Plumb SubtitleParser.Factory into WebvttExtractor

WebvttExtractor will no longer be wrapped in SubtitleTranscodingExtractor, but instead use SubtitleTranscodingExtractorOutput under the hood.

A new constructor will take a boolean parameter to toggle between subtitle parsing during extraction (before the sample queue) or during decoding (after the sample queue).

PiperOrigin-RevId: 597604942

libraries/exoplayer_hls/src/main/java/androidx/media3/exoplayer/hls/BundledHlsMediaChunkExtractor.java

@@ -29,7 +29,6 @@ import androidx.media3.extractor.PositionHolder;
 import androidx.media3.extractor.mp3.Mp3Extractor;
 import androidx.media3.extractor.mp4.FragmentedMp4Extractor;
 import androidx.media3.extractor.text.SubtitleParser;
-import androidx.media3.extractor.text.SubtitleTranscodingExtractor;
 import androidx.media3.extractor.ts.Ac3Extractor;
 import androidx.media3.extractor.ts.Ac4Extractor;
 import androidx.media3.extractor.ts.AdtsExtractor;
@@ -125,13 +124,19 @@ public final class BundledHlsMediaChunkExtractor implements HlsMediaChunkExtract
     Extractor newExtractorInstance;
     // LINT.IfChange(extractor_instantiation)
     if (extractor instanceof WebvttExtractor) {
-      newExtractorInstance =
+      SubtitleParser.Factory webvttSubtitleParserFactory = SubtitleParser.Factory.UNSUPPORTED;
-          new WebvttExtractor(multivariantPlaylistFormat.language, timestampAdjuster);
+      boolean parseSubtitlesDuringExtraction = false;
       if (subtitleParserFactory != null
           && subtitleParserFactory.supportsFormat(multivariantPlaylistFormat)) {
-        newExtractorInstance =
+        webvttSubtitleParserFactory = subtitleParserFactory;
-            new SubtitleTranscodingExtractor(newExtractorInstance, subtitleParserFactory);
+        parseSubtitlesDuringExtraction = true;
       }
+      newExtractorInstance =
+          new WebvttExtractor(
+              multivariantPlaylistFormat.language,
+              timestampAdjuster,
+              webvttSubtitleParserFactory,
+              parseSubtitlesDuringExtraction);
     } else if (extractor instanceof AdtsExtractor) {
       newExtractorInstance = new AdtsExtractor();
     } else if (extractor instanceof Ac3Extractor) {

libraries/exoplayer_hls/src/main/java/androidx/media3/exoplayer/hls/DefaultHlsExtractorFactory.java

@@ -34,7 +34,6 @@ import androidx.media3.extractor.ExtractorInput;
 import androidx.media3.extractor.mp3.Mp3Extractor;
 import androidx.media3.extractor.mp4.FragmentedMp4Extractor;
 import androidx.media3.extractor.text.SubtitleParser;
-import androidx.media3.extractor.text.SubtitleTranscodingExtractor;
 import androidx.media3.extractor.ts.Ac3Extractor;
 import androidx.media3.extractor.ts.Ac4Extractor;
 import androidx.media3.extractor.ts.AdtsExtractor;
@@ -188,12 +187,17 @@ public final class DefaultHlsExtractorFactory implements HlsExtractorFactory {
     // LINT.IfChange(extractor_instantiation)
     switch (fileType) {
       case FileTypes.WEBVTT:
+        SubtitleParser.Factory webvttSubtitleParserFactory = SubtitleParser.Factory.UNSUPPORTED;
+        boolean parseSubtitlesDuringExtraction = false;
         if (subtitleParserFactory != null && subtitleParserFactory.supportsFormat(format)) {
-          return new SubtitleTranscodingExtractor(
+          webvttSubtitleParserFactory = subtitleParserFactory;
-              new WebvttExtractor(format.language, timestampAdjuster), subtitleParserFactory);
+          parseSubtitlesDuringExtraction = true;
-        } else {
-          return new WebvttExtractor(format.language, timestampAdjuster);
         }
+        return new WebvttExtractor(
+            format.language,
+            timestampAdjuster,
+            webvttSubtitleParserFactory,
+            parseSubtitlesDuringExtraction);
       case FileTypes.ADTS:
         return new AdtsExtractor();
       case FileTypes.AC3:

libraries/exoplayer_hls/src/main/java/androidx/media3/exoplayer/hls/WebvttExtractor.java

@@ -31,6 +31,8 @@ import androidx.media3.extractor.ExtractorOutput;
 import androidx.media3.extractor.PositionHolder;
 import androidx.media3.extractor.SeekMap;
 import androidx.media3.extractor.TrackOutput;
+import androidx.media3.extractor.text.SubtitleParser;
+import androidx.media3.extractor.text.SubtitleTranscodingExtractorOutput;
 import androidx.media3.extractor.text.webvtt.WebvttParserUtil;
 import java.io.IOException;
 import java.util.Arrays;
@@ -58,17 +60,38 @@ public final class WebvttExtractor implements Extractor {
   @Nullable private final String language;
   private final TimestampAdjuster timestampAdjuster;
   private final ParsableByteArray sampleDataWrapper;
+  private final SubtitleParser.Factory subtitleParserFactory;
+  private final boolean parseSubtitlesDuringExtraction;
 
   private @MonotonicNonNull ExtractorOutput output;
 
   private byte[] sampleData;
   private int sampleSize;
 
