def handle_starttag(self, tag, attrs):
global cur
a = {}
for attr in attrs:
a[attr[0]] = attr[1]
cur = DOM.Element(cur, tag, a)
if tag in self_closing_tags:
cur.parent.write(cur)
cur = cur.parent
def parse(html):
global cur
cur = DOM.Element(None, 'document')
parser().feed(html)
while cur.tag != 'document':
cur.parent.write(cur)
cur = cur.parent
if not cur.getElementsByTagName('body'):
return parse('<html><body>%s</body></html>' % html)
return cur