__tests__/scraper/scraperLogger.test.js

@@ -0,0 +1,471 @@
+/**
+ * Tests for scraperLogger.js
+ * Following TDD - these tests are written BEFORE implementation
+ */
+
+const { createLogger, LEVELS } = require('../../services/scraperLogger');
+
+describe('scraperLogger', () => {
+  let consoleSpy;
+  let logOutput;
+
+  beforeEach(() => {
+    logOutput = [];
+    consoleSpy = jest.spyOn(console, 'log').mockImplementation((output) => {
+      logOutput.push(output);
+    });
+  });
+
+  afterEach(() => {
+    consoleSpy.mockRestore();
+  });
+
+  describe('createLogger', () => {
+    test('should return an object with info, warn, error methods', () => {
+      const logger = createLogger('test-job-id');
+
+      expect(typeof logger).toBe('object');
+      expect(typeof logger.info).toBe('function');
+      expect(typeof logger.warn).toBe('function');
+      expect(typeof logger.error).toBe('function');
+    });
+
+    test('should include jobId in all log entries', () => {
+      const jobId = 'unique-job-123';
+      const logger = createLogger(jobId);
+
+      logger.info('Test message');
+
+      expect(logOutput.length).toBe(1);
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.jobId).toBe(jobId);
+    });
+  });
+
+  describe('log entry format', () => {
+    test('should include timestamp in ISO format', () => {
+      const logger = createLogger('test-job');
+      const beforeTime = new Date().toISOString();
+
+      logger.info('Test message');
+
+      const afterTime = new Date().toISOString();
+      const parsed = JSON.parse(logOutput[0]);
+
+      expect(parsed.timestamp).toBeDefined();
+      // Verify timestamp is in ISO format
+      expect(parsed.timestamp).toMatch(/^\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}.\d{3}Z$/);
+      // Verify timestamp is within test window
+      expect(parsed.timestamp >= beforeTime).toBe(true);
+      expect(parsed.timestamp <= afterTime).toBe(true);
+    });
+
+    test('should include log level', () => {
+      const logger = createLogger('test-job');
+
+      logger.info('Info message');
+      logger.warn('Warn message');
+      logger.error('Error message');
+
+      expect(logOutput.length).toBe(3);
+
+      const infoEntry = JSON.parse(logOutput[0]);
+      const warnEntry = JSON.parse(logOutput[1]);
+      const errorEntry = JSON.parse(logOutput[2]);
+
+      expect(infoEntry.level).toBe('info');
+      expect(warnEntry.level).toBe('warn');
+      expect(errorEntry.level).toBe('error');
+    });
+
+    test('should include message', () => {
+      const logger = createLogger('test-job');
+
+      logger.info('Test message content');
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.message).toBe('Test message content');
+    });
+
+    test('should include context when provided', () => {
+      const logger = createLogger('test-job');
+      const context = { key: 'value', count: 42 };
+
+      logger.info('Test message', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context).toEqual(context);
+    });
+
+    test('should output valid JSON', () => {
+      const logger = createLogger('test-job');
+
+      logger.info('Test message', { data: 'test' });
+
+      expect(() => JSON.parse(logOutput[0])).not.toThrow();
+    });
+  });
+
+  describe('log levels', () => {
+    test('info() should output with level "info"', () => {
+      const logger = createLogger('test-job');
+
+      logger.info('Info message');
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.level).toBe('info');
+    });
+
+    test('warn() should output with level "warn"', () => {
+      const logger = createLogger('test-job');
+
+      logger.warn('Warning message');
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.level).toBe('warn');
+    });
+
+    test('error() should output with level "error"', () => {
+      const logger = createLogger('test-job');
+
+      logger.error('Error message');
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.level).toBe('error');
+    });
+  });
+
+  describe('LEVELS constant', () => {
+    test('should export LEVELS constant with info, warn, error', () => {
+      expect(LEVELS).toBeDefined();
+      expect(LEVELS.info).toBe('info');
+      expect(LEVELS.warn).toBe('warn');
+      expect(LEVELS.error).toBe('error');
+    });
+  });
+
+  describe('message truncation', () => {
+    test('should truncate messages longer than 1000 characters', () => {
