def on_Save_clicked(self):
minsupp = float(self.doubleSpinBoxMinsupp.text())
minconf = float(self.doubleSpinBoxMinconf.text())
inputName = self.labelInputName.text()
outDir = './OUTPUT/minsup_{}_minconf_{}/{}/'.format(minsupp, minconf, inputName)
freqISName = 'Tap_Phu_Pho_Bien.txt'
maxISName = 'Tap_Phu_Toi_Dai.txt'
rulesName = 'Luat_Ket_Hop.txt'
topTenName = '10_Luat_conf_Cao_Nhat.txt'
make_folder.create_folder(outDir)
freqIS = self.plainTextFreqIS.toPlainText()
maxIS = self.plainTextMaxIS.toPlainText()
rules = self.plainTextRules.toPlainText()
topTenRules = self.plainTextTopten.toPlainText()
if freqIS == "" and maxIS == "" and rules == "" and topTenRules == "":
# QtWidgets.QFileDialog.showEvent(,'Chưa có kết quả để lưu.')
self.labelLog.setText('Chưa có kết quả.')
else:
with open(outDir + freqISName, 'w', encoding = 'utf-8') as freqFile:
freqFile.write(str(freqIS))
with open(outDir + maxISName, 'w', encoding = 'utf-8') as maxFile:
maxFile.write(str(maxIS))
with open(outDir + rulesName, 'w', encoding = 'utf-8') as ruleFile:
ruleFile.write(str(rules))
with open(outDir + topTenName, 'w', encoding = 'utf-8') as topTenFile:
topTenFile.write(str(topTenRules))
self.labelLog.setText('Đã lưu.')
def on_Vectorize_clicked(self):
vectorSize = int(self.spinBoxVecSize.text())
if len(self.listRawSents) == 0:
self.labelLog.setText('Chưa có câu.')
else:
linkFolder = 'outfile/{0}'.format(self.fileInitName)
make_folder.create_folder(linkFolder)
linkModel = linkFolder + '/word2vec.model'
# token sentences --> to list
for sent in self.listRawSents:
tokens = word_tokenize(sent, format='text').split()
words = []
for token in tokens:
if re.match(r'^\w+', token):
words.append(token)
self.listSentToWord.append(words)
# training word2vec cho cái list tách từ.
model = Word2Vec(self.listSentToWord, size=vectorSize, min_count=1)
model.save(linkModel)
sent2vec = Sentence2Vec(linkModel)
listVect = []
for sent in self.listRawSents:
listVect.append(sent2vec.get_vector(sent).tolist())
write_file.list_to_txt(listVect, linkFolder, 'Sent2Vect.txt')
write_file.list_to_txt(self.listSentToWord, linkFolder, 'WordTokenize.txt')
self.labelLog.setText('Đã lưu file vector.')
self.checkVectorize = 1
self.groupBoxKmeans.setEnabled(True)
def list_to_txt(List: list, folderName, name):
make_folder.create_folder(folderName)
if not List:
print('Danh sach rong! ' + name)
return
with open(folderName + '/' + name, 'w', encoding = 'utf-8') as fout:
for item in List:
fout.write('{0}\n'.format(item))
def dict_to_txt(Dict: dict, folderName, name):
make_folder.create_folder(folderName)
if not Dict:
print('Dict rỗng!' + name)
return
with open(folderName + '/' + name, 'w', encoding = 'utf-8') as fout:
for (key, value) in Dict.items():
row = '{0}:{1}\n'.format(key, value)
fout.write(row)
def doubleList_to_txt_continuos(List: list(list()), folderName, name):
make_folder.create_folder(folderName)
if not List:
print('Danh sach rong! ' + name)
return
with open(folderName + name, 'w+', encoding = 'utf-8') as fout:
for iList in List:
for item in iList:
fout.write('{0}\n'.format(item))
fout.write('\n')
def list_to_txt_continuos_with_last_comma(List: list, folderName: str, name: str, seperateType: str):
make_folder.create_folder(folderName)
if not List:
print('Danh sach rong! ' + name)
return
with open(folderName + name, 'a+', encoding = 'utf-8') as fout:
# for item in List:
for index in range(len(List) - 1):
fout.write(('{0}' + seperateType).format(List[index]))
fout.write(('{0}').format(List[-1]))
fout.write('\n')
def list_to_txt_continuos(List: list, folderName: str, name: str, seperateType: str):
make_folder.create_folder(folderName)
if not List:
print('Danh sach rong! ' + name)
return
with open(folderName + name, 'a+', encoding = 'utf-8') as fout:
for item in List:
fout.write(('{0}' + seperateType).format(item))
# fout.write('-!@#$%^&*-'*len(List) + '\n')
# fout.write('\n'*3)
fout.write('\n')
def dict_list_to_txt(Dict: dict, folderName, name):
make_folder.create_folder(folderName)
if not Dict:
print('Dict rỗng!' + name)
return
with open(folderName + '/' + name, 'w', encoding = 'utf-8') as fout:
for (key, values) in Dict.items():
fout.write('{0}:'.format(key))
for indexItem in range(len(values) - 1):
fout.write('{0}, '.format(values[indexItem]) )
fout.write('{0}'.format(values[-1]) )
fout.write('\n')
def list_to_txt_with_last_comma(List: list, folderName, name, splitType: str):
make_folder.create_folder(folderName)
if not List:
print('Danh sach rong! ' + name)
return
with open(folderName + name, 'w', encoding = 'utf-8') as fout:
for itemSet in List:
for index in range(len(itemSet) - 1):
fout.write('{0}{1}'.format(itemSet[index], splitType) )
fout.write('{0}'.format(itemSet[-1]) )
fout.write('\n')
