-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathdataset.py
148 lines (129 loc) · 6.16 KB
/
dataset.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
import sys
import torch
import torch.utils.data as data
import torch.nn as nn
import tables
import json
import random
import numpy as np
import pickle
#from utils import PAD_ID, SOS_ID, EOS_ID, UNK_ID, indexes2sent
class Dataset(data.Dataset):
"""
Dataset that has only positive samples.
"""
def __init__(self, data_dir, f_name, max_name_len, f_api, max_api_len,
f_tokens, max_tok_len, f_descs=None, max_desc_len=None):
self.max_name_len=max_name_len
self.max_api_len=max_api_len
self.max_tok_len=max_tok_len
self.max_desc_len=max_desc_len
# 1. Initialize file path or list of file names.
"""read training data(list of int arrays) from a hdf5 file"""
self.training=True
print("loading data...")
table_name = tables.open_file(data_dir+f_name)
self.names = table_name.get_node('/phrases')[:].astype(np.long)
self.idx_names = table_name.get_node('/indices')[:]
table_api = tables.open_file(data_dir+f_api)
self.apis = table_api.get_node('/phrases')[:].astype(np.long)
self.idx_apis = table_api.get_node('/indices')[:]
table_tokens = tables.open_file(data_dir+f_tokens)
self.tokens = table_tokens.get_node('/phrases')[:].astype(np.long)
self.idx_tokens = table_tokens.get_node('/indices')[:]
if f_descs is not None:
self.training=True
table_desc = tables.open_file(data_dir+f_descs)
self.descs = table_desc.get_node('/phrases')[:].astype(np.long)
self.idx_descs = table_desc.get_node('/indices')[:]
assert self.idx_names.shape[0] == self.idx_apis.shape[0]
assert self.idx_apis.shape[0] == self.idx_tokens.shape[0]
if f_descs is not None:
assert self.idx_names.shape[0]==self.idx_descs.shape[0]
self.data_len = self.idx_names.shape[0]
print("{} entries".format(self.data_len))
def pad_seq(self, seq, maxlen):
if len(seq)<maxlen:
# !!!!! numpy appending is slow. Try to optimize the padding
seq=np.append(seq, [0]*(maxlen-len(seq)))
seq=seq[:maxlen]
return seq
def __getitem__(self, offset):
len, pos = self.idx_names[offset]['length'], self.idx_names[offset]['pos']
name_len=min(int(len),self.max_name_len)
name = self.names[pos: pos+name_len]
name = self.pad_seq(name, self.max_name_len)
len, pos = self.idx_apis[offset]['length'], self.idx_apis[offset]['pos']
api_len = min(int(len), self.max_api_len)
apiseq = self.apis[pos:pos+api_len]
apiseq = self.pad_seq(apiseq, self.max_api_len)
len, pos = self.idx_tokens[offset]['length'], self.idx_tokens[offset]['pos']
tok_len = min(int(len), self.max_tok_len)
tokens = self.tokens[pos:pos+tok_len]
tokens = self.pad_seq(tokens, self.max_tok_len)
if self.training:
len, pos = self.idx_descs[offset]['length'], self.idx_descs[offset]['pos']
good_desc_len = min(int(len), self.max_desc_len)
good_desc = self.descs[pos:pos+good_desc_len]
good_desc = self.pad_seq(good_desc, self.max_desc_len)
rand_offset=random.randint(0, self.data_len-1)
len, pos = self.idx_descs[rand_offset]['length'], self.idx_descs[rand_offset]['pos']
bad_desc_len=min(int(len), self.max_desc_len)
bad_desc = self.descs[pos:pos+bad_desc_len]
bad_desc = self.pad_seq(bad_desc, self.max_desc_len)
return name, name_len, apiseq, api_len, tokens, tok_len, good_desc, good_desc_len, bad_desc, bad_desc_len
return name, name_len, apiseq, api_len, tokens, tok_len
def __len__(self):
return self.data_len
def load_dict(filename):
return json.loads(open(filename, "r").readline())
#return pickle.load(open(filename, 'rb'))
def load_vecs(fin):
"""read vectors (2D numpy array) from a hdf5 file"""
h5f = tables.open_file(fin)
h5vecs= h5f.root.vecs
vecs=np.zeros(shape=h5vecs.shape,dtype=h5vecs.dtype)
vecs[:]=h5vecs[:]
h5f.close()
return vecs
def save_vecs(vecs, fout):
fvec = tables.open_file(fout, 'w')
atom = tables.Atom.from_dtype(vecs.dtype)
filters = tables.Filters(complib='blosc', complevel=5)
ds = fvec.create_carray(fvec.root,'vecs', atom, vecs.shape,filters=filters)
ds[:] = vecs
print('done')
fvec.close()
if __name__ == '__main__':
input_dir='./data/github/'
train_set=CodeSearchDataset(input_dir, 'train.name.h5', 6, 'train.apiseq.h5', 20, 'train.tokens.h5', 30, 'train.desc.h5', 30)
train_data_loader=torch.utils.data.DataLoader(dataset=train_set, batch_size=1, shuffle=False, num_workers=1)
use_set=CodeSearchDataset(input_dir, 'use.name.h5', 6, 'use.apiseq.h5', 20, 'use.tokens.h5', 30)
use_data_loader=torch.utils.data.DataLoader(dataset=use_set, batch_size=1, shuffle=False, num_workers=1)
vocab_api = load_dict(input_dir+'vocab.apiseq.json')
vocab_name = load_dict(input_dir+'vocab.name.json')
vocab_tokens = load_dict(input_dir+'vocab.tokens.json')
vocab_desc = load_dict(input_dir+'vocab.desc.json')
print('============ Train Data ================')
k=0
for batch in train_data_loader:
batch = tuple([t.numpy() for t in batch])
name, name_len, apiseq, api_len, tokens, tok_len, good_desc, good_desc_len, bad_desc, bad_desc_len = batch
k+=1
if k>20: break
print('-------------------------------')
print(indexes2sent(name, vocab_name))
print(indexes2sent(apiseq, vocab_api))
print(indexes2sent(tokens, vocab_tokens))
print(indexes2sent(good_desc, vocab_desc))
print('\n\n============ Use Data ================')
k=0
for batch in use_data_loader:
batch = tuple([t.numpy() for t in batch])
name, name_len, apiseq, api_len, tokens, tok_len = batch
k+=1
if k>20: break
print('-------------------------------')
print(indexes2sent(name, vocab_name))
print(indexes2sent(apiseq, vocab_api))
print(indexes2sent(tokens, vocab_tokens))