Generate tkhd and pasp boxes properly with non-square pixels

- Width/Height in tkhd box should be display resolution instead of
  coded resolution.
- pasp was not generated earlier.
- Assuming SAR as 1:1 if it failed to be extracted successfully.
- Also update test files for mp4_media_parser_unittest.

This fixes Issue #35.

Change-Id: Iedbe6a44465aac6723b6ea2de30aed180821eccd

packager/media/base/video_stream_info.cc

@@ -83,6 +83,11 @@ VideoStreamInfo::VideoStreamInfo(int track_id,
     DVLOG_IF(2, pixel_width_ == 0 || pixel_height_ == 0)
         << "Failed to extract sar_width and sar_height.";
   }
+  if (pixel_width_ == 0 || pixel_height_ == 0) {
+    LOG(WARNING) << "SAR is not extracted successfully. Assuming 1:1.";
+    pixel_width_ = 1;
+    pixel_height_ = 1;
+  }
 }
 
 VideoStreamInfo::~VideoStreamInfo() {}

packager/media/formats/mp4/box_definitions.cc

@@ -372,9 +372,6 @@ bool TrackHeader::ReadWrite(BoxBuffer* buffer) {
     // Set default value for volume, if track is audio, 0x100 else 0.
     if (volume == -1)
       volume = (width != 0 && height != 0) ? 0 : 0x100;
-    // Convert integer to 16.16 fix point.
-    width <<= 16;
-    height <<= 16;
   }
   std::vector<uint8_t> matrix(kUnityMatrix,
                               kUnityMatrix + arraysize(kUnityMatrix));
@@ -386,9 +383,6 @@ bool TrackHeader::ReadWrite(BoxBuffer* buffer) {
          buffer->ReadWriteVector(&matrix, matrix.size()) &&
          buffer->ReadWriteUInt32(&width) &&
          buffer->ReadWriteUInt32(&height));
-  // Convert 16.16 fixed point to integer.
-  width >>= 16;
-  height >>= 16;
   return true;
 }
 
@@ -1001,8 +995,7 @@ bool VideoSampleEntry::ReadWrite(BoxBuffer* buffer) {
          buffer->ReadWriteUInt16(&video_depth) &&
          buffer->ReadWriteInt16(&predefined));
 
-  RCHECK(buffer->PrepareChildren() &&
+  RCHECK(buffer->PrepareChildren());
-         buffer->TryReadWriteChild(&pixel_aspect));
 
   if (format == FOURCC_ENCV) {
     if (buffer->Reading()) {
@@ -1021,6 +1014,7 @@ bool VideoSampleEntry::ReadWrite(BoxBuffer* buffer) {
       (format == FOURCC_ENCV && sinf.format.format == FOURCC_AVC1)) {
     RCHECK(buffer->ReadWriteChild(&avcc));
   }
+  RCHECK(buffer->TryReadWriteChild(&pixel_aspect));
   return true;
 }
 

packager/media/formats/mp4/box_definitions.h

@@ -134,6 +134,8 @@ struct TrackHeader : FullBox {
   int16_t layer;
   int16_t alternate_group;
   int16_t volume;
+  // width and height specify the track's visual presentation size as
+  // fixed-point 16.16 values.
   uint32_t width;
   uint32_t height;
 };

packager/media/formats/mp4/mp4_media_parser_unittest.cc

@@ -114,15 +114,13 @@ TEST_F(MP4MediaParserTest, UnalignedAppend) {
 // the container has a 'pasp' box.
 TEST_F(MP4MediaParserTest, PixelWidthPixelHeightFromPaspBox) {
   // This content has a 'pasp' box that has the aspect ratio.
-  EXPECT_TRUE(ParseMP4File("bear-1280x720.mp4", 512));
+  EXPECT_TRUE(ParseMP4File("bear-640x360-non_square_pixel-with_pasp.mp4", 512));
 
