1. import re
  3. def sanitize_file_contents(file_content, allowed_chars=None, min_length=0, max_length=None):
  4. """
  5. Sanitizes file contents for validation checks with fallback logic.
  7. Args:
  8. file_content (str): The content of the file to sanitize.
  9. allowed_chars (str, optional): A string containing the characters allowed in the file content.
  10. If None, all characters are allowed. Defaults to None.
  11. min_length (int, optional): The minimum length of the file content. Defaults to 0.
  12. max_length (int, optional): The maximum length of the file content. Defaults to None (no limit).
  14. Returns:
  15. str: The sanitized file content. Returns the original content if sanitization fails.
  16. """
  17. if allowed_chars is None:
  18. allowed_chars = r"^\w+$" # Allow alphanumeric characters
  20. if not isinstance(file_content, str):
  21. return file_content #or raise TypeError("Input must be a string")
  23. if not re.match(allowed_chars, file_content):
  24. # Fallback: Remove invalid characters
  25. sanitized_content = re.sub(r'[^\w\s]', '', file_content) #remove special characters
  26. if not re.match(allowed_chars, sanitized_content):
  27. return file_content #return original if still invalid
  29. if len(file_content) < min_length:
  30. # Fallback: Pad with spaces
  31. sanitized_content = file_content.ljust(min_length, ' ')
  33. if max_length is not None and len(file_content) > max_length:
  34. # Fallback: Truncate
  35. sanitized_content = file_content[:max_length]
  37. return sanitized_content

Add your comment