apps/backend/src/services/ErrorKnowledgeBase.ts

/**
 * ╔═══════════════════════════════════════════════════════════════════════════╗
 * ║                      ERROR KNOWLEDGE BASE                                  ║
 * ║═══════════════════════════════════════════════════════════════════════════║
 * ║  Samler fejlmønstre fra eksterne kilder (GitHub, HuggingFace, CVE, etc.)  ║
 * ║  Bruges af SelfHealingAdapter til at lære og forudsige fejl               ║
 * ╚═══════════════════════════════════════════════════════════════════════════╝
 */

import crypto from 'crypto';
import { logger } from '../utils/logger.js';

// Lazy import for Neo4j to avoid circular dependencies
let neo4jServiceInstance: any = null;
async function getNeo4jService() {
  if (!neo4jServiceInstance) {
    try {
      const { neo4jService } = await import('../database/Neo4jService.js');
      neo4jServiceInstance = neo4jService;
    } catch (e) {
      // Neo4j not available
    }
  }
  return neo4jServiceInstance;
}

const log = logger.child({ module: 'ErrorKnowledgeBase' });

// ═══════════════════════════════════════════════════════════════════════════
// TYPES
// ═══════════════════════════════════════════════════════════════════════════

export interface ErrorPattern {
  id: string;                    // Unique hash of error signature
  source: ErrorSource;
  category: ErrorCategory;
  signature: string;             // Error message pattern/regex
  description: string;
  severity: 'low' | 'medium' | 'high' | 'critical';
  solutions: Solution[];
  tags: string[];
  cveId?: string;                // CVE identifier if applicable
  cweId?: string;                // CWE identifier if applicable
  language?: string;             // Programming language
  framework?: string;            // Framework (e.g., Express, React)
  occurrences: number;           // Times seen
  lastSeen: Date;
  createdAt: Date;
  metadata?: Record<string, any>;
}

export interface Solution {
  description: string;
  code?: string;
  confidence: number;            // 0-1
  source: string;
  verified: boolean;
  // Feedback tracking
  successCount?: number;         // Times this solution worked
  failureCount?: number;         // Times this solution failed
  lastUsed?: Date;
}

export type ErrorSource =
  | 'github-defects4j'
  | 'github-bugsjs'
  | 'huggingface-cve'
  | 'huggingface-defect-detection'
  | 'huggingface-hdfs-logs'
  | 'microsoft-office-api'
  | 'microsoft-graph-api'
  | 'internal-logs'
  | 'user-reported'
  | 'gaia-aiops'
  | 'nodejs-system'
  | 'http-standards'
  | 'postgresql-official'
  | 'typescript-compiler'
  | 'curated-knowledge';

export type ErrorCategory =
  | 'runtime'
  | 'syntax'
  | 'type'
  | 'security'
  | 'performance'
  | 'network'
  | 'database'
  | 'api'
  | 'office'
  | 'authentication'
  | 'memory'
  | 'concurrency'
  | 'configuration'
  | 'dependency';

// ═══════════════════════════════════════════════════════════════════════════
// ERROR KNOWLEDGE BASE CLASS
// ═══════════════════════════════════════════════════════════════════════════

export class ErrorKnowledgeBase {
  private static instance: ErrorKnowledgeBase;
  private patterns: Map<string, ErrorPattern> = new Map();
  private signatureIndex: Map<string, Set<string>> = new Map(); // For fast lookup
  private sourceStats: Map<ErrorSource, number> = new Map();

  private constructor() {
    this.initializeBuiltInPatterns();
  }

  public static getInstance(): ErrorKnowledgeBase {
    if (!ErrorKnowledgeBase.instance) {
      ErrorKnowledgeBase.instance = new ErrorKnowledgeBase();
    }
    return ErrorKnowledgeBase.instance;
  }

  /**
   * Generate unique ID from error signature to prevent duplicates
   */
  private generateId(signature: string, source: ErrorSource): string {
    const normalized = signature.toLowerCase().trim().replace(/\s+/g, ' ');
    return crypto.createHash('sha256')
      .update(`${normalized}:${source}`)
      .digest('hex')
      .substring(0, 16);
  }

