Add support for FLAC codec

Implemented according to spec:
  https://github.com/xiph/flac/blob/master/doc/isoflac.txt

Closes #345.

Change-Id: If2e277d8eac4baff5965faa0e13c44c334f6184d

README.md

@@ -32,6 +32,7 @@ Shaka Packager supports:
   |       AAC         |    I / O     |      -       |     I / O    |      I      |
   |  Dolby AC3/EAC3   |    I / O     |      -       |       I      |      -      |
   |       DTS         |    I / O     |      -       |       -      |      -      |
+  |       FLAC        |    I / O     |      -       |       -      |      -      |
   |       Opus        |   *I / O*    |    I / O     |       -      |      -      |
   |      Vorbis       |      -       |    I / O     |       -      |      -      |
 

packager/app/test/packager_test.py

@@ -1071,6 +1071,17 @@ class PackagerFunctionalTest(PackagerAppTest):
     self._VerifyDecryption(self.output[0], 'bear-320x240-opus-golden.mp4')
     self._VerifyDecryption(self.output[1], 'bear-320x240-vp9-golden.mp4')
 
+  def testPackageFlacWithEncryption(self):
+    streams = [
+        self._GetStream(
+            'audio', output_format='mp4', test_file='bear-flac.mp4'),
+    ]
+    flags = self._GetFlags(encryption=True)
+
+    self.assertPackageSuccess(streams, flags)
+    self._CheckTestResults('flac-with-encryption')
+    self._VerifyDecryption(self.output[0], 'bear-flac-golden.mp4')
+
   def testPackageWvmInput(self):
     self.encryption_key = '9248d245390e0a49d483ba9b43fc69c3'
     self.assertPackageSuccess(

packager/app/test/testdata/flac-with-encryption/output.mpd

@@ -0,0 +1,19 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!--Generated with https://github.com/google/shaka-packager version <tag>-<hash>-<test>-->
+<MPD xmlns="urn:mpeg:dash:schema:mpd:2011" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xlink="http://www.w3.org/1999/xlink" xsi:schemaLocation="urn:mpeg:dash:schema:mpd:2011 DASH-MPD.xsd" xmlns:cenc="urn:mpeg:cenc:2013" profiles="urn:mpeg:dash:profile:isoff-on-demand:2011" minBufferTime="PT2S" type="static" mediaPresentationDuration="PT2.739954710006714S">
+  <Period id="0">
+    <AdaptationSet id="0" contentType="audio" subsegmentAlignment="true">
+      <ContentProtection value="cenc" schemeIdUri="urn:mpeg:dash:mp4protection:2011" cenc:default_KID="31323334-3536-3738-3930-313233343536"/>
+      <ContentProtection schemeIdUri="urn:uuid:1077efec-c0b2-4d02-ace3-3c1e52e2fb4b">
+        <cenc:pssh>AAAANHBzc2gBAAAAEHfv7MCyTQKs4zweUuL7SwAAAAExMjM0NTY3ODkwMTIzNDU2AAAAAA==</cenc:pssh>
+      </ContentProtection>
+      <Representation id="0" bandwidth="672924" codecs="flac" mimeType="audio/mp4" audioSamplingRate="44100">
+        <AudioChannelConfiguration schemeIdUri="urn:mpeg:dash:23003:3:audio_channel_configuration:2011" value="2"/>
+        <BaseURL>bear-flac-audio.mp4</BaseURL>
+        <SegmentBase indexRange="983-1050" timescale="44100">
+          <Initialization range="0-982"/>
+        </SegmentBase>
+      </Representation>
+    </AdaptationSet>
+  </Period>
+</MPD>

packager/media/base/audio_stream_info.cc

@@ -37,6 +37,8 @@ std::string AudioCodecToString(Codec codec) {
       return "DTS+";
     case kCodecEAC3:
       return "EAC3";
+    case kCodecFlac:
+      return "FLAC";
     case kCodecOpus:
       return "Opus";
     case kCodecVorbis:
@@ -99,28 +101,30 @@ std::unique_ptr<StreamInfo> AudioStreamInfo::Clone() const {
 std::string AudioStreamInfo::GetCodecString(Codec codec,
                                             uint8_t audio_object_type) {
   switch (codec) {
-    case kCodecVorbis:
-      return "vorbis";
-    case kCodecOpus:
-      return "opus";
     case kCodecAAC:
       return "mp4a.40." + base::UintToString(audio_object_type);
+    case kCodecAC3:
+      return "ac-3";
     case kCodecDTSC:
       return "dtsc";
+    case kCodecDTSE:
+      return "dtse";
     case kCodecDTSH:
       return "dtsh";
     case kCodecDTSL:
       return "dtsl";
-    case kCodecDTSE:
-      return "dtse";
-    case kCodecDTSP:
-      return "dts+";
     case kCodecDTSM:
       return "dts-";
-    case kCodecAC3:
+    case kCodecDTSP:
-      return "ac-3";
+      return "dts+";
     case kCodecEAC3:
       return "ec-3";
+    case kCodecFlac:
+      return "flac";
+    case kCodecOpus:
+      return "opus";
+    case kCodecVorbis:
+      return "vorbis";
     default:
       NOTIMPLEMENTED() << "Codec: " << codec;
       return "unknown";

