def convert_st_to_bow(self, st):
bow = [0] * len(self.words)
tagger = ViPosTagger.postagging(ViTokenizer.tokenize(st))
if not (len(tagger[1]) == 1 and tagger[1][0] == 'Np'
and tagger[0][0] not in SKIP_WORDS):
tagger = ViPosTagger.postagging(ViTokenizer.tokenize(st.lower()))
for i, j in enumerate(tagger[1]):
if j in REPLACE:
tagger[0][i] = REPLACE[tagger[1][i]]
if tagger[0][i] in self.words:
bow[self.words.index(tagger[0][i])] = tagger[0].count(
tagger[0][i])
return np.array(bow)
def separate_sentence(self):
# this file include financial symbols
symbol_arr = []
with open('./data/stockslist.txt', encoding='utf-8') as acro_file:
lines = acro_file.readlines()
for line in lines:
symboli = line.rstrip('\n').split(',')
symbol_arr.append(symboli[0].lower())
self.data = self.execute_special_character(self.data)
sentences = self.data.split('\n')
new_sentences = []
for sentence in sentences:
part_of_sentence = sentence.split(' ')
for part in part_of_sentence:
# file.write(part + '\n')
new_sentences.append(part)
all_words = []
for sentence in new_sentences:
words = ViPosTagger.postagging(ViTokenizer.tokenize(sentence))
words = self.tokenizer_tunning(words, 1)
ssi = 'ssi'
for i, word in enumerate(words):
if (self.is_stop_word(word) == False):
if word not in symbol_arr:
all_words.append(word)
else:
all_words.append(ssi)
return all_words
def res_sentence(self, test_sentence):
test_sentence = ViTokenizer.tokenize(test_sentence)
test_sentence, pos = ViPosTagger.postagging(test_sentence)
new_words, pos = self.process(test_sentence, pos)
X_test = self.sent2features(new_words, pos)
new_tags = self.crf.predict_single(X_test)
st1, st2 = [], []
for i in range(len(new_words)):
if new_tags[i] == 'O':
if new_tags[i - 1] != 'O':
st1.append(new_words[i])
st2.append('O')
print(i)
continue
else:
if i == 0:
st1.append(new_words[i])
st2.append('O')
else:
st1[-1] = st1[-1] + '_' + new_words[i]
elif new_tags[i][0] == 'B':
tag = "" + new_tags[i][2:]
st1.append(new_words[i])
st2.append(tag.upper())
elif new_tags[i][0] == 'I':
st1[-1] = st1[-1] + '_' + new_words[i]
return st1, st2
def sentence_segment(self, text, sw_file='./stopwords', cadidate_pos=['Nc', 'Np', 'S', 'R', 'A', 'C', 'V', 'I']):
"""Store those words only in cadidate_pos"""
# Get stopword
with open(sw_file, 'r') as f:
sw = f.readlines()
for i in range(len(sw)):
sw[i] = sw[i].strip()
# word segment
text = ViTokenizer.tokenize(text)
text = text.replace('‘', ' ')
text = text.replace('’', ' ')
text = text.split('.')
sentences = []
for t in text:
temp = ViPosTagger.postagging(t)
sentence = []
for w,t in zip(temp[0], temp[1]):
if len(w) > 0 and w not in sw and t in cadidate_pos:
sentence.append(w)
sentences.append(sentence)
temp = []
for sentence in sentences:
if len(sentence) >= self.window_size:
temp.append(sentence)
return temp
def text_postag(text):
pos_tag = ViPosTagger.postagging(ViTokenizer.tokenize(text))
dict_tag = {}
for i in range(len(pos_tag[0])):
dict_tag[pos_tag[0][i]] = pos_tag[1][i]
return dict_tag
def pos_pyvi():
file = open("test_gan_nhan.txt", "r", encoding="utf-8")
st = file.readlines()
for i in range(len(st)):
st[i] = list(map(str, st[i].split()))
for j in range(len(st[i])):
st[i][j] = list(map(str, st[i][j].split('/')))
s1 = list([])
kq = list([])
for i in range(len(st)):
tmp = ""
for j in range(len(st[i])):
tmp = tmp + st[i][j][0] + ' '
s1.append(tmp)
s = ViPosTagger.postagging(tmp)
kq.append(list([]))
for j in range(len(s[0])):
kq[i].append(list([s[0][j], s[1][j]]))
H = 0
T = 0
for i in range(len(st)):
for j in range(len(kq[i])):
if (kq[i][j][1] != "CH"):
kq[i][j][1] = kq[i][j][1][0]
if (kq[i][j][1] == "F"):
kq[i][j][1] = "CH"
for j in range(len(kq[i])):
T += 1
if (j < len(st[i]) and st[i][j][1] == kq[i][j][1]):
H += 1
print("Do chinh xac cua gan nhan tu loai cua Pyvi la: " +
str(float(H) / float(T)))
def load_data(data):
train_data = []
for line in data:
line = line.replace(',', '.')