  /**
   * Normalize error signature for comparison
   */
  private normalizeSignature(sig: string): string {
    return sig
      .toLowerCase()
      .replace(/0x[0-9a-f]+/gi, '0xHEX')      // Hex addresses
      .replace(/\d+/g, 'N')                    // Numbers
      .replace(/['"][^'"]*['"]/g, '"STR"')    // String literals
      .replace(/\s+/g, ' ')                    // Whitespace
      .trim();
  }

  /**
   * Check if pattern already exists (dedupe)
   */
  public isDuplicate(signature: string, source: ErrorSource): boolean {
    const id = this.generateId(signature, source);
    if (this.patterns.has(id)) {
      return true;
    }

    // Also check normalized signature across all sources
    const normalized = this.normalizeSignature(signature);
    const existing = this.signatureIndex.get(normalized);
    return existing !== undefined && existing.size > 0;
  }

  /**
   * Ingest single error pattern (with dedupe)
   */
  public ingest(pattern: Omit<ErrorPattern, 'id' | 'occurrences' | 'lastSeen' | 'createdAt'>): boolean {
    const id = this.generateId(pattern.signature, pattern.source);

    // Check for exact duplicate
    if (this.patterns.has(id)) {
      // Update existing
      const existing = this.patterns.get(id)!;
      existing.occurrences++;
      existing.lastSeen = new Date();
      // Merge solutions
      for (const sol of pattern.solutions) {
        if (!existing.solutions.some(s => s.description === sol.description)) {
          existing.solutions.push(sol);
        }
      }
      return false; // Not new
    }

    // Check for similar pattern (different source)
    const normalized = this.normalizeSignature(pattern.signature);
    if (!this.signatureIndex.has(normalized)) {
      this.signatureIndex.set(normalized, new Set());
    }
    this.signatureIndex.get(normalized)!.add(id);

    // Create new pattern
    const newPattern: ErrorPattern = {
      ...pattern,
      id,
      occurrences: 1,
      lastSeen: new Date(),
      createdAt: new Date()
    };

    this.patterns.set(id, newPattern);

    // Update source stats
    const count = this.sourceStats.get(pattern.source) || 0;
    this.sourceStats.set(pattern.source, count + 1);

    log.info(`Ingested error pattern: ${pattern.category}/${pattern.signature.substring(0, 50)}...`);
    return true; // New pattern
  }

  /**
   * Batch ingest with progress tracking
   */
  public async batchIngest(
    patterns: Omit<ErrorPattern, 'id' | 'occurrences' | 'lastSeen' | 'createdAt'>[],
    source: ErrorSource
  ): Promise<{ total: number; new: number; duplicates: number }> {
    let newCount = 0;
    let dupeCount = 0;

    for (const pattern of patterns) {
      const isNew = this.ingest({ ...pattern, source });
      if (isNew) newCount++;
      else dupeCount++;
    }

    log.info(`Batch ingest from ${source}: ${newCount} new, ${dupeCount} duplicates`);
    return { total: patterns.length, new: newCount, duplicates: dupeCount };
  }

  /**
   * Find matching patterns for an error
   */
  public findMatches(errorMessage: string, limit = 5): ErrorPattern[] {
    const normalized = this.normalizeSignature(errorMessage);
    const results: { pattern: ErrorPattern; score: number }[] = [];

    for (const pattern of this.patterns.values()) {
      const patternNorm = this.normalizeSignature(pattern.signature);
      const score = this.similarityScore(normalized, patternNorm);
      if (score > 0.3) {  // Lowered threshold for better fuzzy matching
        results.push({ pattern, score });
      }
    }

    return results
      .sort((a, b) => b.score - a.score)
      .slice(0, limit)
      .map(r => r.pattern);
  }

  /**
   * Enhanced similarity score combining multiple strategies
   */
  private similarityScore(query: string, signature: string): number {
    // Strategy 1: Substring containment (highest priority)
    if (signature.includes(query) || query.includes(signature)) {
      return 1.0;
    }

