Formerly KrillinAI, VideoTranslation and Dubbing Tool for 120+ Languages. Open-source AI workspace for creators, powered by Codex. Video downloader, video translation, create videos, images, audios, and edits with Agents in one place. For your Social Media like YouTube, TikTok, Instagram, etc.
TypeScript agent-skillscodex-clideskop-appdubbingimage-generationlocalizationskillsttsvideo-editingvideo-generationvideo-transcriptionvideo-translation
Support block VTT subtitles
wulien6a5489005949
diff --git a/config/config.go b/config/config.goindex d643238e..e19ba8d3 100644--- a/config/config.go+++ b/config/config.go@@ -22,6 +22,8 @@ type App struct {TranscribeMaxAttempts int `toml:"transcribe_max_attempts"`TranslateMaxAttempts int `toml:"translate_max_attempts"`MaxSentenceLength int `toml:"max_sentence_length"`+ EnableBlockVttBatch bool `toml:"enable_block_vtt_batch"`+ VttBatchSize int `toml:"vtt_batch_size"`Proxy string `toml:"proxy"`ParsedProxy *url.URL `toml:"-"`}@@ -100,6 +102,8 @@ var Conf = Config{TranscribeMaxAttempts: 3,TranslateMaxAttempts: 3,MaxSentenceLength: 70,+ EnableBlockVttBatch: false,+ VttBatchSize: 10,},Server: Server{Host: "127.0.0.1",diff --git a/internal/service/subtitle_service.go b/internal/service/subtitle_service.goindex 8aa6eb7a..24fdad66 100644--- a/internal/service/subtitle_service.go+++ b/internal/service/subtitle_service.go@@ -4,6 +4,7 @@ import ("context""errors""fmt"+ "krillin-ai/config""krillin-ai/internal/dto""krillin-ai/internal/storage""krillin-ai/internal/types"@@ -164,15 +165,52 @@ func (s Service) StartSubtitleTask(req dto.StartVideoSubtitleTaskReq) (*dto.StarURL: req.Url,TaskPtr: stepParam.TaskPtr,}- srtFile, err := s.YouTubeSubtitleSrv.Process(ctx, req)++ // 先下载VTT字幕+ vttFile, err := s.YouTubeSubtitleSrv.downloadYouTubeSubtitle(ctx, req)if err != nil {- // 下载或处理字幕失败,回退到音频转录方式- log.GetLogger().Warn("Failed to process YouTube subtitles with vtt, falling back to audio transcription",+ // 下载失败,回退到音频转录方式+ log.GetLogger().Warn("Failed to download YouTube subtitles, falling back to audio transcription",zap.String("taskId", taskId), zap.Error(err))stepParam.TaskPtr.Status = types.SubtitleTaskStatusFailedstepParam.TaskPtr.FailReason = err.Error()return}+ req.VttFile = vttFile++ // 检测VTT格式类型+ hasWordTimestamps := true // 默认假设有单词级时间戳+ if config.Conf.App.EnableBlockVttBatch {+ // 只有启用了新功能才进行格式检测+ detected, detectErr := s.YouTubeSubtitleSrv.DetectVttFormat(vttFile)+ if detectErr != nil {+ log.GetLogger().Warn("VTT格式检测失败,使用默认处理方式",+ zap.String("taskId", taskId), zap.Error(detectErr))+ } else {+ hasWordTimestamps = detected+ }+ }++ var srtFile string+ if hasWordTimestamps {+ // 使用原有的word-level处理流程(完全不变)+ log.GetLogger().Info("使用word-level VTT处理流程", zap.String("taskId", taskId))+ srtFile, err = s.YouTubeSubtitleSrv.processYouTubeSubtitle(ctx, req)+ } else {+ // 使用新的block-level处理流程+ log.GetLogger().Info("使用block-level VTT处理流程", zap.String("taskId", taskId))+ srtFile, err = s.YouTubeSubtitleSrv.ProcessBlockLevelVtt(ctx, req)+ }++ if err != nil {+ // 处理字幕失败,回退到音频转录方式+ log.GetLogger().Warn("Failed to process YouTube subtitles, falling back to audio transcription",+ zap.String("taskId", taskId), zap.Error(err))+ stepParam.TaskPtr.Status = types.SubtitleTaskStatusFailed+ stepParam.TaskPtr.FailReason = err.Error()+ return+ }+stepParam.BilingualSrtFilePath = srtFileerr = splitSrt(&stepParam)if err != nil {diff --git a/internal/service/translate.go b/internal/service/translate.goindex 4acbf63d..afce1596 100644--- a/internal/service/translate.go+++ b/internal/service/translate.go@@ -10,6 +10,7 @@ import ("krillin-ai/pkg/util""strings""sync"+ "time""go.uber.org/zap")@@ -393,3 +394,270 @@ IMPORTANT: The previous translation was inadequate. Please ensure:return originalPrompt + enhancement}++// BatchTranslateSrtBlocks 批量翻译SRT字幕块(每次处理N个单元)+func (t *Translator) BatchTranslateSrtBlocks(blocks []*util.SrtBlock, originLang, targetLang string, taskPtr *types.SubtitleTask) error {+ if len(blocks) == 0 {+ return nil+ }++ batchSize := config.Conf.App.VttBatchSize+ if batchSize <= 0 {+ batchSize = 10 // 默认值+ }++ originLangCode := types.StandardLanguageCode(originLang)+ targetLangCode := types.StandardLanguageCode(targetLang)++ // 统计有内容的字幕块数量+ validBlocksCount := 0+ for _, block := range blocks {+ if block.OriginLanguageSentence != "" {+ validBlocksCount+++ }+ }++ totalBatches := (validBlocksCount + batchSize - 1) / batchSize++ log.GetLogger().Info("开始批量翻译SRT字幕块",+ zap.Int("总块数", len(blocks)),+ zap.Int("有效块数", validBlocksCount),+ zap.Int("批量大小", batchSize),+ zap.Int("总批次数", totalBatches),+ zap.String("源语言", originLang),+ zap.String("目标语言", targetLang))++ // 分批处理+ currentBatch := 0+ for i := 0; i < len(blocks); i += batchSize {+ end := i + batchSize+ if end > len(blocks) {+ end = len(blocks)+ }++ batch := blocks[i:end]++ // 检查批次中是否有有效内容+ hasValidContent := false+ for _, block := range batch {+ if block.OriginLanguageSentence != "" {+ hasValidContent = true+ break+ }+ }++ if !hasValidContent {+ continue+ }++ currentBatch+++ log.GetLogger().Info("处理批次",+ zap.Int("当前批次", currentBatch),+ zap.Int("总批次", totalBatches),+ zap.Int("批次起始", i+1),+ zap.Int("批次结束", end),+ zap.String("进度", fmt.Sprintf("%d/%d", currentBatch, totalBatches)))++ // 构建批量翻译的输入文本+ var originTexts []string+ for _, block := range batch {+ if block.OriginLanguageSentence != "" {+ originTexts = append(originTexts, block.OriginLanguageSentence)+ }+ }++ if len(originTexts) == 0 {+ continue+ }++ // 调用批量翻译+ translations, err := t.batchTranslateTexts(originTexts, originLangCode, targetLangCode)+ if err != nil {+ log.GetLogger().Error("批量翻译失败,尝试单独翻译",+ zap.Error(err),+ zap.Int("当前批次", currentBatch),+ zap.Int("总批次", totalBatches),+ zap.Int("批次起始", i+1),+ zap.Int("批次结束", end))++ // 失败时回退到单独翻译每个字幕+ for j, block := range batch {+ if block.OriginLanguageSentence == "" {+ continue+ }++ log.GetLogger().Info("单独翻译字幕",+ zap.Int("索引", i+j+1),+ zap.String("文本预览", block.OriginLanguageSentence[:min(len(block.OriginLanguageSentence), 