feat(translate): translate a whole playlist so the next episodes are ready (#16)
Co-authored-by: Tri Vo Khac <git@trivk.dev>
This commit was merged in pull request #16.
This commit is contained in:
1 parent
ad50230829
commit
bcfc2f3f6a
13 files changed
+480
-61
No files matched your search
@@ -1,6 +1,7 @@
|
||||
#include "translationservice.h"
|
||||
|
||||
#include "apikeystore.h"
|
||||
#include "mediainfo.h"
|
||||
#include "qtpotatosettings.h"
|
||||
#include "translationpresets.h"
|
||||
#include "translationstore.h"
|
||||
@@ -69,6 +70,10 @@ struct TranslationService::Job {
|
||||
std::atomic_bool cancelled = false;
|
||||
std::unique_ptr<LlmClient> client;
|
||||
QString mediaTitle;
|
||||
bool playlist = false;
|
||||
// In a playlist: the episode being translated (from 0) and how many there are.
|
||||
int episode = 0;
|
||||
int episodes = 0;
|
||||
};
|
||||
|
||||
struct TranslationService::Batches {
|
||||
@@ -160,6 +165,25 @@ QString TranslationService::glossaryError() const
|
||||
return m_glossaryError;
|
||||
}
|
||||
|
||||
QUrl TranslationService::currentVideo() const
|
||||
{
|
||||
return m_currentVideo;
|
||||
}
|
||||
|
||||
void TranslationService::setCurrentVideo(const QUrl &url)
|
||||
{
|
||||
if (url == m_currentVideo) {
|
||||
return;
|
||||
}
|
||||
m_currentVideo = url;
|
||||
forgetTranslation();
|
||||
// A playlist translation is meant to go on while the episodes play.
|
||||
if (m_job && !m_job->playlist) {
|
||||
cancel();
|
||||
}
|
||||
Q_EMIT currentVideoChanged();
|
||||
}
|
||||
|
||||
void TranslationService::loadGlossary(const QUrl &videoUrl)
|
||||
{
|
||||
const QString directory = videoUrl.isLocalFile() ? QFileInfo(videoUrl.toLocalFile()).absolutePath() : QString();
|
||||
@@ -323,7 +347,25 @@ void TranslationService::translate(const QUrl &videoUrl, const QVariantMap &trac
|
||||
cancel();
|
||||
m_job = std::make_shared<Job>();
|
||||
m_job->mediaTitle = mediaTitle;
|
||||
run(videoUrl, track);
|
||||
run(m_job, videoUrl, track);
|
||||
}
|
||||
|
||||
void TranslationService::translatePlaylist(const QList<QUrl> &urls, const QVariantMap ¤tTrack, const QString ¤tTitle)
|
||||
{
|
||||
cancel();
|
||||
// Start with the episode playing, then the ones after it, which are watched next.
|
||||
QList<QUrl> ordered;
|
||||
const qsizetype current = std::max<qsizetype>(0, urls.indexOf(m_currentVideo));
|
||||
for (qsizetype i = 0; i < urls.size(); ++i) {
|
||||
const QUrl &url = urls.at((current + i) % urls.size());
|
||||
if (!ordered.contains(url)) {
|
||||
ordered.append(url);
|
||||
}
|
||||
}
|
||||
m_job = std::make_shared<Job>();
|
||||
m_job->playlist = true;
|
||||
m_job->episodes = ordered.size();
|
||||
runPlaylist(m_job, ordered, currentTrack, currentTitle);
|
||||
}
|
||||
|
||||
void TranslationService::cancel()
|
||||
@@ -408,10 +450,8 @@ QCoro::Task<Subtitles::ExtractResult> TranslationService::loadCues(std::shared_p
|
||||
});
|
||||
}
|
||||
|
||||
QCoro::Task<> TranslationService::run(QUrl videoUrl, QVariantMap track)
|
||||
QCoro::Task<std::optional<LlmClient::Config>> TranslationService::startJob(std::shared_ptr<Job> job)
|
||||
{
|
||||
const std::shared_ptr<Job> job = m_job;
|
||||
|
||||
setErrorText({});
|
||||
setProgress(0);
|
||||
setBusy(true);
|
||||
@@ -419,29 +459,123 @@ QCoro::Task<> TranslationService::run(QUrl videoUrl, QVariantMap track)
|
||||
|
