boc/quixzoom-capture-pipeline/pilot/ai-result-logger.js

/**
 * QUIXZOOM AI Result Logger
 * 
 * Sparar ALLA AI-resultat för att möjliggöra:
 * - Återträning utan om-annotering
 * - Analys av träffsäkerhet
 * - Kostnadsuppföljning
 * - A/B-testning av modeller
 */

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

class AIResultLogger {
  constructor(baseDir = '/data/ai-results') {
    this.baseDir = baseDir;
    this.ensureDirectory();
    
    // Schema för varje observation
    this.schema = {
      observationId: 'string',      // UUID
      timestamp: 'string',          // ISO 8601
      imagePath: 'string',          // Sökväg till originalbild
      imageHash: 'string',          // SHA-256 hash
      
      // Modell-info
      model: 'string',              // cloud_vision, yolo, grounding_dino, custom
      modelVersion: 'string',       // t.ex. "v1.2.3"
      prompt: 'string',             // Om relevant (text-prompt)
      
      // Detektioner
      detections: [{
        label: 'string',            // AI-föreslagen etikett
        quixzoomLabel: 'string',    // Mappad till vårt vokabulär
        category: 'string',         // lighting, road, utility, etc.
        confidence: 'number',       // 0.0 - 1.0
        bbox: {                     // Bounding box
          x: 'number',
          y: 'number',
          width: 'number',
          height: 'number'
        }
      }],
      
      // Manuell korrigering
      manualCorrection: {
        correctedBy: 'string',      // Zoomer-ID eller admin
        correctedAt: 'string',      // Timestamp
        changes: [{                 // Vad som ändrades
          original: 'string',
          corrected: 'string',
          reason: 'string'
        }]
      },
      
      // Gold label (slutgiltig sanning)
      goldLabel: {
        label: 'string',
        category: 'string',
        bbox: 'object',
        verifiedBy: 'string',
        verifiedAt: 'string'
      },
      
      // Metadata
      region: 'string',             // bangkok, torrevieja, stockholm
      routeId: 'string',            // Gold Route ID
      missionId: 'string',          // Mission ID
      zoomerId: 'string',           // Zoomer ID (om tillämpligt)
      
      // Kostnad
      cost: 'number',               // USD
      latency: 'number',            // ms
      
      // Kvalitet
      quality: {
        aiAccuracy: 'number',       // Jämfört med gold label
        needsReview: 'boolean',     // Kräver manuell granskning
        reviewStatus: 'string'      // pending, approved, rejected
      }
    };
  }

  ensureDirectory() {
    const dirs = [
      this.baseDir,
      path.join(this.baseDir, 'raw'),
      path.join(this.baseDir, 'corrected'),
      path.join(this.baseDir, 'gold'),
      path.join(this.baseDir, 'models')
    ];
    
    for (const dir of dirs) {
      fs.mkdirSync(dir, { recursive: true });
    }
  }

  /**
   * ============================================================
   * HUVUDMETOD: Logga AI-resultat
   * ============================================================
   */

  async logResult(observation) {
    const observationId = observation.observationId || this.generateId();
    const timestamp = new Date().toISOString();
    
    const record = {
      ...observation,
      observationId,
      timestamp,
      loggedAt: timestamp
    };
    
    // Spara rådata
    const rawPath = path.join(this.baseDir, 'raw', `${observationId}.json`);
    fs.writeFileSync(rawPath, JSON.stringify(record, null, 2));
    
    // Spara i modell-specifik logg
    const modelLogPath = path.join(
      this.baseDir, 
      'models', 
      `${record.model || 'unknown'}.jsonl`
    );
    fs.appendFileSync(modelLogPath, JSON.stringify(record) + '\n');
    
    // Uppdatera index
    this.updateIndex(record);
    
    console.log(`[AI Logger] Sparade: ${observationId}`);
    return observationId;
  }

  /**
   * ============================================================
   * LOGGA MANUELL KORRIGERING
   * ============================================================
   */

  async logCorrection(observationId, correction) {
    const rawPath = path.join(this.baseDir, 'raw', `${observationId}.json`);
    
    if (!fs.existsSync(rawPath)) {
      throw new Error(`Observation ${observationId} hittades inte`);
    }
    
    const record = JSON.parse(fs.readFileSync(rawPath, 'utf8'));
    
    record.manualCorrection = {
      ...correction,
      correctedAt: new Date().toISOString()
    };
    
    // Spara korrigerad version
    const correctedPath = path.join(this.baseDir, 'corrected', `${observationId}.json`);
    fs.writeFileSync(correctedPath, JSON.stringify(record, null, 2));
    
    console.log(`[AI Logger] Korrigerade: ${observationId}`);
    return record;
  }

  /**
   * ============================================================
   * LOGGA GOLD LABEL
   * ============================================================
   */

  async logGoldLabel(observationId, goldLabel) {
    const correctedPath = path.join(this.baseDir, 'corrected', `${observationId}.json`);
    const rawPath = path.join(this.baseDir, 'raw', `${observationId}.json`);
    
    const sourcePath = fs.existsSync(correctedPath) ? correctedPath : rawPath;
    const record = JSON.parse(fs.readFileSync(sourcePath, 'utf8'));
    
    record.goldLabel = {
      ...goldLabel,
      verifiedAt: new Date().toISOString()
    };
    