line = line.split('.')
for sentence in line:
sentence = re.sub(r'[():;/%$-@!*&^?><_#+]', ' ', sentence)
sentence = sentence.replace('"', ' ')
sentence = sentence.replace("'", " ")
sentence, pos_tag = ViPosTagger.postagging(sentence)
for index, pos in enumerate(pos_tag):
if pos == 'Np':
sentence[index] = 'Np'
if pos == 'Nc':
sentence[index] = 'Nc'
if pos == 'X':
sentence[index] = 'X'
if pos == 'Ny':
sentence[index] = 'Ny'
if pos == 'M':
sentence[index] = 'M'
sentence = [
word.lower().rstrip('\n') for word in sentence if word != ''
]
if len(sentence) > 1:
train_data.append(sentence)
return train_data
def pyvi_prc(text):
tokens, tags = ViPosTagger.postagging(ViTokenizer.tokenize(text))
result = {}
for i in range(len(tokens)):
tokens[i] = tokens[i].replace('_', ' ')
result[tokens[i]] = tags[i]
return result
def get_POS_feature(text):
tag_pos = ViPosTagger.postagging(text)
vocab = tag_pos[0]
list_pos = tag_pos[1]
result = []
for index,pos in enumerate(list_pos):
result.append(pos)
return result
def add_postag(comment):
comment = comment.lower()
a = ViPosTagger.postagging(ViTokenizer.tokenize(comment))
X = []
for i in range(len(a[0])):
test = a[0][i] + "_" + a[1][i]
X.append(test)
b = " ".join(X)
return b
def pos(query):
query = re.sub(r'[?|$|.|!|<|=|,|\-|\'|\“|\”]', r'', query)
b = ViPosTagger.postagging(ViTokenizer.tokenize(query))
important = ['N', 'Nc', 'Ny', 'Np', 'Nu', 'A', 'V']
result = []
for i in range(len(b[1])):
if (b[1][i] in important):
result.append(b[0][i])
return ' '.join(result)
def get_Word_based_POS(text):
tag_pos = ViPosTagger.postagging(text)
vocab = tag_pos[0]
list_pos = tag_pos[1]
result = []
for index,pos in enumerate(list_pos):
if "N" in pos or "V" in pos or "A" in pos:
result.append(vocab[index])
return result
def processing_text(text):
normalized_text = '. '.join([line.strip() for line in text.split('.')])
tokenized_text, pos_seqs = ViPosTagger.postagging(
ViTokenizer.tokenize(normalized_text))
for i, tag in enumerate(pos_seqs):
if tag in ['Np', 'Nu', 'M']:
tokenized_text[i] = tag
tokens = ' '.join(
[token for token in tokenized_text if token not in string.punctuation])
return tokens
def tokenize(string: str):
tokenized = ViTokenizer.tokenize(string)
words, label = ViPosTagger.postagging(tokenized)
for idx, word in enumerate(words):
word = word.replace('_', " ")
words[idx] = word
words = [w for w in words if w]
return words
def posvt():
m = Frame.m
txt = m.get()
seg = ViPosTagger.postagging(ViTokenizer.tokenize(txt))
print(seg)
pyperclip.copy(" ".join(seg))
root10 = tk.Tk()
label0 = tk.Label(root10, text=seg, font=16)
label0.pack(fill="x")
root10.title('Result(POS-VT)')
root10.mainloop()
def extract_name(text, stopwords):
tokenized_text = ViTokenizer.tokenize(text)
tokenized_text = clean_text(tokenized_text, stopwords)
words, tags = ViPosTagger.postagging(tokenized_text)
res = []
for i in range(len(words)):
if (tags[i] == "Np"):
# print(words[i])
res.append(words[i].replace("_", " "))
return res
def load_new_stopwords(text):
not_labels = ['A', 'L', 'R', 'T', 'E', 'M', 'I']
stop_words = utils.load_stop_words('stopwords.txt')
stop_words += list(string.punctuation)
stop_word_mini = []
text = text.split()
tokens, postag = ViPosTagger.postagging_tokens(text)
for i in range(len(tokens)):
if postag[i] in not_labels:
stop_word_mini.append(tokens[i].lower())
stop_words += stop_word_mini
stop_words = set(stop_words)
return stop_words
def _sanitize(self):
'''
Trims nonessential words such as 'and', 'or', 'for'
Parts of Speech types:
http://www.ling.upenn.edu/courses/Fall_2003/ling001/penn_treebank_pos.html
For vietnamese, see: https://pypi.org/project/pyvi/
'''
words_list = []
if (len(self.query_str) == len(self.query_str.encode('utf-8'))):
tags_to_keep = [
'NN',
'NNS',
'NNP',
'NNPS', # noun types
'VB',
'VBD',
'VBG',
'VBN',
'VBP',
'VBZ', # verb types
'JJ',
'JJR',
'JJS', # adjective types
'RB',
'RBR',
'RBS', # adverbs
'CD',
'FW'
]
tokens = nltk.word_tokenize(self.query_str)
tags = nltk.pos_tag(tokens)
for tag in tags:
if tag[1] in tags_to_keep:
words_list.append(tag[0])
else:
#Vietnamese?
