tga/auto-subtitle
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Allows setting max lines and max line length, enables DeepL, adds user setup bridge

Browse Source
This commit is contained in:
Guillaume Tâche
2024-08-20 21:31:10 +02:00
parent 273a6e996f
commit 44c317f207
49 changed files with 752 additions and 298 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
whisper/base/src/main/java/com/github/gtache/autosubtitle/subtitle/parser/json/whisper/base/JSONSubtitleConverter.java
View File

@@ -48,8 +48,8 @@ public class JSONSubtitleConverter implements SubtitleConverter<SubtitleImpl> {
try {
final var json = gson.fromJson(content, JSONSubtitles.class);
final var subtitles = json.segments().stream().map(s -> {
final var start = (long) s.start() * 1000L;
final var end = (long) s.end() * 1000L;
final var start = (long) (s.start() * 1000L);
final var end = (long) (s.end() * 1000L);
return new SubtitleImpl(s.text(), start, end, null, null);
}).sorted(Comparator.comparing(Subtitle::start).thenComparing(Subtitle::end)).toList();
final var language = Language.getLanguage(json.language());

6
whisper/whisperx/src/main/java/com/github/gtache/autosubtitle/subtitle/extractor/whisperx/WhisperXSubtitleExtractor.java
View File

@@ -55,10 +55,8 @@ public class WhisperXSubtitleExtractor extends AbstractWhisperSubtitleExtractor
args.add("True");
args.add("--compute_type");
args.add("int8");
args.add("--max_line_count");
args.add("2");
args.add("--max_line_width");
args.add("30");
args.add("--threads");
args.add(String.valueOf(Runtime.getRuntime().availableProcessors()));
if (language != Language.AUTO) {
args.add("--language");
args.add(language.iso2());

104
whisper/whisperx/src/main/java/com/github/gtache/autosubtitle/subtitle/parser/json/whisperx/JSONSubtitleConverter.java
View File

