def convert_to_signed(s: Sentence):
conn, a = s.getComponents()
if a is None:
return [s.generate('signtrue')] + s.reduceBrackets()
else:
(_, form) = a
if conn.startswith('not_'):
return [s.generate('signfalse')] + form.reduceBrackets()
else:
return [s.generate('signtrue')] + s.reduceBrackets()
def into_sentence(prefix: list[int], conn_dict: dict[int, tp.Iterable[str]],
var_amount: int, var_type: str, sess) -> Sentence:
s = Sentence([], sess)
variables = []
for _ in range(var_amount):
t = s.generate(var_type)
s.append(t)
variables.append(t)
s = Sentence([], sess)
_into_sentence(s, prefix, conn_dict, variables)
return s
def _into_sentence(s: Sentence, prefix: list[int],
conn_dict: dict[int,
tp.Iterable[str]], variables: list[str]):
l = prefix[0]
if l == 0:
s.append(rchoice(variables))
else:
possible_main = conn_dict[l]
main = s.generate(rchoice(possible_main))
if l == 2:
# INFIX
s.append('(')
_into_sentence(s, prefix[1:], conn_dict, variables)
s.append(main)
_into_sentence(s, prefix[1:], conn_dict, variables)
s.append(')')
else:
s.append(main)
for _ in range(l):
_into_sentence(s, prefix[1:], conn_dict, variables)
def add_prefix(sentence: Sentence, prefix: str, lexem: str = None) -> Sentence:
"""Dodaje prefiks do zdania
:param sentence: Zdanie do modyfikacji
:type sentence: Sentence
:param prefix: Typ prefiksu (`x` w `x_y`)
:type prefix: str
:param lexem: Leksem prefiksu (`y` in `x_y`)
:type lexem: str
:return: Zmieniony prefiks
:rtype: Sentence
"""
token = sentence.generate(prefix) if not lexem else f"{prefix}_{lexem}"
if len(sentence) == 1:
return Sentence([token, *sentence], sentence.S)
new_record = {0: sentence.calcPrecedenceVal(prefix)}
return Sentence(
[token, '(', *sentence, ')'], sentence.S,
{i + 2: j + 1
for i, j in sentence.precedenceBaked.values()} | new_record)
object* = null | object
'''
corpus = '''
article = 这 这个 这部
object = 电影 影片 片子 片儿 他 她 他们 她们 它 它们 男主 男主角 男主角儿 女主 女主角 女主角儿 男配角 女配角 观众 导演
adv = 太 十分 相当 相当的 真是 真的是 很 非常 非常非常 特别地 相当地 令人 让人 竟然 居然
adj = 精彩 吸引人 绝了 值得一看 好看 引人入胜 感人 动容 可怜 愤怒 气人 可恶 垃圾 难看 不真实 不现实 帅 崇拜 美丽 漂亮 可恨 惋惜 牛 厉害 有水平 水平不行 辛苦 水了 差 不咋地 刮目相看 赞 赞了
verb = 出场 上场 出现 消失 牺牲 杀了 打败了 击败 击败了 爱上了 喜欢上 看上了 相中了 憎恨 不喜欢 仇恨 复仇
'''
# adv = 神秘地
# adj = 不正经
snt_grt = Sentence()
snt_grt.set_corpus(corpus)
snt_grt.set_grammar(gram)
sentece_eg = snt_grt.generate('sentence')
print(sentece_eg)
# 2元语法模型
t1 = time()
data_df = pd.read_csv('input/movie_comments.csv')
model = TwoGrams()
model.train(data_df.loc[:20000, 'comment'])
stn = '导演太让人喜欢了'
print('(p,ppl){}={}'.format(stn, model.prob_sentence(stn)))
stn1 = '高晓松是土肥圆？'
print('(p,ppl){}={}'.format(stn1, model.prob_sentence(stn1)))
# best sentence
print(generate_best(snt_grt, model, 100000, 20))
def test_sentence_generator(self):
random_sentence = Sentence()
self.assertEqual(random_sentence.generate(),
("Colorless green ideas sleep furiously"))
from sentence import Sentence
import config as config
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.DEBUG)
if __name__ == '__main__':
# update mimicry model tweet
twitter = Twitter(consumer_key=config.CONSUMER_KEY, consumer_secret=config.CONSUMER_SECRET,
access_token_key=config.ACCESS_TOKEN, access_token_secret=config.ACCESS_TOKEN_SECRET)
twitter.update_status(config.MIMICRY_MODEL)
# setup natural language recognition env.
message = twitter.get_latest_status_text(config.MIMICRY_MODEL)
if message is None:
logging.error("Mimicry target's tweet is None.")
exit()
semantics = Semantics()
similar_words = semantics.get_similar_words(message)
train_posts = twitter.get_all_status_text()
sentence = Sentence()
sentence.learn(train_posts)
generated_message = sentence.generate(similar_words)
twitter.post(generated_message)