disamb.py
6.63 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
#!/usr/bin/env python
#-*- coding:utf-8 -*-
import sys
from django.core.management.base import BaseCommand
from common.util import debug, uniprint, uniopen
from dictionary.util import expand_tag
class Command(BaseCommand):
args = '<nazwa pliku wejściowego>'
help = 'Dezambiguacja tagów z Morfologika'
def handle(self, input_file, **options):
parse_file(input_file)
base_tag = {}
form_categories = {
'adj': ['adj'],
'verb': ['subst:ger', 'verb', 'pact', 'ppas', 'pant', 'pcon'],
'subst': ['subst'],
'adv': ['adv'],
}
def all_tags(form, filter_base=False):
tags = []
for tag, base in zip(form['tags'], form['base']):
if not filter_base or base:
tags += expand_tag(tag)
return tags
def base_tag(tag):
pos = tag.split(':')[0]
if tag.startswith('subst:ger'):
return False
if pos in ('num', 'subst', 'adj'):
return 'nom' in tag
if pos == 'verb':
return 'inf' in tag or 'winien' in tag
if pos in ('refl', 'pact', 'ppas'):
return False
return True
# forms: lista {form: forma, tags: lista możliwych tagów)?
def disamb_lexeme(forms):
base_form_tags = (tag for tag, base
in zip(forms[0]['tags'], forms[0]['base'])
if base and base_tag(tag))
possible_pos = set(tag.split(':')[0] for tag in base_form_tags)
entry = forms[0]['form']
if forms[0]['tags'][0] == '-':
debug(entry, u'nierozpoznana forma podstawowa')
return None, None
if len(possible_pos) == 1:
pos = list(possible_pos)[0]
else:
debug(entry, u'niejednoznaczna część mowy')
return None, None
cats = form_categories.get(pos, [pos])
new_forms = []
other_lexemes = []
for form in forms:
new_tags = []
for tag, base in zip(form['tags'], form['base']):
for cat in cats:
if tag.startswith(cat) and base:
new_tags.append(tag)
break
else: # nie pasowało
if pos == 'verb':
tags = tag.split('+')
fixed = None
if tags[0].startswith('subst'):
fixed = ['subst:ger' + tag[len('subst'):] for tag in
tags]
elif tags[0].startswith('adj'):
start = None
if any(form['form'].endswith(end) for end in
[u'cy', u'ca', u'ce',
u'cą', u'cego', u'cej', u'cemu', u'cym',
u'cych', u'cymi']):
start = 'pact'
else:
start = 'ppas'
if start:
fixed = [start + tag[len('adj'):] for tag in tags]
if fixed:
new_tags.append('+'.join(fixed))
if new_tags:
form['tags'] = new_tags
new_forms.append(form)
else:
if pos == 'adj' and 'adv' in (tag.split(':')[0] for tag in
form['tags']):
form['tags'] = [tag for tag in form['tags'] if
tag.startswith('adv')]
other_lexemes.append([form])
nie_prefix = ''
while forms[0]['form'].startswith(nie_prefix):
nie_prefix += 'nie'
if not form['form'].startswith(nie_prefix):
debug(
u'advadj', '%s %s' % (form['form'], forms[0]['form']))
else:
form['tags'] = [pos + ':irreg']
new_forms.append(form)
#debug(u'odrzucona forma', '%s %s [%s]' %
# (form['form'], ', '.join(form['tags']), pos))
# if len(new_forms[0]['tags']) == 1:
# if pos not in base_tag:
# base_tag[pos] = set()
# base_tag[pos].add(new_forms[0]['tags'][0])
if pos == 'subst':
# ujednoznacznić rodzaj... niezguła, sezamek
genders = set(tag[-1][0] for tag in all_tags(new_forms[0]))
if len(genders) == 1:
gender = list(genders)[0]
else:
genders = set(
tag[-1][0] for tag in all_tags(new_forms[0], filter_base=True))
if len(genders) == 1:
gender = list(genders)[0]
else:
good_genders = []
for gender in genders:
for form in new_forms:
for tag in all_tags(form):
if tag[-1][0] in (gender, 'i'):
break # jest
else: # nie ma
break
else: # ok
good_genders.append(gender)
if len(good_genders) != 1:
debug(entry, u'nie da się ujednoznacznić rodzaju')
return None, None
gender = good_genders[0]
# znamy rodzaj, przesiewamy
for form in new_forms:
good_tags = []
for tag in all_tags(form):
if tag[-1][0] in (gender, 'i') or (
tag[-1] == 'depr' and gender == 'm'):
good_tags.append(':'.join(tag))
if good_tags:
form['tags'] = good_tags
else:
form['tags'] = [pos + ':irreg']
return new_forms, other_lexemes
def print_forms(forms):
for form in forms:
for tag in form['tags']:
uniprint('%s\t%s' % (form['form'], tag))
print
def parse_file(path):
forms = []
for line in uniopen(path):
if line.startswith('Processed '):
break
if line == '':
disambiguated, other_lexemes = disamb_lexeme(forms)
if disambiguated:
print_forms(disambiguated)
for l in other_lexemes:
print_forms(l)
forms = []
else:
form, base, tag = line.split('\t')
if not forms:
entry = form
if not forms or form != forms[-1]['form']:
forms.append({'form': form, 'base': [], 'tags': []})
forms[-1]['tags'].append(tag)
forms[-1]['base'].append(
base == entry or tag == 'adv:comp'
or (
tag.startswith(
'subst:pl') and 'nom' in tag)) # brzydko...
# for pos, tags in base_tag.iteritems():
# uniprint('base %s: %s' % (pos, ', '.join(tags)))
if __name__ == '__main__':
import sys
parse_file(sys.argv[1])