packager/media/base/fourccs.h

@@ -42,6 +42,7 @@ enum FourCC : uint32_t {
   FOURCC_dash = 0x64617368,
   FOURCC_ddts = 0x64647473,
   FOURCC_dec3 = 0x64656333,
+  FOURCC_dfLa = 0x64664c61,
   FOURCC_dinf = 0x64696e66,
   FOURCC_dref = 0x64726566,
   FOURCC_dtsc = 0x64747363,
@@ -57,6 +58,7 @@ enum FourCC : uint32_t {
   FOURCC_enca = 0x656e6361,
   FOURCC_encv = 0x656e6376,
   FOURCC_esds = 0x65736473,
+  FOURCC_fLaC = 0x664c6143,
   FOURCC_free = 0x66726565,
   FOURCC_frma = 0x66726d61,
   FOURCC_ftyp = 0x66747970,

packager/media/base/stream_info.h

@@ -46,6 +46,7 @@ enum Codec {
   kCodecDTSM,
   kCodecDTSP,
   kCodecEAC3,
+  kCodecFlac,
   kCodecOpus,
   kCodecVorbis,
   kCodecAudioMaxPlusOne,

packager/media/formats/mp4/box_definitions.cc

@@ -1774,6 +1774,27 @@ size_t OpusSpecific::ComputeSizeInternal() {
          kOpusMagicSignatureSize;
 }
 
+FlacSpecific::FlacSpecific() {}
+FlacSpecific::~FlacSpecific() {}
+
+FourCC FlacSpecific::BoxType() const {
+  return FOURCC_dfLa;
+}
+
+bool FlacSpecific::ReadWriteInternal(BoxBuffer* buffer) {
+  RCHECK(ReadWriteHeaderInternal(buffer));
+  size_t size = buffer->Reading() ? buffer->BytesLeft() : data.size();
+  RCHECK(buffer->ReadWriteVector(&data, size));
+  return true;
+}
+
+size_t FlacSpecific::ComputeSizeInternal() {
+  // This box is optional. Skip it if not initialized.
+  if (data.empty())
+    return 0;
+  return HeaderSize() + data.size();
+}
+
 AudioSampleEntry::AudioSampleEntry()
     : format(FOURCC_NULL),
       data_reference_index(1),
@@ -1818,6 +1839,7 @@ bool AudioSampleEntry::ReadWriteInternal(BoxBuffer* buffer) {
   RCHECK(buffer->TryReadWriteChild(&dac3));
   RCHECK(buffer->TryReadWriteChild(&dec3));
   RCHECK(buffer->TryReadWriteChild(&dops));
+  RCHECK(buffer->TryReadWriteChild(&dfla));
 
   // Somehow Edge does not support having sinf box before codec_configuration,
   // box, so just do it in the end of AudioSampleEntry. See
@@ -1842,7 +1864,8 @@ size_t AudioSampleEntry::ComputeSizeInternal() {
   return HeaderSize() + sizeof(data_reference_index) + sizeof(channelcount) +
          sizeof(samplesize) + sizeof(samplerate) + sinf.ComputeSize() +
          esds.ComputeSize() + ddts.ComputeSize() + dac3.ComputeSize() +
-         dec3.ComputeSize() + dops.ComputeSize() +
+         dec3.ComputeSize() + dops.ComputeSize() + dfla.ComputeSize() +
+         // Reserved and predefined bytes.
          6 + 8 +  // 6 + 8 bytes reserved.
          4;       // 4 bytes predefined.
 }

packager/media/formats/mp4/box_definitions.h

@@ -342,6 +342,15 @@ struct OpusSpecific : Box {
   uint16_t preskip;
 };
 
+// FLAC specific decoder configuration box:
+//   https://github.com/xiph/flac/blob/master/doc/isoflac.txt
+// We do not care about the actual data inside, which is simply copied over.
+struct FlacSpecific : FullBox {
+  DECLARE_BOX_METHODS(FlacSpecific);
+
+  std::vector<uint8_t> data;
+};
+
 struct AudioSampleEntry : Box {
   DECLARE_BOX_METHODS(AudioSampleEntry);
   // Returns actual format of this sample entry.
@@ -362,6 +371,7 @@ struct AudioSampleEntry : Box {
   AC3Specific dac3;
   EC3Specific dec3;
   OpusSpecific dops;
+  FlacSpecific dfla;
 };
 
 struct WebVTTConfigurationBox : Box {

packager/media/formats/mp4/box_definitions_unittest.cc

@@ -434,6 +434,16 @@ class BoxDefinitionsTestGeneral : public testing::Test {
         dops->opus_identification_header.size() - 1);
   }
 
