1. import re
  3. def index_form_data(form_data):
  4. """
  5. Indexes content of form fields for short-lived tasks.
  7. Args:
  8. form_data (dict): A dictionary where keys are field names and values are the field contents.
  10. Returns:
  11. dict: A dictionary where keys are field names and values are lists of indexed terms.
  12. """
  13. indexed_data = {}
  14. for field_name, content in form_data.items():
  15. if isinstance(content, str): # Process only string values
  16. # Lowercase and remove punctuation for consistent indexing
  17. processed_content = re.sub(r'[^\w\s]', '', content.lower())
  18. terms = processed_content.split() # Split into individual terms
  19. indexed_data[field_name] = terms
  20. else:
  21. indexed_data[field_name] = [] # Handle non-string values gracefully
  22. return indexed_data

Add your comment