SRT Translation MCP Server

#!/usr/bin/env node /** * Conversation Translation Workflow * Detects conversation chunks and translates them individually for better context preservation */ import fs from 'fs'; import { parseSRTFile } from './dist/parsers/srt-parser.js'; import { detectConversations } from './dist/chunking/conversation-detector.js'; class ConversationTranslationWorkflow { constructor() { this.inputFile = '/home/omd/Documents/Mix/SRT-MCP/Example.srt'; this.outputFile = '/home/omd/Documents/Mix/SRT-MCP/Example_Conversation_Translated.srt'; this.targetLanguage = 'es'; // Spanish translation this.sourceLanguage = 'en'; // English source } async processWorkflow() { try { console.log(`🚀 Starting Conversation Translation Workflow...`); console.log(`📁 Input: ${this.inputFile}`); console.log(`🌍 Target Language: ${this.targetLanguage}`); // Step 1: Load and parse SRT file console.log(`\n📖 Step 1: Loading and parsing SRT file...`); const content = fs.readFileSync(this.inputFile, 'utf8'); const parseResult = parseSRTFile(content); if (!parseResult.success || !parseResult.file) { throw new Error('Failed to parse SRT file'); } console.log(`✅ Successfully parsed ${parseResult.file.subtitles.length} subtitle entries`); // Step 2: Detect conversation chunks console.log(`\n🔍 Step 2: Detecting conversation chunks...`); const chunks = detectConversations(parseResult.file.subtitles); console.log(`✅ Detected ${chunks.length} conversation chunks`); // Step 3: Process each chunk with translation console.log(`\n🌐 Step 3: Translating conversation chunks...`); const translatedChunks = []; for (let i = 0; i < chunks.length; i++) { const chunk = chunks[i]; console.log(`\n📝 Processing Chunk ${i + 1}/${chunks.length}:`); console.log(` Range: ${chunk.startIndex}-${chunk.endIndex} (${chunk.subtitles.length} entries)`); console.log(` Speaker: ${chunk.context?.speaker || 'Unknown'}`); console.log(` Context: ${chunk.context?.previousContext || 'N/A'}`); // Translate this chunk const translatedChunk = await this.translateChunk(chunk, i + 1); translatedChunks.push(translatedChunk); console.log(` ✅ Translated ${translatedChunk.subtitles.length} entries`); } // Step 4: Reconstruct the full SRT file console.log(`\n🔧 Step 4: Reconstructing translated SRT file...`); const finalSRT = this.reconstructSRT(translatedChunks); // Step 5: Save the result console.log(`\n💾 Step 5: Saving translated file...`); fs.writeFileSync(this.outputFile, finalSRT, 'utf8'); // Step 6: Generate workflow report console.log(`\n📊 Step 6: Generating workflow report...`); this.generateWorkflowReport(chunks, translatedChunks); console.log(`\n🎉 Workflow completed successfully!`); console.log(`📁 Output saved to: ${this.outputFile}`); } catch (error) { console.error('❌ Workflow error:', error); throw error; } } async translateChunk(chunk, chunkNumber) { // Simulate translation process with context-aware adjustments const translatedSubtitles = []; for (let i = 0; i < chunk.subtitles.length; i++) { const subtitle = chunk.subtitles[i]; const nextSubtitle = chunk.subtitles[i + 1]; const prevSubtitle = chunk.subtitles[i - 1]; // Analyze conversation context for this subtitle const context = this.analyzeSubtitleContext(subtitle, nextSubtitle, prevSubtitle, chunk); // Apply conversation-aware translation const translatedSubtitle = await this.translateSubtitle(subtitle, context, chunkNumber, i + 1); translatedSubtitles.push(translatedSubtitle); } return { ...chunk, subtitles: translatedSubtitles, translated: true, translationLanguage: this.targetLanguage }; } analyzeSubtitleContext(subtitle, nextSubtitle, prevSubtitle, chunk) { return { isFirstInChunk: chunk.subtitles.indexOf(subtitle) === 0, isLastInChunk: chunk.subtitles.indexOf(subtitle) === chunk.subtitles.length - 1, isQuestion: subtitle.text.includes('?'), isExclamation: subtitle.text.includes('!'), isNarration: subtitle.text.includes('<i>'), isDialogue: !subtitle.text.includes('<i>'), hasSpeaker: this.extractSpeaker(subtitle.text) !