Update file meeting_report_prompt.txt

Implemented the general modular framework.

See merge request proj-wise2526-video2document/video2document!22

package-lock.json

@@ -9,7 +9,9 @@
       "version": "1.0.0",
       "license": "ISC",
       "dependencies": {
+        "@types/axios": "^0.9.36",
         "cli-progress": "^3.12.0",
+        "dotenv": "^17.2.3",
         "electron": "^39.1.1",
         "express": "^5.1.0",
         "ffmpeg-static": "^5.2.0",
@@ -149,6 +151,12 @@
       "dev": true,
       "license": "MIT"
     },
+    "node_modules/@types/axios": {
+      "version": "0.9.36",
+      "resolved": "https://registry.npmjs.org/@types/axios/-/axios-0.9.36.tgz",
+      "integrity": "sha512-NLOpedx9o+rxo/X5ChbdiX6mS1atE4WHmEEIcR9NLenRVa5HoVjAvjafwU3FPTqnZEstpoqCaW7fagqSoTDNeg==",
+      "license": "MIT"
+    },
     "node_modules/@types/cacheable-request": {
       "version": "6.0.3",
       "resolved": "https://registry.npmjs.org/@types/cacheable-request/-/cacheable-request-6.0.3.tgz",
@@ -198,6 +206,7 @@
       "resolved": "https://registry.npmjs.org/@types/node/-/node-24.9.2.tgz",
       "integrity": "sha512-uWN8YqxXxqFMX2RqGOrumsKeti4LlmIMIyV0lgut4jx7KQBcBiW6vkDtIBvHnHIquwNfJhk8v2OtmO8zXWHfPA==",
       "license": "MIT",
+      "peer": true,
       "dependencies": {
         "undici-types": "~7.16.0"
       }
@@ -584,6 +593,18 @@
         "node": ">=0.3.1"
       }
     },
+    "node_modules/dotenv": {
+      "version": "17.2.3",
+      "resolved": "https://registry.npmjs.org/dotenv/-/dotenv-17.2.3.tgz",
+      "integrity": "sha512-JVUnt+DUIzu87TABbhPmNfVdBDt18BLOWjMUFJMSi/Qqg7NTYtabbvSNJGOJ7afbRuv9D/lngizHtP7QyLQ+9w==",
+      "license": "BSD-2-Clause",
+      "engines": {
+        "node": ">=12"
+      },
+      "funding": {
+        "url": "https://dotenvx.com"
+      }
+    },
     "node_modules/dunder-proto": {
       "version": "1.0.1",
       "resolved": "https://registry.npmjs.org/dunder-proto/-/dunder-proto-1.0.1.tgz",
@@ -1824,6 +1845,7 @@
       "integrity": "sha512-jl1vZzPDinLr9eUt3J/t7V6FgNEw9QjvBPdysz9KfQDD41fQrC2Y4vKQdiaUpFT4bXlb1RHhLpp8wtm6M5TgSw==",
       "dev": true,
       "license": "Apache-2.0",
+      "peer": true,
       "bin": {
         "tsc": "bin/tsc",
         "tsserver": "bin/tsserver"

package.json

@@ -1,6 +1,8 @@
 {
   "dependencies": {
+    "@types/axios": "^0.9.36",
     "cli-progress": "^3.12.0",
+    "dotenv": "^17.2.3",
     "electron": "^39.1.1",
     "express": "^5.1.0",
     "ffmpeg-static": "^5.2.0",

