clean_texts.py
17.3 KB
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
import argparse
import os
import re
import gcld3
def main():
args = parse_arguments()
if not args.input or not args.output:
print('Error: Input and output must be selected!')
copy_by_year(args.input, args.output)
def parse_arguments():
parser = argparse.ArgumentParser(description='Clean conllup corpora texts from strange sentences.')
parser.add_argument('-o', '--output', help='output directory')
required_arguments = parser.add_argument_group('required arguments')
required_arguments.add_argument('-i', '--input', help='corpora root directory', required=True)
return parser.parse_args()
def copy_by_year(root_directory, out_corpora_directory):
lang_detector = gcld3.NNetLanguageIdentifier(min_num_bytes=0, max_num_bytes=1000)
for root, dirs, files in os.walk(root_directory):
for filename in files:
if filename.endswith('.conllup') or filename.endswith('.conllu'):
src = os.path.join(root, filename)
year = get_year(src)
year_path = os.path.join(out_corpora_directory, year)
os.makedirs(year_path, exist_ok=True)
dst = os.path.join(year_path, filename)
clean_and_save_full(src, dst, year, lang_detector)
def get_year(filepath):
with open(filepath, 'r') as conllup_file:
for line in conllup_file:
line = line.strip()
if is_segment(line):
continue
elif is_metadata(line):
name, value = get_metadata(line)
if name == 'date':
return value.split('-')[0]
return 0
def is_segment(line):
if line and line[0].isdigit():
return True
return False
def is_metadata(line):
if line.startswith('#'):
return True
return False
def get_metadata(line):
name_value_pair = line.split('=', 1)
name = name_value_pair[0].lstrip('#').strip()
value = name_value_pair[1].strip()
return name, value
def is_full_text(filepath):
with open(filepath, 'r') as conllup_file:
for line in conllup_file:
line = line.strip()
if is_segment(line):
continue
elif is_metadata(line):
name, value = get_metadata(line)
if name == 'content_type':
if value.strip() == 'full_text':
return True
else:
return False
return False
def get_first_sentence(filepath):
with open(filepath, 'r') as conllup_file:
for line in conllup_file:
line = line.strip()
if is_segment(line):
continue
elif is_metadata(line):
name, value = get_metadata(line)
if name == 'text':
return value.strip()
return ''
def get_domain(filepath):
doc_id = os.path.basename(filepath).split('.')[0]
if doc_id in ['pl-bn-617666',
'pl-bn-609242',
'pl-bn-609273',
'pl-bn-617396',
'pl-bn-617282',
'pl-bn-617402',
'pl-bn-466202',
'pl-bn-617598']:
return 'Health'
with open(filepath, 'r') as conllup_file:
for line in conllup_file:
line = line.strip()
if is_segment(line):
continue
elif is_metadata(line):
name, value = get_metadata(line)
if name == 'disciplines':
return map_discipline(value.strip())
return 'Other'
def map_discipline(disciplines_val):
disciplines = disciplines_val.split(' | ')
if ('biomedical engineering' in disciplines or 'health sciences' in disciplines or
'medical sciences' in disciplines or 'pharmaceutical sciences' in disciplines or
'physical culture sciences' in disciplines):
return 'Health'
elif 'education' in disciplines:
return 'Education'
elif ('agriculture and gardening' in disciplines or
'forestry sciences' in disciplines or 'veterinary medicine' in disciplines or
'biological sciences' in disciplines or 'zootechnics and fishing' in disciplines):
return 'Nature'
elif 'politics and administration' in disciplines:
return 'Politics'
elif 'economics and finance' in disciplines or 'socio-economic geography and spatial economy' in disciplines:
return 'Economy'
elif ('art sciences' in disciplines or 'fine arts and conservation of works of art' in disciplines or
'science about culture and religion' in disciplines or 'literature' in disciplines or
'philosophy' in disciplines or 'archeology' in disciplines or 'history' in disciplines or
'linguistics' in disciplines):
return 'Culture'
elif ('legal sciences' in disciplines or 'management and quality' in disciplines or
'security sciences' in disciplines or 'social communication and media' in disciplines or
'sociological sciences' in disciplines or 'psychology' in disciplines or 'the canonic law' in disciplines):
return 'Social issues'
elif ('civil engineering and transport' in disciplines or
'environmental engineering, mining and energy' in disciplines or
'automation, electronics and electrical engineering' in disciplines or
'mechanical engineering' in disciplines or
'material engineering' in disciplines or
