1. def sanitize_user_data(user_data):
  2. """
  3. Sanitizes user data for compatibility with older systems.
  4. Replaces potentially problematic values with safe alternatives.
  5. """
  6. sanitized_data = user_data.copy() #Avoid modifying the original dict
  8. # Replace potentially problematic characters in names/titles
  9. for key, value in sanitized_data.items():
  10. if isinstance(value, str):
  11. sanitized_data[key] = value.replace(" ", "_") #Replace spaces with underscores
  12. sanitized_data[key] = sanitized_data[key].replace("<", "&lt;") #Escape less than
  13. sanitized_data[key] = sanitized_data[key].replace(">", "&gt;") #Escape greater than
  14. sanitized_data[key] = sanitized_data[key].replace("\"", "&quot;") #Escape double quote
  15. sanitized_data[key] = sanitized_data[key].replace("'", "&#39;") #Escape single quote
  16. sanitized_data[key] = sanitized_data[key].replace("\\", "&#92;") #Escape backslash
  19. #Handle date formats
  20. if 'birthdate' in sanitized_data:
  21. try:
  22. #Attempt to convert to a standard format. Handle potential errors.
  23. from datetime import datetime
  24. birthdate_str = sanitized_data['birthdate']
  25. sanitized_data['birthdate'] = datetime.strptime(birthdate_str, '%Y-%m-%d').strftime('%Y%m%d') #YYYYMMDD
  26. except (ValueError, TypeError):
  27. sanitized_data['birthdate'] = None #Set to None if conversion fails
  29. #Handle phone numbers (remove special characters)
  30. if 'phone_number' in sanitized_data:
  31. if isinstance(sanitized_data['phone_number'], str):
  32. sanitized_data['phone_number'] = ''.join(filter(str.isdigit, sanitized_data['phone_number'])) #Keep only digits
  34. return sanitized_data

Add your comment