1. import html2text
  2. from bs4 import BeautifulSoup
  3. import json
  5. def convert_html(input_html, output_format="text", default_values=None):
  6. """
  7. Converts HTML to various formats with default values.
  9. Args:
  10. input_html (str): The HTML content to convert.
  11. output_format (str): The desired output format ("text", "markdown", "json"). Defaults to "text".
  12. default_values (dict): Dictionary of default values to use if elements are missing.
  14. Returns:
  15. str or dict: The converted content as a string or dictionary, depending on the output format.
  16. Returns None if the output format is invalid.
  17. """
  19. if output_format == "text":
  20. try:
  21. h = html2text.HTML2Text()
  22. h.ignore_links = True # Optional: ignore links
  23. h.ignore_images = True #Optional : ignore images
  24. text = h.handle(input_html)
  25. return text
  26. except Exception as e:
  27. print(f"Error converting to text: {e}")
  28. return None
  30. elif output_format == "markdown":
  31. try:
  32. soup = BeautifulSoup(input_html, 'html.parser')
  33. return soup.prettify(formatter='html')
  34. except Exception as e:
  35. print(f"Error converting to markdown: {e}")
  36. return None
  38. elif output_format == "json":
  39. try:
  40. soup = BeautifulSoup(input_html, 'html.parser')
  41. data = {}
  42. for element in soup.find_all():
  43. tag = element.name
  44. attributes = element.attrs
  45. text = element.get_text(strip=True)
  47. if tag not in data:
  48. data[tag] = []
  49. data[tag].append({
  50. "attributes": attributes,
  51. "text": text
  52. })
  53. return json.dumps(data, indent=4)
  54. except Exception as e:
  55. print(f"Error converting to JSON: {e}")
  56. return None
  58. else:
  59. print("Invalid output format.")
  60. return None
  63. if __name__ == '__main__':
  64. # Example usage
  65. html_content = """
  66. <html>
  67. <head>
  68. <title>My Page</title>
  69. </head>
  70. <body>
  71. <h1>Hello, world!</h1>
  72. <p>This is a paragraph.</p>
  73. <a href="https://example.com">Example Link</a>
  74. <div>
  75. <p>Some content inside a div.</p>
  76. </div>
  77. </body>
  78. </html>
  79. """
  81. # Convert to text
  82. text_output = convert_html(html_content, output_format="text")
  83. if text_output:
  84. print("Text Output:\n", text_output)
  86. # Convert to markdown
  87. markdown_output = convert_html(html_content, output_format="markdown")
  88. if markdown_output:
  89. print("\nMarkdown Output:\n", markdown_output)
  91. # Convert to JSON
  92. json_output = convert_html(html_content, output_format="json")
  93. if json_output:
  94. print("\nJSON Output:\n", json_output)

Add your comment