load_skladnica_frames.py
14.6 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
#-*- coding:utf-8 -*-
# author: B.Niton
import codecs
import operator
import re
from django.core.management.base import BaseCommand
from dictionary.models import *
from common.util import triple_arg_poss
from django.db.models import Q
SKL_FRAMES_PATH = 'data/ramki_skl_v_130925.txt'
NLEMMA_PATH = 'data/brak_frames.txt'
UNDERLINE_PATH = 'data/underline_frames.txt'
class Command(BaseCommand):
help = 'Creates składnica frame models from składnica frames.'
def handle(self, file_path=SKL_FRAMES_PATH, **options):
load_frames()
def update_arg(arg):
arg = arg.replace(u'mian', u'nom')
arg = arg.replace(u'bier', u'acc')
arg = arg.replace(u'cel', u'dat')
arg = arg.replace(u'dop', u'gen')
arg = arg.replace(u'miej', u'loc')
arg = arg.replace(u'narz', u'inst')
arg = arg.replace(u'pop', u'postp')
for atr in Atribute_Value.objects.filter(related=True):
for main_atr in atr.main_attr_values.all():
arg = arg.replace(u'(%s)' % atr.value, u'(%s)' % main_atr.value)
arg = arg.replace(u'(%s,' % atr.value, u'(%s,' % main_atr.value)
arg = arg.replace(u',%s)' % atr.value, u',%s)' % main_atr.value)
arg = arg.replace(u',%s,' % atr.value, u',%s,' % main_atr.value)
return arg
def possible_args(arg, pos, preps):
possibilities = []
if arg == 'subj':
possibilities.extend(triple_arg_poss('np(str)', ['subj']))
elif arg == 'np(bier)':
possibilities.extend(triple_arg_poss('np(str)', ['', 'obj']))
possibilities.extend(triple_arg_poss(update_arg(arg), ['', 'obj', 'subj']))
possibilities.extend(triple_arg_poss('np(part)', ['', 'obj', 'subj']))
elif arg == 'np(dop)':
possibilities.extend(triple_arg_poss(update_arg(arg), ['', 'obj', 'subj']))
possibilities.extend(triple_arg_poss('np(part)', ['', 'obj', 'subj']))
#if pos == 'ger': # przechodzi tez na biernik
possibilities.extend(triple_arg_poss('np(str)', ['', 'obj']))
possibilities.extend(triple_arg_poss('np(acc)', ['', 'obj', 'subj']))
elif (arg.startswith('prepnp(jak,') or arg.startswith('prepnp(jako,') or
arg.startswith(u'prepnp(niż,')):
prepnp_atr_ls = arg.replace('prepnp(', '').replace(')', '').split(',')
preposition = prepnp_atr_ls[0]
case = prepnp_atr_ls[1]
possibilities.extend(triple_arg_poss(update_arg(arg), ['', 'obj', 'subj']))
if case == 'mian' or case == 'bier':
possibilities.extend(triple_arg_poss('prepnp(' + preposition + ',str)', ['', 'obj', 'subj']))
elif(arg == u'prepnp(na temat,dop)'): # nie znajduje tego w wywleczonych
possibilities.extend(triple_arg_poss(u'comprepnp(na temat)', ['', 'obj', 'subj']))
elif(arg == u"prepnp(w sprawie,dop)"):
possibilities.extend(triple_arg_poss(u'comprepnp(w sprawie)', ['', 'obj', 'subj']))
elif(arg == u"prepnp(z powodu,dop)"):
possibilities.extend(triple_arg_poss(u'comprepnp(z powodu)', ['', 'obj', 'subj']))
elif(arg == u"adjp(mian)"):
possibilities.extend(triple_arg_poss(update_arg(arg), ['', 'obj', 'subj']))
possibilities.extend(triple_arg_poss('adjp(pred)', ['', 'obj', 'subj']))
elif(arg == u"adjp(narz)"):
possibilities.extend(triple_arg_poss(update_arg(arg), ['', 'obj', 'subj']))
possibilities.extend(triple_arg_poss('adjp(pred)', ['', 'obj', 'subj']))
elif(arg.startswith(u'sentp')):
# liczy przecinki by sprawdzic liczbe atrybutow w 'sentp'
number_of_commas = arg.count(u',')
conv_arg = arg.replace(u'pz', u'int')
if(number_of_commas == 0):
possibilities.extend(triple_arg_poss(update_arg(conv_arg.replace(u'sentp', u'cp')), ['', 'obj', 'subj']))