50)]))++ translatedText, err := t.translateSingleText(+ block.OriginLanguageSentence,+ originLangCode,+ targetLangCode)++ if err != nil {+ log.GetLogger().Error("单独翻译失败,使用原文",+ zap.Error(err),+ zap.Int("索引", i+j+1))+ block.TargetLanguageSentence = block.OriginLanguageSentence+ } else {+ block.TargetLanguageSentence = translatedText+ }+ }++ // 更新任务进度+ if taskPtr != nil {+ progress := float64(currentBatch) / float64(totalBatches)+ taskPtr.ProcessPct = 40 + uint8(progress*50)+ }+ continue+ }++ log.GetLogger().Info("批量翻译成功",+ zap.Int("当前批次", currentBatch),+ zap.Int("翻译数量", len(translations)))++ // 将翻译结果赋值回SRT块+ for j, translation := range translations {+ if j < len(batch) {+ batch[j].TargetLanguageSentence = translation+ }+ }++ // 更新任务进度(假设翻译占总进度的50%)+ if taskPtr != nil {+ progress := float64(currentBatch) / float64(totalBatches)+ // 假设翻译在40%-90%之间+ taskPtr.ProcessPct = 40 + uint8(progress*50)+ }+ }++ log.GetLogger().Info("批量翻译完成",+ zap.Int("总块数", len(blocks)),+ zap.Int("处理批次", currentBatch))+ return nil+}++// batchTranslateTexts 批量翻译多个文本(通过单次LLM调用)+func (t *Translator) batchTranslateTexts(texts []string, originLang, targetLang types.StandardLanguageCode) ([]string, error) {+ if len(texts) == 0 {+ return []string{}, nil+ }++ // 构建批量翻译提示词+ var textList strings.Builder+ for i, text := range texts {+ textList.WriteString(fmt.Sprintf("%d. %s\n", i+1, text))+ }++ // 构建语言说明,如果是中文特别强调简体+ targetLangName := types.GetStandardLanguageName(targetLang)+ if targetLang == "zh" || string(targetLang) == "Chinese" {+ targetLangName = "Simplified Chinese (简体中文)"+ }++ prompt := fmt.Sprintf(`You are a professional subtitle translator. Translate the following %d subtitles from %s to %s.++CRITICAL INSTRUCTIONS:+1. Translate naturally and fluently in %s+2. If target language is Chinese, MUST use Simplified Chinese characters (简体中文), NOT Traditional Chinese (繁体中文)+3. Maintain the exact same number of subtitles (%d items)+4. Preserve punctuation and formatting+5. Output ONLY valid JSON, NO markdown code blocks, NO explanations, NO notes+6. Start directly with { and end with }++Input subtitles:+%s++Required JSON format (output ONLY this structure):+{"translations":[{"index":1,"text":"译文1"},{"index":2,"text":"译文2"}]}`,+ len(texts),+ types.GetStandardLanguageName(originLang),+ targetLangName,+ targetLangName,+ len(texts),+ textList.String())++ // 调用LLM+ maxAttempts := 3+ var lastErr error++ for attempt := 0; attempt < maxAttempts; attempt++ {+ log.GetLogger().Info("调用LLM翻译",+ zap.Int("文本数", len(texts)),+ zap.Int("尝试次数", attempt+1))++ response, err := t.chatCompleter.ChatCompletion(prompt)+ if err != nil {+ lastErr = err+ log.GetLogger().Warn("批量翻译LLM调用失败,重试",+ zap.Error(err),+ zap.Int("尝试次数", attempt+1))+ time.Sleep(time.Second * time.Duration(attempt+1))+ continue+ }++ log.GetLogger().Info("LLM返回响应",+ zap.Int("响应长度", len(response)))++ // 解析JSON响应+ cleanResponse := util.CleanMarkdownCodeBlock(response)++ var result struct {+ Translations []struct {+ Index int `json:"index"`+ Text string `json:"text"`+ } `json:"translations"`+ }++ if err := json.Unmarshal([]byte(cleanResponse), &result); err != nil {+ lastErr = fmt.Errorf("解析JSON失败: %w", err)+ if attempt == 0 {+ // 只在第一次失败时记录完整响应+ log.GetLogger().Warn("解析批量翻译响应失败,重试",+ zap.Error(err),+ zap.Int("尝试次数", attempt+1),+ zap.String("清理后响应", cleanResponse[:min(len(cleanResponse), 500)]))+ } else {+ log.GetLogger().Warn("解析批量翻译响应失败,重试",+ zap.Error(err),+ zap.Int("尝试次数", attempt+1))+ }+ continue+ }++ // 验证结果数量+ if len(result.Translations) != len(texts) {+ lastErr = fmt.Errorf("翻译结果数量不匹配: 期望 %d, 实际 %d", len(texts), len(result.Translations))+ log.GetLogger().Warn("翻译结果数量不匹配,重试",+ zap.Error(lastErr),+ zap.Int("尝试次数", attempt+1))+ continue+ }++ // 提取翻译结果+ translations := make([]string, len(texts))+ for _, trans := range result.Translations {+ if trans.Index > 0 && trans.Index <= len(texts) {+ translations[trans.Index-1] = strings.TrimSpace(trans.Text)+ }+ }++ return translations, nil+ }++ return nil, fmt.Errorf("批量翻译失败,已重试%d次: %w", maxAttempts, lastErr)+}++// translateSingleText 翻译单个文本(用作批量翻译失败时的回退)+func (t *Translator) translateSingleText(text string, originLang, targetLang types.StandardLanguageCode) (string, error) {+ prompt := fmt.Sprintf(types.SplitTextPrompt, types.GetStandardLanguageName(targetLang), text)++ translatedText, err := t.translateWithRetry(prompt, text, originLang, targetLang)+ if err != nil {+ return "", fmt.Errorf("单文本翻译失败: %w", err)+ }++ return translatedText, nil+}+diff --git a/internal/service/youtube_subtitle.go b/internal/service/youtube_subtitle.goindex 83c77d8e..a6635f46 100644--- a/internal/service/youtube_subtitle.go+++ b/internal/service/youtube_subtitle.go@@ -1448,6 +1448,49 @@ func (s *YouTubeSubtitleService) writeSrtBlocksToFile(blocks []*util.SrtBlock, sreturn nil}+// writeTargetLanguageSrtFile 写入只包含目标语言的SRT文件+func (s *YouTubeSubtitleService) writeTargetLanguageSrtFile(blocks []*util.SrtBlock, srtFile string) error {+ file, err := os.Create(srtFile)+ if err != nil {+ return fmt.Errorf("failed to create target language SRT file: %w", err)+ }+ defer file.Close()++ for _, block := range blocks {+ // 写入序号+ _, err = file.WriteString(fmt.Sprintf("%d\n", block.Index))+ if err != nil {+ return err+ }++ // 写入时间戳+ _, err = file.WriteString(block.Timestamp + "\n")+ if err != nil {+ return err+ }++ // 只写入目标语言文本+ if block.TargetLanguageSentence != "" {+ _, err = file.WriteString(block.TargetLanguageSentence + "\n\n")+ if err != nil {+ return err+ }+ } else if block.OriginLanguageSentence != "" {+ // 如果没有翻译,使用原语言+ _, err = file.WriteString(block.OriginLanguageSentence + "\n\n")+ if err != nil {+ return err+ }+ }+ }++ log.GetLogger().Info("Target language SRT file written successfully",+ zap.String("文件", srtFile),+ zap.Int("块数", len(blocks)))++ return nil+}+// writeShortMixedSrtFile 生成短字幕文件,基于已拆分的长字幕SRT块func (s *YouTubeSubtitleService) writeShortMixedSrtFile(srtBlocks []*util.SrtBlock, shortSrtFile