-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathpreprocessor.py
330 lines (279 loc) · 13.2 KB
/
preprocessor.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
# -*- coding: utf-8 -*-
"""
Created on Tue Dec 17 22:19:21 2019
@author: admin
"""
import os,re, json, datetime, pickle
from html.parser import HTMLParser
from pyhanlp import *
NLPTokenizer = JClass('com.hankcs.hanlp.tokenizer.NLPTokenizer')
CustomDictionary = JClass("com.hankcs.hanlp.dictionary.CustomDictionary")
class MyHTMLParser(HTMLParser):
text = []
family = []
family_members = []
chapter = ''
previous_data = None
id_ = 0
text_add = False
def handle_starttag(self, tag, attrs):
if 'p' in tag or 'h2' in tag:
self.text_add = True
pass
def handle_endtag(self, tag):
if 'h2' in tag:
self.chapter = self.text[-1][2]
#print("Encountered an end tag :", tag)
def handle_data(self, data):
if len(data.strip())>0 and self.text_add:
data = re.sub('<(.*?)>', "", data.strip())
self.text.append((self.id_, self.chapter, data))
self.id_ += 1
self.text_add = False
if len(data.strip())>0 and \
data.endswith('家族'):
self.family.append(data.strip())
if data.strip().startswith('——') and len(data.strip())>3 and \
len(self.family)>0:
if len(self.family_members)==0 or \
self.family_members[-1][0] != self.family[-1]:
# new family
self.family_members.append((self.family[-1], self.previous_data.strip()))
self.family_members.append((self.family[-1], data.strip()))
if len(data.strip())>0:
self.previous_data = data
return data
#- body:
class Preprocessor():
def __init__(self, parser, sentence_min_length):
self.time = datetime.datetime.strftime(datetime.datetime.now(),"%Y-%m-%d %H:%M:%S")
self.parser = parser
self.sentence_min_length = sentence_min_length
# 把文章的content变为句子list
def tokenize(self):
# load the filtering list:
if os.path.exists("filter_default"):
with open("filter_default",'r',encoding='utf-8') as f:
filters = f.read().splitlines()
f.close()
else:
filters=[]
# split the whole passage into the sentences
# objects are saved in the .textObj:
tok = []
for item in self.textObj:
try:
para = item['content']
para = re.sub('\u3000', "", para) # 空格
para = re.sub('\xa0', "", para) # 空格
para = re.sub(' ', "", para) # 空格
#para = re.sub('([。！？\?])([^”’])', r"\1\n\2", para) # 单字符断句符
#para = re.sub(';', r"\1\n\2", para) # 单字符断句分号
#para = re.sub('(\.{6})([^”’])', r"\1\n\2", para) # 英文省略号
#para = re.sub('(\…{2})([^”’])', r"\1\n\2", para) # 中文省略号
#para = re.sub('([。！？\?][”’])([^，。！？\?])', r'\1\n\2', para)
# 如果双引号前有终止符，那么双引号才是句子的终点，把分句符\n放到双引号后，注意前面的几句都小心保留了双引号
para = para.rstrip() # 段尾如果有多余的\n就去掉它
# 很多规则中会考虑分号;，但是这里我把它忽略不计，破折号、英文双引号等同样忽略，需要的再做些简单调整即可。
res = []
for x in re.split("\n|；",para):
x = x.strip()
if x in res:
continue
if len(x)>=self.sentence_min_length and \
x not in filters and x not in res: # the threshold is 20 words in total.
res.append(x)
#会打乱顺序
#res = list(set(res))
item['content'] = res
if len(res)>0:
tok += [item]
else:
next
except KeyError as e:
print('SolrError: key-<%s> of <%s> is missing. '%(e,item['id'])) # skip this error (no valid text is pumped into the pipe)
next
#print("Process: %s records are tokenized."%(len(tok)))
self.tokenObj = tok
def EnumPathFiles(self, path):
if not os.path.isdir(path):
print('Error:"',path,'" is not a directory or does not exist.')
return
list_dirs = os.walk(path)
path_file=[]
for root, dirs, files in list_dirs:
for file in files:
if os.path.splitext(file)[1] == '.xhtml':
path_file.append({"title":file, "path":os.path.join(root,file)})
return path_file
def inputData(self,INPUT=None, INPUT_PATH=None):
# make sure the input cotains only textual paragraphs:
# for json format data, please decode it before inputting:
if INPUT_PATH!=None:
path_file = self.EnumPathFiles(INPUT_PATH)
data=[]
for item in path_file:
with open(item.get("path"),'r',encoding="utf-8") as f:
data_ = f.read()
f.close()
self.parser.feed(data_)
for i in self.parser.text:
data.append({'content':i[2],'chapter':i[1],
'title':item.get("title")[0:-6],
'id':i[0]})
#data_= "".join([i[1] for i in self.parser.text])
self.parser.text = []
self.parser.id_ = 0
self.parser.chapter = ""
#data.append({'content':data_,'title':item.get("title")[0:-6],'id':'none'} )
self.textObj = data
return
if isinstance(INPUT,str):
if os.path.isfile(INPUT):
with open(INPUT,'r') as f:
data = f.read().splitlines()
f.close()
# - different types:
if INPUT.endswith('.json') or INPUT.endswith('.jsonl'):
data = [eval(re.sub('NaN','"none"',x)) for x in data]
for i in range(len(data)):
data[i]['id'] = 'none'
elif INPUT.endswith('.txt'):
data = [{'content':x,'title':'none','id':'none'} for x in data]
else:
data = None
print("InputError: can only parse .json or .txt")
else:
data = None
print("InputError: wrong directory.")
