media/formats/mp4/mp4_stream_parser_unittest.cc

   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.
   4
   5 #include <algorithm>
   6 #include <string>
   7
   8 #include "base/bind.h"
   9 #include "base/bind_helpers.h"
  10 #include "base/logging.h"
  11 #include "base/memory/ref_counted.h"
  12 #include "base/time/time.h"
  13 #include "media/base/audio_decoder_config.h"
  14 #include "media/base/decoder_buffer.h"
  15 #include "media/base/stream_parser_buffer.h"
  16 #include "media/base/test_data_util.h"
  17 #include "media/base/text_track_config.h"
  18 #include "media/base/video_decoder_config.h"
  19 #include "media/formats/mp4/es_descriptor.h"
  20 #include "media/formats/mp4/mp4_stream_parser.h"
  21 #include "testing/gtest/include/gtest/gtest.h"
  22
  23 using base::TimeDelta;
  24
  25 namespace media {
  26 namespace mp4 {
  27
  28 static const char kCencInitDataType[] = "cenc";
  29
  30 class MP4StreamParserTest : public testing::Test {
  31  public:
  32   MP4StreamParserTest()
  33       : configs_received_(false),
  34         lower_bound_(
  35             DecodeTimestamp::FromPresentationTime(base::TimeDelta::Max())) {
  36     std::set<int> audio_object_types;
  37     audio_object_types.insert(kISO_14496_3);
  38     parser_.reset(new MP4StreamParser(audio_object_types, false));
  39   }
  40
  41  protected:
  42   scoped_ptr<MP4StreamParser> parser_;
  43   bool configs_received_;
  44   DecodeTimestamp lower_bound_;
  45
  46   bool AppendData(const uint8* data, size_t length) {
  47     return parser_->Parse(data, length);
  48   }
  49
  50   bool AppendDataInPieces(const uint8* data, size_t length, size_t piece_size) {
  51     const uint8* start = data;
  52     const uint8* end = data + length;
  53     while (start < end) {
  54       size_t append_size = std::min(piece_size,
  55                                     static_cast<size_t>(end - start));
  56       if (!AppendData(start, append_size))
  57         return false;
  58       start += append_size;
  59     }
  60     return true;
  61   }
  62
  63   void InitF(const StreamParser::InitParameters& params) {
  64     DVLOG(1) << "InitF: dur=" << params.duration.InMilliseconds()
  65              << ", autoTimestampOffset=" << params.auto_update_timestamp_offset;
  66   }
  67
  68   bool NewConfigF(const AudioDecoderConfig& ac,
  69                   const VideoDecoderConfig& vc,
  70                   const StreamParser::TextTrackConfigMap& tc) {
  71     DVLOG(1) << "NewConfigF: audio=" << ac.IsValidConfig()
  72              << ", video=" << vc.IsValidConfig();
  73     configs_received_ = true;
  74     return true;
  75   }
  76
  77   void DumpBuffers(const std::string& label,
  78                    const StreamParser::BufferQueue& buffers) {
  79     DVLOG(2) << "DumpBuffers: " << label << " size " << buffers.size();
  80     for (StreamParser::BufferQueue::const_iterator buf = buffers.begin();
  81          buf != buffers.end(); buf++) {
  82       DVLOG(3) << "  n=" << buf - buffers.begin()
  83                << ", size=" << (*buf)->data_size()
  84                << ", dur=" << (*buf)->duration().InMilliseconds();
  85     }
  86   }
  87
  88   bool NewBuffersF(const StreamParser::BufferQueue& audio_buffers,
  89                    const StreamParser::BufferQueue& video_buffers,
  90                    const StreamParser::TextBufferQueueMap& text_map) {
  91     DumpBuffers("audio_buffers", audio_buffers);
  92     DumpBuffers("video_buffers", video_buffers);
  93
  94     // TODO(wolenetz/acolwell): Add text track support to more MSE parsers. See
  95     // http://crbug.com/336926.
  96     if (!text_map.empty())
  97       return false;
  98
  99     // Find the second highest timestamp so that we know what the
 100     // timestamps on the next set of buffers must be >= than.
 101     DecodeTimestamp audio = !audio_buffers.empty() ?
 102         audio_buffers.back()->GetDecodeTimestamp() : kNoDecodeTimestamp();
