diff --git a/library/src/androidTest/assets/subrip/no_end_timecodes b/library/src/androidTest/assets/subrip/no_end_timecodes new file mode 100644 index 0000000000..0a856c6994 --- /dev/null +++ b/library/src/androidTest/assets/subrip/no_end_timecodes @@ -0,0 +1,11 @@ +1 +00:00:00,000 --> +SubRip doesn't technically allow missing end timecodes. + +2 +00:00:02,345 --> +We interpret it to mean that a subtitle extends to the start of the next one. + +3 +00:00:03,456 --> +Or to the end of the media. \ No newline at end of file diff --git a/library/src/androidTest/java/com/google/android/exoplayer/text/subrip/SubripParserTest.java b/library/src/androidTest/java/com/google/android/exoplayer/text/subrip/SubripParserTest.java index c3f6ed1aaa..05958cf496 100644 --- a/library/src/androidTest/java/com/google/android/exoplayer/text/subrip/SubripParserTest.java +++ b/library/src/androidTest/java/com/google/android/exoplayer/text/subrip/SubripParserTest.java @@ -25,13 +25,14 @@ import java.io.InputStream; */ public final class SubripParserTest extends InstrumentationTestCase { - private static final String TYPICAL_SUBRIP_FILE = "subrip/typical"; - private static final String EMPTY_SUBRIP_FILE = "subrip/empty"; + private static final String EMPTY_FILE = "subrip/empty"; + private static final String TYPICAL_FILE = "subrip/typical"; + private static final String NO_END_TIMECODES_FILE = "subrip/no_end_timecodes"; - public void testParseNullSubripFile() throws IOException { + public void testParseEmptySubripFile() throws IOException { SubripParser parser = new SubripParser(); InputStream inputStream = - getInstrumentation().getContext().getResources().getAssets().open(EMPTY_SUBRIP_FILE); + getInstrumentation().getContext().getResources().getAssets().open(EMPTY_FILE); SubripSubtitle subtitle = parser.parse(inputStream); // Assert that the subtitle is empty. assertEquals(0, subtitle.getEventTimeCount()); @@ -41,7 +42,7 @@ public final class SubripParserTest extends InstrumentationTestCase { public void testParseTypicalSubripFile() throws IOException { SubripParser parser = new SubripParser(); InputStream inputStream = - getInstrumentation().getContext().getResources().getAssets().open(TYPICAL_SUBRIP_FILE); + getInstrumentation().getContext().getResources().getAssets().open(TYPICAL_FILE); SubripSubtitle subtitle = parser.parse(inputStream); // Test event count. @@ -60,4 +61,29 @@ public final class SubripParserTest extends InstrumentationTestCase { assertEquals(3456000, subtitle.getEventTime(3)); } + public void testParseNoEndTimecodes() throws IOException { + SubripParser parser = new SubripParser(); + InputStream inputStream = getInstrumentation().getContext().getResources().getAssets() + .open(NO_END_TIMECODES_FILE); + SubripSubtitle subtitle = parser.parse(inputStream); + + // Test event count. + assertEquals(3, subtitle.getEventTimeCount()); + + // Test first cue. + assertEquals(0, subtitle.getEventTime(0)); + assertEquals("SubRip doesn't technically allow missing end timecodes.", + subtitle.getCues(subtitle.getEventTime(0)).get(0).text.toString()); + + // Test second cue. + assertEquals(2345000, subtitle.getEventTime(1)); + assertEquals("We interpret it to mean that a subtitle extends to the start of the next one.", + subtitle.getCues(subtitle.getEventTime(1)).get(0).text.toString()); + + // Test third cue. + assertEquals(3456000, subtitle.getEventTime(2)); + assertEquals("Or to the end of the media.", + subtitle.getCues(subtitle.getEventTime(2)).get(0).text.toString()); + } + } diff --git a/library/src/main/java/com/google/android/exoplayer/extractor/webm/WebmExtractor.java