def feed(self, raw_data):
assert isinstance(raw_data, TEXT_TYPE), "feed data must be unicode!"
data = raw_data.strip()
# cut out <pre> and <tt> areas block tag areas
data = block_re.sub(self._pre_cut_out, data)
data = inline_re.sub(self._pre_cut_out, data)
# Delete whitespace from html code
data = strip_html(data)
if self.debugging:
print("_" * 79)
print("raw data:")
print(repr(raw_data))
print(" -" * 40)
print("cleaned data:")
print(data)
print("-" * 79)
# print(clean_data.replace(">", ">\n"))
# print("-"*79)
HTMLParser.feed(self, data)
return self.root
def feed(self, raw_data):
assert isinstance(raw_data, TEXT_TYPE), "feed data must be unicode!"
data = raw_data.strip()
# cut out <pre> and <tt> areas block tag areas
data = block_re.sub(self._pre_cut_out, data)
data = inline_re.sub(self._pre_cut_out, data)
# Delete whitespace from html code
data = strip_html(data)
if self.debugging:
print("_" * 79)
print("raw data:")
print(repr(raw_data))
print(" -" * 40)
print("cleaned data:")
print(data)
print("-" * 79)
# print(clean_data.replace(">", ">\n"))
# print("-"*79)
HTMLParser.feed(self, data)
return self.root
def __init__(self, debug=False):
HTMLParser.__init__(self)
self.debugging = debug
if self.debugging:
warnings.warn(
message="Html2Creole debug is on! warn every data append.")
self.result = DebugList(self)
else:
self.result = []
self.blockdata = []
self.root = DocNode("document", None)
self.cur = self.root
self.__list_level = 0
def __init__(self, debug=False):
HTMLParser.__init__(self)
self.debugging = debug
if self.debugging:
warnings.warn(
message="Html2Creole debug is on! warn every data append."
)
self.result = DebugList(self)
else:
self.result = []
self.blockdata = []
self.root = DocNode("document", None)
self.cur = self.root
self.__list_level = 0