 103     DecodeTimestamp video = !video_buffers.empty() ?
 104         video_buffers.back()->GetDecodeTimestamp() : kNoDecodeTimestamp();
 105     DecodeTimestamp second_highest_timestamp =
 106         (audio == kNoDecodeTimestamp() ||
 107          (video != kNoDecodeTimestamp() && audio > video)) ? video : audio;
 108
 109     DCHECK(second_highest_timestamp != kNoDecodeTimestamp());
 110
 111     if (lower_bound_ != kNoDecodeTimestamp() &&
 112         second_highest_timestamp < lower_bound_) {
 113       return false;
 114     }
 115
 116     lower_bound_ = second_highest_timestamp;
 117     return true;
 118   }
 119
 120   void KeyNeededF(const std::string& type,
 121                   const std::vector<uint8>& init_data) {
 122     DVLOG(1) << "KeyNeededF: " << init_data.size();
 123     EXPECT_EQ(kCencInitDataType, type);
 124     EXPECT_FALSE(init_data.empty());
 125   }
 126
 127   void NewSegmentF() {
 128     DVLOG(1) << "NewSegmentF";
 129     lower_bound_ = kNoDecodeTimestamp();
 130   }
 131
 132   void EndOfSegmentF() {
 133     DVLOG(1) << "EndOfSegmentF()";
 134     lower_bound_ =
 135         DecodeTimestamp::FromPresentationTime(base::TimeDelta::Max());
 136   }
 137
 138   void InitializeParser() {
 139     parser_->Init(
 140         base::Bind(&MP4StreamParserTest::InitF, base::Unretained(this)),
 141         base::Bind(&MP4StreamParserTest::NewConfigF, base::Unretained(this)),
 142         base::Bind(&MP4StreamParserTest::NewBuffersF, base::Unretained(this)),
 143         true,
 144         base::Bind(&MP4StreamParserTest::KeyNeededF, base::Unretained(this)),
 145         base::Bind(&MP4StreamParserTest::NewSegmentF, base::Unretained(this)),
 146         base::Bind(&MP4StreamParserTest::EndOfSegmentF,
 147                    base::Unretained(this)),
 148         LogCB());
 149   }
 150
 151   bool ParseMP4File(const std::string& filename, int append_bytes) {
 152     InitializeParser();
 153
 154     scoped_refptr<DecoderBuffer> buffer = ReadTestDataFile(filename);
 155     EXPECT_TRUE(AppendDataInPieces(buffer->data(),
 156                                    buffer->data_size(),
 157                                    append_bytes));
 158     return true;
 159   }
 160 };
 161
 162 TEST_F(MP4StreamParserTest, UnalignedAppend) {
 163   // Test small, non-segment-aligned appends (small enough to exercise
 164   // incremental append system)
 165   ParseMP4File("bear-1280x720-av_frag.mp4", 512);
 166 }
 167
 168 TEST_F(MP4StreamParserTest, BytewiseAppend) {
 169   // Ensure no incremental errors occur when parsing
 170   ParseMP4File("bear-1280x720-av_frag.mp4", 1);
 171 }
 172
 173 TEST_F(MP4StreamParserTest, MultiFragmentAppend) {
 174   // Large size ensures multiple fragments are appended in one call (size is
 175   // larger than this particular test file)
 176   ParseMP4File("bear-1280x720-av_frag.mp4", 768432);
 177 }
 178
 179 TEST_F(MP4StreamParserTest, Flush) {
 180   // Flush while reading sample data, then start a new stream.
 181   InitializeParser();
 182
 183   scoped_refptr<DecoderBuffer> buffer =
 184       ReadTestDataFile("bear-1280x720-av_frag.mp4");
 185   EXPECT_TRUE(AppendDataInPieces(buffer->data(), 65536, 512));
 186   parser_->Flush();
 187   EXPECT_TRUE(AppendDataInPieces(buffer->data(),
 188                                  buffer->data_size(),
 189                                  512));
 190 }
 191
 192 TEST_F(MP4StreamParserTest, Reinitialization) {
 193   InitializeParser();
 194
 195   scoped_refptr<DecoderBuffer> buffer =
 196       ReadTestDataFile("bear-1280x720-av_frag.mp4");
 197   EXPECT_TRUE(AppendDataInPieces(buffer->data(),
 198                                  buffer->data_size(),
 199                                  512));
 200   EXPECT_TRUE(AppendDataInPieces(buffer->data(),
 201                                  buffer->data_size(),
 202                                  512));
 203 }
 204
 205 TEST_F(MP4StreamParserTest, MPEG2_AAC_LC) {
 206   std::set<int> audio_object_types;
 207   audio_object_types.insert(kISO_13818_7_AAC_LC);
 208   parser_.reset(new MP4StreamParser(audio_object_types, false));
 209   ParseMP4File("bear-mpeg2-aac-only_frag.mp4", 512);
 210 }
 211
 212 // Test that a moov box is not always required after Flush() is called.
 213 TEST_F(MP4StreamParserTest, NoMoovAfterFlush) {
 214   InitializeParser();
 215
 216   scoped_refptr<DecoderBuffer> buffer =
 217       ReadTestDataFile("bear-1280x720-av_frag.mp4");
 218   EXPECT_TRUE(AppendDataInPieces(buffer->data(),
 219                                  buffer->data_size(),
 220                                  512));
 221   parser_->Flush();
 222
 223   const int kFirstMoofOffset = 1307;
 224   EXPECT_TRUE(AppendDataInPieces(buffer->data() + kFirstMoofOffset,
 225                                  buffer->data_size() - kFirstMoofOffset,
 226                                  512));
 227 }
 228
 229 // Test an invalid file where there are encrypted samples, but
 230 // SampleAuxiliaryInformation{Sizes|Offsets}Box (saiz|saio) are missing.
 231 // The parser should fail instead of crash. See http://crbug.com/361347
 232 TEST_F(MP4StreamParserTest, MissingSampleAuxInfo) {
 233   InitializeParser();
 234
 235   scoped_refptr<DecoderBuffer> buffer =
 236       ReadTestDataFile("bear-1280x720-a_frag-cenc_missing-saiz-saio.mp4");
 237   EXPECT_FALSE(AppendDataInPieces(buffer->data(), buffer->data_size(), 512));
 238 }
 239
 240 // Test a file where all video samples start with an Access Unit
 241 // Delimiter (AUD) NALU.
 242 TEST_F(MP4StreamParserTest, VideoSamplesStartWithAUDs) {
 243   ParseMP4File("bear-1280x720-av_with-aud-nalus_frag.mp4", 512);
 244 }
 245
 246 // TODO(strobe): Create and test media which uses CENC auxiliary info stored
 247 // inside a private box
 248
 249 }  // namespace mp4
 250 }  // namespace media