||||
auto [config, configError] = co_await loadConfig();
|
||||
if (job->cancelled) {
|
||||
co_return;
|
||||
co_return std::nullopt;
|
||||
}
|
||||
if (configError.isEmpty() && config.model.isEmpty()) {
|
||||
configError = i18n("Choose a model in Settings first.");
|
||||
}
|
||||
if (!configError.isEmpty()) {
|
||||
fail(job, configError);
|
||||
co_return;
|
||||
}
|
||||
if (config.model.isEmpty()) {
|
||||
fail(job, i18n("Choose a model in Settings first."));
|
||||
co_return;
|
||||
co_return std::nullopt;
|
||||
}
|
||||
co_return config;
|
||||
}
|
||||
|
||||
Subtitles::ExtractResult source = co_await loadCues(job, videoUrl, track);
|
||||
QCoro::Task<> TranslationService::run(std::shared_ptr<Job> job, QUrl videoUrl, QVariantMap track)
|
||||
{
|
||||
const std::optional<LlmClient::Config> config = co_await startJob(job);
|
||||
if (!config) {
|
||||
co_return;
|
||||
}
|
||||
const Outcome outcome = co_await translateVideo(job, *config, videoUrl, track);
|
||||
if (job->cancelled) {
|
||||
co_return;
|
||||
}
|
||||
if (!source.error.isEmpty()) {
|
||||
fail(job, source.error);
|
||||
const QString error = outcome.error.isEmpty() ? publish(job, videoUrl, outcome) : outcome.error;
|
||||
if (!error.isEmpty()) {
|
||||
fail(job, error);
|
||||
co_return;
|
||||
}
|
||||
endJob(job);
|
||||
}
|
||||
|
||||
QCoro::Task<> TranslationService::runPlaylist(std::shared_ptr<Job> job, QList<QUrl> urls, QVariantMap currentTrack, QString currentTitle)
|
||||
{
|
||||
const std::optional<LlmClient::Config> config = co_await startJob(job);
|
||||
if (!config) {
|
||||
co_return;
|
||||
}
|
||||
|
||||
int translatedCount = 0;
|
||||
int alreadyTranslatedCount = 0;
|
||||
QStringList problems;
|
||||
for (qsizetype i = 0; i < urls.size(); ++i) {
|
||||
const QUrl videoUrl = urls.at(i);
|
||||
const QString name = videoUrl.fileName();
|
||||
job->episode = int(i);
|
||||
setJobProgress(*job, 0);
|
||||
setJobStatus(*job, i18n("Reading subtitles…"));
|
||||
if (!savedTranslation(videoUrl).isEmpty()) {
|
||||
++alreadyTranslatedCount;
|
||||
continue;
|
||||
}
|
||||
|
||||
QVariantMap track = currentTrack;
|
||||
job->mediaTitle = currentTitle;
|
||||
if (videoUrl != m_currentVideo || currentTrack.isEmpty()) {
|
||||
if (!videoUrl.isLocalFile()) {
|
||||
problems.append(i18nc("@info %1 is a file name", "%1: only subtitles of local files can be translated", name));
|
||||
continue;
|
||||
}
|
||||
const QString path = videoUrl.toLocalFile();
|
||||
const Subtitles::SourceResult source = co_await QtConcurrent::run([path] {
|
||||
return Subtitles::findTranslationSource(path);
|
||||
});
|
||||
if (job->cancelled) {
|
||||
co_return;
|
||||
}
|
||||
if (!source.error.isEmpty() || source.streamIndex < 0) {
|
||||
problems.append(source.error.isEmpty() ? i18nc("@info %1 is a file name", "%1: no text subtitle", name)
|
||||
: i18nc("@info %1 is a file name, %2 an error", "%1: %2", name, source.error));
|
||||
continue;
|
||||
}
|
||||
track = QVariantMap{{u"ffIndex"_s, source.streamIndex}};
|
||||
job->mediaTitle = MediaInfo::cleanTitle(name);
|
||||
}
|
||||
|
||||
const Outcome outcome = co_await translateVideo(job, *config, videoUrl, track);
|
||||
if (job->cancelled) {
|
||||
co_return;
|
||||
}
|
||||
if (!outcome.error.isEmpty() && !outcome.skippable) {
|
||||
// Errors from the API or the disk would repeat for every episode.
