data(), cluster->size());
447 EXPECT_EQ(cluster->size(), result);
448 ASSERT_TRUE(VerifyBuffers(parser_, kDefaultBlockInfo, block_count));
451 TEST_F(WebMClusterParserTest, ParseClusterWithMultipleCalls) {
452 int block_count = arraysize(kDefaultBlockInfo);
453 scoped_ptr<Cluster> cluster(CreateCluster(0, kDefaultBlockInfo, block_count));
455 WebMClusterParser::BufferQueue audio_buffers;
456 WebMClusterParser::BufferQueue video_buffers;
457 const WebMClusterParser::BufferQueue no_text_buffers;
459 const uint8_t* data = cluster->data();
460 int size = cluster->size();
461 int default_parse_size = 3;
462 int parse_size = std::min(default_parse_size, size);
464 while (size > 0) {
465 int result = parser_->Parse(data, parse_size);
466 ASSERT_GE(result, 0);
467 ASSERT_LE(result, parse_size);
469 if (result == 0) {
470 // The parser needs more data so increase the parse_size a little.
471 parse_size += default_parse_size;
472 parse_size = std::min(parse_size, size);
473 continue;
476 AppendToEnd(parser_->GetAudioBuffers(), &audio_buffers);
477 AppendToEnd(parser_->GetVideoBuffers(), &video_buffers);
479 parse_size = default_parse_size;
481 data += result;
482 size -= result;
484 ASSERT_TRUE(VerifyBuffers(audio_buffers, video_buffers,
485 no_text_buffers, kDefaultBlockInfo,
486 block_count));
489 // Verify that both BlockGroups with the BlockDuration before the Block
490 // and BlockGroups with the BlockDuration after the Block are supported
491 // correctly.
492 // Note: Raw bytes are use here because ClusterBuilder only generates
493 // one of these scenarios.
494 TEST_F(WebMClusterParserTest, ParseBlockGroup) {
495 const BlockInfo kBlockInfo[] = {
496 {kAudioTrackNum, 0, 23, false, NULL, 0},
497 {kVideoTrackNum, 33, 34, false, NULL, 0},
499 int block_count = arraysize(kBlockInfo);
501 const uint8_t kClusterData[] = {
502 0x1F, 0x43, 0xB6, 0x75, 0x9B, // Cluster(size=27)
503 0xE7, 0x81, 0x00, // Timecode(size=1, value=0)
504 // BlockGroup with BlockDuration before Block.
505 0xA0, 0x8A, // BlockGroup(size=10)
506 0x9B, 0x81, 0x17, // BlockDuration(size=1, value=23)
507 0xA1, 0x85, 0x81, 0x00, 0x00, 0x00, 0xaa, // Block(size=5, track=1, ts=0)
508 // BlockGroup with BlockDuration after Block.
509 0xA0, 0x8A, // BlockGroup(size=10)
510 0xA1, 0x85, 0x82, 0x00, 0x21, 0x00, 0x55, // Block(size=5, track=2, ts=33)
511 0x9B, 0x81, 0x22, // BlockDuration(size=1, value=34)
513 const int kClusterSize = sizeof(kClusterData);
515 int result = parser_->Parse(kClusterData, kClusterSize);
516 EXPECT_EQ(kClusterSize, result);
517 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo, block_count));
520 TEST_F(WebMClusterParserTest, ParseSimpleBlockAndBlockGroupMixture) {
521 const BlockInfo kBlockInfo[] = {
522 {kAudioTrackNum, 0, 23, true, NULL, 0},
523 {kAudioTrackNum, 23, 23, false, NULL, 0},
524 {kVideoTrackNum, 33, 34, true, NULL, 0},
525 {kAudioTrackNum, 46, 23, false, NULL, 0},
526 {kVideoTrackNum, 67, 33, false, NULL, 0},
528 int block_count = arraysize(kBlockInfo);
529 scoped_ptr<Cluster> cluster(CreateCluster(0, kBlockInfo, block_count));
531 int result = parser_->Parse(cluster->data(), cluster->size());
