/
createdataset.py
509 lines (472 loc) · 17.8 KB
/
createdataset.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
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
from __future__ import division
import math
from random import randint
from random import shuffle
import time
import os
import csv
import sys
import nltk
from nltk.corpus import wordnet as wn
from sklearn.externals import joblib
def read_random_line(f, chunk_size=128): #optimization that is currently not used
import random
with open(f, 'rb') as f_handle:
f_handle.seek(0, os.SEEK_END)
size = f_handle.tell()
i = random.randint(0, size)
while True:
i -= chunk_size
if i < 0:
chunk_size += i
i = 0
f_handle.seek(i, os.SEEK_SET)
chunk = f_handle.read(chunk_size)
i_newline = chunk.rfind(b'\n')
if i_newline != -1:
i += i_newline + 1
break
if i == 0:
break
f_handle.seek(i, os.SEEK_SET)
return f_handle.readline()
def clean_str(c2): #generates the list of utterances from the file
#c2 = c1.read().split('\n')
utterlist = []
for row in c2:
row = row.split('\t')
if row[0] == 'ubotu' or row[0] == 'ubottu' or row[0] == 'ubot3':
return [0,0]
if len("".join(row[3:])) != 0:
utterlist.append("".join(row[3:]))
return utterlist
def getUserList(c2):
userlist = []
for row in c2:
row = row.split('\t')
if row[0] == 'ubotu' or row[0] == 'ubottu' or row[0] == 'ubot3':
return [0,0]
if len("".join(row[3:])) != 0:
userlist.append(row[1])
return userlist
def is_number(s):
try:
float(s)
return True
except ValueError:
return False
def checkValidity(c2, percentage, convo): #checks whether we accept or reject a file
#c2 = c1.read().split('\n')
userlist = []
uniqueuser = {}
for row in c2:
row = row.split('\t')
if len(row)>1:
if len(row[1]) != 0:
userlist.append(row[1])
if row[1] not in uniqueuser:
uniqueuser[row[1]] = 1
else:
uniqueuser[row[1]] += 1
for user,value in uniqueuser.iteritems():
if value < percentage*len(userlist) and len(userlist) >= 1:
return False
self.writeFiles('../deletedfiles.csv', [convo])
return True
def diff_times_in_seconds(t1,t2,date1,date2):
t1 = t1.split(':')
t2 = t2.split(':')
date1 = date1.split('-')
date2 = date2.split('-')
if len(t1)<2 or len(t2)<2 or len(date1)<3 or len(date2)<3:
return 60*60*24 #return 1 day if something goes wrong
if not is_number(t1[0]) or not is_number(t1[1]) or not is_number(t2[0]) or not is_number(t2[1]):
return 60*60*24
if not is_number(date1[0]) or not is_number(date1[1]) or not is_number(date1[2]) or not is_number(date2[0]) or not is_number(date2[1]) or not is_number(date2[2]):
return 60*60*24
h1,m1,s1 = int(t1[0]),int(t1[1]),0#int(t1[2])
h2,m2,s2 = int(t2[0]),int(t2[1]),0#int(t2[2])
d1,mo1,yr1 = int(date1[2]),int(date1[1]),int(date1[0])
d2,mo2,yr2 = int(date2[2]),int(date2[1]),int(date2[0])
t1_secs = s1 + 60*(m1 + 60*(h1 + 24*(d1+ 30*(mo1+12*yr1))))
t2_secs = s2 + 60*(m2 + 60*(h2 + 24*(d2+ 30*(mo2+12*yr2))))
return t2_secs - t1_secs
"""#text = 'Bob \t s1 \n Joe \t s2 \n Joe \t s3 \n Joe \t s3\n Joe \t s3\n Joe \t s3\n Joe \t s3\n Joe \t s3\n Joe \t s3'
path = './dialogs/20/10905.tsv'
with open(path) as c1:
makeTimeArray(c1)
#print checkValidity(c1,0.2)
"""
"""
class NormalizeData:
def __init__(self, readpath, writepath):
self.normdict = {}
self.readpath = readpath
self.folders = [f for f in os.listdir(self.readpath)]
self.writepath = writepath
def createNormDict(self, data):
data = re.split('\n',data)
for row in data:
row = re.split('\t',row)
if row[0] not in self.normdict:
self.normdict[row[0]] = row[1]
def normalize(self, ndata, data, select):
createNormDict(ndata)
for folder in self.folders:
if select == False or int(folder) == 42:
print ' Starting ' + folder + ' folder'
filepath = self.path + folder
files = [f for f in os.listdir(filepath)]
k=0
for convo in files:
k+=1
if k%100 == 0:
print 'Finished ' + str(k) + 'files'
filein = filepath + '/' + convo
with open(filein,'r') as c1:
#c2 = c1.read()
#c3 = re.split('\n',c2)#csv.reader(c1, delimiter='\t')
#utterlist = []
#for row in c3:
# row = re.split('\t',row)
# utterlist.append("".join(row[3:]))
utterlist = clean_str(c1)
for i in xrange(len(utterlist)):
utterlist[i] = re.split(' ',utterlist[i])
for j in xrange(len(utterlist[i])):
if utterlist[i][j] in normdict:
utterlist[i][j] = normdict[utterlist[i][j]]
utterlist[i] = " ".join(utterlist[i])
"""
import os
import re
def preprocess_str(string, TREC=False):
"""
Tokenization/string cleaning for all datasets except for SST.
