Fix avc3 codec string problem

- Added wvm decryption and demuxing test
- Added avc3 test (not strip parameter set nal units)

Change-Id: I28ff4d7f9ef539eafe45cfadfc0912cd11c72b05

packager/app/test/packager_test.py

@@ -354,6 +354,34 @@ class PackagerAppTest(unittest.TestCase):
     self._VerifyDecryption(self.output[0], 'bear-320x240-opus-golden.mp4')
     self._VerifyDecryption(self.output[1], 'bear-320x240-vp9-golden.mp4')
 
+  def testPackageWvmInput(self):
+    self.packager.Package(
+        self._GetStreams(
+            ['0', '1', '2', '3'], test_files=['bear-multi-configs.wvm']),
+        self._GetFlags(
+            decryption=True, encryption_key='9248d245390e0a49d483ba9b43fc69c3'))
+    # Output timescale is 90000.
+    self._DiffGold(self.output[0], 'bear-320x180-v-wvm-golden.mp4')
+    self._DiffGold(self.output[1], 'bear-320x180-a-wvm-golden.mp4')
+    self._DiffGold(self.output[2], 'bear-640x360-v-wvm-golden.mp4')
+    self._DiffGold(self.output[3], 'bear-640x360-a-wvm-golden.mp4')
+    self._DiffGold(self.mpd_output, 'bear-wvm-golden.mpd')
+
+  def testPackageWvmInputWithoutStrippingParameterSetNalus(self):
+    self.packager.Package(
+        self._GetStreams(
+            ['0', '1', '2', '3'], test_files=['bear-multi-configs.wvm']),
+        self._GetFlags(
+            strip_parameter_set_nalus=False,
+            decryption=True,
+            encryption_key='9248d245390e0a49d483ba9b43fc69c3'))
+    # Output timescale is 90000.
+    self._DiffGold(self.output[0], 'bear-320x180-avc3-wvm-golden.mp4')
+    self._DiffGold(self.output[1], 'bear-320x180-a-wvm-golden.mp4')
+    self._DiffGold(self.output[2], 'bear-640x360-avc3-wvm-golden.mp4')
+    self._DiffGold(self.output[3], 'bear-640x360-a-wvm-golden.mp4')
+    self._DiffGold(self.mpd_output, 'bear-avc3-wvm-golden.mpd')
+
   def testPackageWithEncryptionAndRandomIv(self):
     self.packager.Package(
         self._GetStreams(['audio', 'video']),
@@ -582,6 +610,7 @@ class PackagerAppTest(unittest.TestCase):
     return 'mp4'
 
   def _GetFlags(self,
+                strip_parameter_set_nalus=True,
                 encryption=False,
                 protection_scheme=None,
                 vp9_subsample_encryption=True,
@@ -598,6 +627,10 @@ class PackagerAppTest(unittest.TestCase):
                 generate_static_mpd=False,
                 use_fake_clock=True):
     flags = []
+
+    if not strip_parameter_set_nalus:
+      flags += ['--strip_parameter_set_nalus=false']
+
     if widevine_encryption:
       widevine_server_url = ('https://license.uat.widevine.com/cenc'
                              '/getcontentkey/widevine_test')
@@ -624,7 +657,7 @@ class PackagerAppTest(unittest.TestCase):
     if decryption:
       flags += ['--enable_fixed_key_decryption',
                 '--key_id=31323334353637383930313233343536',
-                '--key=32333435363738393021323334353637']
+                '--key=' + encryption_key]
 
     if key_rotation:
       flags.append('--crypto_period_duration=1')