-  // Track ID 2 has the video stream which should have pixel width and height
+  const int kVideoTrackId = 1;
-  // both 1.
+  EXPECT_EQ(8u,
-  const int kVideoTrackId = 2;
-  EXPECT_EQ(1u,
             reinterpret_cast<VideoStreamInfo*>(stream_map_[kVideoTrackId].get())
                 ->pixel_width());
-  EXPECT_EQ(1u,
+  EXPECT_EQ(9u,
             reinterpret_cast<VideoStreamInfo*>(stream_map_[kVideoTrackId].get())
                 ->pixel_height());
 }
@@ -132,13 +130,11 @@ TEST_F(MP4MediaParserTest, PixelWidthPixelHeightFromPaspBox) {
 // No 'pasp' box.
 TEST_F(MP4MediaParserTest,
        PixelWidthPixelHeightFromAVCDecoderConfigurationRecord) {
-  // This file doesn't have pasp. SPS for the video has
+  // This file doesn't have pasp. The stream should extract pixel width and
-  // sar_width = sar_height = 0. So the stream info should return 1 for both
+  // height from SPS.
-  // pixel_width and pixel_height.
+  EXPECT_TRUE(
-  EXPECT_TRUE(ParseMP4File("hb2_v_frag.mp4", 512));
+      ParseMP4File("bear-640x360-non_square_pixel-without_pasp.mp4", 512));
 
-  // Track ID 1 has the video stream which should have pixel width and height
-  // both 1.
   const int kVideoTrackId = 1;
   EXPECT_EQ(8u,
             reinterpret_cast<VideoStreamInfo*>(stream_map_[kVideoTrackId].get())
@@ -156,10 +152,8 @@ TEST_F(MP4MediaParserTest,
   // This file doesn't have pasp. SPS for the video has
   // sar_width = sar_height = 0. So the stream info should return 1 for both
   // pixel_width and pixel_height.
-  EXPECT_TRUE(ParseMP4File("bear-1280x720-av_frag.mp4", 512));
+  EXPECT_TRUE(ParseMP4File("bear-640x360-av_frag.mp4", 512));
 
-  // Track ID 1 has the video stream which should have pixel width and height
-  // both 1.
   const int kVideoTrackId = 1;
   EXPECT_EQ(1u,
             reinterpret_cast<VideoStreamInfo*>(stream_map_[kVideoTrackId].get())

packager/media/formats/mp4/mp4_muxer.cc

@@ -159,8 +159,14 @@ void MP4Muxer::GenerateVideoTrak(const VideoStreamInfo* video_info,
                                  uint32_t track_id) {
   InitializeTrak(video_info, trak);
 
-  trak->header.width = video_info->width();
+  // width and height specify the track's visual presentation size as
-  trak->header.height = video_info->height();
+  // fixed-point 16.16 values.
+  const double sample_aspect_ratio =
+      static_cast<double>(video_info->pixel_width()) /
+      video_info->pixel_height();
+  trak->header.width = video_info->width() * sample_aspect_ratio * 0x10000;
+  trak->header.height = video_info->height() * 0x10000;
+
   trak->media.handler.type = kVideo;
 
   VideoSampleEntry video;
@@ -168,6 +174,10 @@ void MP4Muxer::GenerateVideoTrak(const VideoStreamInfo* video_info,
   video.width = video_info->width();
   video.height = video_info->height();
   video.avcc.data = video_info->extra_data();
+  if (video_info->pixel_width() != 1 || video_info->pixel_height() != 1) {
+    video.pixel_aspect.h_spacing = video_info->pixel_width();
+    video.pixel_aspect.v_spacing = video_info->pixel_height();
+  }
 
   SampleDescription& sample_description =
       trak->media.information.sample_table.description;

packager/media/test/data/README

@@ -44,6 +44,10 @@ bear-640x360-v_frag-cenc.mp4 - A fragmented MP4 version of the video track of be
 [1] 30313233343536373839303132333435
 [2] ebdd62f16814d27b68ef122afce4ae3c
 
+// 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.
+bear-640x360-non_square_pixel-without_pasp.mp4 - A non-square pixel version of the video track of bear-640x360.mp4 without PixelAspectRatio box.
+
 // Container Tests (additional containers derived from bear.ogv)
 bear.ac3    -- created using "avconv -i bear.ogv -f ac3 -b 192k bear.ac3".
 bear.adts   -- created using "avconv -i bear.ogv -f adts -strict experimental bear.adts".

packager/packager.gyp

@@ -97,7 +97,7 @@
         'media/filters/filters.gyp:filters_unittest',
         'media/formats/mp2t/mp2t.gyp:mp2t_unittest',
         'media/formats/mp4/mp4.gyp:mp4_unittest',
-#        'media/formats/wvm/wvm.gyp:wvm_unittest',
+        'media/formats/wvm/wvm.gyp:wvm_unittest',
         'mpd/mpd.gyp:mpd_unittest',
         'packager_test',
       ],