I *just* wrote something that does this. It uses the htmldata module -
you can find that using pypi. It only allows a specific set of html
tags and attempts to close tags not closed. :
from htmldata import tagextract, tagjoin
allowed_tags = ['br', 'b', 'strong', 'em', 'i', 'u', 'tt', 'a', 'big',
'small',
'h2', 'h3', 'h4', 'strike', 'sub', 'sup', 'samp', 's',
'code', 'ins',
'br/',
]
def htmlfilter(intext, allowed_tags=allowed_tags):
"""Given a text entry as input, check it only contains allowed html.
It returns the text with banned html removed.
Uses two functiosn from htmldata :
from htmldata import tagextract, tagjoin
allowed_tags is the list of tags that are allowed.
"""
html = tagextract(intext) #
out_html = []
skip = None
unclosed = []
for entry in html:
if isinstance(entry, basestring):
if skip is not None:
continue
out_html.append(entry)
else:
tag = entry[0]
if skip is not None:
if tag.startswith('/') and tag[1:] == skip:
skip = None
continue
otag = tag
if tag.startswith('/'):
otag = tag[1:]
if otag in allowed_tags:
if tag.startswith('/'):
if otag in unclosed:
unlclosed.remove(otag)
else: # bad html
continue
elif tag not in ['br', '/br', 'hr', '/hr', 'img',
'/img']: # XXXX hardwired - what if we need to add to this ?
unclosed.append(tag)
out_html.append(entry)
continue
if not tag.startswith('/'):
skip = tag
for tag in unclosed:
out_html.append(('/%s' % (tag,), {})) # close any unclosed
tags
return tagjoin(out_html)
###############
I've used it to allow a few html tags to appear in my guestbook
entries. It's not very sophisticated because complex tags like 'div'
and tables aren't allowed.
Best regards,
Fuzzy
http://www.voidspace.org.uk/python 