@@ -1,6 +1,8 @@
package com.github.gtache.autosubtitle.subtitle.parser.json.whisperx;
import com.github.gtache.autosubtitle.Language;
import com.github.gtache.autosubtitle.modules.impl.MaxLineLength;
import com.github.gtache.autosubtitle.modules.impl.MaxLines;
import com.github.gtache.autosubtitle.subtitle.Subtitle;
import com.github.gtache.autosubtitle.subtitle.SubtitleCollection;
import com.github.gtache.autosubtitle.subtitle.converter.ParseException;
@@ -12,10 +14,14 @@ import com.google.gson.Gson;
import javax.inject.Inject;
import javax.inject.Singleton;
import java.nio.file.Path;
import java.util.ArrayList;
import java.util.Comparator;
import java.util.List;
import java.util.Objects;
import java.util.prefs.Preferences;
import java.util.regex.Pattern;
import java.util.stream.Collectors;
import java.util.stream.Stream;
/**
* {@link SubtitleConverter} implementation for JSON files
@@ -23,11 +29,21 @@ import java.util.stream.Collectors;
@Singleton
public class JSONSubtitleConverter implements SubtitleConverter<SubtitleImpl> {
private static final String MAX_LINE_LENGTH = "maxLineLength";
private static final String MAX_LINES = "maxLines";
private static final Pattern SPLIT_PATTERN = Pattern.compile("[ \n]+");
private final Gson gson;
private final Preferences preferences;
private final int defaultMaxLineLength;
private final int defaultMaxLines;
@Inject
JSONSubtitleConverter(final Gson gson) {
JSONSubtitleConverter(final Gson gson, final Preferences preferences, @MaxLineLength final int defaultMaxLineLength, @MaxLines final int defaultMaxLines) {
this.gson = Objects.requireNonNull(gson);
this.preferences = Objects.requireNonNull(preferences);
this.defaultMaxLineLength = defaultMaxLineLength;
this.defaultMaxLines = defaultMaxLines;
}
@Override
@@ -42,10 +58,15 @@ public class JSONSubtitleConverter implements SubtitleConverter<SubtitleImpl> {
public SubtitleCollectionImpl<SubtitleImpl> parse(final String content) throws ParseException {
try {
final var json = gson.fromJson(content, JSONSubtitles.class);
final var subtitles = json.segments().stream().map(s -> {
final var start = (long) s.start() * 1000L;
final var end = (long) s.end() * 1000L;
return new SubtitleImpl(s.text(), start, end, null, null);
final var subtitles = json.segments().stream().flatMap(s -> {
final var start = (long) (s.start() * 1000L);
final var end = (long) (s.end() * 1000L);
if (s.words().isEmpty()) {
return Stream.of(new SubtitleImpl(s.text(), start, end, null, null));
} else {
return splitSubtitle(s);
}
}).sorted(Comparator.comparing(Subtitle::start).thenComparing(Subtitle::end)).toList();
final var language = Language.getLanguage(json.language());
final var subtitlesText = subtitles.stream().map(Subtitle::content).collect(Collectors.joining(""));
@@ -55,6 +76,79 @@ public class JSONSubtitleConverter implements SubtitleConverter<SubtitleImpl> {
}
}
private Stream<SubtitleImpl> splitSubtitle(final JSONSubtitleSegment segment) {
final var maxLineLength = preferences.getInt(MAX_LINE_LENGTH, defaultMaxLineLength);
final var maxLines = preferences.getInt(MAX_LINES, defaultMaxLines);
final var text = segment.text();
if (text.length() <= maxLineLength) {
final var start = (long) (segment.start() * 1000L);
final var end = (long) (segment.end() * 1000L);
return Stream.of(new SubtitleImpl(text.replace("\n", " "), start, end, null, null));
} else if (text.length() <= maxLines * maxLineLength) {
return splitSubtitleLines(segment);
} else {
return splitSubtitleWords(segment);
}
}
private Stream<SubtitleImpl> splitSubtitleLines(final JSONSubtitleSegment segment) {
final var maxLineLength = preferences.getInt(MAX_LINE_LENGTH, defaultMaxLineLength);
final var text = segment.text();
final var split = SPLIT_PATTERN.split(text);
final var builder = new StringBuilder(text.length());
for (final var s : split) {
final var newLength = builder.length() + s.length();
if (areDifferentLines(builder.length(), newLength, maxLineLength)) {
builder.append("\n").append(s);
} else {
builder.append(" ").append(s);
}
}
final var start = (long) (segment.start() * 1000L);
final var end = (long) (segment.end() * 1000L);
return Stream.of(new SubtitleImpl(builder.toString(), start, end, null, null));
}
private static boolean areDifferentLines(final int currentLength, final int newLength, final int maxLength) {
return currentLength / (maxLength + 1) < newLength / (maxLength + 1);
}
private Stream<SubtitleImpl> splitSubtitleWords(final JSONSubtitleSegment segment) {
final var maxLineLength = preferences.getInt(MAX_LINE_LENGTH, defaultMaxLineLength);
final var maxLines = preferences.getInt(MAX_LINES, defaultMaxLines);
final var ret = new ArrayList<SubtitleImpl>(segment.text().length() / (maxLines * maxLineLength));
final var builder = new StringBuilder(maxLines * maxLineLength);
final var words = segment.words();
var currentStart = -1L;
var currentEnd = -1L;
for (final var word : words) {
final var text = word.word();
final var start = (long) (word.start() * 1000L);
final var end = (long) (word.end() * 1000L);
if (currentStart < 0) {
currentStart = start;
}
final var newLength = builder.length() + text.length();
if (newLength > maxLineLength * maxLines) {
final var newSubtitle = new SubtitleImpl(builder.toString(), currentStart, currentEnd, null, null);
ret.add(newSubtitle);
builder.delete(0, builder.length());
builder.append(text);
currentStart = start == 0 ? currentEnd : start;
} else if (areDifferentLines(builder.length(), newLength, maxLineLength)) {
builder.append("\n").append(text);
} else {
builder.append(" ").append(text);
}
currentEnd = end == 0 ? currentEnd : end;
}
if (!builder.isEmpty()) {
final var newSubtitle = new SubtitleImpl(builder.toString(), currentStart, currentEnd, null, null);
ret.add(newSubtitle);
}
return ret.stream();
}
@Override
public boolean canParse(final Path file) {
return file.getFileName().toString().endsWith(".json");

1
whisper/whisperx/src/main/java/module-info.java
View File

@@ -4,6 +4,7 @@
module com.github.gtache.autosubtitle.whisperx {
requires transitive com.github.gtache.autosubtitle.whisper.common;
requires org.apache.logging.log4j;
requires java.prefs;
exports com.github.gtache.autosubtitle.whisperx;
exports com.github.gtache.autosubtitle.setup.whisperx;