Every dataset is lower cased except for TREC
"""
#string = re.sub(r"[^A-Za-z0-9(),!?\'\`]", " ", string)
#string = re.sub(r"\'m", " \'m", string)
#string = re.sub(r"\'s", " \'s", string)
#string = re.sub(r"\'ve", " \'ve", string)
#string = re.sub(r"n\'t", " n\'t", string)
#string = re.sub(r"\'re", " \'re", string)
#string = re.sub(r"\'d", " \'d", string)
#string = re.sub(r"\'ll", " \'ll", string)
string = re.sub(r"`", " ` ", string)
string = re.sub(r",", " , ", string)
#string = re.sub(r"!", " ! ", string)
#string = re.sub(r"\(", " \( ", string)
#string = re.sub(r"\)", " \) ", string)
#string = re.sub(r"\?", " \? ", string)
#string = re.sub(r"\s{2,}", " ", string)
return string.strip()# if TREC else string.strip().lower()
os.environ['CLASSPATH']='.:../dialog/libs/commons-lang3-3.4.jar:../dialog/libs'
from jnius import autoclass
Twokenizer = autoclass('cmu.arktweetnlp.Twokenize')
def tokenize(s, tokenizer=Twokenizer()):
s = preprocess_str(s)
tokens = tokenizer.tokenizeRawTweetText(s.decode('utf-8'))
return [tokens.get(i) for i in xrange(tokens.size())]
from django.core.validators import URLValidator
from django.core.exceptions import ValidationError
val = URLValidator(verify_exists=False)
def is_url(s):
try:
val(s)
return True
except ValidationError:
return False
except ImportError:
return False
def replace_sentence(text):
words = tokenize(text)
sent = nltk.pos_tag(words)
chunks = nltk.ne_chunk(sent, binary=False)
sentence = []
nodelist = ['PERSON','ORGANIZATION','GPE','LOCATION','FACILITY','GSP']
for c,word in map(None, chunks, words):
changed = False
if hasattr(c, 'node'):
if c.node in nodelist:
sentence.append("__%s__" % c.node)
changed = True
if not changed:
if is_url(word):
sentence.append("__URL__")
elif is_number(word):
sentence.append("__NUMBER__")
elif os.path.isabs(word):
sentence.append("__PATH__")
else:
sentence.append(word)
return " ".join(sentence)
"""
path1 = './dialogs/20/10800.tsv'
pathwrite = './'
with open(path1,'r') as c1:
c1 = c1.read()
edit = replace_sentence(c1)
#print c1
#print "-----------------------------------------------"
#print edit
sentence = "john is ryan is Kevin is Bob is Steve is Jacob"
print replace_sentence(sentence)
"""
class CreateDataset:
def __init__(self,path):
self.timelist = []
self.turnlist = []
self.traindic = {}
self.valdic = {}
self.testdic = {}
self.filelist = []
self.path = path
self.folders = [f for f in os.listdir(self.path)]
def makeTimeList(self, c2):
#print c1
#c2 = c1.read().split('\n')
firstind = 0
firstval = c2[0].split('\t')[0]
while len(firstval.split('T'))<2:
firstind += 1
firstval = c2[firstind].split('\t')[0]
lastind = -2
lastval = c2[-2].split('\t')[0]
while len(lastval.split('T')) <2:
lastind -= 1
lastval = c2[lastind].split('\t')[0]
#if len(firstval.split('T')) < 2 or len(lastval.split('T')) <2:
# print firstval
# print lastval
firstdate = firstval.split('T')[0]
firsttime = firstval.split('T')[1].split('Z')[0]
lastdate = lastval.split('T')[0]
lasttime = lastval.split('T')[1].split('Z')[0]
timediff = diff_times_in_seconds(firsttime,lasttime,firstdate,lastdate)
self.timelist.append(timediff)
def generateResponses(self, num_responses, convo, testpct):
fakes = []
i = 0
while i < num_responses:
#for i in xrange(num_responses):
if convo in self.traindic:
num = randint(0,int(len(self.filelist)*(1-2*testpct))-1)
fakefile = self.path + self.filelist[num][1] + '/' + self.filelist[num][0]
elif convo in self.valdic:
