-
Notifications
You must be signed in to change notification settings - Fork 0
/
test.py
184 lines (149 loc) · 7.47 KB
/
test.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
from v2.dataset import tags, Cinnamon_Dataset_Testing_v2, DataLoader
from v2.train import *
from v2.utils.convert import *
from v2.utils.score import *
from collections import Counter
import torch.nn.functional as F
import os, warnings, argparse
warnings.filterwarnings('ignore')
def parse_args(string=None):
parser = argparse.ArgumentParser()
parser.add_argument('test_dataset_path')
parser.add_argument('--batch-size', default=1,
type=int, help='batch size')
parser.add_argument('--gpu', default="0",
type=str, help="0:1080ti 1:1070")
parser.add_argument('--num-workers', default=1,
type=int, help='dataloader num workers')
parser.add_argument('--model', default='blstm',
type=str, help='naive,blstm')
parser.add_argument('--postprocess', action='store_true',
help='do postprocessing ?')
parser.add_argument('--cinnamon-data-path', default='/media/D/ADL2020-SPRING/project/cinnamon/',
type=str, help='cinnamon dataset')
parser.add_argument('--dev_or_test', default='dev',
type=str, help='dev or test set')
parser.add_argument('--load-model', default='./v2/ckpt/blstm/epoch_90.pt',
type=str, help='.pt model file ')
parser.add_argument('--save-result-path', default='./',
type=str, help='.pt model file save dir')
parser.add_argument('--ref-file', default='/media/D/ADL2020-SPRING/project/cinnamon/dev/dev_ref.csv',
type=str, help='calcu score ref file')
parser.add_argument('--score', action='store_true')
args = parser.parse_args() if string is None else parser.parse_args(string)
if not os.path.exists(args.save_result_path): os.makedirs(args.save_result_path)
return args
def fuller(text):
candidate = (ord('0'),ord('1'),ord('2'),ord('3'),ord('4'),ord('5'),ord('6'),ord('7'),ord('8'),ord('9'),
ord('('),ord(')'),ord('~'),)
text_out = ''
for c in text:
if ord(c) in candidate:
text_out += chr(ord(c)+65248)
else:
text_out += c
return text_out
def post_process(value, tag, text):
if tag=='質問箇所TEL/FAX':
return text.replace('イ．','').replace('ア．','').replace('．','').replace(' ','')
elif tag=='質問箇所所属/担当者':
return text.replace('イ．','').replace('ア．','').replace('．','').replace(' ','')
elif tag in ['資格申請送付先',
'資格申請送付先部署/担当者名',
'入札書送付先',
'入札書送付先部署/担当者名',]:
return text.replace('イ．','').replace('ア．','').replace('．','').replace(' ','')
'''
value = value.replace('##l:','ＴＥＬ：').replace('tel:','ＴＥＬ：').replace('Tel:','ＴＥＬ：').replace('TEL:','ＴＥＬ：')
value = value.replace('fax:','ＦＡＸ：').replace('Fax:','ＦＡＸ：').replace('FAX:','ＦＡＸ：')
print(text)
input("")
# 半形 轉 全形
value = fuller(value)
'''
value_ret = ''
for c in value:
if c in text:
c = c
elif chr(ord(c)+65248) in text:
c = chr(ord(c)+65248)
elif ord('a')<=ord(c) and ord(c)<=ord('z'): #小寫轉大寫
if chr(ord(c)-32) in text: #小寫轉大寫 半形
c = chr(ord(c)-32)
elif chr(ord(c)+65248-32) in text: #小寫轉大寫 + 轉全形
c = chr(ord(c)+65248-32)
elif ord('A')<=ord(c) and ord(c)<=ord('Z'): #大寫轉小寫
if chr(ord(c)+32) in text: #大寫轉小寫 半形
c = chr(ord(c)+65248+32)
elif chr(ord(c)+65248+32) in text: #大寫轉小寫 + 轉全形
c = chr(ord(c)+65248+32)
else:
pass
#print(c, text, value)
value_ret += c
return value_ret
def inference(args, tokenizer, dataloader):
if args.model == 'naive':
model = Model()
elif args.model == 'blstm':
model = Model_BLSTM()
model.load_state_dict(torch.load(args.load_model)['state_dict']) #.cuda().eval()
model.eval()
model.cuda()
with torch.no_grad():
total_dataframe = None
data = {'doc':[],'index':[],'ID':[],'Tag':[],'Value':[]}
for iii,(doc, index, ids, _, masks, sample) in enumerate(dataloader):
output = model(ids.cuda())[0]
prob = F.sigmoid(output).cpu()
doc, index, ids, masks, sample = doc[0], index[0], ids[0], masks[0], sample[0]
data['doc'].append(doc)
data['index'].append(index)
data['ID'].append(f"{doc}-{index}")
data['Tag'].append("")
data['Value'].append("")
for i,tag in enumerate(tags):
values = []
for j in range(prob.size(0)):
if prob[j,i] > 0.5:
values.append(ids[j])
if len(values)>0:
value_str = tokenizer.decode(values, skip_special_tokens=True).replace(" ","")
if True or args.postprocess:
value_str = post_process(value_str, tag, sample['text'])
else:
value_str = value_str
# add a tag&value to <Tag> <Value>
if data['Tag'][-1] == "":
data['Tag'][-1] = "{}".format(tag)
data['Value'][-1] = "{}".format(value_str)
else:
data['Tag'][-1] += ";{}".format(tag)
data['Value'][-1] += ";{}".format(value_str)
#total_dataframe = total_dataframe.append(sample) if isinstance(total_dataframe, pd.DataFrame) else sample
print(f'\t[Info] [{iii+1}/{len(dataloader)}]', end=' \r')
total_dataframe = pd.DataFrame(data)
print('\t[Info] finish inference ')
# sort dataframe & save results
total_dataframe = total_dataframe.sort_values(by=['doc','index'], ascending=[True,True])
#total_dataframe = total_dataframe.drop('Page No', axis=1).drop('Parent Index', axis=1).drop('Is Title', axis=1).drop(
# 'Is Table', axis=1).drop('id', axis=1).drop('Index', axis=1)
total_dataframe.to_csv(f'{args.save_result_path}/result.csv', encoding='utf8')
# convert results to submission format
convert(f'{args.save_result_path}/result.csv', f'{args.save_result_path}/submission.csv')
os.remove(f'{args.save_result_path}/result.csv')
# score
if args.score:
s = score(args.ref_file, f'{args.save_result_path}/submission.csv')
print('\t[Info] score:', s)
if __name__ == '__main__':
args = parse_args()
os.environ['CUDA_VISIBLE_DEVICES'] = args.gpu
os.environ['CUDA_LAUNCH_BLOCKING'] = "1"
tokenizer = BertJapaneseTokenizer.from_pretrained(pretrained_weights)#, do_lower_case=True)
dataset = Cinnamon_Dataset_Testing_v2(args.test_dataset_path, tokenizer, tags)
dataloader = DataLoader(dataset,
batch_size=1,
collate_fn=dataset.collate_fn,
shuffle=False)
inference(args, tokenizer, dataloader)