532 EXPECT_EQ(cluster->size(), result);
533 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo, block_count));
536 TEST_F(WebMClusterParserTest, IgnoredTracks) {
537 std::set<int64> ignored_tracks;
538 ignored_tracks.insert(kTextTrackNum);
540 parser_.reset(CreateParserWithIgnoredTracks(ignored_tracks));
542 const BlockInfo kInputBlockInfo[] = {
543 {kAudioTrackNum, 0, 23, true, NULL, 0},
544 {kAudioTrackNum, 23, 23, true, NULL, 0},
545 {kVideoTrackNum, 33, 34, true, NULL, 0},
546 {kTextTrackNum, 33, 99, true, NULL, 0},
547 {kAudioTrackNum, 46, 23, true, NULL, 0},
548 {kVideoTrackNum, 67, 34, true, NULL, 0},
550 int input_block_count = arraysize(kInputBlockInfo);
552 const BlockInfo kOutputBlockInfo[] = {
553 {kAudioTrackNum, 0, 23, true, NULL, 0},
554 {kAudioTrackNum, 23, 23, true, NULL, 0},
555 {kVideoTrackNum, 33, 34, true, NULL, 0},
556 {kAudioTrackNum, 46, 23, true, NULL, 0},
557 {kVideoTrackNum, 67, 34, true, NULL, 0},
559 int output_block_count = arraysize(kOutputBlockInfo);
561 scoped_ptr<Cluster> cluster(
562 CreateCluster(0, kInputBlockInfo, input_block_count));
564 int result = parser_->Parse(cluster->data(), cluster->size());
565 EXPECT_EQ(cluster->size(), result);
566 ASSERT_TRUE(VerifyBuffers(parser_, kOutputBlockInfo, output_block_count));
569 TEST_F(WebMClusterParserTest, ParseTextTracks) {
570 TextTracks text_tracks;
572 text_tracks.insert(std::make_pair(TextTracks::key_type(kTextTrackNum),
573 TextTrackConfig(kTextSubtitles, "", "",
574 "")));
576 parser_.reset(CreateParserWithDefaultDurationsAndOptionalTextTracks(
577 kNoTimestamp(), kNoTimestamp(), text_tracks));
579 const BlockInfo kInputBlockInfo[] = {
580 {kAudioTrackNum, 0, 23, true, NULL, 0},
581 {kAudioTrackNum, 23, 23, true, NULL, 0},
582 {kVideoTrackNum, 33, 34, true, NULL, 0},
583 {kTextTrackNum, 33, 42, false, NULL, 0},
584 {kAudioTrackNum, 46, 23, true, NULL, 0},
585 {kTextTrackNum, 55, 44, false, NULL, 0},
586 {kVideoTrackNum, 67, 34, true, NULL, 0},
588 int input_block_count = arraysize(kInputBlockInfo);
590 scoped_ptr<Cluster> cluster(
591 CreateCluster(0, kInputBlockInfo, input_block_count));
593 int result = parser_->Parse(cluster->data(), cluster->size());
594 EXPECT_EQ(cluster->size(), result);
595 ASSERT_TRUE(VerifyBuffers(parser_, kInputBlockInfo, input_block_count));
598 TEST_F(WebMClusterParserTest, TextTracksSimpleBlock) {
599 TextTracks text_tracks;
601 text_tracks.insert(std::make_pair(TextTracks::key_type(kTextTrackNum),
602 TextTrackConfig(kTextSubtitles, "", "",
603 "")));
605 parser_.reset(CreateParserWithDefaultDurationsAndOptionalTextTracks(
606 kNoTimestamp(), kNoTimestamp(), text_tracks));
608 const BlockInfo kInputBlockInfo[] = {
609 { kTextTrackNum, 33, 42, true },
611 int input_block_count = arraysize(kInputBlockInfo);
613 scoped_ptr<Cluster> cluster(
614 CreateCluster(0, kInputBlockInfo, input_block_count));
616 int result = parser_->Parse(cluster->data(), cluster->size());
617 EXPECT_LT(result, 0);
620 TEST_F(WebMClusterParserTest, ParseMultipleTextTracks) {
621 TextTracks text_tracks;
623 const int kSubtitleTextTrackNum = kTextTrackNum;