== null, speaker: this.extractSpeaker(subtitle.text), chunkSize: chunk.subtitles.length, conversationId: chunk.context?.conversationId, previousContext: chunk.context?.previousContext, nextContext: chunk.context?.nextContext }; } async translateSubtitle(subtitle, context, chunkNumber, subtitleNumber) { // Simulate translation with conversation context const baseTranslation = this.simulateTranslation(subtitle.text); // Apply conversation-aware adjustments let adjustedTranslation = baseTranslation; // Add conversation context markers if (context.isFirstInChunk) { adjustedTranslation = `[Chunk ${chunkNumber}] ${adjustedTranslation}`; } // Handle questions with conversation context if (context.isQuestion) { adjustedTranslation = `[Q] ${adjustedTranslation}`; } // Handle narration with context if (context.isNarration) { adjustedTranslation = `[Narration] ${adjustedTranslation}`; } // Handle speaker identification if (context.hasSpeaker) { adjustedTranslation = `[Speaker: ${context.speaker}] ${adjustedTranslation}`; } // Apply conversation timing adjustments const timingAdjustments = this.calculateConversationTiming(subtitle, context); return { ...subtitle, text: adjustedTranslation, originalText: subtitle.text, translationContext: { chunkNumber, subtitleNumber, isQuestion: context.isQuestion, isNarration: context.isNarration, hasSpeaker: context.hasSpeaker, speaker: context.speaker, timingAdjustments }, startTime: this.adjustTiming(subtitle.startTime, timingAdjustments.startOffset), endTime: this.adjustTiming(subtitle.endTime, timingAdjustments.endOffset) }; } simulateTranslation(text) { // Simulate translation based on content type if (text.includes('?')) { return `[Traducido] ${text.replace(/\?/g, '?')}`; } else if (text.includes('!')) { return `[Traducido] ${text.replace(/!/g, '!')}`; } else if (text.includes('<i>')) { return `<i>[Traducido] ${text.replace(/<i>/g, '').replace(/<\/i>/g, '')}</i>`; } else { return `[Traducido] ${text}`; } } calculateConversationTiming(subtitle, context) { const adjustments = { startOffset: 0, endOffset: 0 }; // Conversation flow timing if (context.isFirstInChunk) { adjustments.startOffset += 100; // Pause before new conversation } if (context.isLastInChunk) { adjustments.endOffset += 200; // Pause after conversation } // Question timing if (context.isQuestion) { adjustments.endOffset += 300; // Longer pause for questions } // Exclamation timing if (context.isExclamation) { adjustments.endOffset += 250; // Pause for exclamations } // Narration timing if (context.isNarration) { adjustments.startOffset += 150; // Pause before narration adjustments.endOffset += 200; // Pause after narration } // Speaker change timing if (context.hasSpeaker) { adjustments.startOffset += 100; // Pause for speaker identification } return adjustments; } adjustTiming(time, offset) { const totalMs = (time.hours * 3600 + time.minutes * 60 + time.seconds) * 1000 + time.milliseconds + offset; const hours = Math.floor(totalMs / 3600000); const minutes = Math.floor((totalMs % 3600000) / 60000); const seconds = Math.floor((totalMs % 60000) / 1000); const milliseconds = totalMs % 1000; return { hours, minutes, seconds, milliseconds }; } extractSpeaker(text) { const speakerPatterns = [ /^([A-Z][a-z]+):\s*(.