Output Empty WebVTT Text Segments

Have the text webvtt pipeline output empty segments rather than skipping
them.

Bug: #324
Change-Id: I175cf22b65b241f6bf566aba2440518f3df95277

packager/media/formats/webvtt/webvtt_pipeline_unittest.cc

@@ -26,6 +26,7 @@ TEST(WebVttTextPipelineTest, SegmentedOutput) {
   const char* kOutput1 = "memory://output/template-1.vtt";
   const char* kOutput2 = "memory://output/template-2.vtt";
   const char* kOutput3 = "memory://output/template-3.vtt";
+  const char* kOutput4 = "memory://output/template-4.vtt";
 
   const char* kInputFile = "memory://input.vtt";
   const char* kInput =
@@ -53,8 +54,14 @@ TEST(WebVttTextPipelineTest, SegmentedOutput) {
       "Thank you.\n";
 
   // Segment One
-  // 00:00:10.000 to 00:00:20.000
+  // 00:00:00.000 to 00:00:10.000
   const char* kExpectedOutput1 =
+      "WEBVTT\n"
+      "\n";
+
+  // Segment Two
+  // 00:00:10.000 to 00:00:20.000
+  const char* kExpectedOutput2 =
       "WEBVTT\n"
       "\n"
       "1\n"
@@ -62,9 +69,9 @@ TEST(WebVttTextPipelineTest, SegmentedOutput) {
       "This blade has a dark past.\n"
       "\n";
 
-  // Segment Two
+  // Segment Three
   // 00:00:20.000 to 00:00:30.000
-  const char* kExpectedOutput2 =
+  const char* kExpectedOutput3 =
       "WEBVTT\n"
       "\n"
       "1\n"
@@ -80,9 +87,9 @@ TEST(WebVttTextPipelineTest, SegmentedOutput) {
       "You're a fool for traveling alone,\nso completely unprepared.\n"
       "\n";
 
-  // Segment Three
+  // Segment Four
   // 00:00:30.000 to 00:00:40.000
-  const char* kExpectedOutput3 =
+  const char* kExpectedOutput4 =
       "WEBVTT\n"
       "\n"
       "3\n"
@@ -128,6 +135,7 @@ TEST(WebVttTextPipelineTest, SegmentedOutput) {
   ASSERT_FILE_STREQ(kOutput1, kExpectedOutput1);
   ASSERT_FILE_STREQ(kOutput2, kExpectedOutput2);
   ASSERT_FILE_STREQ(kOutput3, kExpectedOutput3);
+  ASSERT_FILE_STREQ(kOutput4, kExpectedOutput4);
 }
 }  // namespace media
 }  // namespace shaka

packager/media/formats/webvtt/webvtt_segmenter.cc

@@ -10,7 +10,7 @@ namespace shaka {
 namespace media {
 namespace {
 const size_t kStreamIndex = 0;
-}
+}  // namespace
 
 WebVttSegmenter::WebVttSegmenter(uint64_t segment_duration_ms)
     : segment_duration_ms_(segment_duration_ms) {}
@@ -33,16 +33,18 @@ Status WebVttSegmenter::Process(std::unique_ptr<StreamData> stream_data) {
 }
 
 Status WebVttSegmenter::OnFlushRequest(size_t input_stream_index) {
-  while (segment_map_.size()) {
-    uint64_t segment = segment_map_.begin()->first;
+  // At this point we know that there is a single series of consecutive
+  // segments, all we need to do is run through all of them.
+  for (const auto& pair : segment_map_) {
+    Status status = DispatchSegmentWithSamples(pair.first, pair.second);
 
-    // OnSegmentEnd will remove the segment from the map.
-    Status status = OnSegmentEnd(segment);
     if (!status.ok()) {
       return status;
     }
   }
 
+  segment_map_.clear();
+
   return FlushAllDownstreams();
 }
 
@@ -66,58 +68,58 @@ Status WebVttSegmenter::OnTextSample(std::shared_ptr<const TextSample> sample) {
     return Status::OK;
   }
 
-  // Now that we are accepting this new segment, its starting segment is our
-  // new head segment.
-  head_segment_ = start_segment;
-
   // Add the sample to each segment it spans.
   for (uint64_t segment = start_segment; segment <= ending_segment; segment++) {
     segment_map_[segment].push_back(sample);
   }
 
-  // Output all the segments that come before the start of this cue's first
-  // segment.
-  while (segment_map_.size()) {
-    // Since the segments are in accending order, we can break out of the loop
-    // once we catch-up to the new samples starting segment.
-    const uint64_t segment = segment_map_.begin()->first;
-    if (segment >= start_segment) {
-      break;
+  // Move forward segment-by-segment so that we output empty segments to fill
+  // any segments with no cues.
+  for (uint64_t segment = head_segment_; segment < start_segment; segment++) {
+    auto it = segment_map_.find(segment);
+
+    Status status;
+    if (it == segment_map_.end()) {
+      const WebVttSegmentSamples kNoSamples;
+      status.Update(DispatchSegmentWithSamples(segment, kNoSamples));
+    } else {
+      // We found a segment, output all the samples. Remove it from the map as
+      // we should never need to write to it again.
+      status.Update(DispatchSegmentWithSamples(segment, it->second));
+      segment_map_.erase(it);
     }
 