624 const int kCaptionTextTrackNum = kTextTrackNum + 1;
626 text_tracks.insert(std::make_pair(TextTracks::key_type(kSubtitleTextTrackNum),
627 TextTrackConfig(kTextSubtitles, "", "",
628 "")));
630 text_tracks.insert(std::make_pair(TextTracks::key_type(kCaptionTextTrackNum),
631 TextTrackConfig(kTextCaptions, "", "",
632 "")));
634 parser_.reset(CreateParserWithDefaultDurationsAndOptionalTextTracks(
635 kNoTimestamp(), kNoTimestamp(), text_tracks));
637 const BlockInfo kInputBlockInfo[] = {
638 {kAudioTrackNum, 0, 23, true, NULL, 0},
639 {kAudioTrackNum, 23, 23, true, NULL, 0},
640 {kVideoTrackNum, 33, 34, true, NULL, 0},
641 {kSubtitleTextTrackNum, 33, 42, false, NULL, 0},
642 {kAudioTrackNum, 46, 23, true, NULL, 0},
643 {kCaptionTextTrackNum, 55, 44, false, NULL, 0},
644 {kVideoTrackNum, 67, 34, true, NULL, 0},
645 {kSubtitleTextTrackNum, 67, 33, false, NULL, 0},
647 int input_block_count = arraysize(kInputBlockInfo);
649 scoped_ptr<Cluster> cluster(
650 CreateCluster(0, kInputBlockInfo, input_block_count));
652 int result = parser_->Parse(cluster->data(), cluster->size());
653 EXPECT_EQ(cluster->size(), result);
655 const WebMClusterParser::TextBufferQueueMap& text_map =
656 parser_->GetTextBuffers();
657 for (WebMClusterParser::TextBufferQueueMap::const_iterator itr =
658 text_map.begin();
659 itr != text_map.end();
660 ++itr) {
661 const TextTracks::const_iterator find_result =
662 text_tracks.find(itr->first);
663 ASSERT_TRUE(find_result != text_tracks.end());
664 ASSERT_TRUE(VerifyTextBuffers(parser_, kInputBlockInfo, input_block_count,
665 itr->first, itr->second));
669 TEST_F(WebMClusterParserTest, ParseEncryptedBlock) {
670 scoped_ptr<Cluster> cluster(CreateEncryptedCluster(sizeof(kEncryptedFrame)));
672 parser_.reset(CreateParserWithKeyIdsAndAudioCodec(
673 std::string(), "video_key_id", kUnknownAudioCodec));
674 int result = parser_->Parse(cluster->data(), cluster->size());
675 EXPECT_EQ(cluster->size(), result);
676 ASSERT_EQ(1UL, parser_->GetVideoBuffers().size());
677 scoped_refptr<StreamParserBuffer> buffer = parser_->GetVideoBuffers()[0];
678 VerifyEncryptedBuffer(buffer);
681 TEST_F(WebMClusterParserTest, ParseBadEncryptedBlock) {
682 scoped_ptr<Cluster> cluster(
683 CreateEncryptedCluster(sizeof(kEncryptedFrame) - 1));
685 parser_.reset(CreateParserWithKeyIdsAndAudioCodec(
686 std::string(), "video_key_id", kUnknownAudioCodec));
687 int result = parser_->Parse(cluster->data(), cluster->size());
688 EXPECT_EQ(-1, result);
691 TEST_F(WebMClusterParserTest, ParseInvalidZeroSizedCluster) {
692 const uint8_t kBuffer[] = {
693 0x1F, 0x43, 0xB6, 0x75, 0x80, // CLUSTER (size = 0)
696 EXPECT_EQ(-1, parser_->Parse(kBuffer, sizeof(kBuffer)));
699 TEST_F(WebMClusterParserTest, ParseInvalidUnknownButActuallyZeroSizedCluster) {
700 const uint8_t kBuffer[] = {
701 0x1F, 0x43, 0xB6, 0x75, 0xFF, // CLUSTER (size = "unknown")
702 0x1F, 0x43, 0xB6, 0x75, 0x85, // CLUSTER (size = 5)
705 EXPECT_EQ(-1, parser_->Parse(kBuffer, sizeof(kBuffer)));
708 TEST_F(WebMClusterParserTest, ParseInvalidTextBlockGroupWithoutDuration) {
709 // Text track frames must have explicitly specified BlockGroup BlockDurations.