+  /**
+   * @deprecated Use {@link #WebvttExtractor(String, TimestampAdjuster, SubtitleParser.Factory,
+   *     boolean)} instead.
+   */
+  @Deprecated
   public WebvttExtractor(@Nullable String language, TimestampAdjuster timestampAdjuster) {
+    this(
+        language,
+        timestampAdjuster,
+        SubtitleParser.Factory.UNSUPPORTED,
+        /* parseSubtitlesDuringExtraction= */ false);
+  }
+
+  public WebvttExtractor(
+      @Nullable String language,
+      TimestampAdjuster timestampAdjuster,
+      SubtitleParser.Factory subtitleParserFactory,
+      boolean parseSubtitlesDuringExtraction) {
     this.language = language;
     this.timestampAdjuster = timestampAdjuster;
     this.sampleDataWrapper = new ParsableByteArray();
     sampleData = new byte[1024];
+    this.subtitleParserFactory = subtitleParserFactory;
+    this.parseSubtitlesDuringExtraction = parseSubtitlesDuringExtraction;
   }
 
   // Extractor implementation.
@@ -94,7 +117,10 @@ public final class WebvttExtractor implements Extractor {
 
   @Override
   public void init(ExtractorOutput output) {
-    this.output = output;
+    this.output =
+        parseSubtitlesDuringExtraction
+            ? new SubtitleTranscodingExtractorOutput(output, subtitleParserFactory)
+            : output;
     output.seekMap(new SeekMap.Unseekable(C.TIME_UNSET));
   }
 

libraries/exoplayer_hls/src/test/java/androidx/media3/exoplayer/hls/WebvttExtractorTest.java

@@ -19,6 +19,8 @@ import static com.google.common.truth.Truth.assertThat;
 
 import androidx.media3.common.util.TimestampAdjuster;
 import androidx.media3.extractor.ExtractorInput;
+import androidx.media3.extractor.text.DefaultSubtitleParserFactory;
+import androidx.media3.extractor.text.SubtitleParser;
 import androidx.media3.test.utils.DumpFileAsserts;
 import androidx.media3.test.utils.FakeExtractorInput;
 import androidx.media3.test.utils.FakeExtractorOutput;
@@ -72,7 +74,12 @@ public class WebvttExtractorTest {
     TimestampAdjuster timestampAdjuster = new TimestampAdjuster(/* firstSampleTimestampUs= */ 0);
     // Prime the TimestampAdjuster with a close-ish timestamp (5s before the first cue).
     timestampAdjuster.adjustTsTimestamp(384615190);
-    WebvttExtractor extractor = new WebvttExtractor(/* language= */ null, timestampAdjuster);
+    WebvttExtractor extractor =
+        new WebvttExtractor(
+            /* language= */ null,
+            timestampAdjuster,
+            SubtitleParser.Factory.UNSUPPORTED,
+            /* parseSubtitlesDuringExtraction= */ false);
     // We can't use ExtractorAsserts because WebvttExtractor doesn't fulfill the whole Extractor
     // interface (e.g. throws an exception from seek()).
     FakeExtractorOutput output =
@@ -90,10 +97,42 @@ public class WebvttExtractorTest {
         "extractordumps/webvtt/with_x-timestamp-map_header.dump");
   }
 
+  @Test
+  public void read_handlesLargeCueTimestamps_withSubtitleParsingDuringExtraction()
+      throws Exception {
+    TimestampAdjuster timestampAdjuster = new TimestampAdjuster(/* firstSampleTimestampUs= */ 0);
+    // Prime the TimestampAdjuster with a close-ish timestamp (5s before the first cue).
+    timestampAdjuster.adjustTsTimestamp(384615190);
+    WebvttExtractor extractor =
+        new WebvttExtractor(
+            /* language= */ null,
+            timestampAdjuster,
+            new DefaultSubtitleParserFactory(),
+            /* parseSubtitlesDuringExtraction= */ true);
+
+    FakeExtractorOutput output =
+        TestUtil.extractAllSamplesFromFile(
+            extractor,
+            ApplicationProvider.getApplicationContext(),
+            "media/webvtt/with_x-timestamp-map_header");
+
+    // There are 2 cues in the file which are fed into 2 different samples during extraction
+    // This is different to the parsing-during-decoding flow where the whole file becomes a sample
+    DumpFileAsserts.assertOutput(
+        ApplicationProvider.getApplicationContext(),
+        output,
+        "extractordumps/webvtt/with_x-timestamp-map_header_parsed_during_extraction.dump");
+  }
+
   private static boolean sniffData(byte[] data) throws IOException {
     ExtractorInput input = new FakeExtractorInput.Builder().setData(data).build();
     try {
-      return new WebvttExtractor(/* language= */ null, new TimestampAdjuster(0)).sniff(input);
+      return new WebvttExtractor(
+              /* language= */ null,
+              new TimestampAdjuster(0),
+              SubtitleParser.Factory.UNSUPPORTED,
+              /* parseSubtitlesDuringExtraction= */ false)
+          .sniff(input);
     } catch (EOFException e) {
       return false;
     }

libraries/test_data/src/test/assets/extractordumps/webvtt/with_x-timestamp-map_header_parsed_during_extraction.dump

@@ -0,0 +1,20 @@
+seekMap:
+  isSeekable = false
+  duration = UNSET TIME
+  getPosition(0) = [[timeUs=0, position=0]]
+numberOfTracks = 1
+track 0:
+  total output bytes = 2248
+  sample count = 2
+  format 0:
+    sampleMimeType = application/x-media3-cues
+    codecs = text/vtt
+  sample 0:
+    time = 5000155
+    flags = 1
+    data = length 1124, hash E87709B3
+  sample 1:
+    time = 6754155
+    flags = 1
+    data = length 1124, hash 5C8A8288
+tracksEnded = true