def get_cut_words2(comment):
stop_words = get_stop_list("stop_words")
positive_words = get_stop_list("positive_dict")
negative_words = get_stop_list("negative_dict")
words = cut_word.CutWord(comment).cut()
return [
word for word in words if word in positive_words
or word in negative_words or word not in stop_words
]
def __init__(self):
QtGui.QMainWindow.__init__(self)
Ui_MainWindow.__init__(self)
self.setupUi(self)
self.RMMButton.clicked.connect(self.rmm)
self.MMButton.clicked.connect(self.mm)
self.CW = cut_word.CutWord()
# self.setWindowTitle("Darlin")
self.setWindowTitle(QtCore.QString(u"Darlin 陈文达 吴嘉琳 翁靖达"))
def get_data_from_sql():
"""获取数据"""
mysql_cn = MySQLdb.connect(host='10.250.30.158',
port=3306,
user='******',
passwd='88888888',
db='zjy_test')
sql = "SELECT content,sort FROM language_filterdata where sort!=0 and sort!=3"
cursor = mysql_cn.cursor()
cursor.execute(sql)
alldata = cursor.fetchall()
cursor.close()
mysql_cn.close()
r = '{机器型号:[\s\S]*?}|回复：[\s\S]*'
alldata = [(cut_word.CutWord(re.sub(r, '', data[0].decode("utf-8")),
'zh').cut(), data[1]) for data in alldata]
return alldata
def get_cut_words(comments):
stop_words = get_stop_list("stop_words")
positive_words = get_stop_list("positive_dict")
negative_words = get_stop_list("negative_dict")
words = []
for comment in comments:
print comment
cut_words = cut_word.CutWord(comment, 'zh').cut()
new_cut_words = [
word for word in cut_words if word in positive_words
or word in negative_words or word not in stop_words
]
words += new_cut_words
print "-" * 50
print "cut words:", "/".join(new_cut_words)
print "=" * 100
return words