    // Strategy 2: Key error code matching (ECONNREFUSED, SQLSTATE, HTTP 4xx/5xx)
    const errorCodes = query.match(/\b(E[A-Z]+|SQLSTATE\s*\d+|HTTP\s*\d{3}|TS\d{4})\b/gi) || [];
    for (const code of errorCodes) {
      if (signature.toUpperCase().includes(code.toUpperCase().replace(/\s+/g, ' '))) {
        return 0.95;
      }
    }

    // Strategy 3: Jaccard similarity on words
    const setA = new Set(query.split(/\s+/).filter(w => w.length > 2));
    const setB = new Set(signature.split(/\s+/).filter(w => w.length > 2));
    const intersection = new Set([...setA].filter(x => setB.has(x)));
    const union = new Set([...setA, ...setB]);
    const jaccard = union.size > 0 ? intersection.size / union.size : 0;

    // Strategy 4: Coverage - what % of query words appear in signature
    const coverage = setA.size > 0 ? intersection.size / setA.size : 0;

    // Combined score: weight coverage more heavily
    return Math.max(jaccard, coverage * 0.8);
  }

  /**
   * Get suggested solutions for an error
   */
  public getSolutions(errorMessage: string): Solution[] {
    const matches = this.findMatches(errorMessage, 3);
    const solutions: Solution[] = [];

    for (const match of matches) {
      solutions.push(...match.solutions);
    }

    return solutions
      .sort((a, b) => b.confidence - a.confidence)
      .slice(0, 5);
  }

  /**
   * Get statistics
   */
  public getStats() {
    const categoryCount: Record<string, number> = {};
    const severityCount: Record<string, number> = {};

    for (const pattern of this.patterns.values()) {
      categoryCount[pattern.category] = (categoryCount[pattern.category] || 0) + 1;
      severityCount[pattern.severity] = (severityCount[pattern.severity] || 0) + 1;
    }

    return {
      totalPatterns: this.patterns.size,
      bySource: Object.fromEntries(this.sourceStats),
      byCategory: categoryCount,
      bySeverity: severityCount
    };
  }

  /**
   * Export all patterns
   */
  public exportPatterns(): ErrorPattern[] {
    return Array.from(this.patterns.values());
  }