packager/app/test/testdata/bear-avc3-wvm-golden.mpd

@@ -0,0 +1,36 @@
+<?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="PT2.73607S">
+  <Period id="0">
+    <AdaptationSet id="0" contentType="video" maxWidth="640" maxHeight="360" frameRate="90000/3003" subsegmentAlignment="true" par="16:9">
+      <Representation id="0" bandwidth="354240" codecs="avc3.64000d" mimeType="video/mp4" sar="1:1" width="320" height="180">
+        <BaseURL>output_0.mp4</BaseURL>
+        <SegmentBase indexRange="818-885" timescale="90000">
+          <Initialization range="0-817"/>
+        </SegmentBase>
+      </Representation>
+      <Representation id="2" bandwidth="888442" codecs="avc3.64001e" mimeType="video/mp4" sar="1:1" width="640" height="360">
+        <BaseURL>output_2.mp4</BaseURL>
+        <SegmentBase indexRange="819-886" timescale="90000">
+          <Initialization range="0-818"/>
+        </SegmentBase>
+      </Representation>
+    </AdaptationSet>
+    <AdaptationSet id="1" contentType="audio" subsegmentAlignment="true">
+      <Representation id="1" bandwidth="132362" codecs="mp4a.40.2" mimeType="audio/mp4" audioSamplingRate="44100">
+        <AudioChannelConfiguration schemeIdUri="urn:mpeg:dash:23003:3:audio_channel_configuration:2011" value="2"/>
+        <BaseURL>output_1.mp4</BaseURL>
+        <SegmentBase indexRange="757-824" timescale="90000">
+          <Initialization range="0-756"/>
+        </SegmentBase>
+      </Representation>
+      <Representation id="3" bandwidth="136806" codecs="mp4a.40.2" mimeType="audio/mp4" audioSamplingRate="44100">
+        <AudioChannelConfiguration schemeIdUri="urn:mpeg:dash:23003:3:audio_channel_configuration:2011" value="2"/>
+        <BaseURL>output_3.mp4</BaseURL>
+        <SegmentBase indexRange="757-824" timescale="90000">
+          <Initialization range="0-756"/>
+        </SegmentBase>
+      </Representation>
+    </AdaptationSet>
+  </Period>
+</MPD>

packager/app/test/testdata/bear-wvm-golden.mpd

@@ -0,0 +1,36 @@
+<?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="PT2.73607S">
+  <Period id="0">
+    <AdaptationSet id="0" contentType="video" maxWidth="640" maxHeight="360" frameRate="90000/3003" subsegmentAlignment="true" par="16:9">
+      <Representation id="0" bandwidth="353919" codecs="avc1.64000d" mimeType="video/mp4" sar="1:1" width="320" height="180">
+        <BaseURL>output_0.mp4</BaseURL>
+        <SegmentBase indexRange="822-889" timescale="90000">
+          <Initialization range="0-821"/>
+        </SegmentBase>
+      </Representation>
+      <Representation id="2" bandwidth="888108" codecs="avc1.64001e" mimeType="video/mp4" sar="1:1" width="640" height="360">
+        <BaseURL>output_2.mp4</BaseURL>
+        <SegmentBase indexRange="823-890" timescale="90000">
+          <Initialization range="0-822"/>
+        </SegmentBase>
+      </Representation>
+    </AdaptationSet>
+    <AdaptationSet id="1" contentType="audio" subsegmentAlignment="true">
+      <Representation id="1" bandwidth="132362" codecs="mp4a.40.2" mimeType="audio/mp4" audioSamplingRate="44100">
+        <AudioChannelConfiguration schemeIdUri="urn:mpeg:dash:23003:3:audio_channel_configuration:2011" value="2"/>
+        <BaseURL>output_1.mp4</BaseURL>
+        <SegmentBase indexRange="757-824" timescale="90000">
+          <Initialization range="0-756"/>
+        </SegmentBase>
+      </Representation>
+      <Representation id="3" bandwidth="136806" codecs="mp4a.40.2" mimeType="audio/mp4" audioSamplingRate="44100">
+        <AudioChannelConfiguration schemeIdUri="urn:mpeg:dash:23003:3:audio_channel_configuration:2011" value="2"/>
+        <BaseURL>output_3.mp4</BaseURL>
+        <SegmentBase indexRange="757-824" timescale="90000">
+          <Initialization range="0-756"/>
+        </SegmentBase>
+      </Representation>
+    </AdaptationSet>
+  </Period>
+</MPD>

packager/media/codecs/avc_decoder_configuration_record.cc

@@ -88,18 +88,21 @@ bool AVCDecoderConfigurationRecord::ParseInternal() {
   return true;
 }
 
