def read_block(self, stream):
# Decide which lexical element we're in.
lexelt_num = bisect.bisect_right(self._lexelt_starts,
stream.tell()) - 1
lexelt = self._lexelts[lexelt_num]
instance_lines = []
in_instance = False
while True:
line = stream.readline()
if line == '':
assert instance_lines == []
return []
# Start of a lexical element?
if line.lstrip().startswith('<lexelt'):
lexelt_num += 1
m = re.search('item=("[^"]+"|\'[^\']+\')', line)
assert m is not None # <lexelt> has no 'item=...'
lexelt = m.group(1)[1:-1]
if lexelt_num < len(self._lexelts):
assert lexelt == self._lexelts[lexelt_num]
else:
self._lexelts.append(lexelt)
self._lexelt_starts.append(stream.tell())
# Start of an instance?
if line.lstrip().startswith('<instance'):
assert instance_lines == []
in_instance = True
# Body of an instance?
if in_instance:
instance_lines.append(line)
# End of an instance?
if line.lstrip().startswith('</instance'):
xml_block = '\n'.join(instance_lines)
xml_block = _fixXML(xml_block)
inst = ElementTree.fromstring(xml_block)
return [self._parse_instance(inst, lexelt)]
def read_block(self, stream):
# Decide which lexical element we're in.
lexelt_num = bisect.bisect_right(self._lexelt_starts, stream.tell())-1
lexelt = self._lexelts[lexelt_num]
instance_lines = []
in_instance = False
while True:
line = stream.readline()
if line == '':
assert instance_lines == []
return []
# Start of a lexical element?
if line.lstrip().startswith('<lexelt'):
lexelt_num += 1
m = re.search('item=("[^"]+"|\'[^\']+\')', line)
assert m is not None # <lexelt> has no 'item=...'
lexelt = m.group(1)[1:-1]
if lexelt_num < len(self._lexelts):
assert lexelt == self._lexelts[lexelt_num]
else:
self._lexelts.append(lexelt)
self._lexelt_starts.append(stream.tell())
# Start of an instance?
if line.lstrip().startswith('<instance'):
assert instance_lines == []
in_instance = True
# Body of an instance?
if in_instance:
instance_lines.append(line)
# End of an instance?
if line.lstrip().startswith('</instance'):
xml_block = '\n'.join(instance_lines)
xml_block = _fixXML(xml_block)
inst = ElementTree.fromstring(xml_block)
return [self._parse_instance(inst, lexelt)]
def parse_from_string(self, string):
raw = ElementTree.fromstring(string)
self.parse_tree(raw)
def fromstring(s):
return CRFInfo._read(ElementTree.fromstring(s))
def fromstring(s):
return CRFInfo._read(ElementTree.fromstring(s))
