#!/usr/bin/env node
/**
 * Multi-Perspective Audio Analyzer
 * Spawns multiple Gemini instances to analyze audio from different angles
 * 
 * Usage: node multi-analyze.js <audio-file.wav>
 */

const fs = require('fs');
const path = require('path');

const GEMINI_KEY = 'AIzaSyClMlVU3Z1jh1UBxTRn25yesH8RU1q_umY';
const GEMINI_URL = `https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=${GEMINI_KEY}`;

// Different analysis perspectives
const PERSPECTIVES = [
  {
    name: 'Transcription',
    prompt: 'Listen to this audio and transcribe EXACTLY what is said. Return ONLY the spoken text/numbers, nothing else. Be precise.'
  },
  {
    name: 'Captcha-Focused',
    prompt: 'This is an audio captcha. Listen carefully and identify the letters, numbers, or words being spoken. Ignore any background noise. Return ONLY the captcha answer.'
  },
  {
    name: 'Phonetic',
    prompt: 'Listen to this audio and write out what you hear phonetically. If it sounds like letters being spelled out, list each letter. If numbers, list each digit.'
  },
  {
    name: 'Noise-Filtered',
    prompt: 'This audio may have distortion or background noise (common in captchas). Focus on the human voice and transcribe what is being said. Ignore beeps, static, or music.'
  }
];

async function analyzeWithPerspective(audioBase64, perspective) {
  const startTime = Date.now();
  
  try {
    const response = await fetch(GEMINI_URL, {
      method: 'POST',
      headers: { 'Content-Type': 'application/json' },
      body: JSON.stringify({
        contents: [{
          parts: [
            { text: perspective.prompt },
            { inline_data: { mime_type: 'audio/wav', data: audioBase64 } }
          ]
        }]
      })
    });
    
    const data = await response.json();
    const result = data.candidates?.[0]?.content?.parts?.[0]?.text || 'NO RESPONSE';
    const elapsed = Date.now() - startTime;
    
    return {
      perspective: perspective.name,
      result: result.trim(),
      elapsed: `${elapsed}ms`
    };
  } catch (err) {
    return {
      perspective: perspective.name,
      result: `ERROR: ${err.message}`,
      elapsed: 'N/A'
    };
  }
}

async function main() {
  const audioFile = process.argv[2];
  
  if (!audioFile) {
    console.log('Usage: node multi-analyze.js <audio-file.wav>');
    console.log('');
    console.log('Or pipe audio: cat audio.wav | node multi-analyze.js -');
    process.exit(1);
  }
  
  let audioBuffer;
  if (audioFile === '-') {
    // Read from stdin
    const chunks = [];
    for await (const chunk of process.stdin) {
      chunks.push(chunk);
    }
    audioBuffer = Buffer.concat(chunks);
  } else {
    if (!fs.existsSync(audioFile)) {
      console.error(`File not found: ${audioFile}`);
      process.exit(1);
    }
    audioBuffer = fs.readFileSync(audioFile);
  }
  
  const audioBase64 = audioBuffer.toString('base64');
  
  console.log('🎧 Spawning', PERSPECTIVES.length, 'Gemini analyzers...\n');
  
  // Run all perspectives in parallel
  const results = await Promise.all(
    PERSPECTIVES.map(p => analyzeWithPerspective(audioBase64, p))
  );
  
  // Display results
  console.log('=' .repeat(60));
  for (const r of results) {
    console.log(`\n📊 ${r.perspective} (${r.elapsed}):`);
    console.log(`   "${r.result}"`);
  }
  console.log('\n' + '='.repeat(60));
  
  // Find consensus
  const answers = results.map(r => r.result.toLowerCase().replace(/[^a-z0-9]/g, ''));
  const counts = {};
  for (const a of answers) {
    counts[a] = (counts[a] || 0) + 1;
  }
  
  const sorted = Object.entries(counts).sort((a, b) => b[1] - a[1]);
  if (sorted.length > 0 && sorted[0][1] > 1) {
    console.log(`\n🎯 CONSENSUS (${sorted[0][1]}/${results.length} agree): "${sorted[0][0]}"`);
  } else {
    console.log('\n⚠️  No consensus - results vary. Check individual outputs.');
  }
}

main().catch(console.error);