b/library/src/main/java/com/google/android/exoplayer/extractor/webm/WebmExtractor.java index ee5fc71662..0b3f06840f 100644 --- a/library/src/main/java/com/google/android/exoplayer/extractor/webm/WebmExtractor.java +++ b/library/src/main/java/com/google/android/exoplayer/extractor/webm/WebmExtractor.java @@ -73,6 +73,8 @@ public final class WebmExtractor implements Extractor { private static final String CODEC_ID_AAC = "A_AAC"; private static final String CODEC_ID_MP3 = "A_MPEG/L3"; private static final String CODEC_ID_AC3 = "A_AC3"; + private static final String CODEC_ID_SUBRIP = "S_TEXT/UTF8"; + private static final int VORBIS_MAX_INPUT_SIZE = 8192; private static final int OPUS_MAX_INPUT_SIZE = 5760; private static final int MP3_MAX_INPUT_SIZE = 4096; @@ -98,6 +100,7 @@ public final class WebmExtractor implements Extractor { private static final int ID_SIMPLE_BLOCK = 0xA3; private static final int ID_BLOCK_GROUP = 0xA0; private static final int ID_BLOCK = 0xA1; + private static final int ID_BLOCK_DURATION = 0x9B; private static final int ID_REFERENCE_BLOCK = 0xFB; private static final int ID_TRACKS = 0x1654AE6B; private static final int ID_TRACK_ENTRY = 0xAE; @@ -131,12 +134,39 @@ public final class WebmExtractor implements Extractor { private static final int ID_CUE_TIME = 0xB3; private static final int ID_CUE_TRACK_POSITIONS = 0xB7; private static final int ID_CUE_CLUSTER_POSITION = 0xF1; + private static final int ID_LANGUAGE = 0x22B59C; private static final int LACING_NONE = 0; private static final int LACING_XIPH = 1; private static final int LACING_FIXED_SIZE = 2; private static final int LACING_EBML = 3; + /** + * A template for the prefix that must be added to each subrip sample. The 12 byte end timecode + * starting at {@link #SUBRIP_PREFIX_END_TIMECODE_OFFSET} is set to a dummy value, and must be + * replaced with the duration of the subtitle. + *

+ * Equivalent to the UTF-8 string: "1\n00:00:00,000 --> 00:00:00,000\n". + */ + private static final byte[] SUBRIP_PREFIX = new byte[] {49, 10, 48, 48, 58, 48, 48, 58, 48, 48, + 44, 48, 48, 48, 32, 45, 45, 62, 32, 48, 48, 58, 48, 48, 58, 48, 48, 44, 48, 48, 48, 10}; + /** + * A special end timecode indicating that a subtitle should be displayed until the next subtitle, + * or until the end of the media in the case of the last subtitle. + *

+ * Equivalent to the UTF-8 string: " ". + */ + private static final byte[] SUBRIP_TIMECODE_EMPTY = + new byte[] {32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32}; + /** + * The byte offset of the end timecode in {@link #SUBRIP_PREFIX}. + */ + private static final int SUBRIP_PREFIX_END_TIMECODE_OFFSET = 19; + /** + * The length in bytes of a timecode in a subrip prefix. + */ + private static final int SUBRIP_TIMECODE_LENGTH = 12; + private final EbmlReader reader; private final VarintReader varintReader; private final SparseArray tracks; @@ -148,6 +178,7 @@ public final class WebmExtractor implements Extractor { private final ParsableByteArray vorbisNumPageSamples; private final ParsableByteArray seekEntryIdBytes; private final ParsableByteArray sampleStrippedBytes; + private final ParsableByteArray subripSample; private long segmentContentPosition = UNKNOWN; private long segmentContentSize = UNKNOWN; @@ -178,13 +209,13 @@ public final class WebmExtractor implements Extractor { // Block reading state. private int blockState; private long blockTimeUs; + private long blockDurationUs; private int blockLacingSampleIndex; private int blockLacingSampleCount; private int[] blockLacingSampleSizes; private int blockTrackNumber; private int