prompt/meeting_report_prompt.txt

@@ -0,0 +1,109 @@
+Generate a structured meeting report in MARKDOWN using STRUCTURE and STYLE.
+Output ONLY the final .md document — no meta comments, no explanations.
+
+Follow exactly the STRUCTURE defined below.
+Follow exactly the STYLE rules.
+Use timestamps in HH:MM:SS format.
+If information is missing, use: UNKLAR:<reason>.
+
+==================== STRUCTURE & RULES ====================
+
+{
+  "FORMAT": "markdown",
+
+  "STRUCTURE": {
+    "titlepage": [
+      "title",
+      "date",
+      "start",
+      "end",
+      "duration",
+      "location",
+      "host",
+      "participants"
+    ],
+
+    "toc": "[section](#anchor) — HH:MM:SS",
+
+    "section": {
+      "h2": "<topic> — HH:MM:SS",
+      "summary": "exactly 1 concise sentence",
+      "key_points": "maximum 5 bullet points; quotes optional",
+      "decisions": "list items formatted as: decision text | owner | due date",
+      "actions": "markdown table: id | task | owner | due | status"
+    },
+
+    "exec_summary": "exactly 3 short sentences",
+
+    "consolidated": [
+      "decisions",
+      "actions"
+    ],
+
+    "appendix": "optional"
+  },
+
+  "STYLE": {
+    "tone": "neutral, concise, professional",
+    "ts_format": "HH:MM:SS",
+    "no_meta": true
+  },
+
+  "PROCESS": {
+    "timestamps": "use transcript timestamps if present; otherwise estimate minimal",
+    "speakers": "use names if available; else Speaker X",
+    "long_transcripts": "split → summarize → merge",
+    "unclear": "UNKLAR:<reason>"
+  },
+
+  "PROMPT_SNIPPET": "Generate meeting report in markdown using STRUCTURE and STYLE. Output only the report."
+}
+
+============================================================
+
+Insert all generated content into the following MARKDOWN TEMPLATE:
+
+# {{title}}
+
+**Date:** {{date}}  
+**Start:** {{start}}  
+**End:** {{end}}  
+**Duration:** {{duration}}  
+**Location:** {{location}}  
+**Host:** {{host}}  
+**Participants:** {{participants}}
+
+---
+
+## Table of Contents
+{{toc}}
+
+---
+
+## Executive Summary
+{{exec_summary}}
+
+---
+
+## Sections
+{{sections}}
+
+---
+
+## Consolidated Decisions
+{{consolidated_decisions}}
+
+---
+
+## Consolidated Actions
+{{consolidated_actions}}
+
+---
+
+## Appendix
+{{appendix}}
+
+============================================================
+
+Final Requirement:
+Output ONLY the completed Markdown meeting report.

scripts/commands.txt

@@ -1,5 +0,0 @@
-npx ts-node ./extract.ts /Users/mikehughes/Downloads/Testvideo/Kurzgesagt.mov
-npx ts-node ./transcribe.ts ../storage/audio/Kurzgesagt.wav
-
-npx ts-node ./extract.ts /Users/mikehughes/Downloads/Testvideo/GitLabMeeting.mov
-npx ts-node ./transcribe.ts ../storage/audio/GitLabMeeting.wav

scripts/extract.ts

@@ -1,23 +0,0 @@
-#!/usr/bin/env ts-node
-
-import { extractAudioFromVideo } from "../services/modules/extraction/ffmpegExtractor.ts";
-
-const videoPath = process.argv[2];
-
-if (!videoPath) {
-  console.error("Usage: ts-node extractAudio.ts <videoPath>");
-  process.exit(1);
-}
-
-(async () => {
-  try {
-    console.log(`Extracting audio from: ${videoPath}`);
-
-    await extractAudioFromVideo(videoPath); // Call the extraction function (ffmpegExtractor.ts in services/modules/extraction)
-
-    console.log("Audio extraction completed successfully.");
-  } catch (err) {
-    console.error("Audio extraction failed:", err);
-    process.exit(1);
-  }
-})();

scripts/transcribe.ts

@@ -1,18 +0,0 @@
-import { whisperLocal } from "../services/modules/transcription/local/whisperLocal.ts";
-
-const audioPath = process.argv[2];
-if (!audioPath) {
-  console.error("Please provide an audio file path as argument.");
-  process.exit(1);
-}
-
-const whisper = new whisperLocal();
-
-(async () => {
-  try {
-    const text = await whisper.transcribe(audioPath);
-    console.log(text);
-  } catch (err) {
-    console.error("Transcription failed:", err);
-  }
-})();

