1. import json
  2. from urllib.parse import urlparse
  4. def extract_urls_from_config(config_file):
  5. """
  6. Extracts URLs from a configuration file.
  8. Args:
  9. config_file (str): Path to the configuration file (JSON).
  11. Returns:
  12. list: A list of extracted URLs. Returns an empty list if the file
  13. doesn't exist or is invalid.
  14. """
  15. try:
  16. with open(config_file, 'r') as f:
  17. config = json.load(f)
  18. except FileNotFoundError:
  19. print(f"Error: Configuration file not found: {config_file}")
  20. return []
  21. except json.JSONDecodeError:
  22. print(f"Error: Invalid JSON format in {config_file}")
  23. return []
  25. urls = []
  26. # Iterate through the configuration data. Assumes 'urls' key.
  27. if 'urls' in config:
  28. for url_data in config['urls']:
  29. if isinstance(url_data, str): # Handle simple string URLs
  30. urls.append(url_data)
  31. elif isinstance(url_data, dict) and 'value' in url_data: #Handle dictionary with a value
  32. urls.append(url_data['value'])
  33. elif isinstance(url_data, list): #Handle list of strings
  34. for item in url_data:
  35. if isinstance(item, str):
  36. urls.append(item)
  37. else:
  38. print(f"Warning: Unexpected URL data format: {url_data}")
  39. return urls
  42. def validate_url(url):
  43. """
  44. Validates a URL using urllib.parse.urlparse.
  46. Args:
  47. url (str): The URL to validate.
  49. Returns:
  50. bool: True if the URL is valid, False otherwise.
  51. """
  52. try:
  53. result = urlparse(url)
  54. return all([result.scheme, result.netloc]) # Check scheme and network location
  55. except:
  56. return False
  60. def main():
  61. """
  62. Main function to demonstrate URL extraction.
  63. """
  64. config_file = 'config.json' # Replace with your config file name
  65. extracted_urls = extract_urls_from_config(config_file)
  67. valid_urls = [url for url in extracted_urls if validate_url(url)]
  69. print("Extracted URLs:")
  70. for url in valid_urls:
  71. print(url)
  75. if __name__ == "__main__":
  76. main()

Add your comment