1. import re
  3. def normalize_headers(headers):
  4. """
  5. Normalizes request headers, handling case, whitespace, and basic sanity checks.
  7. Args:
  8. headers (dict): A dictionary of request headers.
  10. Returns:
  11. dict: A dictionary of normalized headers. Returns an empty dictionary if input is invalid.
  12. """
  14. if not isinstance(headers, dict):
  15. print("Error: Input must be a dictionary.")
  16. return {}
  18. normalized_headers = {}
  19. for key, value in headers.items():
  20. if not isinstance(key, str):
  21. print(f"Warning: Invalid header key: {key}. Skipping.") #Log warning instead of error
  22. continue
  24. if not isinstance(value, str):
  25. normalized_headers[key] = str(value) #Convert to string if not already
  26. continue
  28. # Normalize case to lowercase
  29. normalized_key = key.lower()
  30. normalized_value = value.lower()
  32. # Remove leading/trailing whitespace
  33. normalized_key = normalized_key.strip()
  34. normalized_value = normalized_value.strip()
  36. #Sanity check: Remove potentially problematic characters
  37. normalized_key = re.sub(r'[^a-z0-9_]+', '', normalized_key)
  38. normalized_value = re.sub(r'[^a-z0-9_.,\s-]+', '', normalized_value) #Allow periods, commas, spaces and hyphens
  40. if not normalized_key:
  41. print(f"Warning: Empty header key after sanitization: {key}. Skipping.")
  42. continue
  44. normalized_headers[normalized_key] = normalized_value
  46. return normalized_headers

Add your comment