data = pd.read_excel("../data/dataset.xlsx")
list_label = []
train_list, dev_list, test_list = [],[],[]
data_value = data.values
for i in range(len(data_value)):
    if data_value[i][0] not in list_label:
        list_label.append(data_value[i][0])
for j in range(len(list_label)):
    s = data.loc[data["
 
c1"] == list_label[j]]
    s = s.loc[:,['c1','c2']]     #c1為標籤，c2為具體的資料
    s = s.ix[:].values
    s = s.tolist()
    s_random = random.sample(s,len(s))
    train_list = train_list +s_random[:int(0.8*len(s_random))]
    dev_list = dev_list + s_random[int(0.8*len(s_random)):int(0.9*len(s_random))]
    test_list 
 
= test_list + s_random[int(0.9*len(s_random)):]
def write_file (s,f_file):
    s = random.sample(s,len(s))
    f1 = open(f_file,'w',encoding='utf-8')
    sen_str = ''
    labe_cata = []
    for i in range(len(s)):
        sen_str += '__label__' + s[i][0]
        labe_cata.append(s[i][0])
        sen_str 
 
+= "\t"
        for j in s[i][1]:
            sen_str += j + ' '
        sen_str.strip()
        sen_str += '\n'
    print(len(set(labe_cata)))
    f1.write(sen_str)
    f1.close()

write_file(train_list, '../data/train.txt')
write_file(dev_list, '../data/val.txt')
write_file(test_list, '../data/test.txt')