string, sentences []Sentence) error {file, err := os.Create(shortSrtFile)@@ -2429,3 +2472,139 @@ func max(a, b int) int {}return b}++// DetectVttFormat 检测VTT文件格式类型,返回是否包含单词级时间戳+// 返回值: true=word-level (有行内时间戳), false=block-level (仅块级时间戳)+// 导出为公开方法以便测试+func (s *YouTubeSubtitleService) DetectVttFormat(vttFile string) (bool, error) {+ file, err := os.Open(vttFile)+ if err != nil {+ return false, fmt.Errorf("无法打开VTT文件: %w", err)+ }+ defer file.Close()++ scanner := bufio.NewScanner(file)+ // 匹配单词级行内时间戳的正则表达式+ wordTimeRegex := regexp.MustCompile(`<(\d{2}:\d{2}:\d{2}\.\d{3})>`)++ lineCount := 0+ maxLinesToCheck := 100 // 检查前100行即可判断格式++ for scanner.Scan() && lineCount < maxLinesToCheck {+ line := scanner.Text()+ lineCount++++ // 如果发现行内时间戳,说明是word-level格式+ if wordTimeRegex.MatchString(line) {+ log.GetLogger().Info("检测到word-level VTT格式(包含单词级时间戳)",+ zap.String("file", vttFile))+ return true, nil+ }+ }++ if err := scanner.Err(); err != nil {+ return false, fmt.Errorf("读取VTT文件错误: %w", err)+ }++ log.GetLogger().Info("检测到block-level VTT格式(仅块级时间戳)",+ zap.String("file", vttFile))+ return false, nil+}++// ProcessBlockLevelVtt 处理块级时间戳的VTT文件(无单词级时间戳)+// 流程: VTT → SRT → 批量翻译 → 双语字幕+func (s *YouTubeSubtitleService) ProcessBlockLevelVtt(ctx context.Context, req *YoutubeSubtitleReq) (string, error) {+ log.GetLogger().Info("开始处理block-level VTT字幕",+ zap.String("taskId", req.TaskId),+ zap.String("vttFile", req.VttFile))++ // 更新进度:开始处理+ if req.TaskPtr != nil {+ req.TaskPtr.ProcessPct = 20+ }++ // 1. 转换VTT到临时SRT文件+ tempSrtFile := filepath.Join(req.TaskBasePath, "temp_block_level.srt")+ err := util.ConvertBlockVttToSrt(req.VttFile, tempSrtFile)+ if err != nil {+ return "", fmt.Errorf("VTT转SRT失败: %w", err)+ }+ log.GetLogger().Info("VTT转SRT完成", zap.String("srtFile", tempSrtFile))++ // 更新进度:VTT转换完成+ if req.TaskPtr != nil {+ req.TaskPtr.ProcessPct = 30+ }++ // 2. 生成原文SRT文件(origin_language_srt.srt)+ originSrtFile := filepath.Join(req.TaskBasePath, types.SubtitleTaskOriginLanguageSrtFileName)+ // 直接复制临时SRT作为原文字幕+ originData, err := os.ReadFile(tempSrtFile)+ if err != nil {+ return "", fmt.Errorf("读取临时SRT失败: %w", err)+ }+ err = os.WriteFile(originSrtFile, originData, 0644)+ if err != nil {+ return "", fmt.Errorf("写入原文SRT失败: %w", err)+ }+ log.GetLogger().Info("生成原文SRT完成", zap.String("originSrtFile", originSrtFile))++ // 更新进度:原文SRT生成完成+ if req.TaskPtr != nil {+ req.TaskPtr.ProcessPct = 40+ }++ // 3. 解析SRT文件+ srtBlocks, err := util.ParseSrtFile(tempSrtFile)+ if err != nil {+ return "", fmt.Errorf("解析SRT文件失败: %w", err)+ }+ log.GetLogger().Info("解析SRT完成", zap.Int("字幕块数", len(srtBlocks)))++ // 4. 批量翻译(40%-90%的进度在BatchTranslateSrtBlocks内部更新)+ err = s.translator.BatchTranslateSrtBlocks(srtBlocks, req.OriginLanguage, req.TargetLanguage, req.TaskPtr)+ if err != nil {+ return "", fmt.Errorf("批量翻译失败: %w", err)+ }+ log.GetLogger().Info("批量翻译完成", zap.Int("翻译块数", len(srtBlocks)))++ // 更新进度:翻译完成+ if req.TaskPtr != nil {+ req.TaskPtr.ProcessPct = 90+ }++ // 5. 