+)$/, /^([A-Z][A-Z\s]+):\s*(.+)$/, /^([A-Z][a-z]+)\s*-\s*(.+)$/, /<b>Speaker (\d+):<\/b>/i ]; for (const pattern of speakerPatterns) { const match = text.match(pattern); if (match) { return match[1].trim(); } } return null; } reconstructSRT(translatedChunks) { const allSubtitles = []; translatedChunks.forEach(chunk => { allSubtitles.push(...chunk.subtitles); }); // Sort by index to maintain order allSubtitles.sort((a, b) => a.index - b.index); // Convert to SRT format return allSubtitles.map(subtitle => { const startTime = this.formatTime(subtitle.startTime); const endTime = this.formatTime(subtitle.endTime); return `${subtitle.index}\n${startTime} --> ${endTime}\n${subtitle.text}\n`; }).join('\n'); } formatTime(time) { return `${time.hours.toString().padStart(2, '0')}:${time.minutes.toString().padStart(2, '0')}:${time.seconds.toString().padStart(2, '0')},${time.milliseconds.toString().padStart(3, '0')}`; } generateWorkflowReport(originalChunks, translatedChunks) { const report = { workflow: 'Conversation Translation Workflow', inputFile: this.inputFile, outputFile: this.outputFile, targetLanguage: this.targetLanguage, sourceLanguage: this.sourceLanguage, processing: { totalChunks: originalChunks.length, totalSubtitles: translatedChunks.reduce((sum, chunk) => sum + chunk.subtitles.length, 0), averageChunkSize: (translatedChunks.reduce((sum, chunk) => sum + chunk.subtitles.length, 0) / originalChunks.length).toFixed(1) }, translation: { translatedChunks: translatedChunks.filter(chunk => chunk.translated).length, questionsTranslated: translatedChunks.reduce((sum, chunk) => sum + chunk.subtitles.filter(sub => sub.translationContext?.isQuestion).length, 0), narrationTranslated: translatedChunks.reduce((sum, chunk) => sum + chunk.subtitles.filter(sub => sub.translationContext?.isNarration).length, 0), speakersIdentified: translatedChunks.reduce((sum, chunk) => sum + chunk.subtitles.filter(sub => sub.translationContext?.hasSpeaker).length, 0) }, timing: { totalAdjustments: translatedChunks.reduce((sum, chunk) => sum + chunk.subtitles.reduce((subSum, sub) => subSum + (sub.translationContext?.timingAdjustments?.startOffset || 0) + (sub.translationContext?.timingAdjustments?.endOffset || 0), 0), 0), averageAdjustment: 0 } }; report.timing.averageAdjustment = report.timing.totalAdjustments / report.processing.totalSubtitles; console.log(`\n📊 WORKFLOW REPORT:`); console.log(` Total Chunks: ${report.processing.totalChunks}`); console.log(` Total Subtitles: ${report.processing.totalSubtitles}`); console.log(` Average Chunk Size: ${report.processing.averageChunkSize} entries`); console.log(` Questions Translated: ${report.translation.questionsTranslated}`); console.log(` Narration Translated: ${report.translation.narrationTranslated}`); console.log(` Speakers Identified: ${report.translation.speakersIdentified}`); console.log(` Total Timing Adjustments: ${report.timing.totalAdjustments}ms`); console.log(` Average Adjustment: ${report.timing.averageAdjustment.toFixed(1)}ms`); // Save detailed report fs.writeFileSync('/home/omd/Documents/Mix/SRT-MCP/conversation-translation-report.json', JSON.stringify(report, null, 2)); console.log(`\n📄 Detailed report saved to conversation-translation-report.json`); } } // Main execution async function main() { const workflow = new ConversationTranslationWorkflow(); await workflow.processWorkflow(); } if (import.meta.url === `file://${process.argv[1]}`) { main().catch(console.error); } export default ConversationTranslationWorkflow;