-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathpreprocess.py
199 lines (160 loc) · 6.84 KB
/
preprocess.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
#-*- coding:utf-8 –*-
import json
import h5py
import argparse
import numpy as np
from nltk.tokenize import word_tokenize
import nltk
from unidecode import unidecode
import os
import re
import pdb
import random
import sys
import torch
from itertools import count
parser = argparse.ArgumentParser()
# Input files
parser.add_argument('-input_json', default='train.json', help='Input json file')
parser.add_argument('-output_h5', default='train.bpe.h5', help='Output hdf5 file')
# Options
# parser.add_argument('-vocab', default='/home/xwshi/easymoses_workspace2/reinforcement_learning/nmt/vi-en-baseline/corpus/vi_bpe.vocab.pt', help='path to dataset, now json file')
parser.add_argument('-vocab', default='/home/xwshi/easymoses_workspace2/reinforcement_learning/source_code/demo_data/nhfx/nhfx_bpe.vocab.pt', help=' ')
# parser.add_argument('-filename', default='/home/xwshi/easymoses_workspace2/reinforcement_learning/source_code/demo_data/nmpt.bpe', help=' ')
# parser.add_argument('-filename', default='nmpt.bpe', help=' ')
parser.add_argument('-max_src_len', default=50, type=int, help='Max length of sources')
parser.add_argument('-options_num', default=100, type=int, help='Max length of sources')
parser.add_argument('-max_tgt_len', default=50, type=int, help='Max length of target')
parser.add_argument('-test', action="store_true", help='')
args = parser.parse_args()
def load_from_vocab(vocabfile):
"""
Load Field objects from `vocab.pt` file.
"""
vocab = torch.load(vocabfile)
vocab = dict(vocab)
# print (len(vocab))
print (len(vocab['src']), type(vocab['src']))
print (len(vocab['tgt']), type(vocab['tgt']))
return vocab
def tokenize_data(data):
'''
Tokenize captions, questions and answers
Also maintain word count if required
'''
src_toks, tgt_toks = [], []
# print 'Tokenizing questions...'
for src in data['sources']:
src_tok = src.strip().split()
src_toks.append(src_tok)
# print 'Tokenizing answers...'
for tgt in data['targets']:
tgt_tok = tgt.strip().split()
tgt_toks.append(tgt_tok)
return src_toks, tgt_toks
def encode_vocab(source, target, vocab):
src_idx, tgt_idx = [], []
src_stoi = vocab['src'].stoi
# for k in src_stoi:
# if k not in ['<unk>', '<blank>']:
# src_stoi[k] += 30000
tgt_stoi = vocab['tgt'].stoi
# print("src_stoi['<unk>']", type(src_stoi['<unk>']), src_stoi['<unk>'])
# exit()
for txt in source:
# print("txt", txt)
ind = [src_stoi.get(w, src_stoi['<unk>']) for w in txt]
# print("ind", ind)
# exit()
src_idx.append(ind)
for txt in target:
# print("txt", txt)
ind = [tgt_stoi.get(w, tgt_stoi['<unk>']) for w in txt]
# print("ind", ind)
# exit()
tgt_idx.append(ind)
return src_idx, tgt_idx
def create_mats(data, vocab, src_idx, tgt_idx, params):
# N = len(data['paras'])
# data = data['data']
options_number = params.options_num
N = len(data['paras'])
print ("data size", N)
# exit()
# num_round = 1
max_src_len = params.max_src_len
max_tgt_len = params.max_tgt_len
sources = np.zeros([N, max_src_len], dtype='uint32')
target = np.zeros([N, max_tgt_len], dtype='uint32')
target_index = np.zeros([N], dtype='uint32')
source_len = np.zeros([N], dtype='uint32')
target_len = np.zeros([N], dtype='uint32')
options = np.zeros([N, options_number], dtype='uint32')
idx = 0
for i, pa in enumerate(data['paras']):
for j, para in enumerate(pa['para']):
# print("idx", idx)
# print(para)
# print(para['source'])
src_len = len(src_idx[para['source']][0:max_src_len])
# exit()
source_len[idx] = src_len
sources[idx,:src_len] = src_idx[para['source']][0:src_len]
if not params.test == "test":
tgt_len = len(tgt_idx[para['target']][0:max_tgt_len])
target_len[idx] = tgt_len
target[idx,:tgt_len] = tgt_idx[para['target']][0:tgt_len]
target_index[idx] = para['ref_index']
options[idx][:len(para['target_options'])] = para['target_options']
# print ("options", options[idx])
# print ("target_index", target_index[idx])
idx += 1
print("data size", idx)
options_list = np.zeros([len(tgt_idx), max_tgt_len], dtype='uint32')
options_len = np.zeros(len(tgt_idx), dtype='uint32')
for i, tgt in enumerate(tgt_idx):
options_len[i] = len(tgt[0:max_tgt_len])
options_list[i][0:options_len[i]] = tgt[0:max_tgt_len]
return sources, source_len, target, target_len, options, options_list, options_len, target_index
if __name__ == "__main__":
# print 'Reading json...'
# data_train = json.load(open(args.input_json_train, 'r'))
vocab = load_from_vocab(args.vocab)
# exit()
# src_lines = open(args.filename+'.'+args.src_id, 'r').readlines()
# tgt_lines = open(args.filename+'.'+args.tgt_id, 'r').readlines()
data = json.load(open(args.input_json, 'r'))
split_type = data['split']
print("split_type", split_type)
# src_tok_train, tgt_tok_train = tokenize_data(data_train)
src_tok, tgt_tok = tokenize_data(data['data'])
# print (src_tok[200])
# print (tgt_tok[200])
# print ("src_tok", len(src_tok))
# print ("tgt_tok", len(tgt_tok))
print ('Encoding based on vocabulary...')
# src_idx_train, tgt_idx_train = encode_vocab(src_tok_train, tgt_tok_train, wtoi)
src_idx, tgt_idx = encode_vocab(src_tok, tgt_tok, vocab)
src, src_len, tgt, tgt_len, opt, opt_list, opt_len, tgt_index = create_mats(data['data'], vocab, src_idx, tgt_idx, args)
# src, src_len, tgt, tgt_len, opt, opt_list, opt_len, tgt_index = create_mats(data, src_idx, tgt_idx, args)
test_number = 50
# print("src_tok", src_tok[test_number])
# print("tgt_index", tgt_index[test_number])
# print("opt", opt[test_number])
# print("opt", opt[test_number][tgt_index[test_number]])
# print(tgt_tok[opt[test_number][tgt_index[test_number]]])
# print(tgt_idx[opt[test_number][tgt_index[test_number]]])
# print(tgt_idx[test_number])
# print(opt_list[test_number])
print ('Saving hdf5...')
f = h5py.File(args.output_h5, 'w')
f.create_dataset('src', dtype='uint32', data=src)
f.create_dataset('src_len', dtype='uint32', data=src_len)
f.create_dataset('tgt', dtype='uint32', data=tgt)
f.create_dataset('tgt_len', dtype='uint32', data=tgt_len)
f.create_dataset('tgt_index', dtype='uint32', data=tgt_index)
f.create_dataset('opt', dtype='uint32', data=opt)
f.create_dataset('opt_len', dtype='uint32', data=opt_len)
f.create_dataset('opt_list', dtype='uint32', data=opt_list)
f.close()
#data_toks, src_inds, tgt_inds = encode_vocab(data_toks, src_toks, tgt_toks, word2ind)