Align DecodedOutput with BatchDecodedOutput (#458)

Author: NicolasHug

src/torchcodec/decoders/_core/VideoDecoder.cpp

@@ -872,10 +872,8 @@ VideoDecoder::DecodedOutput VideoDecoder::convertAVFrameToDecodedOutput(
   output.streamIndex = streamIndex;
   auto& streamInfo = streams_[streamIndex];
   TORCH_CHECK(streamInfo.stream->codecpar->codec_type == AVMEDIA_TYPE_VIDEO);
-  output.pts = frame->pts;
   output.ptsSeconds =
       ptsToSeconds(frame->pts, formatContext_->streams[streamIndex]->time_base);
-  output.duration = getDuration(frame);
   output.durationSeconds = ptsToSeconds(
       getDuration(frame), formatContext_->streams[streamIndex]->time_base);
   // TODO: we should fold preAllocatedOutputTensor into RawDecodedOutput.

src/torchcodec/decoders/_core/VideoDecoder.h

@@ -174,12 +174,8 @@ class VideoDecoder {
     // The stream index of the decoded frame. Used to distinguish
     // between streams that are of the same type.
     int streamIndex;
-    // The presentation timestamp of the decoded frame in time base.
-    int64_t pts;
     // The presentation timestamp of the decoded frame in seconds.
     double ptsSeconds;
-    // The duration of the decoded frame in time base.
-    int64_t duration;
     // The duration of the decoded frame in seconds.
     double durationSeconds;
   };

test/decoders/VideoDecoderTest.cpp

@@ -172,12 +172,10 @@ TEST_P(VideoDecoderTest, ReturnsFirstTwoFramesOfVideo) {
   torch::Tensor tensor0FromOurDecoder = output.frame;
   EXPECT_EQ(tensor0FromOurDecoder.sizes(), std::vector<long>({3, 270, 480}));
   EXPECT_EQ(output.ptsSeconds, 0.0);
-  EXPECT_EQ(output.pts, 0);
   output = ourDecoder->getNextFrameNoDemux();
   torch::Tensor tensor1FromOurDecoder = output.frame;
   EXPECT_EQ(tensor1FromOurDecoder.sizes(), std::vector<long>({3, 270, 480}));
   EXPECT_EQ(output.ptsSeconds, 1'001. / 30'000);
-  EXPECT_EQ(output.pts, 1001);
 
   torch::Tensor tensor0FromFFMPEG =
       readTensorFromDisk("nasa_13013.mp4.stream3.frame000000.pt");