710 TextTracks text_tracks;
712 text_tracks.insert(std::make_pair(TextTracks::key_type(kTextTrackNum),
713 TextTrackConfig(kTextSubtitles, "", "",
714 "")));
716 parser_.reset(CreateParserWithDefaultDurationsAndOptionalTextTracks(
717 kNoTimestamp(), kNoTimestamp(), text_tracks));
719 const BlockInfo kBlockInfo[] = {
720 { kTextTrackNum, 33, -42, false },
722 int block_count = arraysize(kBlockInfo);
723 scoped_ptr<Cluster> cluster(CreateCluster(0, kBlockInfo, block_count));
724 int result = parser_->Parse(cluster->data(), cluster->size());
725 EXPECT_LT(result, 0);
728 TEST_F(WebMClusterParserTest, ParseWithDefaultDurationsSimpleBlocks) {
729 InSequence s;
730 ResetParserToHaveDefaultDurations();
732 EXPECT_LT(kTestAudioFrameDefaultDurationInMs, 23);
733 EXPECT_LT(kTestVideoFrameDefaultDurationInMs, 33);
735 const BlockInfo kBlockInfo[] = {
736 {kAudioTrackNum, 0, kTestAudioFrameDefaultDurationInMs, true, NULL, 0},
737 {kAudioTrackNum, 23, kTestAudioFrameDefaultDurationInMs, true, NULL, 0},
738 {kVideoTrackNum, 33, kTestVideoFrameDefaultDurationInMs, true, NULL, 0},
739 {kAudioTrackNum, 46, kTestAudioFrameDefaultDurationInMs, true, NULL, 0},
740 {kVideoTrackNum, 67, kTestVideoFrameDefaultDurationInMs, true, NULL, 0},
741 {kAudioTrackNum, 69, kTestAudioFrameDefaultDurationInMs, true, NULL, 0},
742 {kVideoTrackNum, 100, kTestVideoFrameDefaultDurationInMs, true, NULL, 0},
745 int block_count = arraysize(kBlockInfo);
746 scoped_ptr<Cluster> cluster(CreateCluster(0, kBlockInfo, block_count));
748 // Send slightly less than the full cluster so all but the last block is
749 // parsed. Though all the blocks are simple blocks, none should be held aside
750 // for duration estimation prior to end of cluster detection because all the
751 // tracks have DefaultDurations.
752 int result = parser_->Parse(cluster->data(), cluster->size() - 1);
753 EXPECT_GT(result, 0);
754 EXPECT_LT(result, cluster->size());
755 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo, block_count - 1));
757 parser_->Reset();
759 // Now parse a whole cluster to verify that all the blocks will get parsed.
760 result = parser_->Parse(cluster->data(), cluster->size());
761 EXPECT_EQ(cluster->size(), result);
762 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo, block_count));
765 TEST_F(WebMClusterParserTest, ParseWithoutAnyDurationsSimpleBlocks) {
766 InSequence s;
768 // Absent DefaultDuration information, SimpleBlock durations are derived from
769 // inter-buffer track timestamp delta if within the cluster. Duration for the
770 // last block in a cluster is estimated independently for each track in the
771 // cluster. For video tracks we use the maximum seen so far. For audio we use
772 // the the minimum.
773 // TODO(chcunningham): Move audio over to use the maximum.
774 const BlockInfo kBlockInfo1[] = {
775 {kAudioTrackNum, 0, 23, true, NULL, 0},
776 {kAudioTrackNum, 23, 22, true, NULL, 0},
777 {kVideoTrackNum, 33, 33, true, NULL, 0},
778 {kAudioTrackNum, 45, 23, true, NULL, 0},
779 {kVideoTrackNum, 66, 34, true, NULL, 0},
780 // Estimated from minimum audio dur
781 {kAudioTrackNum, 68, 22, true, NULL, 0},
782 // Estimated from maximum video dur
783 {kVideoTrackNum, 100, 34, true, NULL, 0},
786 int block_count1 = arraysize(kBlockInfo1);
787 scoped_ptr<Cluster> cluster1(CreateCluster(0, kBlockInfo1, block_count1));
789 // Send slightly less than the first full cluster so all but the last video
790 // block is parsed. Verify the last fully parsed audio and video buffer are
791 // both missing from the result (parser should hold them aside for duration
792 // estimation prior to end of cluster detection in the absence of
793 // DefaultDurations.)
