1. import re
  2. import logging
  4. logging.basicConfig(level=logging.INFO)
  7. def clean_form_data(data, dry_run=False):
  8. """
  9. Cleans data from web forms for hypothesis validation.
  11. Args:
  12. data (dict): A dictionary containing the form data.
  13. dry_run (bool): If True, only print changes without modifying the data.
  15. Returns:
  16. dict: The cleaned data.
  17. """
  18. cleaned_data = {}
  19. for key, value in data.items():
  20. if isinstance(value, str):
  21. # Remove leading/trailing whitespace
  22. value = value.strip()
  23. # Convert to lowercase
  24. value = value.lower()
  25. # Remove special characters (keep alphanumeric and spaces)
  26. value = re.sub(r'[^a-zA-Z0-9\s]', '', value)
  27. cleaned_data[key] = value
  28. elif isinstance(value, (int, float)):
  29. cleaned_data[key] = value # Keep numeric values as they are
  30. elif value is None:
  31. cleaned_data[key] = None # keep null values
  32. else:
  33. cleaned_data[key] = str(value) # Convert other types to strings
  34. logging.warning(f"Type conversion detected for key: {key}")
  36. if dry_run:
  37. print("Dry run mode: Changes will not be applied.")
  38. print("Original Data:", data)
  39. print("Cleaned Data:", cleaned_data)
  40. return data
  41. else:
  42. return cleaned_data
  45. if __name__ == '__main__':
  46. # Example usage
  47. form_data = {
  48. "name": " John Doe! ",
  49. "email": "john.doe@example.com",
  50. "age": "30",
  51. "city": None,
  52. "phone": "123-456-7890",
  53. "other_field": 123.45
  54. }
  56. cleaned_data = clean_form_data(form_data, dry_run=True)
  57. print("Cleaned Data:", cleaned_data)
  59. cleaned_data = clean_form_data(form_data)
  60. print("Cleaned Data:", cleaned_data)

Add your comment