+      const logger = createLogger('test-job');
+      const longMessage = 'x'.repeat(1500);
+
+      logger.info(longMessage);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.message.length).toBeLessThanOrEqual(1020); // 1000 + "... [truncated]"
+      expect(parsed.message).toContain('... [truncated]');
+    });
+
+    test('should not truncate messages of 1000 characters or less', () => {
+      const logger = createLogger('test-job');
+      const shortMessage = 'x'.repeat(1000);
+
+      logger.info(shortMessage);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.message).toBe(shortMessage);
+      expect(parsed.message).not.toContain('... [truncated]');
+    });
+
+    test('should handle empty message', () => {
+      const logger = createLogger('test-job');
+
+      logger.info('');
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.message).toBe('');
+    });
+
+    test('should handle null or undefined message', () => {
+      const logger = createLogger('test-job');
+
+      logger.info(null);
+      logger.info(undefined);
+
+      expect(logOutput.length).toBe(2);
+      // Should not throw and should handle gracefully
+    });
+  });
+
+  describe('MongoDB connection string redaction', () => {
+    test('should redact username and password from mongodb:// URIs', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        uri: 'mongodb://admin:secretpassword123@localhost:27017/mydb'
+      };
+
+      logger.info('Database connection', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.uri).toBe('mongodb://[user]:[REDACTED]@localhost:27017/mydb');
+      expect(parsed.context.uri).not.toContain('admin');
+      expect(parsed.context.uri).not.toContain('secretpassword123');
+    });
+
+    test('should redact username and password from mongodb+srv:// URIs', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        uri: 'mongodb+srv://myuser:mypass@cluster0.example.mongodb.net/testdb'
+      };
+
+      logger.info('Database connection', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.uri).toBe('mongodb+srv://[user]:[REDACTED]@cluster0.example.mongodb.net/testdb');
+      expect(parsed.context.uri).not.toContain('myuser');
+      expect(parsed.context.uri).not.toContain('mypass');
+    });
+
+    test('should handle connection string without credentials', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        uri: 'mongodb://localhost:27017/mydb'
+      };
+
+      logger.info('Database connection', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.uri).toBe('mongodb://localhost:27017/mydb');
+    });
+  });
+
+  describe('sensitive key redaction', () => {
+    test('should redact keys containing "password"', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        password: 'secret123',
+        userPassword: 'anotherSecret',
+        password_hash: 'hashed'
+      };
+
+      logger.info('User data', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.password).toBe('[REDACTED]');
+      expect(parsed.context.userPassword).toBe('[REDACTED]');
+      expect(parsed.context.password_hash).toBe('[REDACTED]');
+    });
+
+    test('should redact keys containing "secret"', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        secret: 'mysecret',
+        clientSecret: 'secret123',
+        SECRET_KEY: 'key'
+      };
+
+      logger.info('Config data', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.secret).toBe('[REDACTED]');
+      expect(parsed.context.clientSecret).toBe('[REDACTED]');
+      // Case-insensitive check
+    });
+
+    test('should redact keys containing "token"', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        token: 'abc123',
+        accessToken: 'token456',
+        refresh_token: 'refresh789'
+      };
+
+      logger.info('Auth data', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.token).toBe('[REDACTED]');
+      expect(parsed.context.accessToken).toBe('[REDACTED]');
+      expect(parsed.context.refresh_token).toBe('[REDACTED]');
+    });
+
+    test('should redact keys containing "apikey"', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        apikey: 'key123',
+        apiKey: 'key456',
+        api_key: 'key789' // Note: underscores may not match 'apikey'
+      };
+
+      logger.info('API data', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.apikey).toBe('[REDACTED]');
+      expect(parsed.context.apiKey).toBe('[REDACTED]');