'architecture and urban planning' in disciplines or
'informatics' in disciplines or
'chemical engineering' in disciplines or
'chemical sciences' in disciplines or
'mathematics' in disciplines or
'physical sciences' in disciplines or
'technical IT and telecommunications' in disciplines or
'food and nutrition technology' in disciplines or
'Earth and the environment sciences' in disciplines):
return 'Science'
return disciplines_val
def clean_and_save_full(src, dst, year, lang_detector):
cleaned_lines = []
paragraph = []
sentence = []
publishing_company = get_publishing_company(src)
journal = get_journal(src)
domain = get_domain(src)
title = get_title(src)
first_sentence_text = get_first_sentence(src)
full_text = is_full_text(src)
article_id = os.path.basename(src).split('.')[0]
if domain == 'Other':
pass
elif first_sentence_text in ['Artykuł nie zawiera streszczenia', 'nie dotyczy', 'Metryczka wydawnicza numeru.',
'Materiały bibliograficzne', 'Brak abstraktu w języku polskim'] and \
not full_text:
pass
elif publishing_company == 'Uniwersytet Jana Długosza w Częstochowie. Wydawnictwo Uniwersytetu Jana Długosza w Częstochowie' and \
journal == 'Edukacja Muzyczna' and year in ['2010', '2011', '2012', '2013', '2014'] and full_text:
pass
elif publishing_company == 'Uniwersytet Jana Długosza w Częstochowie. Wydawnictwo Uniwersytetu Jana Długosza w Częstochowie' and \
journal == 'Prace Naukowe Akademii im. Jana Długosza w Częstochowie. Technika, Informatyka, Inżynieria Bezpieczeństwa' and \
year in ['2013', '2014', '2015', '2016'] and full_text:
pass
elif publishing_company == 'Stowarzyszenie Geodetów Polskich' and \
journal == 'Archiwum Fotogrametrii, Kartografii i Teledetekcji' and \
year in ['2000', '2002', '2004'] and full_text:
pass
elif publishing_company == 'Uniwersytet Śląski' and \
journal == 'Studia Politicae Universitatis Silesiensis' and \
year in ['2006'] and full_text:
pass
elif publishing_company == 'Polskie Towarzystwo Rusycystyczne' and \
journal == 'Przegląd Rusycystyczny' and \
year in ['2010'] and full_text:
pass
elif publishing_company == 'Uniwersytet im. Adama Mickiewicza w Poznaniu' and \
journal == 'Rocznik Integracji Europejskiej' and \
year in ['2010'] and full_text:
pass
elif not get_authors(src):
print('Missing authors:', article_id)
pass
elif not journal:
print('Missing journal:', article_id)
pass
elif not title:
print('Missing title:', article_id)
pass
else:
with open(src, 'r') as conllup_file:
first_sentence = True
for line in conllup_file:
line = line.strip()
if is_segment(line):
sentence.append(line)
elif is_metadata(line):
name, value = get_metadata(line)
if value == '[UNKNOWN]':
pass
elif name == 'newpar id':
if sentence and sentence_is_fine(sentence, lang_detector, first_sentence, full_text, title):
sentence.append('')
paragraph.extend(sentence)
first_sentence = False
sentence = []
if len(paragraph) > 1:
cleaned_lines.extend(paragraph)
paragraph = [line]
elif name == 'sent_id':
if sentence and sentence_is_fine(sentence, lang_detector, first_sentence, full_text, title):
sentence.append('')
paragraph.extend(sentence)
first_sentence = False
sentence = [line]
elif name == 'text':
sentence.append(line)
elif name == 'type':
cleaned_lines.append(line)
cleaned_lines.append(f'# domain = {domain}')
else:
cleaned_lines.append(line)
if sentence and sentence_is_fine(sentence, lang_detector, first_sentence, full_text, title):
sentence.append('')
paragraph.extend(sentence)
if len(paragraph) > 1:
cleaned_lines.extend(paragraph)
if not cleaned_lines[-1]:
cleaned_lines.append('')
contains_polish_signs = False
for line in cleaned_lines:
if is_metadata(line):
name, value = get_metadata(line)
if name == 'text' and ('ą' in value or 'ł' in value):
contains_polish_signs = True
break
if not cleaned_lines[-1].strip():
cleaned_lines.pop()
if len(cleaned_lines) >= 100 and contains_polish_signs:
with open(dst, 'w') as dst_file:
dst_file.write('\n'.join(cleaned_lines))
def get_publishing_company(filepath):
with open(filepath, 'r') as conllup_file:
for line in conllup_file:
line = line.strip()
if is_segment(line):
continue
elif is_metadata(line):
name, value = get_metadata(line)
if name == 'publishing_company':
return value.strip()
return ''
def get_authors(filepath):
with open(filepath, 'r') as conllup_file:
for line in conllup_file:
line = line.strip()
if is_segment(line):
continue
elif is_metadata(line):
name, value = get_metadata(line)
if name == 'authors':
return value.strip()