794 int result = parser_->Parse(cluster1->data(), cluster1->size() - 1);
795 EXPECT_GT(result, 0);
796 EXPECT_LT(result, cluster1->size());
797 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo1, block_count1 - 3));
798 EXPECT_EQ(3UL, parser_->GetAudioBuffers().size());
799 EXPECT_EQ(1UL, parser_->GetVideoBuffers().size());
801 parser_->Reset();
803 // Now parse the full first cluster and verify all the blocks are parsed.
804 result = parser_->Parse(cluster1->data(), cluster1->size());
805 EXPECT_EQ(cluster1->size(), result);
806 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo1, block_count1));
808 // Verify that the estimated frame duration is tracked across clusters for
809 // each track.
810 const BlockInfo kBlockInfo2[] = {
811 // Estimate carries over across clusters
812 {kAudioTrackNum, 200, 22, true, NULL, 0},
813 // Estimate carries over across clusters
814 {kVideoTrackNum, 201, 34, true, NULL, 0},
817 int block_count2 = arraysize(kBlockInfo2);
818 scoped_ptr<Cluster> cluster2(CreateCluster(0, kBlockInfo2, block_count2));
819 result = parser_->Parse(cluster2->data(), cluster2->size());
820 EXPECT_EQ(cluster2->size(), result);
821 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo2, block_count2));
824 TEST_F(WebMClusterParserTest, ParseWithoutAnyDurationsBlockGroups) {
825 InSequence s;
827 // Absent DefaultDuration and BlockDuration information, BlockGroup block
828 // durations are derived from inter-buffer track timestamp delta if within the
829 // cluster. Duration for the last block in a cluster is estimated
830 // independently for each track in the cluster. For video tracks we use the
831 // maximum seen so far. For audio we use the the minimum.
832 // TODO(chcunningham): Move audio over to use the maximum.
833 const BlockInfo kBlockInfo1[] = {
834 {kAudioTrackNum, 0, -23, false, NULL, 0},
835 {kAudioTrackNum, 23, -22, false, NULL, 0},
836 {kVideoTrackNum, 33, -33, false, NULL, 0},
837 {kAudioTrackNum, 45, -23, false, NULL, 0},
838 {kVideoTrackNum, 66, -34, false, NULL, 0},
839 // Estimated from minimum audio dur
840 {kAudioTrackNum, 68, -22, false, NULL, 0},
841 // Estimated from maximum video dur
842 {kVideoTrackNum, 100, -34, false, NULL, 0},
845 int block_count1 = arraysize(kBlockInfo1);
846 scoped_ptr<Cluster> cluster1(CreateCluster(0, kBlockInfo1, block_count1));
848 // Send slightly less than the first full cluster so all but the last video
849 // block is parsed. Verify the last fully parsed audio and video buffer are
850 // both missing from the result (parser should hold them aside for duration
851 // estimation prior to end of cluster detection in the absence of
852 // DefaultDurations.)
853 int result = parser_->Parse(cluster1->data(), cluster1->size() - 1);
854 EXPECT_GT(result, 0);
855 EXPECT_LT(result, cluster1->size());
856 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo1, block_count1 - 3));
857 EXPECT_EQ(3UL, parser_->GetAudioBuffers().size());
858 EXPECT_EQ(1UL, parser_->GetVideoBuffers().size());
860 parser_->Reset();
862 // Now parse the full first cluster and verify all the blocks are parsed.
863 result = parser_->Parse(cluster1->data(), cluster1->size());
864 EXPECT_EQ(cluster1->size(), result);
865 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo1, block_count1));
867 // Verify that the estimated frame duration is tracked across clusters for
868 // each track.