+    });
+
+    test('should redact keys containing "authorization"', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        authorization: 'Bearer token123',
+        Authorization: 'Basic base64string'
+      };
+
+      logger.info('Header data', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.authorization).toBe('[REDACTED]');
+    });
+
+    test('should not redact non-sensitive keys', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        username: 'john',
+        email: 'john@example.com',
+        count: 42
+      };
+
+      logger.info('User info', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.username).toBe('john');
+      expect(parsed.context.email).toBe('john@example.com');
+      expect(parsed.context.count).toBe(42);
+    });
+  });
+
+  describe('circular reference handling', () => {
+    test('should handle circular references in context', () => {
+      const logger = createLogger('test-job');
+      const context = { name: 'test' };
+      context.self = context; // Create circular reference
+
+      // Should not throw
+      expect(() => {
+        logger.info('Circular test', context);
+      }).not.toThrow();
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.name).toBe('test');
+      expect(parsed.context.self).toBe('[Circular]');
+    });
+
+    test('should handle deeply nested circular references', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        level1: {
+          level2: {
+            level3: {}
+          }
+        }
+      };
+      context.level1.level2.level3.back = context.level1;
+
+      expect(() => {
+        logger.info('Deep circular test', context);
+      }).not.toThrow();
+
+      // Should be valid JSON output
+      expect(() => JSON.parse(logOutput[0])).not.toThrow();
+    });
+  });
+
+  describe('Buffer handling', () => {
+    test('should represent Buffer objects as "[Buffer: N bytes]"', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        data: Buffer.from('Hello World')
+      };
+
+      logger.info('Buffer data', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.data).toBe('[Buffer: 11 bytes]');
+    });
+
+    test('should handle empty Buffer', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        data: Buffer.alloc(0)
+      };
+
+      logger.info('Empty buffer', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.data).toBe('[Buffer: 0 bytes]');
+    });
+
+    test('should handle large Buffer', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        data: Buffer.alloc(1024 * 1024) // 1MB buffer
+      };
+
+      logger.info('Large buffer', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.data).toBe('[Buffer: 1048576 bytes]');
+    });
+  });
+
+  describe('edge cases', () => {
+    test('should handle empty context object', () => {
+      const logger = createLogger('test-job');
+
+      logger.info('Test message', {});
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context).toEqual({});
+    });
+
+    test('should handle context not provided', () => {
+      const logger = createLogger('test-job');
+
+      logger.info('Test message');
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context).toEqual({});
+    });
+
+    test('should handle null context', () => {
+      const logger = createLogger('test-job');
+
+      logger.info('Test message', null);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context).toEqual({});
+    });
+
+    test('should handle undefined values in context', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        defined: 'value',
+        undefinedValue: undefined
+      };
+
+      logger.info('Test message', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.defined).toBe('value');
+      // undefined is typically converted to null in JSON
+      expect(parsed.context.undefinedValue).toBeNull();
+    });
+
+    test('should handle nested objects in context', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        outer: {
+          inner: {
+            value: 'deep'
+          }
+        }
+      };
+
+      logger.info('Nested test', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.outer.inner.value).toBe('deep');
+    });
+
+    test('should handle arrays in context', () => {
+      const logger = createLogger('test-job');
+      const context = {
+        items: [1, 2, 3, 'four']
+      };
+
+      logger.info('Array test', context);
+
