tests/html5lib-vs-clean.py

   1 #!/usr/bin/python
   2
   3 import lxml.html, lxml.html.clean
   4 from lxml.html import html5parser
   5
   6 html ="""<html><body>
   7 <!-- a comment -->
   8 </body></html>
   9 """
  10
  11 #tree = lxml.html.document_fromstring(html)
  12 tree = html5parser.document_fromstring(html)
  13
  14 cleaner = lxml.html.clean.Cleaner()
  15 cleaner(tree)
  16