869 const BlockInfo kBlockInfo2[] = {
870 {kAudioTrackNum, 200, -22, false, NULL, 0},
871 {kVideoTrackNum, 201, -34, false, NULL, 0},
874 int block_count2 = arraysize(kBlockInfo2);
875 scoped_ptr<Cluster> cluster2(CreateCluster(0, kBlockInfo2, block_count2));
876 result = parser_->Parse(cluster2->data(), cluster2->size());
877 EXPECT_EQ(cluster2->size(), result);
878 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo2, block_count2));
881 // TODO(wolenetz): Is parser behavior correct? See http://crbug.com/363433.
882 TEST_F(WebMClusterParserTest,
883 ParseWithDefaultDurationsBlockGroupsWithoutDurations) {
884 InSequence s;
885 ResetParserToHaveDefaultDurations();
887 EXPECT_LT(kTestAudioFrameDefaultDurationInMs, 23);
888 EXPECT_LT(kTestVideoFrameDefaultDurationInMs, 33);
890 const BlockInfo kBlockInfo[] = {
891 {kAudioTrackNum, 0, -kTestAudioFrameDefaultDurationInMs, false, NULL, 0},
892 {kAudioTrackNum, 23, -kTestAudioFrameDefaultDurationInMs, false, NULL, 0},
893 {kVideoTrackNum, 33, -kTestVideoFrameDefaultDurationInMs, false, NULL, 0},
894 {kAudioTrackNum, 46, -kTestAudioFrameDefaultDurationInMs, false, NULL, 0},
895 {kVideoTrackNum, 67, -kTestVideoFrameDefaultDurationInMs, false, NULL, 0},
896 {kAudioTrackNum, 69, -kTestAudioFrameDefaultDurationInMs, false, NULL, 0},
897 {kVideoTrackNum,
898 100,
899 -kTestVideoFrameDefaultDurationInMs,
900 false,
901 NULL,
905 int block_count = arraysize(kBlockInfo);
906 scoped_ptr<Cluster> cluster(CreateCluster(0, kBlockInfo, block_count));
908 // Send slightly less than the full cluster so all but the last block is
909 // parsed. None should be held aside for duration estimation prior to end of
910 // cluster detection because all the tracks have DefaultDurations.
911 int result = parser_->Parse(cluster->data(), cluster->size() - 1);
912 EXPECT_GT(result, 0);
913 EXPECT_LT(result, cluster->size());
914 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo, block_count - 1));
916 parser_->Reset();
918 // Now parse a whole cluster to verify that all the blocks will get parsed.
919 result = parser_->Parse(cluster->data(), cluster->size());
920 EXPECT_EQ(cluster->size(), result);
921 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo, block_count));
924 TEST_F(WebMClusterParserTest,
925 ParseDegenerateClusterYieldsHardcodedEstimatedDurations) {
926 const BlockInfo kBlockInfo[] = {
928 kAudioTrackNum,
930 WebMClusterParser::kDefaultAudioBufferDurationInMs,
931 true
932 }, {
933 kVideoTrackNum,
935 WebMClusterParser::kDefaultVideoBufferDurationInMs,
936 true
940 int block_count = arraysize(kBlockInfo);
941 scoped_ptr<Cluster> cluster(CreateCluster(0, kBlockInfo, block_count));
942 int result = parser_->Parse(cluster->data(), cluster->size());
943 EXPECT_EQ(cluster->size(), result);
944 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo, block_count));
947 TEST_F(WebMClusterParserTest,
948 ParseDegenerateClusterWithDefaultDurationsYieldsDefaultDurations) {
949 ResetParserToHaveDefaultDurations();
951 const BlockInfo kBlockInfo[] = {
952 { kAudioTrackNum, 0, kTestAudioFrameDefaultDurationInMs, true },
953 { kVideoTrackNum, 0, kTestVideoFrameDefaultDurationInMs, true },
956 int block_count = arraysize(kBlockInfo);
957 scoped_ptr<Cluster> cluster(CreateCluster(0, kBlockInfo, block_count));
958 int result = parser_->Parse(cluster->data(), cluster->size());
959 EXPECT_EQ(cluster->size(), result);
960 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo, block_count));
963 TEST_F(WebMClusterParserTest, ReadOpusDurationsSimpleBlockAtEndOfCluster) {
964 // Reset parser to expect Opus codec audio.