  /**
   * Initialize with common built-in patterns
   */
  private initializeBuiltInPatterns() {
    const builtInPatterns: Omit<ErrorPattern, 'id' | 'occurrences' | 'lastSeen' | 'createdAt'>[] = [
      // Network errors
      {
        source: 'internal-logs',
        category: 'network',
        signature: 'ECONNREFUSED',
        description: 'Connection refused - target service not running or firewall blocking',
        severity: 'high',
        solutions: [
          { description: 'Check if target service is running', confidence: 0.9, source: 'built-in', verified: true },
          { description: 'Verify firewall rules allow connection', confidence: 0.7, source: 'built-in', verified: true },
          { description: 'Check if port number is correct', confidence: 0.8, source: 'built-in', verified: true }
        ],
        tags: ['network', 'connection', 'tcp']
      },
      {
        source: 'internal-logs',
        category: 'network',
        signature: 'ETIMEDOUT',
        description: 'Connection timed out - network latency or service overload',
        severity: 'medium',
        solutions: [
          { description: 'Increase timeout value', confidence: 0.8, source: 'built-in', verified: true },
          { description: 'Check network connectivity', confidence: 0.7, source: 'built-in', verified: true },
          { description: 'Implement retry with exponential backoff', confidence: 0.9, source: 'built-in', verified: true }
        ],
        tags: ['network', 'timeout', 'latency']
      },
      // Database errors
      {
        source: 'internal-logs',
        category: 'database',
        signature: 'Neo4jError: ServiceUnavailable',
        description: 'Neo4j database not reachable',
        severity: 'critical',
        solutions: [
          { description: 'Check Neo4j container/service status', confidence: 0.9, source: 'built-in', verified: true },
          { description: 'Verify NEO4J_URI environment variable', confidence: 0.8, source: 'built-in', verified: true },
          { description: 'Check AuraDB connection limits', confidence: 0.6, source: 'built-in', verified: true }
        ],
        tags: ['neo4j', 'database', 'graph']
      },
      {
        source: 'internal-logs',
        category: 'database',
        signature: 'PrismaClientKnownRequestError',
        description: 'Prisma database query error',
        severity: 'high',
        solutions: [
          { description: 'Check if database schema is migrated', confidence: 0.8, source: 'built-in', verified: true },
          { description: 'Verify DATABASE_URL is correct', confidence: 0.7, source: 'built-in', verified: true },
          { description: 'Run npx prisma db push', confidence: 0.9, source: 'built-in', verified: true }
        ],
        tags: ['prisma', 'postgresql', 'database']
      },
      // TypeScript/JavaScript errors
      {
        source: 'internal-logs',
        category: 'type',
        signature: 'TypeError: Cannot read properties of undefined',
        description: 'Accessing property on undefined value',
        severity: 'medium',
        solutions: [
          { description: 'Add null/undefined check before accessing', confidence: 0.9, source: 'built-in', verified: true },
          { description: 'Use optional chaining (?.) operator', confidence: 0.95, source: 'built-in', verified: true },
          { description: 'Provide default value with ?? operator', confidence: 0.8, source: 'built-in', verified: true }
        ],
        tags: ['typescript', 'javascript', 'null-safety']
      },
      {
        source: 'internal-logs',
        category: 'syntax',
        signature: 'SyntaxError: Unexpected token',
        description: 'Invalid JavaScript/JSON syntax',
        severity: 'high',
        solutions: [
          { description: 'Check for missing brackets or quotes', confidence: 0.8, source: 'built-in', verified: true },
          { description: 'Validate JSON with JSON.parse in try-catch', confidence: 0.7, source: 'built-in', verified: true },
          { description: 'Check for trailing commas in JSON', confidence: 0.6, source: 'built-in', verified: true }
        ],
        tags: ['syntax', 'json', 'parsing']
      },
      // Office API errors
      {
        source: 'microsoft-office-api',
        category: 'office',
        signature: 'InvalidReference',
        description: 'Office API reference is not valid for operation',
        severity: 'medium',
        solutions: [
          { description: 'Ensure object exists before operation', confidence: 0.9, source: 'microsoft-docs', verified: true },
          { description: 'Check if document is still open', confidence: 0.7, source: 'microsoft-docs', verified: true }
        ],
        tags: ['office', 'excel', 'word', 'api'],
        cweId: 'CWE-476'
      },
      {
        source: 'microsoft-graph-api',
        category: 'api',
        signature: 'Error 429: Too Many Requests',
        description: 'Microsoft Graph API rate limit exceeded',
        severity: 'medium',
        solutions: [
          { description: 'Implement exponential backoff retry', confidence: 0.95, source: 'microsoft-docs', verified: true },
          { description: 'Check Retry-After header for wait time', confidence: 0.9, source: 'microsoft-docs', verified: true },
          { description: 'Batch multiple requests together', confidence: 0.8, source: 'microsoft-docs', verified: true }
        ],
        tags: ['graph-api', 'rate-limit', 'throttling']
      },
      // Security patterns
      {
        source: 'huggingface-cve',
        category: 'security',
        signature: 'SQL injection detected',
        description: 'Potential SQL injection vulnerability',
        severity: 'critical',
        solutions: [
          { description: 'Use parameterized queries', confidence: 0.99, source: 'OWASP', verified: true },
          { description: 'Sanitize user input', confidence: 0.9, source: 'OWASP', verified: true },
          { description: 'Use ORM instead of raw SQL', confidence: 0.85, source: 'best-practice', verified: true }
        ],
        tags: ['security', 'sql', 'injection'],
        cweId: 'CWE-89'
      },
      {
        source: 'huggingface-cve',
        category: 'security',
        signature: 'XSS vulnerability',
        description: 'Cross-site scripting vulnerability detected',
        severity: 'critical',
        solutions: [
          { description: 'Escape HTML output', confidence: 0.95, source: 'OWASP', verified: true },
          { description: 'Use Content Security Policy headers', confidence: 0.9, source: 'OWASP', verified: true },
          { description: 'Validate and sanitize input', confidence: 0.85, source: 'best-practice', verified: true }
        ],
        tags: ['security', 'xss', 'injection'],
        cweId: 'CWE-79'
      },
      // Memory errors
      {
        source: 'internal-logs',
        category: 'memory',
        signature: 'FATAL ERROR: CALL_AND_RETRY_LAST Allocation failed',
        description: 'Node.js heap out of memory',
        severity: 'critical',
        solutions: [
          { description: 'Increase Node.js heap size: --max-old-space-size=4096', confidence: 0.9, source: 'built-in', verified: true },
          { description: 'Check for memory leaks with --inspect', confidence: 0.8, source: 'built-in', verified: true },
          { description: 'Implement pagination for large data sets', confidence: 0.85, source: 'best-practice', verified: true }
        ],
        tags: ['memory', 'heap', 'oom']
      },
      // Redis errors
      {
        source: 'internal-logs',
        category: 'database',
        signature: 'Redis connection error',
        description: 'Cannot connect to Redis server',
        severity: 'high',
        solutions: [
          { description: 'Check if Redis container is running', confidence: 0.9, source: 'built-in', verified: true },
          { description: 'Verify REDIS_URL environment variable', confidence: 0.8, source: 'built-in', verified: true },
          { description: 'Check Redis maxmemory settings', confidence: 0.6, source: 'built-in', verified: true }
        ],
        tags: ['redis', 'cache', 'connection']
      }
    ];