|
||||
fail(job, i18nc("@info %1 is a file name, %2 an error", "Stopped translating the playlist at %1: %2", name, outcome.error));
|
||||
co_return;
|
||||
}
|
||||
const QString error = outcome.error.isEmpty() ? publish(job, videoUrl, outcome) : outcome.error;
|
||||
if (error.isEmpty()) {
|
||||
++translatedCount;
|
||||
} else {
|
||||
problems.append(i18nc("@info %1 is a file name, %2 an error", "%1: %2", name, error));
|
||||
}
|
||||
}
|
||||
|
||||
endJob(job);
|
||||
if (!problems.isEmpty()) {
|
||||
setErrorText(i18np("Translated %2 of %3 episodes. Skipped one:\n%4",
|
||||
"Translated %2 of %3 episodes. Skipped %1:\n%4",
|
||||
problems.size(),
|
||||
translatedCount + alreadyTranslatedCount,
|
||||
urls.size(),
|
||||
problems.join(u'\n')));
|
||||
}
|
||||
Q_EMIT playlistTranslated(translatedCount, alreadyTranslatedCount, int(problems.size()));
|
||||
}
|
||||
|
||||
QCoro::Task<TranslationService::Outcome> TranslationService::translateVideo(std::shared_ptr<Job> job, LlmClient::Config config, QUrl videoUrl, QVariantMap track)
|
||||
{
|
||||
setJobProgress(*job, 0);
|
||||
setJobStatus(*job, i18n("Reading subtitles…"));
|
||||
Subtitles::ExtractResult source = co_await loadCues(job, videoUrl, track);
|
||||
if (job->cancelled) {
|
||||
co_return Outcome{};
|
||||
}
|
||||
if (!source.error.isEmpty()) {
|
||||
co_return Outcome::failed(source.error, true);
|
||||
}
|
||||
const QList<SubtitleCue> cues = Subtitles::mergeRepeated(source.cues);
|
||||
if (cues.isEmpty()) {
|
||||
fail(job, i18n("This subtitle track has no text to translate."));
|
||||
co_return;
|
||||
co_return Outcome::failed(i18n("This subtitle track has no text to translate."), true);
|
||||
}
|
||||
|
||||
// Episodes in one folder share a glossary. Streams have no folder, so they get none.
|
||||
@@ -449,24 +583,30 @@ QCoro::Task<> TranslationService::run(QUrl videoUrl, QVariantMap track)
|
||||
const bool useGlossary = !seriesDirectory.isEmpty();
|
||||
GlossaryStore::Glossary glossary = useGlossary ? GlossaryStore::load(seriesDirectory) : GlossaryStore::Glossary{};
|
||||
const QString episode = videoUrl.isLocalFile() ? QFileInfo(videoUrl.toLocalFile()).completeBaseName() : job->mediaTitle;
|
||||
const auto saveGlossaryGroups = [&]() {
|
||||
// Returns an error message.
|
||||
const auto saveGlossaryGroups = [&]() -> QString {
|
||||
QString error;
|
||||
if (!GlossaryStore::save(glossary, &error)) {
|
||||
fail(job, i18n("Could not save the translation glossary: %1", error));
|
||||
return false;
|
||||
return i18n("Could not save the translation glossary: %1", error);
|
||||
}
|
||||
m_glossary = glossary;
|
||||
m_glossaryEpisode = episode;
|
||||
Q_EMIT glossaryChanged();
|
||||
return true;
|
||||
// Another episode of the playlist only updates the terms of the open series.
|
||||
if (!job->playlist || videoUrl == m_currentVideo) {
|
||||
m_glossary = glossary;
|
||||
m_glossaryEpisode = episode;
|
||||
Q_EMIT glossaryChanged();
|
||||
} else if (m_glossary.directory == glossary.directory) {
|
||||
m_glossary = glossary;
|
||||
Q_EMIT glossaryChanged();
|
||||
}
|
||||
return {};
|
||||
};
|
||||
if (useGlossary) {
|
||||
// New terms belong to the episode being translated. Migrate legacy or
|
||||
// imported entries without provenance to this first episode too. This
|
||||
// happens before the cache shortcut and never changes translation text.
