Output Empty MP4 WebVtt Cues

When a gap is found in the text stream, the WebVtt to Mp4 converter will
now output the special empty vtt cue.

Change-Id: I8be88c6b7589aa120a2215e1e4b8e98031fe326d
Closes: #324

packager/media/formats/webvtt/webvtt_to_mp4_handler.cc

@@ -129,14 +129,10 @@ Status WebVttToMp4Handler::ProcessUpToTime(uint64_t cutoff_time) {
     // Get the time range for which the current active state is valid.
     const uint64_t previous_change = next_change_;
     next_change_ = actions_.top()->time();
-    // The only time that |previous_change| and |next_change_| should ever break
-    // the rule |next_change_ > previous_change| is at the start where
-    // |previous_change| and |next_change_| are both zero.
-    DCHECK((previous_change == 0 && next_change_ == 0) ||
-           next_change_ > previous_change);
 
-    // Send out the active group. If there is nothing in the active group, then
+    if (next_change_ > previous_change) {
-    // an empty cue is sent.
+      // Send out the active group. If there is nothing in the active group,
+      // then an empty cue is sent.
       Status status =
           active_.size()
               ? MergeAndSendSamples(active_, previous_change, next_change_)
@@ -145,6 +141,13 @@ Status WebVttToMp4Handler::ProcessUpToTime(uint64_t cutoff_time) {
       if (!status.ok()) {
         return status;
       }
+    } else {
+      // The only time that |previous_change| and |next_change_| should ever
+      // break the rule |next_change_ > previous_change| is at the start where
+      // |previous_change| and |next_change_| are both zero.
+      DCHECK_EQ(previous_change, 0u);
+      DCHECK_EQ(next_change_, 0u);
+    }
 
     // STAGE 2: Move to the next state.
     while (actions_.size() && actions_.top()->time() == next_change_) {
@@ -180,8 +183,19 @@ Status WebVttToMp4Handler::MergeAndSendSamples(
 
 Status WebVttToMp4Handler::SendEmptySample(uint64_t start_time,
                                            uint64_t end_time) {
-  // TODO(vaage): Dispatch an empty vtt sample.
+  DCHECK_GT(end_time, start_time);
-  return Status::OK;
+
+  box_writer_.Clear();
+
+  mp4::VTTEmptyCueBox box;
+  box.Write(&box_writer_);
+
+  std::shared_ptr<MediaSample> sample =
+      MediaSample::CopyFrom(box_writer_.Buffer(), box_writer_.Size(), true);
+  sample->set_pts(start_time);
+  sample->set_dts(start_time);
+  sample->set_duration(end_time - start_time);
+  return DispatchMediaSample(kTrackId, std::move(sample));
 }
 
 uint64_t WebVttToMp4Handler::NextActionId() {

packager/media/formats/webvtt/webvtt_to_mp4_handler_unittest.cc

@@ -63,32 +63,104 @@ class WebVttToMp4HandlerTest : public MediaHandlerTestBase {
   std::shared_ptr<TestableWebVttToMp4Handler> mp4_handler_;
 };
 
-// Verify the cues are grouped correctly when the cues do not overlap at all.
+// Verify that when the stream starts at a non-zero value, the gap at the
-//
+// start will be filled.
-// [----A---]  [---B---]
+// |    [----A----]
-TEST_F(WebVttToMp4HandlerTest, NoOverlap) {
+TEST_F(WebVttToMp4HandlerTest, NonZeroStartTime) {
-  const int64_t kStart[] = {0, 1100};
+  const int64_t kGapStart = 0;
-  const int64_t kEnd[] = {1000, 2100};
+  const int64_t kGapEnd = 100;
+  const int64_t kGapDuration = kGapEnd - kGapStart;
+
+  const char* kSampleId = kId[0];
+  const char* kSamplePayload = kPayload[0];
+  const int64_t kSampleStart = kGapEnd;
+  const int64_t kSampleDuration = 500;
+  const int64_t kSampleEnd = kSampleStart + kSampleDuration;
 
   {
     testing::InSequence s;
 