生成目标语言SRT文件(target_language_srt.srt)+ targetSrtFile := filepath.Join(req.TaskBasePath, types.SubtitleTaskTargetLanguageSrtFileName)+ err = s.writeTargetLanguageSrtFile(srtBlocks, targetSrtFile)+ if err != nil {+ return "", fmt.Errorf("写入目标语言SRT失败: %w", err)+ }+ log.GetLogger().Info("生成目标语言SRT完成", zap.String("targetSrtFile", targetSrtFile))++ // 更新进度:目标语言SRT生成完成+ if req.TaskPtr != nil {+ req.TaskPtr.ProcessPct = 95+ }++ // 6. 生成双语字幕文件(bilingual_srt.srt)+ bilingualSrtFile := filepath.Join(req.TaskBasePath, types.SubtitleTaskBilingualSrtFileName)+ err = s.writeSrtBlocksToFile(srtBlocks, bilingualSrtFile)+ if err != nil {+ return "", fmt.Errorf("写入双语字幕失败: %w", err)+ }+ log.GetLogger().Info("生成双语字幕完成", zap.String("bilingualSrtFile", bilingualSrtFile))++ // 更新进度:完成+ if req.TaskPtr != nil {+ req.TaskPtr.ProcessPct = 100+ }++ log.GetLogger().Info("block-level VTT处理完成",+ zap.String("taskId", req.TaskId),+ zap.String("输出文件", bilingualSrtFile))++ // 清理临时文件+ os.Remove(tempSrtFile)++ return bilingualSrtFile, nil+}diff --git a/internal/types/subtitle_task.go b/internal/types/subtitle_task.goindex 5dd031bf..ed562c93 100644--- a/internal/types/subtitle_task.go+++ b/internal/types/subtitle_task.go@@ -22,7 +22,7 @@ package typesvar SplitTextPrompt = `你是一个语言处理专家,专注于自然语言处理和翻译任务。按照以下步骤和要求,以最大程度实现字幕的准确和高质量翻译:-1. 将原句翻译为%s,确保译文流畅、自然,达到专业翻译水平,保持意思相同。+1. 将原句翻译为%s,确保译文流畅、自然,达到专业翻译水平,保持意思相同。**如果目标语言是中文,必须使用简体中文,不能使用繁体中文。**2. 严格依据标点符号(逗号: ,,、句号:。.、问号:??等)将内容拆分成单独的句子,并依据以下规则确保拆分长度较短:- 每个句子在保证句意完整的情况下尽可能短,适中的字幕长短能提供舒适的观看体验。- 根据连词(例如 "and", "but", "which", "when", "so", "所以", "但是", "因此", "考虑到" 等)进一步拆分句子,得到较短的结果。@@ -51,7 +51,7 @@ var SplitTextPrompt = `你是一个语言处理专家,专注于自然语言处// 带有语气词过滤的拆分Promptvar SplitTextPromptWithModalFilter = `你是一个语言处理专家,专注于自然语言处理和翻译任务。按照以下步骤和要求,以最大程度实现字幕的准确和高质量翻译:-1. 将原句翻译为%s,确保译文流畅、自然,达到专业翻译水平,保持意思相同。+1. 将原句翻译为%s,确保译文流畅、自然,达到专业翻译水平,保持意思相同。**如果目标语言是中文,必须使用简体中文,不能使用繁体中文。**2. 严格依据标点符号(逗号: ,,、句号:。.、问号:??等)将内容拆分成单独的句子,并依据以下规则确保拆分长度较短:- 每个句子在保证句意完整的情况下尽可能短,适中的字幕长短能提供舒适的观看体验。- 根据连词(例如 "and", "but", "which", "when", "so", "所以", "但是", "因此", "考虑到" 等)进一步拆分句子,得到较短的结果。@@ -187,13 +187,14 @@ Use "Previous Sentences" to understand context and maintain coherence.**Critical Rules**:1. OUTPUT MUST BE A SINGLE LINE: only the translated text-2. Translate naturally and idiomatically - avoid word-for-word literal translation-3. Remove stuttering/repeated words (e.g., "I I I'm" → translate as "I'm")-4. Filter out filler words (um, uh, er, ah, oh, mm, hmm, etc.) - do NOT translate them-5. Use proper punctuation marks in the target language (for Chinese: use ,。!? not spaces)-6. Keep the original meaning but express it smoothly and naturally in the target language-7. If sentence is incomplete/fragmentary, keep it that way but translate fluently-8. IGNORE the "Next Sentences" - they are for reference only+2. If translating to Chinese, MUST use Simplified Chinese characters (简体中文), NOT Traditional Chinese (繁体中文)+3. Translate naturally and idiomatically - avoid