1. import argparse
  2. from bs4 import BeautifulSoup
  3. import json
  5. def validate_dom(html_content, schema_file):
  6. """
  7. Validates a DOM structure against a JSON schema.
  9. Args:
  10. html_content (str): The HTML content to validate.
  11. schema_file (str): Path to the JSON schema file.
  13. Returns:
  14. bool: True if the DOM is valid, False otherwise.
  15. """
  16. try:
  17. soup = BeautifulSoup(html_content, 'html.parser')
  18. except Exception as e:
  19. print(f"Error parsing HTML: {e}")
  20. return False
  22. with open(schema_file, 'r') as f:
  23. schema = json.load(f)
  25. def validate_element(element, schema_node):
  26. """Recursively validates an element against the schema."""
  27. if schema_node['type'] == 'string':
  28. if element.get('data-id') != schema_node['value']:
  29. print(f"Validation error: Element '{element.name}' data-id should be '{schema_node['value']}'")
  30. return False
  31. elif schema_node['type'] == 'attribute':
  32. attr_name = schema_node['attribute']
  33. attr_value = element.get(attr_name)
  34. if attr_value != schema_node['value']:
  35. print(f"Validation error: Element '{element.name}' attribute '{attr_name}' should be '{schema_node['value']}'")
  36. return False
  37. elif schema_node['type'] == 'children':
  38. if 'children' not in schema_node:
  39. return True # no children required
  40. for child in element.children:
  41. if not validate_element(child, schema_node['children']):
  42. return False
  43. return True
  44. elif schema_node['type'] == 'tag':
  45. if element.name != schema_node['value']:
  46. print(f"Validation error: Element '{element.name}' should be '{schema_node['value']}'")
  47. return False
  48. for attr in schema_node['attributes']:
  49. attr_name = attr['name']
  50. attr_value = element.get(attr_name)
  51. if attr_value != attr['value']:
  52. print(f"Validation error: Element '{element.name}' attribute '{attr_name}' should be '{attr['value']}'")
  53. return False
  54. for child in element.children:
  55. if not validate_element(child, schema_node['children']):
  56. return False
  57. return True
  58. else:
  59. print(f"Unknown schema type: {schema_node['type']}")
  60. return False
  62. return validate_element(soup.body, schema) # Start validation from the body
  64. def main():
  65. """Parses command-line arguments and performs DOM validation."""
  66. parser = argparse.ArgumentParser(description='Validate DOM structure against a JSON schema.')
  67. parser.add_argument('html_file', help='Path to the HTML file.')
  68. parser.add_argument('schema_file', help='Path to the JSON schema file.')
  69. args = parser.parse_args()
  71. try:
  72. with open(args.html_file, 'r') as f:
  73. html_content = f.read()
  74. except FileNotFoundError:
  75. print(f"Error: HTML file not found: {args.html_file}")
  76. return
  78. if validate_dom(html_content, args.schema_file):
  79. print("DOM is valid.")
  80. else:
  81. print("DOM is invalid.")
  83. if __name__ == "__main__":
  84. main()

Add your comment