elif(number_of_commas == 1):
possibilities.extend(triple_arg_poss(update_arg(conv_arg.replace(u'sentp', u'ncp')), ['', 'obj', 'subj']))
elif(number_of_commas == 2):
possibilities.extend(triple_arg_poss(update_arg(conv_arg.replace(u'sentp', u'prepncp')), ['', 'obj', 'subj']))
elif(arg == u'advp'):
possibilities.extend(triple_arg_poss(u'xp(_)', ['', 'obj', 'subj']))
possibilities.extend(triple_arg_poss(u'advp(pron)', ['', 'obj', 'subj']))
possibilities.extend(triple_arg_poss(u'advp(misc)', ['', 'obj', 'subj']))
possibilities.extend(triple_arg_poss(u'advp(pred)', ['', 'obj', 'subj']))
#possibilities.extend(triple_arg_poss(u'advp(locat)', ['', 'obj', 'subj']))
#possibilities.extend(triple_arg_poss(u'advp(abl)', ['', 'obj', 'subj']))
#possibilities.extend(triple_arg_poss(u'advp(adl)', ['', 'obj', 'subj']))
#possibilities.extend(triple_arg_poss(u'advp(perl)', ['', 'obj', 'subj']))
#possibilities.extend(triple_arg_poss(u'advp(temp)', ['', 'obj', 'subj']))
#possibilities.extend(triple_arg_poss(u'advp(dur)', ['', 'obj', 'subj']))
#possibilities.extend(triple_arg_poss(u'advp(mod)', ['', 'obj', 'subj']))
for prepnp in preps:
q_prepnps = []
possibilities.extend(prepnp['poss_args'])
for prep_arg in prepnp['poss_args']:
q_prepnps.append(Q(realizations=prep_arg))
if q_prepnps:
xps = Argument.objects.filter(type='xp').filter(reduce(operator.or_, q_prepnps)).all()
for xp in xps:
possibilities.extend(triple_arg_poss(xp.text_rep, ['', 'obj', 'subj']))
elif(arg == u'prepnp(przez,bier)'):
possibilities.extend(triple_arg_poss(update_arg(arg), ['', 'obj', 'subj']))
#if pos == 'ger' or pos == 'ppas':
possibilities.extend(triple_arg_poss('np(str)', ['subj']))
elif arg.startswith('infp('):
conv_arg = copy.deepcopy(arg)
conv_arg = conv_arg.replace('(nd)', '(imperf)')
conv_arg = conv_arg.replace('(dk)', '(perf)')
possibilities.extend(triple_arg_poss(update_arg(conv_arg), ['', 'obj', 'subj']))
possibilities.extend(triple_arg_poss('infp(_)', ['', 'obj', 'subj']))
elif(arg == u''):
possibilities.extend(triple_arg_poss('E', ['subj']))
possibilities.extend(triple_arg_poss('np(str)', ['subj']))
else:
possibilities.extend(triple_arg_poss(update_arg(arg), ['', 'obj', 'subj']))
return possibilities
def mark_sub_frames(lemma):
print smart_str(lemma)
for frame in lemma.skladnica_frames.all():
frame.sub_frames.clear()
frame.visible = True
frame.save()
for frame in lemma.skladnica_frames.all():
print frame
bigger_frames = lemma.skladnica_frames.all()
bigger_frames = bigger_frames.filter(reflex=frame.reflex).exclude(pk=frame.pk)
for arg in frame.arguments.all():
bigger_frames = bigger_frames.filter(arguments__text_id=arg.text_id)
if bigger_frames.count() > 0:
frame.visible = False
frame.save()
for bigger_frame in bigger_frames.all():
bigger_frame.sub_frames.add(frame)
for bigger_frame in bigger_frames.all():
if(bigger_frame.arguments.count() == frame.arguments.count()):
if not frame.skladnica_frames.annotate(arguments_count=Count('arguments')).filter(arguments_count__gt=
frame.arguments.count()).count() > 0:
bigger_frame.visible = False
bigger_frame.save()
frame.visible = True
frame.save()
for sub_frame in frame.sub_frames.all():
if sub_frame.arguments.count() < frame.arguments.count():
sub_frame.visible = False
sub_frame.save()
if(frame.old_frame_value == '' and
lemma.skladnica_frames.count() != lemma.skladnica_frames.filter(old_frame_value='').count()):
frame.visible = False
frame.save()
def load_frames():
print 'Be patient, it can take a while.'
