def fromstring(cls, html):
encoding = get_encodings_from_content(html)
encoding = encoding and encoding[0] or None
if not encoding:
html = encodeValue(html)
doc = lxml.html.fromstring(html)
else:
html = smart_str(html, encoding=encoding)
parser = lxml.html.HTMLParser(encoding=encoding)
doc = lxml.html.fromstring(html, parser=parser)
return doc
def stringToNode(cls, html):
html = encodeValue(html)
return lxml.html.fromstring(html)
def fromstring(cls, html):
html = encodeValue(html)
doc = etree.parse(StringIO(html),
etree.HTMLParser(recover=True)).getroot()
return doc
def fromstring(cls, html):
from lxml.html import soupparser
html = encodeValue(html)
doc = soupparser.fromstring(html)
return doc
def fromstring(self, html):
from lxml.html import soupparser
html = encodeValue(html)
self.doc = soupparser.fromstring(html)
return self.doc