+  void Fill(FlacSpecific* dfla) {
+    const uint8_t kFlacData[] = {0x50, 0x11, 0x60};
+    dfla->data.assign(std::begin(kFlacData), std::end(kFlacData));
+  }
+
+  void Modify(FlacSpecific* dfla) {
+    const uint8_t kFlacData[] = {0x50, 0x11, 0x40};
+    dfla->data.assign(std::begin(kFlacData), std::end(kFlacData));
+  }
+
   void Fill(AudioSampleEntry* enca) {
     enca->format = FOURCC_enca;
     enca->data_reference_index = 2;
@@ -1234,6 +1244,21 @@ TEST_F(BoxDefinitionsTest, OpusSampleEntry) {
   ASSERT_EQ(entry, entry_readback);
 }
 
+TEST_F(BoxDefinitionsTest, FlacSampleEntry) {
+  AudioSampleEntry entry;
+  entry.format = FOURCC_fLaC;
+  entry.data_reference_index = 2;
+  entry.channelcount = 5;
+  entry.samplesize = 16;
+  entry.samplerate = 44100;
+  Fill(&entry.dfla);
+  entry.Write(this->buffer_.get());
+
+  AudioSampleEntry entry_readback;
+  ASSERT_TRUE(ReadBack(&entry_readback));
+  ASSERT_EQ(entry, entry_readback);
+}
+
 TEST_F(BoxDefinitionsTest, CompactSampleSize_FieldSize16) {
   CompactSampleSize stz2;
   stz2.field_size = 16;

packager/media/formats/mp4/mp4_media_parser.cc

@@ -89,6 +89,8 @@ Codec FourCCToCodec(FourCC fourcc) {
       return kCodecAC3;
     case FOURCC_ec_3:
       return kCodecEAC3;
+    case FOURCC_fLaC:
+      return kCodecFlac;
     default:
       return kUnknownCodec;
   }
@@ -425,12 +427,12 @@ bool MP4MediaParser::ParseMoov(BoxReader* reader) {
           break;
         case FOURCC_dtsc:
           FALLTHROUGH_INTENDED;
+        case FOURCC_dtse:
+          FALLTHROUGH_INTENDED;
         case FOURCC_dtsh:
           FALLTHROUGH_INTENDED;
         case FOURCC_dtsl:
           FALLTHROUGH_INTENDED;
-        case FOURCC_dtse:
-          FALLTHROUGH_INTENDED;
         case FOURCC_dtsm:
           codec_config = entry.ddts.extra_data;
           max_bitrate = entry.ddts.max_bitrate;
@@ -448,6 +450,11 @@ bool MP4MediaParser::ParseMoov(BoxReader* reader) {
           num_channels = static_cast<uint8_t>(GetEc3NumChannels(codec_config));
           sampling_frequency = entry.samplerate;
           break;
+        case FOURCC_fLaC:
+          codec_config = entry.dfla.data;
+          num_channels = entry.channelcount;
+          sampling_frequency = entry.samplerate;
+          break;
         case FOURCC_Opus:
           codec_config = entry.dops.opus_identification_header;
           num_channels = entry.channelcount;

packager/media/formats/mp4/mp4_muxer.cc

@@ -73,6 +73,8 @@ FourCC CodecToFourCC(Codec codec, H26xStreamFormat h26x_stream_format) {
       return FOURCC_dtsm;
     case kCodecEAC3:
       return FOURCC_ec_3;
+    case kCodecFlac:
+      return FOURCC_fLaC;
     case kCodecOpus:
       return FOURCC_Opus;
     default:
@@ -367,6 +369,9 @@ void MP4Muxer::GenerateAudioTrak(const AudioStreamInfo* audio_info,
     case kCodecEAC3:
       audio.dec3.data = audio_info->codec_config();
       break;
+    case kCodecFlac:
+      audio.dfla.data = audio_info->codec_config();
+      break;
     case kCodecOpus:
       audio.dops.opus_identification_header = audio_info->codec_config();
       break;

packager/media/test/data/README

@@ -41,6 +41,10 @@ bear-640x360-trailing-moov-additional-mdat.mp4  - Same content, but with moov bo
 bear-640x360-av_frag.mp4  - Same content, but in fragmented mp4.
 bear-640x360-aac_lc-silent_right.mp4  - Audio only, stereo, but right channel is silent, with AAC-LC profile.
 bear-640x360-aac_he-silent_right.mp4  - Same as above, but with AAC-HE profile.
+bear-flac.mp4 - Unfragmented audio-only 44.1kHz FLAC in MP4 file, created using:
+  ffmpeg -i bear-1280x720.mp4 -map 0:0 -acodec flac -strict -2 bear-flac.mp4
+  Note, "-strict -2" was required because current ffmpeg libavformat version
+  57.75.100 indicates that flac in MP4 support is experimental.
 
 // Non square pixels.
 bear-640x360-non_square_pixel-with_pasp.mp4 - A non-square pixel version of the video track of bear-640x360.mp4 with PixelAspectRatio box.