word-for-word literal translation+4. Remove stuttering/repeated words (e.g., "I I I'm" → translate as "I'm")+5. Filter out filler words (um, uh, er, ah, oh, mm, hmm, etc.) - do NOT translate them+6. Use proper punctuation marks in the target language (for Chinese: use ,。!? not spaces)+7. Keep the original meaning but express it smoothly and naturally in the target language+8. If sentence is incomplete/fragmentary, keep it that way but translate fluently+9. IGNORE the "Next Sentences" - they are for reference only**Context**:[Previous Sentences]diff --git a/pkg/util/youtube_subtitle_helper.go b/pkg/util/youtube_subtitle_helper.goindex 71d68f28..17b569c6 100644--- a/pkg/util/youtube_subtitle_helper.go+++ b/pkg/util/youtube_subtitle_helper.go@@ -213,6 +213,79 @@ func ConvertVttToSrt(inputPath, outputPath string) error {return os.WriteFile(outputPath, []byte(srtContent.String()), 0644)}+// ConvertBlockVttToSrt converts block-level VTT (without word-level timestamps) to SRT format+func ConvertBlockVttToSrt(inputPath, outputPath string) error {+ contentBytes, err := os.ReadFile(inputPath)+ if err != nil {+ return fmt.Errorf("failed to read VTT file: %w", err)+ }+ content := string(contentBytes)+ lines := strings.Split(content, "\n")++ timestampRegex := regexp.MustCompile(`^(\d{2}:\d{2}:\d{2})\.(\d{3})\s-->\s(\d{2}:\d{2}:\d{2})\.(\d{3})`)+ tagRegex := regexp.MustCompile(`<[^>]*>`)++ var srtBlocks []struct {+ startTime string+ endTime string+ text string+ }++ for i := 0; i < len(lines); {+ line := strings.TrimSpace(lines[i])++ // Skip header lines+ if line == "" || strings.HasPrefix(line, "WEBVTT") || strings.HasPrefix(line, "Kind:") || strings.HasPrefix(line, "Language:") {+ i+++ continue+ }++ // Check for timestamp line+ if matches := timestampRegex.FindStringSubmatch(line); len(matches) == 5 {+ startTime := fmt.Sprintf("%s,%s", matches[1], matches[2])+ endTime := fmt.Sprintf("%s,%s", matches[3], matches[4])++ i+++ var subtitleLines []string++ // Collect all subtitle lines until empty line+ for i < len(lines) && strings.TrimSpace(lines[i]) != "" {+ cleanLine := strings.TrimSpace(tagRegex.ReplaceAllString(lines[i], ""))+ if cleanLine != "" {+ subtitleLines = append(subtitleLines, cleanLine)+ }+ i+++ }++ // Merge multiple lines with space+ if len(subtitleLines) > 0 {+ text := strings.Join(subtitleLines, " ")+ srtBlocks = append(srtBlocks, struct {+ startTime string+ endTime string+ text string+ }{+ startTime: startTime,+ endTime: endTime,+ text: text,+ })+ }+ } else {+ i+++ }+ }++ // Write SRT file+ var srtContent strings.Builder+ for i, block := range srtBlocks {+ srtContent.WriteString(fmt.Sprintf("%d\n", i+1))+ srtContent.WriteString(fmt.Sprintf("%s --> %s\n", block.startTime, block.endTime))+ srtContent.WriteString(block.text + "\n\n")+ }++ return os.WriteFile(outputPath, []byte(srtContent.String()), 0644)+}+func ParseVttTime(timeStr string) (float64, error) {// VTT format: HH:MM:SS.ms or MM:SS.msparts := strings.Split(timeStr, ":")