tags_to_keep = ['N', 'Ny', 'Np', 'V', 'A', 'R', 'M', 'X']
tokens = ViTokenizer.tokenize(self.query_str)
tags = ViPosTagger.postagging(tokens)
for index in range(len(tags[0])):
if tags[1][index] in tags_to_keep:
words_list.append(tags[0][index].replace('_', ' '))
new_query_str = ' '.join(words_list)
if len(new_query_str) == 0:
new_query_str = self.query_str
self.deleted_words += len(self.query_str.split()) - len(
new_query_str.split())
self.query_str = new_query_str
def create_and_train(self):
for key, value in self.json_data.items():
if 'patterns' not in value:
continue
for pattern in value['patterns']:
tagger = ViPosTagger.postagging(ViTokenizer.tokenize(pattern))
w = []
for i, j in enumerate(tagger[1]):
if j in REPLACE:
tagger[0][i] = REPLACE[j]
if j not in POS_TAG and tagger[0][i] not in STOP_WORDS:
w.append(tagger[0][i])
self.words.extend(w)
self.documents.append((w, key))
self.classes.append(key)
self.words = sorted(list(set(self.words)))
training = []
for doc in self.documents:
st_out = [0] * len(self.words)
for w in doc[0]:
st_out[self.words.index(w)] = doc[0].count(w)
class_out = [0] * len(self.classes)
class_out[self.classes.index(doc[1])] = 1
training.append([st_out, class_out])
random.shuffle(training)
training = np.array(training)
self.train_x = list(training[:, 0])
self.train_y = list(training[:, 1])
pickle.dump(
{
"documents": self.documents,
"classes": self.classes,
"words": self.words
}, open(SAVE_FILE, "wb"))
self.model = Model()
self.model.train(self.train_x, self.train_y)
def get_topic_word_in_sentence(sentence):
topic_words, other_words = [], []
# token = list(ViTokenizer.tokenize(setence).split())
token = list(
requests.post(url=url_token, data={
"text": sentence
}).text.split())
tokens = " ".join((t for t in token if len(t) > 1))
token_pos = ViPosTagger.postagging(tokens)
for i in range(len(token_pos[0])):
# print(token_pos[0][i], token_pos[1][i])
if token_pos[1][i] in noun:
topic_words.append(token_pos[0][i])
else:
other_words.append(token_pos[0][i])
return topic_words, other_words
def locKiTuDacBiet(s):
s1 = ""
s2 = ""
if s != 'href' and s != 'class' and s != 'hashtag-link' and s != '\n' and s != 'thì' and s != 'là' and s != 'ở' and s != 'đi' and s != 'tao' and s != 'mày' and s != 'cây' and s != 'đến' and s != 'vừng' and s != 'bán' and s != 'đồ ăn' and s != 'Đồ ăn' and s != 'cơm_chiên' and s != 'vô' and s != 'cách' and s != 'đây' and s != 'Vị_trí' and s != 'bánh_bao' and s != 'Kem' and s != 'từ' and s != 'ngoài' and s != 'vô' and s != 'của' and s != 'xe' and s != 'thứ' and s != 'hôm' and s != 'đó' and s != 'kho' and s != 'quẹt' and s != 'buổi_sáng' and s != 'Xe_đẩy' and s != 'decor' and s != 'i' and s != 'o' and s != 'đươ' and s != 'c' and s != 'n' and s != 'cu' and s != '_' and s != 'service' and s != 'Menu' and s != 'bad' and s != 'ㅠ' and s != 'bill' and s != 'Matcha' and s != 'green' and s != 'almond' and s != 'chocolate' and s != 'PERFECT' and s != 'kpop' and s != 'SG' and s != 'upstair' and s != 'driving' and s != 'in' and s != 'to' and s != 'check' and s != 'say' and s != 'ran':
for i in range(len(s)):