|
||||
glossary = GlossaryStore::assignUnassigned(glossary, episode);
|
||||
if (!saveGlossaryGroups()) {
|
||||
co_return;
|
||||
if (const QString error = saveGlossaryGroups(); !error.isEmpty()) {
|
||||
co_return Outcome::failed(error);
|
||||
}
|
||||
}
|
||||
const QString sourceSrt = Subtitles::toSrt(cues);
|
||||
@@ -488,12 +628,11 @@ QCoro::Task<> TranslationService::run(QUrl videoUrl, QVariantMap track)
|
||||
QString outputPath = TranslationStore::cachePath(cacheDir(), videoUrl, configKey, suffix);
|
||||
|
||||
if (QFileInfo::exists(outputPath)) {
|
||||
finish(job, videoUrl, suffix, configKey, outputPath, true);
|
||||
co_return;
|
||||
co_return Outcome::ready(outputPath, suffix, configKey, true);
|
||||
}
|
||||
|
||||
if (useGlossary && QtPotatoSettings::self()->autoBuildGlossary()) {
|
||||
setStatusText(i18n("Building glossary…"));
|
||||
setJobStatus(*job, i18n("Building glossary…"));
|
||||
QStringList extractionLines;
|
||||
QHash<QString, bool> seen;
|
||||
for (const SubtitleCue &cue : cues) {
|
||||
@@ -518,7 +657,7 @@ QCoro::Task<> TranslationService::run(QUrl videoUrl, QVariantMap track)
|
||||
job->client = std::make_unique<LlmClient>(m_network, config);
|
||||
const LlmClient::BatchResult extracted = co_await job->client->extractGlossary(sampled, job->mediaTitle);
|
||||
if (job->cancelled) {
|
||||
co_return;
|
||||
co_return Outcome{};
|
||||
}
|
||||
if (extracted.error.isEmpty()) {
|
||||
QList<GlossaryStore::Entry> additions;
|
||||
@@ -529,17 +668,16 @@ QCoro::Task<> TranslationService::run(QUrl videoUrl, QVariantMap track)
|
||||
// Suggestions extracted while translating this episode belong to
|
||||
// this episode. Existing terms keep their original ownership.
|
||||
glossary = GlossaryStore::merge(glossary, additions, episode);
|
||||
if (!saveGlossaryGroups()) {
|
||||
co_return;
|
||||
if (const QString error = saveGlossaryGroups(); !error.isEmpty()) {
|
||||
co_return Outcome::failed(error);
|
||||
}
|
||||
} else {
|
||||
setStatusText(i18n("Glossary unavailable; translating without it."));
|
||||
setJobStatus(*job, i18n("Glossary unavailable; translating without it."));
|
||||
}
|
||||
configKey = configKeyFor(glossary);
|
||||
outputPath = TranslationStore::cachePath(cacheDir(), videoUrl, configKey, suffix);
|
||||
if (QFileInfo::exists(outputPath)) {
|
||||
finish(job, videoUrl, suffix, configKey, outputPath, true);
|
||||
co_return;
|
||||
co_return Outcome::ready(outputPath, suffix, configKey, true);
|
||||
}
|
||||
}
|
||||
|
||||
@@ -567,7 +705,7 @@ QCoro::Task<> TranslationService::run(QUrl videoUrl, QVariantMap track)
|
||||
batches->lines = uniqueLines;
|
||||
batches->translations.resize(uniqueLines.size());
|
||||
batches->batchSize = QtPotatoSettings::self()->batchSize();
|
||||
setStatusText(i18n("Translating… %1 of %2 lines", 0, uniqueLines.size()));
|
||||
setLinesStatus(*job, 0, uniqueLines.size());
|
||||
|
||||
// Tasks start running as soon as they are created; awaiting them just waits for the end.
|
||||
std::vector<QCoro::Task<>> workers;
|
||||
@@ -578,11 +716,10 @@ QCoro::Task<> TranslationService::run(QUrl videoUrl, QVariantMap track)
|
||||
co_await std::move(worker);
|
||||
}
|
||||
if (job->cancelled) {
|
||||
co_return;
|
||||
co_return Outcome{};
|
||||
}
|
||||
if (!batches->error.isEmpty()) {
|
||||
fail(job, batches->error);
|
||||
co_return;
|
||||
co_return Outcome::failed(batches->error);
|
||||
}
|
||||
|
||||
QList<SubtitleCue> output = cues;
|
||||
@@ -593,36 +730,64 @@ QCoro::Task<> TranslationService::run(QUrl videoUrl, QVariantMap track)
|
||||
QDir().mkpath(QFileInfo(outputPath).absolutePath());
|
||||
QSaveFile file(outputPath);
|
||||
if (!file.open(QIODevice::WriteOnly) || file.write(Subtitles::toSrt(output).toUtf8()) < 0 || !file.commit()) {
|
||||
fail(job, i18n("Could not save the translation to %1: %2", outputPath, file.errorString()));
|
||||
co_return;
|
||||
co_return Outcome::failed(i18n("Could not save the translation to %1: %2", outputPath, file.errorString()));
|
||||
}
|
||||
|
||||
finish(job, videoUrl, suffix, configKey, outputPath, false);
|
||||
co_return Outcome::ready(outputPath, suffix, configKey, false);
|
||||
}
|
||||
|
||||
void TranslationService::finish(const std::shared_ptr<Job> &job,
|
||||
const QUrl &videoUrl,
|
||||
const QString &suffix,
|
||||
const QString &configKey,
|
||||
const QString &cacheFile,
|
||||
bool fromCache)
|
||||
QString TranslationService::publish(const std::shared_ptr<Job> &job, const QUrl &videoUrl, const Outcome &outcome)
|
||||
{
|
||||
QFile file(cacheFile);
|
||||
QFile file(outcome.cacheFile);
|
||||
if (!file.open(QIODevice::ReadOnly)) {
|
||||
fail(job, i18n("Could not read the translation from %1: %2", cacheFile, file.errorString()));
|
||||
return;
|
||||
return i18n("Could not read the translation from %1: %2", outcome.cacheFile, file.errorString());
|
||||
}
|
||||
// The cache keeps every version; the published copy is the one players load.
