def handle_endtag(self, name):
if self.removing_trees:
if name == self.removing_trees[-1]:
self.removing_trees.pop()
return
if name not in self.valid_tags:
if (name in taginfo.tags and taginfo.type(name) not in tags_to_divize):
return
name = 'div'
self.buffer += '</%s>' % name
def handle_startendtag(self, name, attrs):
if self.removing_trees:
return
if name not in self.valid_tags:
if (name in taginfo.tags and taginfo.type(name) not in tags_to_divize):
return
name = 'div'
attrs = [(k, v) for k, v in attrs if self.is_attr_allowed(name, k)]
attrs.append(('shucker', self.get_uid()))
self.buffer += '<%s%s/>' % (name, attrs_to_html(attrs))
def handle_starttag(self, name, attrs):
if self.removing_trees:
return
if name not in self.valid_tags:
if name in taginfo.tags and taginfo.type(name) in tags_to_empty:
self.removing_trees.append(name)
if (name in taginfo.tags and taginfo.type(name) not in tags_to_divize):
return
name = 'div'
attrs = [(k, v) for k, v in attrs if self.is_attr_allowed(name, k)]
attrs = dict(attrs)
if 'href' in attrs:
attrs['href'] = re.compile('(java|vb)script:.*', re.I).sub('#',
attrs['href'])
attrs = attrs.items()
attrs.append(('shucker', self.get_uid()))
self.buffer += '<%s%s>' % (name, attrs_to_html(attrs))
