def _parse(self, str):
""" Parses the text data from an XML element defined by tag.
"""
str = replace_entities(str)
str = strip_tags(str)
str = collapse_spaces(str)
return str
def _parse(self, str):
""" Parses the text data from an XML element defined by tag.
"""
str = replace_entities(str)
str = strip_tags(str)
str = collapse_spaces(str)
return str
def load(self, data):
data = replace_entities(data)
try:
BeautifulSoup.__init__(self, data)
except UnicodeEncodeError:
self.error = PageUnicodeError()
BeautifulSoup.__init__(self, "")
except:
self.error = PageParseError()
BeautifulSoup.__init__(self, "")
def load(self, data):
data = replace_entities(data)
try:
BeautifulSoup.__init__(self, data)
except UnicodeEncodeError:
self.error = PageUnicodeError()
BeautifulSoup.__init__(self, "")
except:
self.error = PageParseError()
BeautifulSoup.__init__(self, "")
def _parse(self, e, tag):
""" Parses the text data from an XML element defined by tag.
"""
tags = e.getElementsByTagName(tag)
children = tags[0].childNodes
if len(children) != 1: return None
assert children[0].nodeType == xml.dom.minidom.Element.TEXT_NODE
s = children[0].nodeValue
s = format_data(s)
s = replace_entities(s)
return s
def _parse(self, e, tag):
""" Parses the text data from an XML element defined by tag.
"""
tags = e.getElementsByTagName(tag)
children = tags[0].childNodes
if len(children) != 1:
return None
assert children[0].nodeType == xml.dom.minidom.Element.TEXT_NODE
s = children[0].nodeValue
s = format_data(s)
s = replace_entities(s)
return s