    // Beräkna accuracy
    if (record.detections && record.detections.length > 0) {
      const aiLabel = record.detections[0].quixzoomLabel;
      record.quality = {
        aiAccuracy: aiLabel === goldLabel.label ? 1.0 : 0.0,
        needsReview: false,
        reviewStatus: 'approved'
      };
    }
    
    // Spara gold version
    const goldPath = path.join(this.baseDir, 'gold', `${observationId}.json`);
    fs.writeFileSync(goldPath, JSON.stringify(record, null, 2));
    
    console.log(`[AI Logger] Gold label: ${observationId} (${record.quality?.aiAccuracy === 1.0 ? '✅ Korrekt' : '❌ Fel'})`);
    return record;
  }

  /**
   * ============================================================
   * INDEX & STATISTIK
   * ============================================================
   */

  updateIndex(record) {
    const indexPath = path.join(this.baseDir, 'index.json');
    let index = { observations: [], stats: {} };
    
    if (fs.existsSync(indexPath)) {
      index = JSON.parse(fs.readFileSync(indexPath, 'utf8'));
    }
    
    index.observations.push({
      id: record.observationId,
      timestamp: record.timestamp,
      model: record.model,
      region: record.region,
      hasCorrection: !!record.manualCorrection,
      hasGoldLabel: !!record.goldLabel
    });
    
    // Uppdatera stats
    index.stats = this.calculateStats(index.observations);
    
    fs.writeFileSync(indexPath, JSON.stringify(index, null, 2));
  }

  calculateStats(observations) {
    const stats = {
      total: observations.length,
      byModel: {},
      byRegion: {},
      corrected: 0,
      goldLabeled: 0
    };
    
    for (const obs of observations) {
      stats.byModel[obs.model] = (stats.byModel[obs.model] || 0) + 1;
      stats.byRegion[obs.region] = (stats.byRegion[obs.region] || 0) + 1;
      if (obs.hasCorrection) stats.corrected++;
      if (obs.hasGoldLabel) stats.goldLabeled++;
    }
    
    return stats;
  }

  /**
   * ============================================================
   * EXPORT FÖR TRÄNING
   * ============================================================
   */

  exportForTraining(outputDir, options = {}) {
    fs.mkdirSync(outputDir, { recursive: true });
    
    // Hämta alla gold labels
    const goldDir = path.join(this.baseDir, 'gold');
    const goldFiles = fs.readdirSync(goldDir).filter(f => f.endsWith('.json'));
    
    const trainingData = [];
    
    for (const file of goldFiles) {
      const record = JSON.parse(fs.readFileSync(path.join(goldDir, file), 'utf8'));
      
      if (options.model && record.model !== options.model) continue;
      if (options.region && record.region !== options.region) continue;
      
      trainingData.push({
        imagePath: record.imagePath,
        imageHash: record.imageHash,
        label: record.goldLabel.label,
        category: record.goldLabel.category,
        bbox: record.goldLabel.bbox,
        originalDetections: record.detections
      });
    }
    
    // Spara
    const outputPath = path.join(outputDir, `training-data-${Date.now()}.json`);
    fs.writeFileSync(outputPath, JSON.stringify(trainingData, null, 2));
    
    console.log(`[AI Logger] Exporterade ${trainingData.length} gold labels till: ${outputPath}`);
    return trainingData;
  }

  /**
   * ============================================================
   * STATISTIK
   * ============================================================
   */

  getStats() {
    const indexPath = path.join(this.baseDir, 'index.json');
    
    if (!fs.existsSync(indexPath)) {
      return { total: 0 };
    }
    
    return JSON.parse(fs.readFileSync(indexPath, 'utf8')).stats;
  }

  generateId() {
    return `obs-${Date.now()}-${Math.random().toString(36).substr(2, 9)}`;
  }
}

module.exports = AIResultLogger;

// Demo
if (require.main === module) {
  const logger = new AIResultLogger('/tmp/ai-results-demo');
  
  console.log('╔════════════════════════════════════════════════════════════╗');
  console.log('║     AI RESULT LOGGER — DEMO                               ║');
  console.log('╚════════════════════════════════════════════════════════════╝\n');
  
  console.log('Schema för varje observation:');
  console.log('  - observationId (UUID)');
  console.log('  - imagePath + imageHash');
  console.log('  - model + modelVersion');
  console.log('  - detections (label, confidence, bbox)');
  console.log('  - manualCorrection');
  console.log('  - goldLabel');
  console.log('  - region, routeId, missionId');
  console.log('  - cost, latency');
  console.log('  - quality (accuracy, needsReview)');
  console.log('');
  console.log('Användning:');
  console.log('  const logger = new AIResultLogger("/data/ai-results");');
  console.log('  await logger.logResult(observation);');
  console.log('  await logger.logCorrection(id, correction);');
  console.log('  await logger.logGoldLabel(id, goldLabel);');
  console.log('  logger.exportForTraining("/data/training/");');
  console.log('');
  console.log('✅ AI Result Logger redo!');
}