if s[i] != '!' and s[i] != '/' and s[i] != '.' and s[
i] != "'": #and s[i]!='['and s[i]!=']'and s[i]!='|'and s[i]!='{'and s[i]!='}'and s[i]!=';'and s[i]!=','
s1 += s[i]
s1.rstrip("\n")
s3 = ViPosTagger.postagging(ViTokenizer.tokenize(u"%s" % s1))
for i in range(len(s3[0])):
if s3[1][i] != 'N' and s3[1][i] != 'Np' and s3[1][i] != 'P' and s3[1][
i] != 'E' and s3[1][i] != 'T' and s3[1][i] != 'L' and s3[1][
i] != 'M':
s2 += s3[0][i] + " "
return s2
def fetch_data(st, t):
tagger = ViPosTagger.postagging(ViTokenizer.tokenize(st))
for i, j in enumerate(tagger[1]):
if j == 'Np' and t == 'NAME':
return tagger[0][i]
if j == 'M':
if t == 'NUMBER':
return tagger[0][i]
if t == 'DATE':
str_date = tagger[0][i]
try:
return datetime.datetime.strptime(str_date, "%d/%m/%Y")
except ValueError:
return (
"Bạn chưa nhập đúng định dạng ngày theo ngày/tháng/năm hoặc ngày nhập không hợp lệ",
1)
return False
def get_kqxs(st, session, data):
tagger = ViPosTagger.postagging(ViTokenizer.tokenize(st.title()))
if tagger[1][-1] == 'Np':
p = tagger[0][-1]
url = 'http://xskt.com.vn/rss-feed/'
rss = {
'Bắc': 'mien-bac-xsmb.rss',
'Nam': 'mien-nam-xsmn.rss',
'Trung': 'mien-trung-xsmt.rss',
'Bình_Định': 'binh-dinh-xsbdi.rss',
'Đắc_Lắk': 'dac-lac-xsdlk.rss',
'Đà Nẵng': 'da-nang-xsdng.rss',
'Đắc_Nông': 'dac-nong-xsdno.rss',
'TP.HCM': 'tp-hcm-xshcm.rss',
'Sài_Gòn': 'tp-hcm-xshcm.rss',
'Hcm': 'tp-hcm-xshcm.rss',
'Quảng_Ngãi': 'quang-ngai-xsqng.rss',
'Quảng_Nam': 'quang-nam-xsqnm.rss'
}
try:
link = url + rss[p]
except KeyError:
p = no_accent_vietnamese(p.lower().replace("_", " "))
vt = ""
for i in p.split():
vt += i[0]
link = url + p.replace(" ", "-") + ("-xs{0}.rss").format(vt)
feed = feedparser.parse(link)
content = ""
if len(feed['items']) == 0:
feed = feedparser.parse(url + rss['Bắc'])
for item in feed['items']:
content += "<div><strong>" + item['title'] + "</strong></div>"
content += "<div>" + item['summary'].replace("[", "<br/>[").replace(
"8:", " 8:") + "</div>"
return content
def get_bmi(st, track, data):
st = st.lower()
try:
if st.index("cm"):
st = st.replace("cm", " cm")
except ValueError:
st = st.replace("m", " m")
st = st.replace("kg", " kg")
arr = []
tagger = ViPosTagger.postagging(ViTokenizer.tokenize(st))
for i, j in enumerate(tagger[1]):
if j == 'M':
arr.append(float(tagger[0][i]))
if j == 'Nu' and tagger[0][i] == 'cm':
arr[0] = arr[0] / 100
if (len(arr) < 2):
return ({"text": "Cho em chỉ số về chiều cao (m) và cân nặng (kg)"}, 1)
arr = sorted(arr)
h = arr[0]
w = arr[1]
bmi = w / (h * h)
if (bmi < 18.5):
tt = "gầy"
elif (bmi < 24.9):
tt = "bình thường"
elif (bmi < 29.9):
tt = "hơi béo"
elif (bmi < 34.9):
tt = "béo phì cấp độ 1"
elif (bmi < 39.9):
tt = "béo phì cấp độ 2"
else:
tt = "béo phì cấp độ 3"
return ({
"text":
"Chỉ số BMI của bạn là: %.2f.\nTình trạng hiện tại: %s" % (bmi, tt)
}, 1)
def get_L_sentences(self, s):
from pyvi import ViTokenizer, ViPosTagger
LSeg, LPOS = ViPosTagger.postagging(ViTokenizer.tokenize(s))
LRtn = []
for i, (segment, pos) in enumerate(zip(LSeg, LPOS), start=1):
LRtn.append(
CubeItem(index=i,
word=segment.replace('_', ' '),
