def __init__(self, pcfg=None):
if pcfg is None:
pcfg = build_model()
self.pcfg = pcfg
self.tokenizer = PennTreebankTokenizer()
if nltk_is_available:
self.parse = self.nltk_parse
else:
self.parse = self.raw_parse
class Parser(object):
def __init__(self, pcfg=None):
if pcfg is None:
pcfg = build_model()
self.pcfg = pcfg
self.tokenizer = PennTreebankTokenizer()
if nltk_is_available:
self.parse = self.nltk_parse
else:
self.parse = self.raw_parse
def norm_parse(self, sentence):
words = self.tokenizer.tokenize(sentence)
if is_cap_word(words[0]):
words[0] = words[0].lower()
norm_words = []
for word in words:
if isinstance(word, tuple):
# This is already a word normalized to the Treebank conventions
norm_words.append(word)
else:
# rare words normalization
norm_words.append((self.pcfg.norm_word(word), word))
return CKY(self.pcfg, norm_words)
def raw_parse(self, sentence):
tree = self.norm_parse(sentence)
un_chomsky_normal_form(tree)
return tree
def nltk_parse(self, sentence):
return nltk_tree(self.raw_parse(sentence))
def __init__(self, pcfg=None):
if pcfg is None:
pcfg = build_model()
self.pcfg = pcfg
self.tokenizer = PennTreebankTokenizer()
if nltk_is_available:
self.parse = self.nltk_parse
else:
self.parse = self.raw_parse
def tokenize_for_parsing(string):
tokenizer = PennTreebankTokenizer()
return tokenizer.tokenize(string)
