def __init__(self, parser_file = 'englishPCFG.ser.gz',
parser_options=['-maxLength', '80', '-retainTmpSubcategories']):
"""@param parser_file: path to the serialised parser model
(e.g. englishPCFG.ser.gz)
@param parser_options: options
"""
assert os.path.exists(parser_file)
options = Options()
options.setOptions(parser_options)
self.lp = LexicalizedParser.getParserFromFile(parser_file, options)
tlp = PennTreebankLanguagePack()
self.gsf = tlp.grammaticalStructureFactory()
self.lemmer = Morphology()
self.word_token_factory = WordTokenFactory()
self.parser_query = None
def __init__(self, parser_file,
parser_options=['-maxLength', '80', '-retainTmpSubcategories']):
"""@param parser_file: path to the serialised parser model
(e.g. englishPCFG.ser.gz)
@param parser_options: options
"""
assert os.path.exists(parser_file)
options = Options()
options.setOptions(parser_options)
self.lp = LexicalizedParser.getParserFromFile(parser_file, options)
tlp = PennTreebankLanguagePack()
self.gsf = tlp.grammaticalStructureFactory()
self.lemmer = Morphology()
self.word_token_factory = WordTokenFactory()
self.parser_query = None
def __init__(self):
tlp = PennTreebankLanguagePack()
self.gsf = tlp.grammaticalStructureFactory()