-    for (size_t i = kA; i <= kB; i++) {
+    // Empty Cue to fill gap
-      EXPECT_CALL(*mp4_handler_, OnWriteCue(kId[i], kNoSettings, kPayload[i]));
     EXPECT_CALL(*Output(kOutputIndex),
-                  OnProcess(IsMediaSample(kStreamIndex, kStart[i],
+                OnProcess(IsMediaSample(kStreamIndex, kGapStart, kGapDuration,
-                                          kEnd[i] - kStart[i], !kEncrypted)));
+                                        !kEncrypted)));
-    }
+
+    // Sample
+    EXPECT_CALL(*mp4_handler_,
+                OnWriteCue(kSampleId, kNoSettings, kSamplePayload));
+    EXPECT_CALL(*Output(kOutputIndex),
+                OnProcess(IsMediaSample(kStreamIndex, kSampleStart,
+                                        kSampleDuration, !kEncrypted)));
 
     EXPECT_CALL(*Output(kOutputIndex), OnFlush(kStreamIndex));
   }
 
-  for (size_t i = kA; i <= kB; i++) {
   ASSERT_OK(Input(kInputIndex)
                 ->Dispatch(StreamData::FromTextSample(
-                      kStreamIndex,
+                    kStreamIndex, GetTextSample(kSampleId, kSampleStart,
-                      GetTextSample(kId[i], kStart[i], kEnd[i], kPayload[i]))));
+                                                kSampleEnd, kSamplePayload))));
+
+  ASSERT_OK(Input(kInputIndex)->FlushAllDownstreams());
 }
+
+// Verify the cues are grouped correctly when the cues do not overlap at all.
+// An empty cue should be inserted between the two as there is a gap.
+//
+// [----A---]  [---B---]
+TEST_F(WebVttToMp4HandlerTest, NoOverlap) {
+  const int64_t kDuration = 1000;
+
+  const char* kSample1Id = kId[0];
+  const char* kSample1Payload = kPayload[0];
+  const int64_t kSample1Start = 0;
+  const int64_t kSample1End = kSample1Start + kDuration;
+
+  // Make sample 2 be just a little after sample 1.
+  const char* kSample2Id = kId[1];
+  const char* kSample2Payload = kPayload[1];
+  const int64_t kSample2Start = kSample1End + 100;
+  const int64_t kSample2End = kSample2Start + kDuration;
+
+  const int64_t kGapStart = kSample1End;
+  const int64_t kGapDuration = kSample2Start - kSample1End;
+
+  {
+    testing::InSequence s;
+
+    // Sample 1
+    EXPECT_CALL(*mp4_handler_,
+                OnWriteCue(kSample1Id, kNoSettings, kSample1Payload));
+    EXPECT_CALL(*Output(kOutputIndex),
+                OnProcess(IsMediaSample(kStreamIndex, kSample1Start, kDuration,
+                                        !kEncrypted)));
+
+    // Empty Cue to fill gap
+    EXPECT_CALL(*Output(kOutputIndex),
+                OnProcess(IsMediaSample(kStreamIndex, kGapStart, kGapDuration,
+                                        !kEncrypted)));
+
+    // Sample 2
+    EXPECT_CALL(*mp4_handler_,
+                OnWriteCue(kSample2Id, kNoSettings, kSample2Payload));
+    EXPECT_CALL(*Output(kOutputIndex),
+                OnProcess(IsMediaSample(kStreamIndex, kSample2Start, kDuration,
+                                        !kEncrypted)));
+
+    EXPECT_CALL(*Output(kOutputIndex), OnFlush(kStreamIndex));
+  }
+
+  ASSERT_OK(
+      Input(kInputIndex)
+          ->Dispatch(StreamData::FromTextSample(
+              kStreamIndex, GetTextSample(kSample1Id, kSample1Start,
+                                          kSample1End, kSample1Payload))));
+
+  ASSERT_OK(
+      Input(kInputIndex)
+          ->Dispatch(StreamData::FromTextSample(
+              kStreamIndex, GetTextSample(kSample2Id, kSample2Start,
+                                          kSample2End, kSample2Payload))));
+
   ASSERT_OK(Input(kInputIndex)->FlushAllDownstreams());
 }