services/modules/transcription-local/whisperLocal.ts

@@ -5,7 +5,7 @@ import { fileURLToPath } from "url"; // To handle __dirname in ES modules
 
 const __filename = fileURLToPath(import.meta.url); // Get current file path
 const __dirname = path.dirname(__filename); // Get current directory path
-const transcriptsDir = path.resolve(__dirname, "../../storage/transcriptions");
+const transcriptsDir = path.resolve(__dirname, "../../../storage/transcriptions");
 
 
 export class whisperLocal { // is called by transcribe.ts
@@ -26,7 +26,6 @@ export class whisperLocal { // is called by transcribe.ts
 
   async transcribe(audioPath: string): Promise<string> { //asyncronous function to transcribe audio
     return new Promise((resolve, reject) => {
-        const transcriptsDir = path.resolve(__dirname, "../../../../storage/transcripts"); //storage directory for transcripts
 
         if (!fs.existsSync(transcriptsDir)) { //if transcripts directory does not exist, create it
             fs.mkdirSync(transcriptsDir, { recursive: true });

services/modules/transcription-remote/assembly.ts

@@ -0,0 +1,133 @@
+import 'dotenv/config';
+import axios from 'axios';
+import fs from 'fs';
+import path from 'path';
+import { fileURLToPath } from 'url';
+
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = path.dirname(__filename);
+
+const API_KEY = process.env.ASSEMBLYAI_API_KEY;
+const BASE_URL = 'https://api.assemblyai.com/v2';
+
+/**
+ * Uploads audio file to AssemblyAI
+ */
+async function uploadAudio(audioPath: string): Promise<string> {
+  const audioData = fs.readFileSync(audioPath);
+  
+  const response = await axios.post<{ upload_url: string }>(`${BASE_URL}/upload`, audioData, {
+    headers: {
+      'authorization': API_KEY,
+      'content-type': 'application/octet-stream'
+    }
+  });
+  
+  return response.data.upload_url;
+}
+
+/**
+ * Extract a session id (basename without extension) from a local path or a URL
+ */
+function getSessionId(inputPath: string): string {
+  try {
+    const parsed = new URL(inputPath);
+    const base = path.basename(parsed.pathname);
+    return base.replace(/\.[^.]+$/, '');
+  } catch (err) {
+    // not a URL, treat as local path
+    return path.basename(inputPath, path.extname(inputPath));
+  }
+}
+
+/**
+ * Creates transcription job with speaker diarization
+ */
+async function createTranscript(audioUrl: string): Promise<string> {
+  const response = await axios.post<{ id: string }>(`${BASE_URL}/transcript`, {
+    audio_url: audioUrl,
+    speaker_labels: true,
+    language_detection: true
+  }, {
+    headers: {
+      'authorization': API_KEY,
+      'content-type': 'application/json'
+    }
+  });
+  
+  return response.data.id;
+}
+
+/**
+ * Polls transcript status until completed
+ */
+async function pollTranscript(transcriptId: string): Promise<any> {
+  while (true) {
+    const response = await axios.get<any>(`${BASE_URL}/transcript/${transcriptId}`, {
+      headers: { 'authorization': API_KEY }
+    });
+    
+    const status = response.data.status;
+    
+    if (status === 'completed') {
+      return response.data;
+    } else if (status === 'error') {
+      throw new Error(`Transcription failed: ${response.data.error}`);
+    }
+    
+    // Wait 3 seconds before next poll
+    await new Promise(resolve => setTimeout(resolve, 3000));
+  }
+}
+
+/**
+ * Saves transcript to storage
+ */
+function saveTranscript(transcript: any, sessionId: string): void {
+  const outputDir = path.join(__dirname, '..', '..', '..', 'storage', 'transcripts');
+  
+  if (!fs.existsSync(outputDir)) {
+    fs.mkdirSync(outputDir, { recursive: true });
+  }
+  
+  const outputPath = path.join(outputDir, `${sessionId}.json`);
+  fs.writeFileSync(outputPath, JSON.stringify(transcript, null, 2));
+  
+  console.log(`✅ Transcript saved: ${outputPath}`);
+}
+
+export default {
+  name: "assembly",
+  type: "transcription",
+  displayname: "AssemblyAI",
+  run: async (audioPath: string) => {
+    try {
+      // Determine if audioPath is an external URL or a local file
+      let audioUrl: string;
+      if (/^https?:\/\//i.test(audioPath)) {
+        console.log('🔗 Using external audio URL...');
+        audioUrl = audioPath;
+      } else {
+        console.log('🔄 Uploading local audio...');
+        if (!fs.existsSync(audioPath)) {
+          throw new Error(`Audio file not found: ${audioPath}`);
+        }
+        audioUrl = await uploadAudio(audioPath);
+      }
+      
+      console.log('🔄 Creating transcript job...');
+      const transcriptId = await createTranscript(audioUrl);
+      
+      console.log('⏳ Waiting for transcription...');
+      const transcript = await pollTranscript(transcriptId);
+      
+  const sessionId = getSessionId(audioPath);
+      saveTranscript(transcript, sessionId);
+      
+      return transcript;
+    } catch (error: any) {
+      console.error('❌ Transcription error:', error.message);
+      throw error;
+    }
+  }
+};