-    // Output the segment. If there is an error, there is no reason to continue.
-    Status status = OnSegmentEnd(segment);
+    // If we fail to output a single sample, just stop.
     if (!status.ok()) {
       return status;
     }
   }
 
+  // Jump ahead to the start of this segment as we should never have any samples
+  // start before |start_segment|.
+  head_segment_ = start_segment;
+
   return Status::OK;
 }
 
-Status WebVttSegmenter::OnSegmentEnd(uint64_t segment) {
+Status WebVttSegmenter::DispatchSegmentWithSamples(
+    uint64_t segment,
+    const WebVttSegmentSamples& samples) {
   Status status;
-
-  const auto found = segment_map_.find(segment);
-  if (found != segment_map_.end()) {
-    for (const auto& sample : found->second) {
+  for (const auto& sample : samples) {
     status.Update(DispatchTextSample(kStreamIndex, sample));
   }
 
-    // Stop storing the segment.
-    segment_map_.erase(found);
+  // Only send the segment info if all the samples were successful.
+  if (!status.ok()) {
+    return status;
   }
 
-  // Only send the segment info if all the samples were accepted.
-  if (status.ok()) {
   std::shared_ptr<SegmentInfo> info = std::make_shared<SegmentInfo>();
   info->start_timestamp = segment * segment_duration_ms_;
   info->duration = segment_duration_ms_;
 
-    status.Update(DispatchSegmentInfo(kStreamIndex, std::move(info)));
-  }
-
-  return status;
+  return DispatchSegmentInfo(kStreamIndex, std::move(info));
 }
 }  // namespace media
 }  // namespace shaka

packager/media/formats/webvtt/webvtt_segmenter.h

@@ -30,19 +30,20 @@ class WebVttSegmenter : public MediaHandler {
   WebVttSegmenter(const WebVttSegmenter&) = delete;
   WebVttSegmenter& operator=(const WebVttSegmenter&) = delete;
 
+  using WebVttSample = std::shared_ptr<const TextSample>;
+  using WebVttSegmentSamples = std::vector<WebVttSample>;
+
   Status InitializeInternal() override;
 
   Status OnTextSample(std::shared_ptr<const TextSample> sample);
 
-  Status OnSegmentEnd(uint64_t segment);
+  Status DispatchSegmentWithSamples(uint64_t segment,
+                                    const WebVttSegmentSamples& samples);
 
   uint64_t segment_duration_ms_;
 
-  using WebVttSample = std::shared_ptr<const TextSample>;
-  using WebVttSegment = std::vector<WebVttSample>;
-
   // Mapping of segment number to segment.
-  std::map<uint64_t, WebVttSegment> segment_map_;
+  std::map<uint64_t, WebVttSegmentSamples> segment_map_;
   uint64_t head_segment_ = 0;
 };
 

packager/media/formats/webvtt/webvtt_segmenter_unittest.cc

@@ -142,9 +142,13 @@ TEST_F(WebVttSegmenterTest, CreatesSegmentsForCues) {
 //           |          |
 //           |          | [---B---]
 //           |          |
-TEST_F(WebVttSegmenterTest, SkipsEmptySegments) {
+TEST_F(WebVttSegmenterTest, OutputsEmptySegments) {
   const uint64_t kSampleDuration = kSegmentDuration / 2;
 
+  const int64_t kSegment1Start = kStartTime;
+  const int64_t kSegment2Start = kSegment1Start + kSegmentDuration;
+  const int64_t kSegment3Start = kSegment2Start + kSegmentDuration;
+
   {
     testing::InSequence s;
 
@@ -157,19 +161,25 @@ TEST_F(WebVttSegmenterTest, SkipsEmptySegments) {
                                kNoSettings, kPayload[0])));
     EXPECT_CALL(
         *Output(kOutputIndex),
-        OnProcess(IsSegmentInfo(kStreamIndex, kStartTimeSigned,
-                                kSegmentDuration, !kSubSegment, !kEncrypted)));
+        OnProcess(IsSegmentInfo(kStreamIndex, kSegment1Start, kSegmentDuration,
+                                !kSubSegment, !kEncrypted)));
 
-    // Segment Two
+    // Segment Two (empty segment)
+    EXPECT_CALL(
+        *Output(kOutputIndex),
+        OnProcess(IsSegmentInfo(kStreamIndex, kSegment2Start, kSegmentDuration,
+                                !kSubSegment, !kEncrypted)));
+
+    // Segment Three
     EXPECT_CALL(*Output(kOutputIndex),
                 OnProcess(IsTextSample(
                     kId[1], kStartTime + 2 * kSegmentDuration,
                     kStartTime + 2 * kSegmentDuration + kSampleDuration,
                     kNoSettings, kPayload[1])));
-    EXPECT_CALL(*Output(kOutputIndex),
-                OnProcess(IsSegmentInfo(
-                    kStreamIndex, kStartTimeSigned + 2 * kSegmentDuration,
-                    kSegmentDuration, !kSubSegment, !kEncrypted)));
+    EXPECT_CALL(
+        *Output(kOutputIndex),
+        OnProcess(IsSegmentInfo(kStreamIndex, kSegment3Start, kSegmentDuration,
+                                !kSubSegment, !kEncrypted)));
 
     EXPECT_CALL(*Output(kOutputIndex), OnFlush(kStreamIndex));
   }