lemma=segment.replace('_', ' '),
upos=DPOSToCube[pos],
xpos='',
attrs='',
head='',
label='',
space_after='_'
if i != len(LSeg) - 1 else 'SpaceAfter=No'))
return [LRtn]
def get_data_for_training(self):
with open(self.PATH, encoding='utf-8') as json_file:
training_data = json.load(json_file)
for pattern in training_data:
wTV = ViPosTagger.postagging(
ViTokenizer.tokenize((pattern['sentence'])))
self.words.extend(wTV[0])
self.documents.append((wTV[0], pattern['class']))
if pattern['class'] not in self.classes:
self.classes.append(pattern['class'])
self.output_empty = [0] * len(self.classes)
self.words = [
w.lower() for w in self.words if w not in self.ignore_words
]
self.words = sorted(list(set(self.words)))
# prepare for predicting
self.load_w()
def Process_V_Sentence(sent):
#tach tu va lay nhan cho tu
processed_sent = []
list_tagged_word = ViPosTagger.postagging(ViTokenizer.tokenize(sent))
list_word = list_tagged_word[0] #lay ra cac tu
list_pos = list_tagged_word[1] #lay ra cac nhan tuong ung voi cac tu
for i in range(len(list_word)):
#them tu vao cau
if list_pos[i] not in ['F', 'P']:
if list_pos[i] not in ['Ny', 'Np']:
list_word[i] = list_word[i].lower()
list_word[i] = list_word[i].replace('_', ' ')
processed_sent.append(list_word[i])
#loai bo cac stop word trong cau
processed_sent = Eliminate_V_Stop_Word("v_stopwords.txt", processed_sent)
return processed_sent
def Get_List_V_Sent(paragraph):
list_sents = []
words_in_sent = []
list_tagged_word = ViPosTagger.postagging(ViTokenizer.tokenize(paragraph))
list_word = list_tagged_word[0] #lay ra cac tu
list_pos = list_tagged_word[1] #lay ra cac nhan tuong ung voi cac tu
for i in range(len(list_word)):
#them tu vao cau
if list_pos[i] != 'F':
list_word[i] = list_word[i].replace('_', ' ')
words_in_sent.append(list_word[i])
if (list_pos[i] == 'F') and (list_word[i]
in '.!?'): #neu gap dau ket thuc cau
sentence = ' '.join(words_in_sent)
sentence += list_word[i] #them dau ket thuc cau
list_sents.append(sentence)
words_in_sent = []
return list_sents
def create_pos(sentence):
sentence = preprocess(sentence)
pos = ViPosTagger.postagging(ViTokenizer.tokenize(sentence))
result = []
final = []
for i in range(len(pos[0])):
result.append((pos[0][i], pos[-1][i]))
result = postprocess_pos(result)
i = 0
while i < len(result):
if i < len(result) - 2 and pos_tags[result[i][-1]] == pos_tags[result[i+2][-1]] == "numeral" and pos_tags[result[i+1][-1]] == "punctuation":
temp = (result[i][0] + result[i+1][0] +
result[i+2][0], name_to_pos["numeral"])
i += 3
else:
temp = result[i]
i += 1
final.append(temp)
return final
def remove_stopword_sent(self, sent):
new_tokens = []
new_pos = []
print("before res:", sent)
sent = self.restore_acronym(sent)
print("after res:", sent)
s = ViPosTagger.postagging(ViTokenizer.tokenize(sent))
for i in range(len(s[0])):
if s[0][i] in self.stock_code:
new_tokens.append(s[0][i])
new_pos.append("Np")
elif self.is_stop_word(s[0][i]):
if (s[0][i] == "như"):
print("S[i]", s[0][i])
a = 0
else:
new_tokens.append(s[0][i])
new_pos.append(s[1][i])
return (new_tokens, new_pos)