forked from ShuyangCao/open-ended_question_ontology
-
Notifications
You must be signed in to change notification settings - Fork 0
/
create_template_data.py
186 lines (160 loc) · 7.64 KB
/
create_template_data.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
import json
import argparse
from collections import defaultdict
from concurrent.futures import ProcessPoolExecutor
from fairseq.data.encoders.gpt2_bpe import GPT2BPE, GPT2BPEConfig
import regex as re
tokenizer = re.compile(r"""'s|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+""")
bpe_encoder = GPT2BPE(GPT2BPEConfig()).bpe
NP_TOKEN = 44320
VERB_TOKEN = 45003
ADJP_TOKEN = 45199
ADVP_TOKEN = 45544
OTHER_TOKEN = 45545
type_words = {
'1': set(),
'2': set(['or']),
'3': set(['mean']),
'5': set(['many', 'much', 'long', 'take', 'get']),
'7': set(['good', 'best', 'find', 'anyone', 'get']),
'8': set(['difference', 'best', 'better', 'and', 'or']),
'9': set(['think', 'would', 'like', 'anyone']),
'10': set(['people']),
'11': set(['happens', 'would', 'affect', 'happen', 'effects', 'effect']),
'13': set(['get', 'way', 'make', 'best', 'know'])
}
def process_one(line, type):
data = json.loads(line)
type_word = type_words[type]
abstracted_words = []
template_bpe_tokens = []
template_fill_tgt_bpe_tokens = []
template_fill_tgt_words = []
for abs_sent in data['abs_sents']:
curr_temp = False
curr_constituency = None
curr_temp_length = 0
for wid, word in enumerate(abs_sent['words']):
curr_token_bpe_tokens = []
if wid != 0:
word_text = ' ' + word['word']
else:
word_text = word['word']
word_text = word_text[0].upper() + word_text[1:]
for token in re.findall(tokenizer, word_text):
token = ''.join(bpe_encoder.byte_encoder[b] for b in token.encode('utf-8'))
curr_token_bpe_tokens.extend([bpe_encoder.encoder[bpe_token] for bpe_token in bpe_encoder.bpe(token).split(' ')])
template_fill_tgt_bpe_tokens.extend(curr_token_bpe_tokens)
template_fill_tgt_words.append(word_text.strip())
if word_text.strip().lower() in type_word:
word['abstract'] = False
word['replaced_constituency'] = None
if wid + 1 < len(abs_sent['words']) and word['pos'] == 'DT':
next_word = abs_sent['words'][wid + 1]['word']
if type == '7' and next_word.lower() in ['good', 'best']:
word['abstract'] = False
word['replaced_constituency'] = None
elif type == '8' and next_word.lower() in ['difference', 'best']:
word['abstract'] = False
word['replaced_constituency'] = None
elif type == '10' and next_word.lower() == 'people':
word['abstract'] = False
word['replaced_constituency'] = None
elif type == '11' and next_word.lower() in ['effects', 'effect']:
word['abstract'] = False
word['replaced_constituency'] = None
# print(curr_temp_length, curr_token_bpe_tokens)
if word['abstract'] and not curr_temp:
curr_temp = True
curr_constituency = word['replaced_constituency']
# assert len(curr_token_bpe_tokens) == 0
curr_temp_length += len(curr_token_bpe_tokens)
if word['replaced_constituency'] is not None:
if curr_constituency[0] == 'NP':
template_bpe_tokens.append(NP_TOKEN)
elif curr_constituency[0] == 'ADJP':
template_bpe_tokens.append(ADJP_TOKEN)
elif curr_constituency[0] == 'ADVP':
template_bpe_tokens.append(ADVP_TOKEN)
else:
raise NotImplementedError
abstracted_words.append('<TEMP-{}>'.format(word['replaced_constituency'][0]))
elif word['pos'] in ['VB', 'VBD', 'VBG', 'VBN', 'VBP', 'VBZ']:
template_bpe_tokens.append(VERB_TOKEN)
curr_constituency = 'VERB'
abstracted_words.append('<TEMP-VERB>')
else:
template_bpe_tokens.append(OTHER_TOKEN)
abstracted_words.append('<TEMP-OTHER>')
elif not word['abstract']:
curr_temp_length = 0
abstracted_words.append(word_text.strip())
template_bpe_tokens.extend(curr_token_bpe_tokens)
curr_temp = False
curr_constituency = None
elif curr_constituency == 'VERB' and word['replaced_constituency'] is None:
if word['pos'] not in ['VB', 'VBD', 'VBG', 'VBN', 'VBP', 'VBZ']:
curr_temp_length = len(curr_token_bpe_tokens)
curr_constituency = None
abstracted_words.append('<TEMP-OTHER>')
template_bpe_tokens.append(OTHER_TOKEN)
else:
curr_temp_length += len(curr_token_bpe_tokens)
elif curr_constituency is None and word['replaced_constituency'] is None \
and word['pos'] in ['VB', 'VBD', 'VBG', 'VBN', 'VBP', 'VBZ']:
curr_temp_length = len(curr_token_bpe_tokens)
curr_constituency = 'VERB'
abstracted_words.append('<TEMP-VERB>')
template_bpe_tokens.append(VERB_TOKEN)
elif curr_constituency != word['replaced_constituency']:
curr_constituency = word['replaced_constituency']
curr_temp_length = len(curr_token_bpe_tokens)
if word['replaced_constituency'] is not None:
abstracted_words.append('<TEMP-{}>'.format(word['replaced_constituency'][0]))
if curr_constituency[0] == 'NP':
template_bpe_tokens.append(NP_TOKEN)
elif curr_constituency[0] == 'ADJP':
template_bpe_tokens.append(ADJP_TOKEN)
elif curr_constituency[0] == 'ADVP':
template_bpe_tokens.append(ADVP_TOKEN)
else:
raise NotImplementedError
elif word['pos'] in ['VB', 'VBD', 'VBG', 'VBN', 'VBP', 'VBZ']:
abstracted_words.append('<TEMP-VERB>')
template_bpe_tokens.append(VERB_TOKEN)
curr_constituency = 'VERB'
else:
template_bpe_tokens.append(OTHER_TOKEN)
abstracted_words.append('<TEMP-OTHER>')
else:
curr_temp_length += len(curr_token_bpe_tokens)
return {
'source': abstracted_words,
'bpe.source': template_bpe_tokens,
'target': template_fill_tgt_words,
'bpe.target': template_fill_tgt_bpe_tokens
}
def main():
parser = argparse.ArgumentParser()
parser.add_argument('jsonl_file')
parser.add_argument('type')
parser.add_argument('output_prefix')
args = parser.parse_args()
with open(args.type) as f:
question_types = [line.strip() for line in f]
with ProcessPoolExecutor() as executor:
futures = []
with open(args.jsonl_file) as f:
for i, line in enumerate(f):
futures.append(executor.submit(process_one, line, question_types[i]))
results = [future.result() for future in futures]
dict_results = defaultdict(list)
for result in results:
for k, v in result.items():
dict_results[k].append(v)
for k, v in dict_results.items():
with open(args.output_prefix + '.' + k, 'w') as f:
for sample_v in v:
f.write(' '.join(str(ele) for ele in sample_v) + '\n')
if __name__ == '__main__':
main()