elif isinstance(INPUT,list):
data = [x for x in INPUT if len(x)>20] # default threshold:
else:
data = None
print("InputError: wrong directory.")
self.textObj = data
def wordFilter(self,mode='keyword',*words):
if len(words) < 1: # no specific keywords loaded
# by default, use the local keywords list:
if os.path.exists('keywords_default'):
with open('keywords_default','r',encoding='utf-8') as f:
kws = f.read().splitlines()
f.close()
else:
kws =[]
else:
# input: keywords list:
kws = [x for x in words] # serialize the words
# pre-filtering by [filter_log]:
if "filter_log" in os.listdir():
with open("filter_log","r") as f:
flog = f.read().splitlines()
flog = [x.split(">>")[-1] for x in flog]
f.close()
else:
flog = [""]
# filter the logged records:
# two ways of filtering the sentences that we need:
tokens = [x['content'] for x in self.tokenObj] # serialize
snippets = []
# - keywords stats:
if mode == "keyword":
# len(tokens)： the length of all passage
for i in range(0,len(tokens)):
tks = tokens[i]
# tks: [sentence1, sentence2,...]
tks = list(set(tks)) # simply remove the "totally alike" replicates
count = 0 #the number of keywords in this passages
#tk = []
for t in tks:
for k in kws:
if k in t:
count+=1
#tk += [t]
else:
next
print("Process: [%s] in stack."%(self.tokenObj[i]['id']))
snippets += [{"id":self.tokenObj[i]['id'],
"title":self.tokenObj[i]['title'],
"chapter":self.tokenObj[i]['chapter'],
"snip":re.sub('[\r\n\t\xa0]','',x),
'hanlp_tokens': [str(i).split("/")[0] for i in list(NLPTokenizer.segment(re.sub('[\r\n\t\xa0]','',x)))],
'hanlp_pos': [str(i).split("/")[1] for i in list(NLPTokenizer.segment(re.sub('[\r\n\t\xa0]','',x)))]} for x in tks ]
else:
pass # remained: !!!!!
# save in tokens
print("Process: %s tokens are filtered."%(len(snippets)))
self.tokens = snippets
def tokenAsAnnotation(self,path=None,name="sample.txt"):
if path is None:
path = os.getcwd()
if name not in os.listdir():
f = open(name,'w', encoding="utf-8")
f.close()
with open(path+name,'a') as f:
for item in self.tokens:
# convert to .jsonl for prodigy's annotation:
if name.endswith('.json') or name.endswith('.jsonl'):
sn = json.dumps({"text":item['snip'],
"meta":{"source":item['title'],
"chapter":item['chapter'],
'id': item['id']}})
f.write(sn + "\n")
else:
sn = item['snip']
f.write(sn + "\n")
f.close()
else:
if name not in os.listdir(path):
f = open(name,'w', encoding="utf-8")
f.close()
with open(os.path.join(path, name),'w', encoding="utf-8") as f:
for item in self.tokens:
if name.endswith('.json') or name.endswith('.jsonl'):
sn = json.dumps({"text":item['snip'],
'hanlp_tokens': item['hanlp_tokens'],
'hanlp_pos': item['hanlp_pos'],
"meta":{"source":item['title'],
"chapter":item['chapter'],
'id': item['id']}})
f.write(sn + "\n")
else:
sn = item['snip']
f.write(sn + "\n")
f.close()
# -----
# print("""Process: annotation tokens are ready, \n \t see <%s> in [%s]."""%(name,path))
def main():
os.chdir(os.getcwd())
data_path = os.path.join(os.getcwd(),'raw_data')
output_path = os.path.join(os.getcwd(),'preprocessed_data')
#characters = pickle.load(open(output_path + '/characters.pkl','rb'))
with open(output_path +'/literal_vocabulary','r', encoding='utf-8') as f:
lines = f.readlines()
characters = [i.strip().split(":")[1] for i in lines]
tag = [i.strip().split(":")[0] for i in lines]
#characters_att = pickle.load(open(output_path + '/characters_att.pkl','rb'))
for t, cs in zip(tag, characters):
print(cs, t)
if t=='true_entity':
CustomDictionary.insert(cs, t+" 2048")
else:
CustomDictionary.insert(cs, t+" 1024")
with open(data_path + '/stopwords/哈工大停用词表.txt','r',
encoding="utf-8") as f:
stopwords = [s.strip() for s in f.readlines()]
stopwords.append('…')
for s in stopwords:
CustomDictionary.insert(s, "stopwords 1024")
if os.path.exists(output_path)==False:
os.mkdir(output_path)
parser = MyHTMLParser()
sentence_min_length = 5
proc = Preprocessor(parser, sentence_min_length) # solr starts by default:
proc.inputData(INPUT_PATH=data_path)
proc.tokenize()
proc.wordFilter()
proc.tokenAsAnnotation(path=output_path,
name="preprocessed_data.jsonl")
if __name__ == "__main__":
main()