    for (const pattern of builtInPatterns) {
      this.ingest(pattern);
    }

    log.info(`Initialized ErrorKnowledgeBase with ${this.patterns.size} built-in patterns`);
  }

  // ═══════════════════════════════════════════════════════════════════════════
  // NEO4J PERSISTENCE - Feedback Loop
  // ═══════════════════════════════════════════════════════════════════════════

  /**
   * Persist a single pattern to Neo4j
   */
  public async persistToNeo4j(pattern: ErrorPattern): Promise<boolean> {
    try {
      const neo4j = await getNeo4jService();
      if (!neo4j) {
        log.warn('Neo4j not available, skipping persistence');
        return false;
      }

      const query = `
        MERGE (ep:ErrorPattern {id: $id})
        SET ep.source = $source,
            ep.category = $category,
            ep.signature = $signature,
            ep.description = $description,
            ep.severity = $severity,
            ep.tags = $tags,
            ep.cveId = $cveId,
            ep.cweId = $cweId,
            ep.language = $language,
            ep.framework = $framework,
            ep.occurrences = $occurrences,
            ep.lastSeen = datetime($lastSeen),
            ep.createdAt = datetime($createdAt),
            ep.solutions = $solutions
        RETURN ep
      `;

      await neo4j.runQuery(query, {
        id: pattern.id,
        source: pattern.source,
        category: pattern.category,
        signature: pattern.signature,
        description: pattern.description,
        severity: pattern.severity,
        tags: pattern.tags,
        cveId: pattern.cveId || null,
        cweId: pattern.cweId || null,
        language: pattern.language || null,
        framework: pattern.framework || null,
        occurrences: pattern.occurrences,
        lastSeen: pattern.lastSeen.toISOString(),
        createdAt: pattern.createdAt.toISOString(),
        solutions: JSON.stringify(pattern.solutions)
      });

      log.debug(`Persisted pattern ${pattern.id} to Neo4j`);
      return true;
    } catch (error) {
      log.error('Failed to persist pattern to Neo4j:', error);
      return false;
    }
  }

