def html(self):
try:
return self.html5lib.parseFragment(self.content, treebuilder="etree")
except ImportError as err:
raise ImproperlyConfigured("Error while importing html5lib: %s" % err)
except Exception as err:
raise ParserError("Error while initializing Parser: %s" % err)
class Html5LibParser(ParserBase):
def __init__(self, content):
super(Html5LibParser, self).__init__(content)
import html5lib
self.html5lib = html5lib
def _serialize(self, elem):
fragment = self.html5lib.treebuilders.simpletree.DocumentFragment()
fragment.appendChild(elem)
return self.html5lib.serialize(fragment,
quote_attr_values=True,
omit_optional_tags=False)
def _find(self, *names):
for node in self.html.childNodes:
if node.type == 5 and node.name in names:
yield node
@cached_property
def html(self):
try:
return self.html5lib.parseFragment(self.content)
except ImportError, err:
raise ImproperlyConfigured("Error while importing html5lib: %s" %
err)
except Exception, err:
raise ParserError("Error while initializing Parser: %s" % err)
class BeautifulSoupParser(ParserBase):
@cached_property
def soup(self):
try:
from BeautifulSoup import BeautifulSoup
return BeautifulSoup(self.content)
except ImportError, err:
raise ImproperlyConfigured(
"Error while importing BeautifulSoup: %s" % err)
except Exception, err:
raise ParserError("Error while initializing Parser: %s" % err)
def __init__(self, content):
HTMLParser.__init__(self)
self.content = content
self._css_elems = []
self._js_elems = []
self._current_tag = None
try:
self.feed(self.content)
self.close()
except Exception, err:
raise ParserError("Error while initializing HtmlParser: %s" % err)
def soup(self):
try:
if six.PY3:
from bs4 import BeautifulSoup
else:
from BeautifulSoup import BeautifulSoup
return BeautifulSoup(self.content)
except ImportError as err:
raise ImproperlyConfigured(
"Error while importing BeautifulSoup: %s" % err)
except Exception as err:
raise ParserError("Error while initializing Parser: %s" % err)
class LxmlParser(ParserBase):
def __init__(self, content):
try:
from lxml.html import fromstring, soupparser
from lxml.etree import tostring
self.fromstring = fromstring
self.soupparser = soupparser
self.tostring = tostring
except ImportError, err:
raise ImproperlyConfigured("Error while importing lxml: %s" % err)
except Exception, err:
raise ParserError("Error while initializing Parser: %s" % err)
def __init__(self, content):
try:
from lxml.html import fromstring
from lxml.etree import tostring
except ImportError as err:
raise ImproperlyConfigured("Error while importing lxml: %s" % err)
except Exception as err:
raise ParserError("Error while initializing parser: %s" % err)
self.fromstring = fromstring
self.tostring = tostring
super(LxmlParser, self).__init__(content)
def __init__(self, content):
six.moves.html_parser.HTMLParser.__init__(self, **HTML_PARSER_ARGS)
self.content = content
self._css_elems = []
self._js_elems = []
self._current_tag = None
try:
self.feed(self.content)
self.close()
except Exception as err:
lineno = err.lineno
line = self.content.splitlines()[lineno]
raise ParserError("Error while initializing HtmlParser: %s (line: %s)" % (err, repr(line)))
def __init__(self, content):
try:
from lxml.html import fromstring
from lxml.etree import tostring
except ImportError as err:
raise ImproperlyConfigured("Error while importing lxml: %s" % err)
except Exception as err:
raise ParserError("Error while initializing parser: %s" % err)
if not six.PY3:
# soupparser uses Beautiful Soup 3 which does not run on python 3.x
try:
from lxml.html import soupparser
except ImportError as err:
soupparser = None
except Exception as err:
raise ParserError("Error while initializing parser: %s" % err)
else:
soupparser = None
self.soupparser = soupparser
self.fromstring = fromstring
self.tostring = tostring
super(LxmlParser, self).__init__(content)
class LxmlParser(ParserBase):
@cached_property
def tree(self):
content = '<root>%s</root>' % self.content
try:
from lxml.html import fromstring, soupparser
from lxml.etree import tostring
self.tostring = tostring
tree = fromstring(content)
try:
ignore = tostring(tree, encoding=unicode)
except UnicodeDecodeError:
tree = soupparser.fromstring(content)
except ImportError, err:
raise ImproperlyConfigured("Error while importing lxml: %s" % err)
except Exception, err:
raise ParserError("Error while initializing Parser: %s" % err)
def tree(self):
try:
from lxml import html
from lxml.etree import tostring
except ImportError, e:
raise ParserError("Error while initializing Parser: %s" % e)
def soup(self):
try:
from BeautifulSoup import BeautifulSoup
except ImportError, e:
raise ParserError("Error while initializing Parser: %s" % e)