-std::string AVCDecoderConfigurationRecord::GetCodecString() const {
-  return GetCodecString(profile_indication_, profile_compatibility_,
-                        avc_level_);
+std::string AVCDecoderConfigurationRecord::GetCodecString(
+    FourCC codec_fourcc) const {
+  return GetCodecString(codec_fourcc, profile_indication_,
+                        profile_compatibility_, avc_level_);
 }
 
 std::string AVCDecoderConfigurationRecord::GetCodecString(
+    FourCC codec_fourcc,
     uint8_t profile_indication,
     uint8_t profile_compatibility,
     uint8_t avc_level) {
   const uint8_t bytes[] = {profile_indication, profile_compatibility,
                            avc_level};
-  return "avc1." + base::ToLowerASCII(base::HexEncode(bytes, arraysize(bytes)));
+  return FourCCToString(codec_fourcc) + "." +
+         base::ToLowerASCII(base::HexEncode(bytes, arraysize(bytes)));
 }
 
 }  // namespace media

packager/media/codecs/avc_decoder_configuration_record.h

@@ -12,6 +12,7 @@
 #include <vector>
 
 #include "packager/base/macros.h"
+#include "packager/media/base/fourccs.h"
 #include "packager/media/codecs/decoder_configuration_record.h"
 
 namespace shaka {
@@ -24,7 +25,7 @@ class AVCDecoderConfigurationRecord : public DecoderConfigurationRecord {
   ~AVCDecoderConfigurationRecord() override;
 
   /// @return The codec string.
-  std::string GetCodecString() const;
+  std::string GetCodecString(FourCC codec_fourcc) const;
 
   uint8_t version() const { return version_; }
   uint8_t profile_indication() const { return profile_indication_; }
@@ -37,7 +38,8 @@ class AVCDecoderConfigurationRecord : public DecoderConfigurationRecord {
 
   /// Static version of GetCodecString.
   /// @return The codec string.
-  static std::string GetCodecString(uint8_t profile_indication,
+  static std::string GetCodecString(FourCC codec_fourcc,
+                                    uint8_t profile_indication,
                                     uint8_t profile_compatibility,
                                     uint8_t avc_level);
 

packager/media/codecs/avc_decoder_configuration_record_unittest.cc

@@ -32,7 +32,8 @@ TEST(AVCDecoderConfigurationRecordTest, Success) {
   EXPECT_EQ(8u, avc_config.pixel_width());
   EXPECT_EQ(9u, avc_config.pixel_height());
 
-  EXPECT_EQ("avc1.64001e", avc_config.GetCodecString());
+  EXPECT_EQ("avc1.64001e", avc_config.GetCodecString(FOURCC_avc1));
+  EXPECT_EQ("avc3.64001e", avc_config.GetCodecString(FOURCC_avc3));
 }
 
 TEST(AVCDecoderConfigurationRecordTest, FailsOnInvalidNaluLengthSize) {
@@ -56,8 +57,8 @@ TEST(AVCDecoderConfigurationRecordTest, FailOnInsufficientData) {
 }
 
 TEST(AVCDecoderConfigurationRecordTest, GetCodecString) {
-  EXPECT_EQ("avc1.123456",
-            AVCDecoderConfigurationRecord::GetCodecString(0x12, 0x34, 0x56));
+  EXPECT_EQ("avc1.123456", AVCDecoderConfigurationRecord::GetCodecString(
+                               FOURCC_avc1, 0x12, 0x34, 0x56));
 }
 