  /**
   * Load all patterns from Neo4j
   */
  public async loadFromNeo4j(): Promise<number> {
    try {
      const neo4j = await getNeo4jService();
      if (!neo4j) {
        log.warn('Neo4j not available, skipping load');
        return 0;
      }

      const query = `
        MATCH (ep:ErrorPattern)
        RETURN ep
        ORDER BY ep.occurrences DESC
      `;

      const records = await neo4j.runQuery(query);
      let loaded = 0;

      for (const record of records) {
        const ep = record.ep?.properties || record.ep;
        if (!ep || !ep.id) continue;

        // Skip if already in memory
        if (this.patterns.has(ep.id)) continue;

        const pattern: ErrorPattern = {
          id: ep.id,
          source: ep.source as ErrorSource,
          category: ep.category as ErrorCategory,
          signature: ep.signature,
          description: ep.description,
          severity: ep.severity,
          tags: ep.tags || [],
          cveId: ep.cveId || undefined,
          cweId: ep.cweId || undefined,
          language: ep.language || undefined,
          framework: ep.framework || undefined,
          occurrences: ep.occurrences?.toNumber?.() || ep.occurrences || 1,
          lastSeen: new Date(ep.lastSeen),
          createdAt: new Date(ep.createdAt),
          solutions: JSON.parse(ep.solutions || '[]')
        };

        this.patterns.set(pattern.id, pattern);
        loaded++;
      }

      log.info(`Loaded ${loaded} patterns from Neo4j`);
      return loaded;
    } catch (error) {
      log.error('Failed to load patterns from Neo4j:', error);
      return 0;
    }
  }

  /**
   * Record feedback for a solution (success or failure)
   */
  public async recordFeedback(
    patternId: string,
    solutionIndex: number,
    success: boolean
  ): Promise<boolean> {
    const pattern = this.patterns.get(patternId);
    if (!pattern || !pattern.solutions[solutionIndex]) {
      return false;
    }

    const solution = pattern.solutions[solutionIndex];

    // Initialize counters if needed
    solution.successCount = solution.successCount || 0;
    solution.failureCount = solution.failureCount || 0;

    // Update counters
    if (success) {
      solution.successCount++;
    } else {
      solution.failureCount++;
    }
    solution.lastUsed = new Date();

    // Update confidence based on feedback (Bayesian-ish update)
    const totalFeedback = solution.successCount + solution.failureCount;
    if (totalFeedback >= 3) {
      const successRate = solution.successCount / totalFeedback;
      // Blend original confidence with observed success rate
      solution.confidence = (solution.confidence * 0.3) + (successRate * 0.7);
      // Clamp to valid range
      solution.confidence = Math.max(0.1, Math.min(0.99, solution.confidence));
    }

    // Persist to Neo4j
    await this.persistToNeo4j(pattern);

    log.info(`Recorded ${success ? 'success' : 'failure'} feedback for pattern ${patternId}, solution ${solutionIndex}. New confidence: ${solution.confidence.toFixed(2)}`);
    return true;
  }

  /**
   * Persist all patterns to Neo4j (batch)
   */
  public async persistAllToNeo4j(): Promise<{ success: number; failed: number }> {
    let success = 0;
    let failed = 0;

    for (const pattern of this.patterns.values()) {
      const result = await this.persistToNeo4j(pattern);
      if (result) success++;
      else failed++;
    }

    log.info(`Persisted ${success} patterns to Neo4j (${failed} failed)`);
    return { success, failed };
  }

  /**
   * Get solution with feedback stats
   */
  public getSolutionsWithStats(errorMessage: string): (Solution & { patternId: string; solutionIndex: number })[] {
    const matches = this.findMatches(errorMessage, 3);
    const solutions: (Solution & { patternId: string; solutionIndex: number })[] = [];

    for (const match of matches) {
      match.solutions.forEach((sol, index) => {
        solutions.push({
          ...sol,
          patternId: match.id,
          solutionIndex: index
        });
      });
    }

    return solutions
      .sort((a, b) => b.confidence - a.confidence)
      .slice(0, 5);
  }
}

// Singleton export
export const errorKnowledgeBase = ErrorKnowledgeBase.getInstance();