num = randint(int(len(self.filelist)*(1-2*testpct)),int(len(self.filelist)*(1-testpct))-1)
fakefile = self.path + self.filelist[num][1] + '/' + self.filelist[num][0]
else:
num = randint(int(len(self.filelist)*(1-testpct)),len(self.filelist)-1)
fakefile = self.path + self.filelist[num][1] + '/' + self.filelist[num][0]
#fakes.append(read_random_line(fakefile).split('\t')[3:])
with open(fakefile,'r') as c1:
utterlist = clean_str(c1)
#c = c1.read().split('\n')
#c = c[randint(0,len(c)-1)].split('\t')
#c2 = "".join(c[3:])
#c2 = c2.strip()
c2 = utterlist[randint(0,len(utterlist)-1)]
if isinstance(c2,basestring) == False:
break
c2 = c2.split()
if len(c2) > 1:
fakes.append(c2)
i += 1
return [replace_sentence(s) for s in fakes]
def createDicts(self, testpct, trainfiles = None, valfiles = None, testfiles = None):
print 'Creating dictionary of training, validation, and test sets'
if trainfiles == None:
for folder in self.folders:
if int(folder) > 2:
filepath = self.path + folder
for f in os.listdir(filepath):
self.filelist.append([f, folder])
shuffle(self.filelist)
for i in xrange(int(len(self.filelist)*(1-2*testpct))):
self.traindic[self.filelist[i][0]] = self.filelist[i][1]
self.writeFiles('../trainfiles.csv', [self.filelist[i]])
for i in xrange(int(len(self.filelist)*(1-2*testpct)),int(len(self.filelist)*(1-testpct))):
self.valdic[self.filelist[i][0]] = self.filelist[i][1]
#self.writeFiles('../valfiles.csv', [self.filelist[i]])
for i in xrange(int(len(self.filelist)*(1-testpct)),len(self.filelist)):
self.testdic[self.filelist[i][0]] = self.filelist[i][1]
#self.writeFiles('../testfiles.csv', [self.filelist[i]])
else:
with open(trainfiles,'r') as c1:
c1 = c1.read()
for f,folder in c1:
self.filelist.append([f,folder])
self.traindic[f] = folder
with open(valfiles,'r') as c1:
c1 = c1.read()
for f,folder in c1:
self.filelist.append([f,folder])
self.traindic[f] = folder
with open(testfiles,'r') as c1:
c1 = c1.read()
for f,folder in c1:
self.filelist.append([f,folder])
self.traindic[f] = folder
def writeFiles(self, filename, data, listbool=False):
csvname = self.path + filename
with open(csvname,'a+') as out:
csv_out = csv.writer(out)
for row in data:
if listbool:
for col in row:
csv_out.writerow(col)
else:
csv_out.writerow(row)
def concatUtter(self, utterlist,userlist):
utterlist_new = []
i = 0
while i<len(utterlist):
utter = utterlist[i]
if i == len(utterlist) - 1:
utterlist_new.append(utter)
break
j = i+1
while userlist[i] == userlist[j] and j<len(userlist):
utter = utter + joinsentence + utterlist[j]
j += 1
if j == len(userlist):
break
i = j
utterlist_new.append(utter)
return utterlist_new
def sortFiles(self, max_context_size=20, num_options_train=2, num_options_test=2, testpct=0.1, filesperprint=100, elimpct=0.2, badfiles=False):
self.createDicts(testpct)
print 'Finished dictionaries, making data files'
firstline = [['Context','Response','Correct']]
self.writeFiles('../trainset.csv', firstline)
self.writeFiles('../valset.csv', firstline)
self.writeFiles('../testset.csv', firstline)
trainexamples = 0
testexamples = 0
valexamples = 0
traindata = []
valdata = []
testdata = []
for folder in self.folders:
if int(folder) > 2:
print ' Starting ' + folder + ' folder'
filepath = self.path + folder
files = [f for f in os.listdir(filepath)]
k=0
for convo in files:
#print convo
k+=1
if k%100 == 0:
print 'Finished ' + str(k) + 'files'
#if k % 1000 == 0:
# break
filein = filepath + '/' + convo