blockTrackNumberLength; private int blockFlags; - private byte[] blockEncryptionKeyId; // Sample reading state. private int sampleBytesRead; @@ -212,6 +243,7 @@ public final class WebmExtractor implements Extractor { nalStartCode = new ParsableByteArray(NalUnitUtil.NAL_START_CODE); nalLength = new ParsableByteArray(4); sampleStrippedBytes = new ParsableByteArray(); + subripSample = new ParsableByteArray(); } @Override @@ -274,6 +306,7 @@ public final class WebmExtractor implements Extractor { case ID_SEEK_POSITION: case ID_TIMECODE_SCALE: case ID_TIME_CODE: + case ID_BLOCK_DURATION: case ID_PIXEL_WIDTH: case ID_PIXEL_HEIGHT: case ID_TRACK_NUMBER: @@ -293,6 +326,7 @@ public final class WebmExtractor implements Extractor { return EbmlReader.TYPE_UNSIGNED_INT; case ID_DOC_TYPE: case ID_CODEC_ID: + case ID_LANGUAGE: return EbmlReader.TYPE_STRING; case ID_SEEK_ID: case ID_CONTENT_COMPRESSION_SETTINGS: @@ -397,7 +431,7 @@ public final class WebmExtractor implements Extractor { if (!sampleSeenReferenceBlock) { blockFlags |= C.SAMPLE_FLAG_SYNC; } - outputSampleMetadata(tracks.get(blockTrackNumber), blockTimeUs); + commitSampleToOutput(tracks.get(blockTrackNumber), blockTimeUs); blockState = BLOCK_STATE_START; return; case ID_CONTENT_ENCODING: @@ -531,6 +565,9 @@ public final class WebmExtractor implements Extractor { case ID_TIME_CODE: clusterTimecodeUs = scaleTimecodeToUs(value); return; + case ID_BLOCK_DURATION: + blockDurationUs = scaleTimecodeToUs(value); + return; default: return; } @@ -560,6 +597,9 @@ public final class WebmExtractor implements Extractor { case ID_CODEC_ID: currentTrack.codecId = value; return; + case ID_LANGUAGE: + currentTrack.language = value; + return; default: return; } @@ -597,6 +637,7 @@ public final class WebmExtractor implements Extractor { if (blockState == BLOCK_STATE_START) { blockTrackNumber = (int) varintReader.readUnsignedVarint(input, false, true); blockTrackNumberLength = varintReader.getLastLength(); + blockDurationUs = UNKNOWN; blockState = BLOCK_STATE_HEADER; scratch.reset(); } @@ -698,7 +739,6 @@ public final class WebmExtractor implements Extractor { || (id == ID_SIMPLE_BLOCK && (scratch.data[2] & 0x80) == 0x80); blockFlags = (isKeyframe ? C.SAMPLE_FLAG_SYNC : 0) | (isInvisible ? C.SAMPLE_FLAG_DECODE_ONLY : 0); - blockEncryptionKeyId = track.encryptionKeyId; blockState = BLOCK_STATE_DATA; blockLacingSampleIndex = 0; } @@ -709,7 +749,7 @@ public final class WebmExtractor implements Extractor { writeSampleData(input, track, blockLacingSampleSizes[blockLacingSampleIndex]); long sampleTimeUs = this.blockTimeUs + (blockLacingSampleIndex * track.defaultSampleDurationNs) / 1000; - outputSampleMetadata(track, sampleTimeUs); + commitSampleToOutput(track, sampleTimeUs); blockLacingSampleIndex++; } blockState = BLOCK_STATE_START; @@ -725,8 +765,11 @@ public final class WebmExtractor implements Extractor { } } - private void outputSampleMetadata(Track track, long timeUs) { - track.output.sampleMetadata(timeUs, blockFlags, sampleBytesWritten, 0, blockEncryptionKeyId); + private void commitSampleToOutput(Track track, long timeUs) { + if (CODEC_ID_SUBRIP.equals(track.codecId)) { + writeSubripSample(track); + } + track.output.sampleMetadata(timeUs, blockFlags, sampleBytesWritten, 0, track.encryptionKeyId); sampleRead = true; resetSample(); } @@ -758,6 +801,21 @@ public final class WebmExtractor implements Extractor { private void writeSampleData(ExtractorInput input, Track track, int size) throws IOException, InterruptedException { + if (CODEC_ID_SUBRIP.equals(track.codecId)) { + int sizeWithPrefix = SUBRIP_PREFIX.length + size; + if (subripSample.capacity() < sizeWithPrefix) { + // Initialize subripSample to contain the required prefix and have space to hold a subtitle + // twice as long as this one. + subripSample.data = Arrays.copyOf(SUBRIP_PREFIX, sizeWithPrefix + size); + } + input.readFully(subripSample.data, SUBRIP_PREFIX.length, size); + subripSample.setPosition(0); + subripSample.setLimit(sizeWithPrefix); + // Defer writing the data to the track output. We need to modify the sample data by setting + // the correct end timecode, which we might not have yet. + return; + } + TrackOutput output = track.output; if (!sampleEncodingHandled) { if (track.hasContentEncryption) { @@ -834,6 +892,33 @@ public final class WebmExtractor implements Extractor { } } + private void writeSubripSample(Track track) { + setSubripSampleEndTimecode(subripSample.data, blockDurationUs); + // Note: If we ever want to support DRM protected subtitles then we'll need to output the + // appropriate encryption data here. + track.output.sampleData(subripSample, subripSample.limit()); + sampleBytesWritten += subripSample.limit(); + } + + private static void setSubripSampleEndTimecode(byte[] subripSampleData, long timeUs) { + byte[] timeCodeData; + if (timeUs == UNKNOWN) { + timeCodeData = SUBRIP_TIMECODE_EMPTY; + } else { + int hours = (int) (timeUs / 3600000000L); + timeUs -= (hours * 3600000000L); + int minutes = (int) (timeUs / 60000000); + timeUs -= (minutes * 60000000); + int seconds = (int) (timeUs / 1000000); + timeUs -= (seconds * 1000000); + int milliseconds = (int) (timeUs / 1000); + timeCodeData = String.format("%02d:%02d:%02d,%03d", hours, minutes, seconds, milliseconds) + .getBytes(); + } + System.arraycopy(timeCodeData, 0, subripSampleData, SUBRIP_PREFIX_END_TIMECODE_OFFSET, + SUBRIP_TIMECODE_LENGTH); + } + /** * Writes {@code length} bytes of sample data into {@code target} at {@code offset}, consisting of * pending {@link #sampleStrippedBytes} and any remaining data read from {@code input}. @@ -948,7 +1033,8 @@ public final class WebmExtractor implements Extractor { || CODEC_ID_VORBIS.equals(codecId) || CODEC_ID_AAC.equals(codecId) || CODEC_ID_MP3.equals(codecId) - || CODEC_ID_AC3.equals(codecId); + || CODEC_ID_AC3.equals(codecId) + || CODEC_ID_SUBRIP.equals(codecId); } /** @@ -1032,6 +1118,9 @@ public final class WebmExtractor implements Extractor { public long codecDelayNs = 0; public long seekPreRollNs = 0; + // Text elements. + private String language = "eng"; + // Set when the output is initialized. nalUnitLengthFieldLength is only set for H264/H265. public TrackOutput output; public int nalUnitLengthFieldLength; @@ -1097,6 +1186,9 @@ public final class WebmExtractor implements Extractor { case CODEC_ID_AC3: mimeType = MimeTypes.AUDIO_AC3; break; + case CODEC_ID_SUBRIP: + mimeType = MimeTypes.APPLICATION_SUBRIP; + break; default: throw new ParserException("Unrecognized codec identifier."); } @@ -1108,6 +1200,8 @@ public final class WebmExtractor implements Extractor { } else if (MimeTypes.isVideo(mimeType)) { format = MediaFormat.createVideoFormat(mimeType, MediaFormat.NO_VALUE, maxInputSize, durationUs, width, height, 0, initializationData); + } else if (MimeTypes.APPLICATION_SUBRIP.equals(mimeType)) { + format = MediaFormat.createTextFormat(mimeType, MediaFormat.NO_VALUE, language, durationUs); } else { throw new ParserException("Unexpected MIME type."); } diff --git a/library/src/main/java/com/google/android/exoplayer/text/subrip/SubripParser.java b/library/src/main/java/com/google/android/exoplayer/text/subrip/SubripParser.java index bfc646f457..742ec542bf 100644 --- a/library/src/main/java/com/google/android/exoplayer/text/subrip/SubripParser.java +++ b/library/src/main/java/com/google/android/exoplayer/text/subrip/SubripParser.java @@ -39,7 +39,7 @@ import java.util.regex.Pattern; */ public final class SubripParser implements SubtitleParser { - private static final Pattern SUBRIP_TIMING_LINE = Pattern.compile("(.