import os
path = '/Users/v_yangwenyi01/Downloads/infile_obem3.txt'
log_path = "/home/public/djangoWeb/web_mysite/datasemanteme/infile_obem2.txt"
os.system("python trans_obems2label_for_siyuan.py infile_obem.txt %s" % (log_path))
with open(path, 'r') as fin:
data = fin.read().splitlines(True)
with open(path, 'w') as fout:
fout.writelines(data[1:])
a_list = open(path, 'r').readlines()
a_list = [a for a in a_list if a.strip()]
a_list = set(a_list)
a_list = list(a_list)
train_list = []
for sents in a_list:
items = sents.split('\t')
if len(items) == 6:
lable_items_key = items[3] + ',' + items[4] + ','+','
elif len(items) > 6:
lable_items_key=items[3]+','+items[4]+','+items[5]+','+items[6]
else:
label = items[3] + ',' + items[4] + ',' + items[5]+','+items[6]
train_list.append(Semanteme(type='semanteme', periods=1000002, train=items[1], label=items[3]+','+items[4]+','+items[5]))
Semanteme.objects.using("tndatadb").bulk_create(train_list)