1. import json
  2. import hashlib
  4. def verify_user_records(filepath, record_hash_field='record_hash'):
  5. """
  6. Verifies the integrity of user records stored in a JSON file.
  8. Args:
  9. filepath (str): The path to the JSON file containing user records.
  10. record_hash_field (str): The field name in each record that stores the hash.
  12. Returns:
  13. list: A list of user records that have integrity issues (hash mismatch).
  14. Returns an empty list if all records are valid.
  15. """
  17. integrity_issues = []
  18. try:
  19. with open(filepath, 'r') as f:
  20. records = json.load(f)
  21. except FileNotFoundError:
  22. print(f"Error: File not found at {filepath}")
  23. return []
  24. except json.JSONDecodeError:
  25. print(f"Error: Invalid JSON format in {filepath}")
  26. return []
  28. for record in records:
  29. try:
  30. record_hash = record.get(record_hash_field)
  32. if record_hash is None:
  33. print(f"Warning: Record missing '{record_hash_field}'. Skipping integrity check.")
  34. integrity_issues.append(record) #Consider as invalid since hash is missing.
  35. continue
  37. # Recalculate the hash from the record data
  38. data_string = json.dumps(record, sort_keys=True).encode('utf-8')
  39. calculated_hash = hashlib.sha256(data_string).hexdigest()
  41. if calculated_hash != record_hash:
  42. print(f"Integrity issue: Record with ID {record.get('id', 'Unknown')} has incorrect hash.")
  43. integrity_issues.append(record)
  45. except Exception as e:
  46. print(f"Error processing record: {e}")
  47. integrity_issues.append(record)
  49. return integrity_issues
  51. if __name__ == '__main__':
  52. # Example usage:
  53. filepath = 'user_records.json' # Replace with your file path
  55. # Create a dummy user_records.json for testing
  56. dummy_data = [
  57. {"id": 1, "name": "Alice", "email": "alice@example.com", "record_hash": "e5b7e3b94d1e91946a34287b7428472a67873790e49b3556727c55f124821580"},
  58. {"id": 2, "name": "Bob", "email": "bob@example.com", "record_hash": "2cedc9c46a91b88377062b3d851c44e478762110454f131302fc2115001760c9"},
  59. {"id": 3, "name": "Charlie", "email": "charlie@example.com", "record_hash": "e5b7e3b94d1e91946a34287b7428472a67873790e49b3556727c55f124821580"}, #Duplicate hash
  60. {"id": 4, "name": "David", "email": "david@example.com"} #Missing record_hash
  61. ]
  62. with open(filepath, 'w') as f:
  63. json.dump(dummy_data, f, indent=4)
  65. integrity_issues = verify_user_records(filepath)
  67. if integrity_issues:
  68. print("Integrity issues found:")
  69. for record in integrity_issues:
  70. print(record)
  71. else:
  72. print("All records are valid.")

Add your comment