965 parser_.reset(CreateParserWithKeyIdsAndAudioCodec(std::string(),
966 std::string(), kCodecOpus));
968 int loop_count = 0;
969 for (const auto* packet_ptr : BuildAllOpusPackets()) {
970 const BlockInfo kBlockInfo[] = {{kAudioTrackNum,
972 packet_ptr->duration_ms(),
973 true, // Make it a SimpleBlock.
974 packet_ptr->data(),
975 packet_ptr->size()}};
977 int block_count = arraysize(kBlockInfo);
978 scoped_ptr<Cluster> cluster(CreateCluster(0, kBlockInfo, block_count));
979 int result = parser_->Parse(cluster->data(), cluster->size());
980 EXPECT_EQ(cluster->size(), result);
981 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo, block_count));
982 loop_count++;
985 // Test should minimally cover all the combinations of config and frame count.
986 ASSERT_GE(loop_count, kNumPossibleOpusConfigs * kMaxOpusPacketFrameCount);
989 TEST_F(WebMClusterParserTest, PreferOpusDurationsOverBlockDurations) {
990 // Reset parser to expect Opus codec audio.
991 parser_.reset(CreateParserWithKeyIdsAndAudioCodec(std::string(),
992 std::string(), kCodecOpus));
994 int loop_count = 0;
995 for (const auto* packet_ptr : BuildAllOpusPackets()) {
996 // Setting BlockDuration != Opus duration to see which one the parser uses.
997 int block_duration_ms = packet_ptr->duration_ms() + 10;
999 BlockInfo block_infos[] = {{kAudioTrackNum,
1001 block_duration_ms,
1002 false, // Not a SimpleBlock.
1003 packet_ptr->data(),
1004 packet_ptr->size()}};
1006 int block_count = arraysize(block_infos);
1007 scoped_ptr<Cluster> cluster(CreateCluster(0, block_infos, block_count));
1008 int result = parser_->Parse(cluster->data(), cluster->size());
1009 EXPECT_EQ(cluster->size(), result);
1011 // BlockInfo duration will be used to verify buffer duration, so changing
1012 // duration to be that of the Opus packet to verify it was preferred.
1013 block_infos[0].duration = packet_ptr->duration_ms();
1015 ASSERT_TRUE(VerifyBuffers(parser_, block_infos, block_count));
1016 loop_count++;
1019 // Test should minimally cover all the combinations of config and frame count.
1020 ASSERT_GE(loop_count, kNumPossibleOpusConfigs * kMaxOpusPacketFrameCount);
1023 // Tests that BlockDuration is used to set duration on buffer rather than
1024 // encoded duration in Opus packet (or hard coded duration estimates). Encoded
1025 // Opus duration is usually preferred but cannot be known when encrypted.
1026 TEST_F(WebMClusterParserTest, DontReadEncodedDurationWhenEncrypted) {
1027 // Non-empty dummy value signals encryption is active for audio.
1028 std::string audio_encryption_id("audio_key_id");
1030 // Reset parser to expect Opus codec audio and use audio encryption key id.
1031 parser_.reset(CreateParserWithKeyIdsAndAudioCodec(audio_encryption_id,
1032 std::string(), kCodecOpus));
1034 // Single Block with BlockDuration and encrypted data.
1035 const BlockInfo kBlockInfo[] = {{kAudioTrackNum,
1037 kTestAudioFrameDefaultDurationInMs,
1038 false, // Not a SimpleBlock
1039 kEncryptedFrame, // Encrypted frame data
1040 arraysize(kEncryptedFrame)}};
1042 int block_count = arraysize(kBlockInfo);
1043 scoped_ptr<Cluster> cluster(CreateCluster(0, kBlockInfo, block_count));
1044 int result = parser_->Parse(cluster->data(), cluster->size());
1045 EXPECT_EQ(cluster->size(), result);
1047 // Will verify that duration of buffer matches that of BlockDuration.
1048 ASSERT_TRUE(VerifyBuffers(parser_, kBlockInfo, block_count));
1051 } // namespace media