*)\\s+-->\\s+(.*)"); + private static final Pattern SUBRIP_TIMING_LINE = Pattern.compile("(\\S*)\\s*-->\\s*(\\S*)"); private static final Pattern SUBRIP_TIMESTAMP = Pattern.compile("(?:(\\d+):)?(\\d+):(\\d+),(\\d+)"); @@ -54,6 +54,7 @@ public final class SubripParser implements SubtitleParser { ArrayList cues = new ArrayList<>(); LongArray cueTimesUs = new LongArray(); BufferedReader reader = new BufferedReader(new InputStreamReader(inputStream, C.UTF8_NAME)); + boolean haveEndTimecode; String currentLine; while ((currentLine = reader.readLine()) != null) { @@ -65,11 +66,16 @@ public final class SubripParser implements SubtitleParser { } // Read and parse the timing line. + haveEndTimecode = false; currentLine = reader.readLine(); Matcher matcher = SUBRIP_TIMING_LINE.matcher(currentLine); if (matcher.find()) { - cueTimesUs.add(parseTimestampUs(matcher.group(1))); - cueTimesUs.add(parseTimestampUs(matcher.group(2))); + cueTimesUs.add(parseTimecode(matcher.group(1))); + String endTimecode = matcher.group(2); + if (!TextUtils.isEmpty(endTimecode)) { + haveEndTimecode = true; + cueTimesUs.add(parseTimecode(matcher.group(2))); + } } else { throw new ParserException("Expected timing line: " + currentLine); } @@ -85,6 +91,9 @@ public final class SubripParser implements SubtitleParser { Spanned text = Html.fromHtml(textBuilder.toString()); cues.add(new Cue(text)); + if (haveEndTimecode) { + cues.add(null); + } } Cue[] cuesArray = new Cue[cues.size()]; @@ -98,7 +107,7 @@ public final class SubripParser implements SubtitleParser { return MimeTypes.APPLICATION_SUBRIP.equals(mimeType); } - private static long parseTimestampUs(String s) throws NumberFormatException { + private static long parseTimecode(String s) throws NumberFormatException { Matcher matcher = SUBRIP_TIMESTAMP.matcher(s); if (!matcher.matches()) { throw new NumberFormatException("has invalid format"); diff --git a/library/src/main/java/com/google/android/exoplayer/text/subrip/SubripSubtitle.java b/library/src/main/java/com/google/android/exoplayer/text/subrip/SubripSubtitle.java index 497628f88e..b10804311b 100644 --- a/library/src/main/java/com/google/android/exoplayer/text/subrip/SubripSubtitle.java +++ b/library/src/main/java/com/google/android/exoplayer/text/subrip/SubripSubtitle.java @@ -32,8 +32,8 @@ import java.util.List; private final long[] cueTimesUs; /** - * @param cues The cues in the subtitle. - * @param cueTimesUs Interleaved cue start and end times, in microseconds. + * @param cues The cues in the subtitle. Null entries may be used to represent empty cues. + * @param cueTimesUs The cue times, in microseconds. */ public SubripSubtitle(Cue[] cues, long[] cueTimesUs) { this.cues = cues; @@ -69,11 +69,11 @@ import java.util.List; @Override public List getCues(long timeUs) { int index = Util.binarySearchFloor(cueTimesUs, timeUs, true, false); - if (index == -1 || index % 2 == 1) { - // timeUs is earlier than the start of the first cue, or corresponds to a gap between cues. + if (index == -1 || cues[index] == null) { + // timeUs is earlier than the start of the first cue, or we have an empty cue. return Collections.emptyList(); } else { - return Collections.singletonList(cues[index / 2]); + return Collections.singletonList(cues[index]); } }