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

Rework to avoid using preferences object to retrieve options

Browse Source
This commit is contained in:
Guillaume Tâche
2024-09-22 21:59:10 +02:00
parent 7f99c48e2c
commit c59619da2d
115 changed files with 2294 additions and 765 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
whisper/whisperx/src/main/java/com/github/gtache/autosubtitle/modules/subtitle/parser/json/whisperx/WhisperXJsonModule.java
View File

@@ -1,8 +1,8 @@
package com.github.gtache.autosubtitle.modules.subtitle.parser.json.whisperx;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.github.gtache.autosubtitle.subtitle.converter.SubtitleConverter;
import com.github.gtache.autosubtitle.subtitle.parser.json.whisperx.JSONSubtitleConverter;
import com.google.gson.Gson;
import dagger.Binds;
import dagger.Module;
import dagger.Provides;
@@ -25,7 +25,7 @@ public abstract class WhisperXJsonModule {
@Provides
@Singleton
static Gson providesGson() {
return new Gson();
static ObjectMapper providesObjectMapper() {
return new ObjectMapper();
}
}

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

@@ -5,7 +5,7 @@ import com.github.gtache.autosubtitle.impl.OS;
import com.github.gtache.autosubtitle.modules.setup.whisper.WhisperVenvPath;
import com.github.gtache.autosubtitle.process.ProcessRunner;
import com.github.gtache.autosubtitle.subtitle.converter.SubtitleConverterProvider;
import com.github.gtache.autosubtitle.subtitle.extractor.ExtractionModel;
import com.github.gtache.autosubtitle.subtitle.extractor.ExtractOptions;
import com.github.gtache.autosubtitle.subtitle.extractor.SubtitleExtractor;
import com.github.gtache.autosubtitle.subtitle.extractor.whisper.AbstractWhisperSubtitleExtractor;
import com.github.gtache.autosubtitle.whisper.WhisperModels;
@@ -28,7 +28,9 @@ public class WhisperXSubtitleExtractor extends AbstractWhisperSubtitleExtractor
}
@Override
protected List<String> createArgs(final Path path, final Language language, final ExtractionModel model, final Path outputDir) {
protected List<String> createArgs(final Path path, final ExtractOptions options, final Path outputDir) {
final var model = options.model();
final var language = options.language();
final var args = new ArrayList<String>();
args.add(getPythonPath().toString());
args.add("-m");

58
whisper/whisperx/src/main/java/com/github/gtache/autosubtitle/subtitle/parser/json/whisperx/JSONSubtitleConverter.java
View File

@@ -1,16 +1,17 @@
package com.github.gtache.autosubtitle.subtitle.parser.json.whisperx;
import com.fasterxml.jackson.core.JsonProcessingException;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.github.gtache.autosubtitle.Language;
import com.github.gtache.autosubtitle.VideoInfo;
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.FormatException;
import com.github.gtache.autosubtitle.subtitle.converter.FormatOptions;
import com.github.gtache.autosubtitle.subtitle.converter.ParseException;
import com.github.gtache.autosubtitle.subtitle.converter.ParseOptions;
import com.github.gtache.autosubtitle.subtitle.converter.SubtitleConverter;
import com.github.gtache.autosubtitle.subtitle.impl.SubtitleCollectionImpl;
import com.github.gtache.autosubtitle.subtitle.impl.SubtitleImpl;
import com.google.gson.Gson;
import javax.inject.Inject;
import javax.inject.Singleton;
@@ -19,7 +20,6 @@ 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;
@@ -30,42 +30,38 @@ import java.util.stream.Stream;
@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;
private final ObjectMapper mapper;
@Inject
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;
JSONSubtitleConverter(final ObjectMapper mapper) {
this.mapper = Objects.requireNonNull(mapper);
}
@Override
public String format(final SubtitleCollection<?> collection, final VideoInfo videoInfo) {
public String format(final SubtitleCollection<?> collection, final FormatOptions options) throws FormatException {
final var segments = collection.subtitles().stream().map(s -> new JSONSubtitleSegment(s.start() / (double) 1000,
s.end() / (double) 1000, s.content(), List.of())).toList();
final var subtitles = new JSONSubtitles(segments, collection.language().iso2());
return gson.toJson(subtitles);
try {
return mapper.writeValueAsString(subtitles);
} catch (final JsonProcessingException e) {
throw new FormatException(e);
}
}
@Override
public SubtitleCollectionImpl<SubtitleImpl> parse(final String content) throws ParseException {
public SubtitleCollectionImpl<SubtitleImpl> parse(final String content, final ParseOptions options) throws ParseException {
try {
final var json = gson.fromJson(content, JSONSubtitles.class);
final var json = mapper.readValue(content, JSONSubtitles.class);
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);
return splitSubtitle(s, options);
}
}).sorted(Comparator.comparing(Subtitle::start).thenComparing(Subtitle::end)).toList();
@@ -77,23 +73,23 @@ 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);
private static Stream<SubtitleImpl> splitSubtitle(final JSONSubtitleSegment segment, final ParseOptions options) {
final var maxLineLength = options.maxLineLength();
final var maxLines = options.maxLines();
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);
return splitSubtitleLines(segment, options);
} else {
return splitSubtitleWords(segment);
return splitSubtitleWords(segment, options);
}
}
private Stream<SubtitleImpl> splitSubtitleLines(final JSONSubtitleSegment segment) {
final var maxLineLength = preferences.getInt(MAX_LINE_LENGTH, defaultMaxLineLength);
private static Stream<SubtitleImpl> splitSubtitleLines(final JSONSubtitleSegment segment, final ParseOptions options) {
final var maxLineLength = options.maxLineLength();
final var text = segment.text();
final var split = SPLIT_PATTERN.split(text);
final var builder = new StringBuilder(text.length());
@@ -116,9 +112,9 @@ public class JSONSubtitleConverter implements SubtitleConverter<SubtitleImpl> {
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);
private static Stream<SubtitleImpl> splitSubtitleWords(final JSONSubtitleSegment segment, final ParseOptions options) {
final var maxLineLength = options.maxLineLength();
final var maxLines = options.maxLines();
final var ret = new ArrayList<SubtitleImpl>(segment.text().length() / (maxLines * maxLineLength));
final var builder = new StringBuilder(maxLines * maxLineLength);
final var words = segment.words();

3
whisper/whisperx/src/main/java/com/github/gtache/autosubtitle/subtitle/parser/json/whisperx/JSONSubtitles.java
View File

@@ -1,6 +1,9 @@
package com.github.gtache.autosubtitle.subtitle.parser.json.whisperx;
import com.fasterxml.jackson.annotation.JsonIgnoreProperties;
import java.util.List;
@JsonIgnoreProperties("word_segments")
public record JSONSubtitles(List<JSONSubtitleSegment> segments, String language) {
}

2
whisper/whisperx/src/main/java/module-info.java
View File

@@ -3,8 +3,8 @@
*/
module com.github.gtache.autosubtitle.whisperx {
requires transitive com.github.gtache.autosubtitle.whisper.common;
requires transitive com.fasterxml.jackson.databind;
requires org.apache.logging.log4j;
requires java.prefs;
exports com.github.gtache.autosubtitle.whisperx;
exports com.github.gtache.autosubtitle.setup.whisperx;