+      const parsed = JSON.parse(logOutput[0]);
+      expect(parsed.context.items).toEqual([1, 2, 3, 'four']);
+    });
+  });
+});

services/scraperLogger.js

@@ -0,0 +1,164 @@
+/**
+ * Scraper-specific structured JSON logger
+ * Produces one JSON object per line to stdout
+ */
+
+const LEVELS = {
+  info: 'info',
+  warn: 'warn',
+  error: 'error'
+};
+
+/**
+ * Create a logger instance scoped to a job ID
+ * @param {string} jobId - Job identifier for correlation
+ * @returns {Object} Logger object with info, warn, error methods
+ */
+function createLogger(jobId) {
+  /**
+   * Internal log function
+   * @param {string} level - Log level
+   * @param {string} message - Log message
+   * @param {Object} context - Additional context data
+   */
+  const log = (level, message, context = {}) => {
+    const entry = {
+      timestamp: new Date().toISOString(),
+      level,
+      message: truncateMessage(message),
+      jobId,
+      context: sanitizeContext(context)
+    };
+
+    // Output as single-line JSON
+    console.log(JSON.stringify(entry));
+  };
+
+  return {
+    info: (message, context) => log(LEVELS.info, message, context),
+    warn: (message, context) => log(LEVELS.warn, message, context),
+    error: (message, context) => log(LEVELS.error, message, context)
+  };
+}
+
+/**
+ * Truncate message to prevent log bloat
+ * @param {string} message - Message to truncate
+ * @param {number} maxLength - Maximum length (default 1000)
+ * @returns {string} Truncated message
+ */
+function truncateMessage(message, maxLength = 1000) {
+  if (message === null || message === undefined) {
+    return '';
+  }
+  const str = String(message);
+  if (str.length <= maxLength) {
+    return str;
+  }
+  return str.substring(0, maxLength) + '... [truncated]';
+}
+
+/**
+ * Recursively process an object to handle Buffers, circular references, and sensitive data
+ * @param {*} obj - Object to process
+ * @param {WeakSet} seen - Set of seen objects for circular reference detection
+ * @returns {*} Processed value
+ */
+function processValue(obj, seen = new WeakSet()) {
+  // Handle null/undefined
+  if (obj === null) {
+    return null;
+  }
+  if (obj === undefined) {
+    return null;
+  }
+
+  // Handle Buffer BEFORE checking for object (Buffer is an object)
+  if (Buffer.isBuffer(obj)) {
+    return `[Buffer: ${obj.length} bytes]`;
+  }
+
+  // Handle strings - check for MongoDB connection strings
+  if (typeof obj === 'string') {
+    if (/mongodb(\+srv)?:\/\//.test(obj)) {
+      return redactConnectionString(obj);
+    }
+    return obj;
+  }
+
+  // Handle primitives
+  if (typeof obj !== 'object') {
+    return obj;
+  }
+
+  // Handle circular references
+  if (seen.has(obj)) {
+    return '[Circular]';
+  }
+  seen.add(obj);
+
+  // Handle arrays
+  if (Array.isArray(obj)) {
+    return obj.map(item => processValue(item, seen));
+  }
+
+  // Handle plain objects
+  const result = {};
+  const sensitiveKeys = ['password', 'secret', 'token', 'apikey', 'authorization'];
+
+  for (const key of Object.keys(obj)) {
+    // Check for sensitive keys
+    if (sensitiveKeys.some(k => key.toLowerCase().includes(k))) {
+      result[key] = '[REDACTED]';
+    } else {
+      result[key] = processValue(obj[key], seen);
+    }
+  }
+
+  return result;
+}
+
+/**
+ * Sanitize context object for safe logging
+ * - Remove circular references
+ * - Redact sensitive data
+ * - Handle special types (Buffer, undefined)
+ * @param {Object} context - Context object
+ * @returns {Object} Sanitized context
+ */
+function sanitizeContext(context) {
+  if (!context || typeof context !== 'object') {
+    return {};
+  }
+
+  try {
+    return processValue(context);
+  } catch (error) {
+    // If sanitization fails, return empty context
+    return { sanitizationError: 'Failed to sanitize context' };
+  }
+}
+
+/**
+ * Redact credentials from MongoDB connection string
+ * @param {string} uri - Connection string
+ * @returns {string} Redacted string
+ */
+function redactConnectionString(uri) {
+  try {
+    // Match mongodb://user:pass@host or mongodb+srv://user:pass@host
+    return uri.replace(
+      /mongodb(\+srv)?:\/\/([^:]+):([^@]+)@/,
+      'mongodb$1://[user]:[REDACTED]@'
+    );
+  } catch {
+    return '[REDACTED CONNECTION STRING]';
+  }
+}
+
+module.exports = {
+  createLogger,
+  LEVELS,
+  // Export internal functions for testing
+  redactConnectionString
+};