-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy path7.use_model.py
192 lines (159 loc) · 5.05 KB
/
7.use_model.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
185
186
187
188
189
190
191
192
import os
# os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
# import tensorflow as tf
import transformers
# tf.compat.v1.logging.set_verbosity(tf.compat.v1.logging.ERROR)
from transformers import GPT2LMHeadModel, GPT2Tokenizer
from time import sleep
NEWLINECHAR = "<N>"
transformers.logging.set_verbosity_error()
tokenizer = GPT2Tokenizer.from_pretrained('tokenizer')
os.environ["USE_TORCH"] = "True"
tokenizer.add_special_tokens({
"eos_token": "</s>",
"bos_token": "<s>",
"unk_token": "<unk>",
"pad_token": "<pad>",
"mask_token": "<mask>"
})
# model = GPT2LMHeadModel.from_pretrained("/home/ise/pzy/AUTOCoder/GPyT_3/checkpoint-2500").to("cuda")
model = GPT2LMHeadModel.from_pretrained("GPyT_3/").to("cuda")
def encode_newlines(inp):
"""
Args:
inp: 用户输入的代码段
Returns:
对输入进行换行符替换操作后，输出
"""
return inp.replace("\n", NEWLINECHAR)
def decode_newlines(inp):
"""
Args:
inp: 模型输出的未经处理的语句
Returns:
输出将鬼画符替换为换行符后的结果
"""
return inp.replace(NEWLINECHAR, "\n")
def generate(code, max_length=100):
"""Takes input code, replaces newline chars with <N>,
tokenizes, feeds thru model, decodes,
then reformats the newlines back in"""
newlinechar = "<N>"
converted = code.replace("\n", newlinechar)
tokenized = tokenizer.encode(converted, return_tensors='pt').to("cuda")
# resp = model.generate(tokenized, max_length=max_length, num_beams=3, num_return_sequences=3).to("cuda")
resp = model.generate(tokenized, max_length=max_length).to("cuda")
decoded = tokenizer.decode(resp[0])
reformatted = decoded.replace("<N>", "\n")
'''l = len(resp)
reformatted = []
for i in range(0, l):
decoded = tokenizer.decode(resp[i]) + "<N>"
reformatted.append(decoded.replace("<N>", "\n"))
ans = "".join(reformatted)'''
return reformatted
def auto_complete(inp, maxlength=100):
"""
Args:
inp: 用户输入的代码段
maxlength: 处理字符的上限,默认为100
Returns:
模型生成的新代码
"""
try:
model_output = generate(inp, maxlength)
sequence = model_output['sequences'][0]
decoded = decode_newlines(tokenizer.decode(sequence))
return decoded
except TypeError as e:
# print("TypeError:", e)
return ""
def stop_at_repeat(model_out):
lines = model_out.splitlines(True)
no_repeat = ""
for l in lines:
if no_repeat.count(l) == 0 or l == "\n":
no_repeat += l
else:
# print(l)
pass
return no_repeat
def next_line_only(original, model_out):
orig_nl = original.count("\n")
one_more_lines = [l for l in model_out.splitlines(True)][:orig_nl + 1]
one_more_line = "".join(one_more_lines)
return one_more_line
def count(i):
return 2
def postprocess(origin_output):
processed_output = origin_output.replace("N>", "\n")
processed_output = processed_output.replace(">#include", "\n#include")
return processed_output
while True:
mode = input(
"请输入数字,选择启动方式：0、命令行格式（加强版）1、命令行格式 2、后台钩子模式: 3、生成器模式 4、多段预输入模式")
if mode == "0":
try:
inpu = input(">>> ")
m = generate(inpu)
predict = stop_at_repeat(m)
predict = postprocess(predict)
print("Autocompleted:")
print()
print(predict)
except RuntimeError:
pass
except IndexError:
pass
elif mode == "1":
try:
inpu = input(">>> ")
predict = generate(inpu)
predict = postprocess(predict)
print("Autocompleted:")
print()
print(predict)
except RuntimeError:
pass
except IndexError:
pass
elif mode == "2":
try:
with open("keyboard.txt", 'r') as f:
inpu = f.read()
predict = stop_at_repeat(inpu)
print(predict)
except RuntimeError:
pass
sleep(5)
os.system('cls')
elif mode == "3":
try:
inpu = input(">>> ")
m = generate(inpu)
predict = next_line_only(inpu, m)
predict = postprocess(predict)
print("Autocompleted:")
print()
print(predict)
except RuntimeError:
pass
except IndexError:
pass
elif mode == "4":
try:
inpu = '''#include <QtCore/QCoreApplication>
#include <QtCore/QDebug>
int main(int argc, char** argv)
{
QCoreApplication app(argc, argv);'''
predict = generate(inpu)
predict = stop_at_repeat(predict)
predict = postprocess(predict)
print("Autocompleted:")
print()
print(predict)
except RuntimeError:
pass
except IndexError:
pass