services/pipeline/jobs/transcribeLatest.ts

@@ -0,0 +1,52 @@
+// services/pipeline/jobs/transcribeLatest.ts
+import path from 'path';
+import fs from 'fs';
+import assembly from '../../modules/transcription/assembly';
+
+/**
+ * Finds the most recently modified .wav file in storage/audio/
+ */
+function getLatestWav(): string {
+  const audioDir = path.join(process.cwd(), 'storage', 'audio');
+  const files = fs.readdirSync(audioDir).filter(f => f.toLowerCase().endsWith('.wav'));
+  if (files.length === 0) throw new Error('⚠️ No .wav file found in storage/audio');
+
+  const newest = files
+    .map(f => ({ f, t: fs.statSync(path.join(audioDir, f)).mtimeMs }))
+    .sort((a, b) => b.t - a.t)[0].f;
+
+  return path.join(audioDir, newest);
+}
+
+/**
+ * Full transcription pipeline according to the defined workflow:
+ * 1. Audio Upload → AssemblyAI
+ * 2. Job Creation (transcript_id)
+ * 3. Polling Status (queued → processing → completed)
+ * 4. Download Transcript JSON
+ * 5. Storage: /transcripts/{session_id}.json
+ */
+async function main() {
+  const audioPath = getLatestWav();
+
+  console.log('1️⃣ Audio Upload → AssemblyAI');
+  console.log('   Source:', audioPath);
+
+  console.log('2️⃣ Job Creation (transcript_id)');
+  console.log('3️⃣ Polling Status (queued → processing → completed)');
+  console.log('4️⃣ Download Transcript JSON');
+  console.log('5️⃣ Storage: /transcripts/{session_id}.json');
+
+  // Execute the transcription process via the AssemblyAI module
+  const result = await assembly.run(audioPath);
+
+  console.log('✅ Transcription completed successfully');
+  console.log('🆔 Transcript ID:', result.id);
+  console.log('📁 Transcript file saved under: storage/transcripts/');
+}
+
+// Entry point
+main().catch((err) => {
+  console.error('❌ Transcription pipeline failed:', err.message || err);
+  process.exit(1);
+});

test/integration/assembly.test.js

@@ -0,0 +1,14 @@
+import 'dotenv/config';
+import assemblyModule from '../../services/modules/transcription-remote/assembly.ts';
+
+// Test: URL passed as argument OR local file ./storage/audio/test.wav
+const audioPath = process.argv[2] || './storage/audio/test.wav';
+
+assemblyModule.run(audioPath)
+  .then(result => {
+    console.log('✅ Success!');
+    console.log('Transcript ID:', result.id);
+  })
+  .catch(error => {
+    console.error('❌ Error:', error?.message || error);
+  });