-
Notifications
You must be signed in to change notification settings - Fork 0
/
data_utils.py
214 lines (182 loc) · 8.05 KB
/
data_utils.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
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
#!/usr/bin/python
#-*-coding:utf-8 -*-
#Version : 1.0
#Filename : data_utils.py
from __future__ import print_function
import csv
import os
import logging
logger = logging.getLogger(__name__)
class Example(object):
def __init__(self, guid, text_a, label=None, meta=None, att=None):
self.guid = guid
self.text_a = text_a
self.text_b = None
self.label = label
self.att = att
self.aux_label = []
if meta is not None:
for no in range(att):
if str(no) in meta:
self.aux_label.append("1")
else:
self.aux_label.append("0")
def __str__(self):
text = ""
text += "guid: {}\n".format(self.guid)
text += "text: {}\n".format(self.text_a)
text += "label: {}".format(self.label)
return text
class DataProcessor(object):
"""Base class for data converters for sequence classification data sets."""
def __init__(self, data_dir, num_labels, num_attrs, label_probs=False):
self.data_dir = data_dir
self.num_labels = num_labels
self.num_attrs = num_attrs
self.label_probs = label_probs
def get_train_examples(self):
"""Gets a collection of `InputExample`s for the train set."""
return self._create_examples(
self._read_tsv(os.path.join(self.data_dir, "train.tsv")))
def get_dev_examples(self):
"""Gets a collection of `InputExample`s for the dev set."""
return self._create_examples(
self._read_tsv(os.path.join(self.data_dir, "dev.tsv")))
def get_test_examples(self, input_file):
"""Gets a collection of `InputExample`s for prediction."""
return self._create_examples(
self._read_tsv(input_file))
def get_labels(self):
"""Gets the list of labels for this data set."""
return [str(i) for i in range(self.num_labels)]
def _read_tsv(self, input_file, quotechar=None):
"""Reads a tab separated value file."""
with open(input_file, "r") as f:
reader = csv.reader((line.replace('\0','') for line in f), delimiter="\t", quotechar=quotechar)
lines = []
for line in reader:
lines.append(line)
return lines
def _create_examples(self, lines):
examples = []
for i, line in enumerate(lines):
if len(line) == 1:
examples.append(Example(i, line[0]))
elif len(line) == 2:
label = line[1].split()
# assert len(label) == self.num_labels, "the number of labels does not match the predicted probs"
if len(label) > 1:
examples.append(Example(i, line[0], [float(l) for l in label]))
else:
examples.append(Example(i, line[0], label[0]))
else:
examples.append(Example(i, line[0], line[1], line[2], self.num_attrs))
return examples
class AG_data(DataProcessor):
@classmethod
def get_ag_data(cls, data_dir):
return cls(data_dir, num_labels=4, num_attrs=5)
class Blog_data(DataProcessor):
@classmethod
def get_blog_data(cls, data_dir):
return cls(data_dir, num_labels=10, num_attrs=2)
class TP_data(DataProcessor):
@classmethod
def get_tp_data(cls, data_dir):
return cls(data_dir, num_labels=5, num_attrs=2)
class TPUK_data(DataProcessor):
@classmethod
def get_tp_data(cls, data_dir):
return cls(data_dir, num_labels=5, num_attrs=2)
class YELP_data(DataProcessor):
@classmethod
def get_yelp_data(cls, data_dir):
return cls(data_dir, num_labels=2, num_attrs=0)
def get_processors(data_dir):
get_data = {"ag": lambda : AG_data.get_ag_data(data_dir),
"blog": lambda : Blog_data.get_blog_data(data_dir),
"tp": lambda : TP_data.get_tp_data(data_dir),
"tpuk": lambda : TPUK_data.get_tp_data(data_dir),
"ag_full": lambda : AG_data.get_ag_data(data_dir),
"yelp": lambda : YELP_data.get_yelp_data(data_dir),
}
return get_data
class InputFeatures(object):
"""
A single set of features of data.
Args:
input_ids: Indices of input sequence tokens in the vocabulary.
attention_mask: Mask to avoid performing attention on padding token indices.
Mask values selected in ``[0, 1]``:
Usually ``1`` for tokens that are NOT MASKED, ``0`` for MASKED (padded) tokens.
token_type_ids: Segment token indices to indicate first and second portions of the inputs.
label: Label corresponding to the input
"""
def __init__(self, guid, input_ids, attention_mask=None, token_type_ids=None, label=None, aux_label=None):
self.guid = guid
self.input_ids = input_ids
self.attention_mask = attention_mask
self.token_type_ids = token_type_ids
self.label = label
self.aux_label = aux_label
def convert_examples_to_features(
examples,
tokenizer,
max_length=512,
label_list=None,
pad_on_left=False,
pad_token=0,
pad_token_segment_id=0,
mask_padding_with_zero=True,
):
label_map = {label: i for i, label in enumerate(label_list)}
aux_label_map = {"0": 0, "1": 1}
features = []
for (ex_index, example) in enumerate(examples):
inputs = tokenizer.encode_plus(example.text_a, example.text_b, add_special_tokens=True, max_length=max_length, truncation=True)
if "token_type_ids" in inputs:
input_ids, token_type_ids = inputs["input_ids"], inputs["token_type_ids"]
else:
input_ids, token_type_ids = inputs["input_ids"], None
# The mask has 1 for real tokens and 0 for padding tokens. Only real
# tokens are attended to.
attention_mask = [1 if mask_padding_with_zero else 0] * len(input_ids)
# Zero-pad up to the sequence length.
padding_length = max_length - len(input_ids)
if pad_on_left:
input_ids = ([pad_token] * padding_length) + input_ids
attention_mask = ([0 if mask_padding_with_zero else 1] * padding_length) + attention_mask
if token_type_ids is not None:
token_type_ids = ([pad_token_segment_id] * padding_length) + token_type_ids
else:
input_ids = input_ids + ([pad_token] * padding_length)
attention_mask = attention_mask + ([0 if mask_padding_with_zero else 1] * padding_length)
if token_type_ids is not None:
token_type_ids = token_type_ids + ([pad_token_segment_id] * padding_length)
assert len(input_ids) == max_length, "Error with input length {} vs {}".format(len(input_ids), max_length)
assert len(attention_mask) == max_length, "Error with input length {} vs {}".format(
len(attention_mask), max_length
)
if token_type_ids is not None:
assert len(token_type_ids) == max_length, "Error with input length {} vs {}".format(
len(token_type_ids), max_length
)
label = label_map[example.label] if type(example.label) == str else example.label
aux_label = [aux_label_map[l] for l in example.aux_label] if example.aux_label is not None else example.aux_label
if ex_index < 5:
logger.info("*** Example ***")
logger.info("guid: %s" % (example.guid))
logger.info("input_ids: %s" % " ".join([str(x) for x in input_ids]))
logger.info("attention_mask: %s" % " ".join([str(x) for x in attention_mask]))
if token_type_ids is not None:
logger.info("token_type_ids: %s" % " ".join([str(x) for x in token_type_ids]))
logger.info("label: {} ".format(label))
logger.info("auxilary label: {} ".format(aux_label))
features.append(
InputFeatures(
guid=example.guid,
input_ids=input_ids, attention_mask=attention_mask, token_type_ids=token_type_ids,
label=label, aux_label=aux_label
)
)
return features