with open(filein,'r') as c1:
ctemp = c1
c2 = c1.read().split('\n')
utterlist = clean_str(c2)
userlist = getUserList(c2)
if badfiles: #for making badfiles.csv
utterlist = []
namedict = {}
for row in c2:
row = row.split('\t')
if len("".join(row[3:])) != 0:
utterlist.append("".join(row[3:]))
if len(row) < 4 and len(row[0]) != 0:
namedict['error'] = 0
if len(row) > 3:
if len(row[2]) != 0:
namedict[row[2]] = 0
namedict[row[1]] = 0
if len(row[1]) == 0:
namedict['error'] = 0
if len(namedict) > 2:
self.writeFiles('../badfiles.csv', [[filein]])
utterlist_orig = utterlist
for i in xrange(len(utterlist)): #parses each sentence
utterlist[i] = replace_sentence(utterlist[i])
if checkValidity(c2,elimpct,convo):
#print utterlist
utterlist = self.concatUtter(utterlist,userlist)
if len(utterlist)<3:
#print convo
#print utterlist
self.writeFiles('../badfiles.csv',[[convo]])
else:
if utterlist[0] != utterlist[1]: #checks for ubotu utterance, and for 'good' dialogue
self.turnlist.append(len(utterlist))
self.makeTimeList(c2)
if convo in self.traindic:
for i in xrange(2,len(utterlist) - 1):
context = utterlist[max(0,i - max_context_size):i]
context = joinstring.join(context)
response = utterlist[i]
fakes = self.generateResponses(num_options_train - 1, convo, testpct)
data = [[context, response, 1]]
for fake in fakes:
data.append([context, fake, 0])
traindata.append(data)
else:
#generate a context window size, following the approximate distribution of the training set
contextsize = int((max_context_size*10)/randint(max_context_size/2,max_context_size*10)) + 1 + 1 #last +1 for prediction sentence
if contextsize > len(utterlist):
contextsize = len(utterlist)
for i in xrange(0,int((len(utterlist))/contextsize)):
j = i*contextsize
context = utterlist[j:j + contextsize - 1]
context = joinstring.join(context)
response = utterlist[j + contextsize - 1]
fakes = self.generateResponses(num_options_test - 1, convo, testpct)
data = [[context, response, 1]]
for fake in fakes:
data.append([context, fake, 0])
if convo in self.valdic:
valdata.append(data)
self.writeFiles('../valfiles.csv', [[convo,contextsize-1]])
else:
testdata.append(data)
self.writeFiles('../testfiles.csv', [[convo,contextsize-1]])
if k % filesperprint == 0 or k == len(files):
if traindata != []:
self.writeFiles('../trainset.csv', traindata, listbool=True)
if valdata != []:
self.writeFiles('../valset.csv', valdata, listbool=True)
if testdata != []:
self.writeFiles('../testset.csv', testdata, listbool=True)
traindata = []
valdata = []
testdata = []
self.writeFiles('../timelist.csv',[timelist])
self.writeFiles('../turnlist.csv',[turnlist])
global joinstring
global joinsentence
joinstring = ' __EOS__ '
joinsentence = '. '
data1 = CreateDataset('./dialogs/')
data1.sortFiles()
#data1.sortFiles(20,2,2,0.1,50,False)
"""
def makeTimeList(c2):
#print c1
#c2 = c1.read().split('\n')
print c2
firstval = c2[0].split('\t')[0]
lastval = c2[-2].split('\t')[0]
if len(firstval.split('T')) ==1 or len(lastval.split('T')) ==1:
print firstval
print lastval
firstdate = firstval.split('T')[0]
firsttime = firstval.split('T')[1].split('Z')[0]
lastdate = lastval.split('T')[0]
lasttime = lastval.split('T')[1].split('Z')[0]
timediff = diff_times_in_seconds(firsttime,lasttime,firstdate,lastdate)
print timediff
#self.timelist.append(timediff)
filein = './dialogs/10/1.tsv'
with open(filein,'r') as c1:
c2 = c1.read().split('\n')
makeTimeList(c2)
print checkValidity(c2,0.2,'a')
"""