|
||||
const QString registryFile = TranslationStore::registryFile();
|
||||
TranslationStore::Registry registry = TranslationStore::load(registryFile);
|
||||
const TranslationStore::PublishResult published = TranslationStore::publish(registry, videoUrl, suffix, configKey, file.readAll(), cacheFile);
|
||||
const TranslationStore::PublishResult published =
|
||||
TranslationStore::publish(registry, videoUrl, outcome.suffix, outcome.configKey, file.readAll(), outcome.cacheFile);
|
||||
TranslationStore::save(registryFile, registry);
|
||||
|
||||
// Other episodes of a playlist load their translation when they are opened.
|
||||
if (!job->playlist || videoUrl == m_currentVideo) {
|
||||
m_lastTranslationPath = published.path;
|
||||
Q_EMIT translated(QUrl::fromLocalFile(published.path), outcome.fromCache);
|
||||
}
|
||||
return {};
|
||||
}
|
||||
|
||||
void TranslationService::endJob(const std::shared_ptr<Job> &job)
|
||||
{
|
||||
if (m_job != job) {
|
||||
return;
|
||||
}
|
||||
m_job.reset();
|
||||
m_lastTranslationPath = published.path;
|
||||
setBusy(false);
|
||||
setStatusText({});
|
||||
Q_EMIT translated(QUrl::fromLocalFile(published.path), fromCache);
|
||||
}
|
||||
|
||||
void TranslationService::setJobStatus(const Job &job, const QString &text)
|
||||
{
|
||||
if (!job.playlist) {
|
||||
setStatusText(text);
|
||||
return;
|
||||
}
|
||||
setStatusText(i18nc("@info:status %1 and %2 are numbers, %3 is what is being done",
|
||||
"Episode %1 of %2: %3",
|
||||
job.episode + 1,
|
||||
job.episodes,
|
||||
text));
|
||||
}
|
||||
|
||||
void TranslationService::setLinesStatus(const Job &job, qsizetype done, qsizetype total)
|
||||
{
|
||||
// The line count would not fit after the episode number; the progress bar shows it.
|
||||
setJobStatus(job, job.playlist ? i18n("Translating…") : i18n("Translating… %1 of %2 lines", done, total));
|
||||
}
|
||||
|
||||
void TranslationService::setJobProgress(const Job &job, double episodeProgress)
|
||||
{
|
||||
setProgress(job.playlist ? (job.episode + episodeProgress) / job.episodes : episodeProgress);
|
||||
}
|
||||
|
||||
QString TranslationService::savedTranslation(const QUrl &videoUrl) const
|
||||
@@ -716,8 +881,8 @@ QCoro::Task<> TranslationService::translateBatches(std::shared_ptr<Job> job, std
|
||||
batches->translations[start + i] = result.translations.at(i);
|
||||
}
|
||||
batches->doneLines += result.translations.size();
|
||||
setProgress(double(batches->doneLines) / batches->lines.size());
|
||||
setStatusText(i18n("Translating… %1 of %2 lines", batches->doneLines, batches->lines.size()));
|
||||
setJobProgress(*job, double(batches->doneLines) / batches->lines.size());
|
||||
setLinesStatus(*job, batches->doneLines, batches->lines.size());
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
Reference in new issue
Block a user