 }  // namespace media

packager/media/formats/mp2t/es_parser_h264.cc

@@ -148,12 +148,16 @@ bool EsParserH264::UpdateVideoDecoderConfig(int pps_id) {
 
   const uint8_t nalu_length_size =
       H26xByteToUnitStreamConverter::kUnitStreamNaluLengthSize;
+  const H26xStreamFormat stream_format = stream_converter()->stream_format();
+  const FourCC codec_fourcc =
+      stream_format == H26xStreamFormat::kNalUnitStreamWithParameterSetNalus
+          ? FOURCC_avc3
+          : FOURCC_avc1;
   last_video_decoder_config_ = std::make_shared<VideoStreamInfo>(
-      pid(), kMpeg2Timescale, kInfiniteDuration, kCodecH264,
-      stream_converter()->stream_format(),
-      AVCDecoderConfigurationRecord::GetCodecString(decoder_config_record[1],
-                                                    decoder_config_record[2],
-                                                    decoder_config_record[3]),
+      pid(), kMpeg2Timescale, kInfiniteDuration, kCodecH264, stream_format,
+      AVCDecoderConfigurationRecord::GetCodecString(
+          codec_fourcc, decoder_config_record[1], decoder_config_record[2],
+          decoder_config_record[3]),
       decoder_config_record.data(), decoder_config_record.size(), coded_width,
       coded_height, pixel_width, pixel_height, 0, nalu_length_size,
       std::string(), false);

packager/media/formats/mp4/mp4_media_parser.cc

@@ -520,7 +520,7 @@ bool MP4MediaParser::ParseMoov(BoxReader* reader) {
             LOG(ERROR) << "Failed to parse avcc.";
             return false;
           }
-          codec_string = avc_config.GetCodecString();
+          codec_string = avc_config.GetCodecString(actual_format);
           nalu_length_size = avc_config.nalu_length_size();
 
           if (coded_width != avc_config.coded_width() ||

packager/media/formats/wvm/wvm_media_parser.cc

@@ -861,7 +861,13 @@ bool WvmMediaParser::Output(bool output_encrypted_sample) {
                 return false;
               }
             }
-            video_stream_info->set_codec_string(avc_config.GetCodecString());
+            const FourCC codec_fourcc =
+                byte_to_unit_stream_converter_.stream_format() ==
+                        H26xStreamFormat::kNalUnitStreamWithParameterSetNalus
+                    ? FOURCC_avc3
+                    : FOURCC_avc1;
+            video_stream_info->set_codec_string(
+                avc_config.GetCodecString(codec_fourcc));
 
             if (avc_config.pixel_width() != video_stream_info->pixel_width() ||
                 avc_config.pixel_height() !=

packager/media/formats/wvm/wvm_media_parser_unittest.cc

@@ -215,7 +215,7 @@ TEST_F(WvmMediaParserTest, ParseMultiConfigWvm) {
   EXPECT_CALL(*key_source_, GetKey(_, _))
       .WillOnce(DoAll(SetArgPointee<1>(encryption_key_), Return(Status::OK)));
   Parse(kMultiConfigWvmFile);
-  EXPECT_EQ(6u, stream_map_.size());
+  ASSERT_EQ(4u, stream_map_.size());
 
   ASSERT_EQ(kStreamVideo, stream_map_[0]->stream_type());
   VideoStreamInfo* video_info = reinterpret_cast<VideoStreamInfo*>(
@@ -242,18 +242,6 @@ TEST_F(WvmMediaParserTest, ParseMultiConfigWvm) {
   EXPECT_EQ("mp4a.40.2", audio_info->codec_string());
   EXPECT_EQ(2u, audio_info->num_channels());
   EXPECT_EQ(44100u, audio_info->sampling_frequency());
-
-  ASSERT_EQ(kStreamVideo, stream_map_[4]->stream_type());
-  video_info = reinterpret_cast<VideoStreamInfo*>(stream_map_[4].get());
-  EXPECT_EQ("avc1.64001f", video_info->codec_string());
-  EXPECT_EQ(1280u, video_info->width());
-  EXPECT_EQ(720u, video_info->height());
-
-  ASSERT_EQ(kStreamAudio, stream_map_[5]->stream_type());
-  audio_info = reinterpret_cast<AudioStreamInfo*>(stream_map_[5].get());
-  EXPECT_EQ("mp4a.40.2", audio_info->codec_string());
-  EXPECT_EQ(2u, audio_info->num_channels());
-  EXPECT_EQ(48000u, audio_info->sampling_frequency());
 }
 
 }  // namespace wvm