1. import json
  2. import os
  3. from pathlib import Path
  4. from bs4 import BeautifulSoup
  6. def attach_metadata(html_file_path, metadata, override=None):
  7. """
  8. Attaches metadata to an HTML file for sandbox usage with manual overrides.
  10. Args:
  11. html_file_path (str): Path to the HTML file.
  12. metadata (dict): Dictionary containing the metadata to attach.
  13. override (dict, optional): Dictionary containing metadata overrides. Defaults to None.
  14. """
  15. try:
  16. with open(html_file_path, 'r', encoding='utf-8') as f:
  17. html_content = f.read()
  18. except FileNotFoundError:
  19. print(f"Error: File not found at {html_file_path}")
  20. return
  22. soup = BeautifulSoup(html_content, 'html.parser')
  24. # Apply default metadata
  25. for key, value in metadata.items():
  26. if key == 'title':
  27. soup.title.string = str(value) if value else ''
  28. elif key == 'description':
  29. soup.find('meta', attrs={'name': 'description'})['content'] = str(value) if value else ''
  30. elif key == 'keywords':
  31. soup.find('meta', attrs={'name': 'keywords'})['content'] = str(value) if value else ''
  32. elif key == 'og:title':
  33. soup.find('meta', attrs={'property': 'og:title'})['content'] = str(value) if value else ''
  34. elif key == 'og:description':
  35. soup.find('meta', attrs={'property': 'og:description'})['content'] = str(value) if value else ''
  36. elif key == 'og:type':
  37. soup.find('meta', attrs={'property': 'og:type'})['content'] = str(value) if value else ''
  38. elif key == 'og:image':
  39. soup.find('meta', attrs={'property': 'og:image'})['content'] = str(value) if value else ''
  41. # Apply overrides
  42. if override:
  43. for key, value in override.items():
  44. if key == 'title':
  45. soup.title.string = str(value) if value else ''
  46. elif key == 'description':
  47. soup.find('meta', attrs={'name': 'description'})['content'] = str(value) if value else ''
  48. elif key == 'keywords':
  49. soup.find('meta', attrs={'name': 'keywords'})['content'] = str(value) if value else ''
  50. elif key == 'og:title':
  51. soup.find('meta', attrs={'property': 'og:title'})['content'] = str(value) if value else ''
  52. elif key == 'og:description':
  53. soup.find('meta', attrs={'property': 'og:description'})['content'] = str(value) if value else ''
  54. elif key == 'og:type':
  55. soup.find('meta', attrs={'property': 'og:type'})['content'] = str(value) if value else ''
  56. elif key == 'og:image':
  57. soup.find('meta', attrs={'property': 'og:image'})['content'] = str(value) if value else ''
  59. # Write the modified HTML back to the file
  60. try:
  61. with open(html_file_path, 'w', encoding='utf-8') as f:
  62. f.write(soup.prettify()) # Use prettify for better formatting
  63. except Exception as e:
  64. print(f"Error writing to file: {e}")
  68. if __name__ == '__main__':
  69. # Example Usage
  70. html_file = 'example.html'
  72. # Create a dummy HTML file for testing
  73. with open(html_file, 'w', encoding='utf-8') as f:
  74. f.write('<html><head><title>Original Title</title></head><body><h1>Hello World</h1></body></html>')
  76. default_metadata = {
  77. 'title': 'Sandbox Page',
  78. 'description': 'This is a sandbox page.',
  79. 'keywords': 'sandbox, html, metadata',
  80. 'og:title': 'Sandbox Page',
  81. 'og:description': 'This is a sandbox page.',
  82. 'og:type': 'website',
  83. 'og:image': 'https://example.com/image.jpg'
  84. }
  86. override_metadata = {
  87. 'title': 'Overridden Title

Add your comment