return ''
def get_journal(filepath):
with open(filepath, 'r') as conllup_file:
for line in conllup_file:
line = line.strip()
if is_segment(line):
continue
elif is_metadata(line):
name, value = get_metadata(line)
if name == 'journal':
return value.strip()
return ''
def get_title(filepath):
with open(filepath, 'r') as conllup_file:
for line in conllup_file:
line = line.strip()
if is_segment(line):
continue
elif is_metadata(line):
name, value = get_metadata(line)
if name == 'title':
return value.strip()
return ''
def get_type(filepath):
with open(filepath, 'r') as conllup_file:
for line in conllup_file:
line = line.strip()
if is_segment(line):
continue
elif is_metadata(line):
name, value = get_metadata(line)
if name == 'type':
return value.strip()
return ''
def get_id(filepath):
with open(filepath, 'r') as conllup_file:
for line in conllup_file:
line = line.strip()
if is_segment(line):
continue
elif is_metadata(line):
name, value = get_metadata(line)
if name == 'newdoc id':
return value.strip()
return ''
def is_bib_start(text):
if re.search('^\d*\s*(Bibliografia|Literatura|LITERATURA)', text):
return True
return False
def sentence_is_fine(sentence_lines, lang_detector, first_sentence, full_text, title):
first_segment = sentence_lines[2]
_, text = get_metadata(sentence_lines[1])
if not full_text and first_sentence and title.startswith(text):
return True
if not all_parentheses_closed(text):
return False
if text[0].islower():
return False
if text[0].isdigit():
return False
if text[-1] not in ['!', '?', '.', '…']:
return False
if first_segment.split()[3] == 'PUNCT' or first_segment.split()[4] == 'interp':
return False
if not correct_root(sentence_lines):
return False
if wrong_start(text):
return False
if unwanted_word(text):
return False
if not contains_verb(sentence_lines):
return False
if to_much_short_words(text):
return False
if to_much_upper(text):
return False
if lang_detector.FindLanguage(text).language != 'pl':
return False
for sign in ['∏', 'ê', 'Ê', '´', 'ç', 'ƒ', '˝', 'à', '¹', 'æ', '¿', '³', 'ñ', 'Ŝ', 'œ', '', '�', 'Œ']:
if sign in text:
return False
return True
def all_parentheses_closed(sequence):
stack = []
opening = set('([{')
closing = set(')]}')
pair = {')': '(', ']': '[', '}': '{'}
for i in sequence:
if i in opening:
stack.append(i)
if i in closing:
if not stack:
return False
elif stack.pop() != pair[i] :
return False
else:
continue
if not stack :
return True
else:
return False
def correct_root(sentence_lines):
root_count = 0
for line in sentence_lines:
if line.strip() and line[0] != '#':
cols = line.split('\t')
if 'root' == cols[7]:
root_count += 1
if cols[6] != '0':
return False
if root_count == 1:
return True
return False
def to_much_upper(text):
upper_count = 0
for sign in text:
if sign.isupper():
upper_count += 1
if float(upper_count)/float(len(text)) >= 0.2:
return True
return False
def to_much_short_words(text):
short_count = 0
words = text.split()
for word in words:
if len(word) == 1:
short_count += 1
if float(short_count)/float(len(words)) >= 0.5:
return True
return False
def unwanted_word(text):
words = text.split()
for word in words:
if len(word) > 1 and (word.startswith('-') or word.endswith('-') or word.startswith('‑') or word.endswith('‑')):
return True
elif word in ['(red.)', 'Wydawnictwo', 'wyd.', 'grantu', 'issue', 'volume', 'Journal', 'journal',
'STRESZCZENIE:', 'KLUCZOWE:', '(bud.', 'Druk.']:
return True
elif re.search('^\d\d\d\d:', word):
return True
return False
def wrong_start(text):
if re.search('^([Rr]y[sc]\.?\s+\d|Rysunek\s+\d\.|Tabela\s+\d|Tab\.?\s+\d|Recenzował:|Słowa\s+kluczowe:|Źródło:|'
'STRESZCZENIE:|Vol\.\s+\d|Streszczenie:|Dane\s+autorów:|WSTĘP|Streszczenie\s+[A-ZĄĆĘŁŃÓŚŹŻ]|'
'Key\s+words:|Keywords:|Explanation:|Pobrano\s+z:|E-mail:|[XIV]+\.|Zeszyt\s+|Roczniki\s+|Literatura:|'
'Wstęp\s+|Lp\.\s+|A b s tr a k t\s+|Adres\s+do\s+korespondencji:|Zesz\.\s+|Przegl\.|Zagłęb\.|'
'WęWęW\s+giel|Retyk\s+i\s+lias|Górny\s+dogger|Pobrane\s+z\s+czasopisma\s+|ORCID:|'
'Autor\s+korespondujący:|Materiały\s+konf\.)',
text):
return True
return False
def contains_verb(sentence_lines):
for line in sentence_lines:
if line.strip() and line[0] != '#':
cols = line.split('\t')
if cols[3] in ['VERB', 'AUX']:
return True
return False
def text_hard_error(sentence_lines):
_, text = get_metadata(sentence_lines[1])
for sign in ['∏', 'ê', 'Ê', '´', 'ç', 'ƒ', '˝', 'à', '¹', 'æ', '¿', '³', 'ñ', 'Ŝ', 'œ', '']:
if sign in text:
return True
return False
if __name__ == '__main__':
main()