# for lemma in Lemma.objects.filter(old=False): ### UWAGA NA TO, POTEM WYWALIC
# for frame in lemma.skladnica_frames.all():
# frame.delete()
try:
f = codecs.open(SKL_FRAMES_PATH, "rt", 'utf-8')
nlemmafile = codecs.open(NLEMMA_PATH, 'wt', 'utf-8')
underfile = codecs.open(UNDERLINE_PATH, 'wt', 'utf-8')
example_source_obj = NKJP_Source.objects.get(source=u'korpus ręcznie anotowany (1.2M segmentów)')
example_opinion_obj = NKJP_Opinion.objects.get(opinion=u'dobry')
try:
for line in f:
line_pattern = re.compile(ur"^([^\s]+)[\s]*([^\s]+)[\s]*((\[[^\]]*\])|(_))[\s]*(\[[^\]]*\])[\s]*\[sent_id='([^']*)'\][\s]*\[text=([^\]]*)\][\s]*$")
m = line_pattern.match(line)
if not m:
print smart_str(line)
continue
if m:
lemma_str = m.group(1).strip()
pos = m.group(2).strip()
frame_str = m.group(3).strip()
prep_str = m.group(6).strip()
sent_id_str = m.group(7).strip()
example_str = m.group(8).strip().strip("'")
print smart_str(lemma_str)
lemmas = Lemma.objects.filter(entry=lemma_str, old=False) # TODO powinno objac tez pewnie lematy ze starych wersji w kontroli zmian
if lemmas.count() == 0:
nlemmafile.write(line.strip() + u'\n')
continue
if frame_str == '_':
underfile.write(line.strip() + u'\n')
frame_str = '[]'
tokens = frame_str.replace('[', '').replace(']', '').split(',')
preps = []
prep_obj = None
if not prep_str == '[]':
for prep in prep_str.replace('[', '').replace(']', '').split(';'):
if not prep.startswith('cat='):
preps.append({'arg': prep,
'poss_args': possible_args(prep, pos, preps),
'poss_positions': []})
if preps:
prep_obj, created = Skladnica_Argument.objects.get_or_create(text_rep=preps[0]['arg'],
text_id=preps[0]['arg'],
prepnp=None)
prep_obj.possible_args.add(*preps[0]['poss_args'])
args_ls = []
arg_str = ''
for tok in tokens:
arg_str += tok.strip() + ','
if (('(' in arg_str and ')' in arg_str) or
(not ('(' in arg_str) and not (')' in arg_str))):
args_ls.append(arg_str.strip().rstrip(','))
arg_str = ''
skladnica_args_ls = []
args_ls.sort()
for arg in args_ls:
possibilities = []
if arg != 'sie':
if prep_obj:
prep_text_rep = prep_obj.text_rep
else:
prep_text_rep = ''
possibilities = possible_args(arg, pos, preps)
if arg == 'advp' and prep_text_rep:
arg_obj, created = Skladnica_Argument.objects.get_or_create(text_rep=arg,
text_id=arg+'_'+prep_text_rep,
prepnp=prep_obj)
else:
arg_obj, created = Skladnica_Argument.objects.get_or_create(text_rep=arg,
text_id=arg,
prepnp=None)
arg_obj.possible_args.add(*possibilities)
skladnica_args_ls.append(arg_obj) #prepnp=prep_obj)
if 'subj' in args_ls:
property_obj = Old_Frame_Property.objects.get(name='V')
else:
property_obj = Old_Frame_Property.objects.get(name='Q')
if 'sie' in args_ls:
reflex = True
else:
reflex = False
nkjp_example_obj = NKJP_Example(sentence=example_str, source=example_source_obj,
comment='', opinion=example_opinion_obj) # nowododany warunek wyszukiwania pole wyszukiwania
nkjp_example_obj.save()
skladnica_frame_obj = Old_Frame(property=property_obj,
reflex=reflex,
example=nkjp_example_obj,
pos_tag=pos,
old_frame_value='+'.join(args_ls),
sent_id=sent_id_str)
skladnica_frame_obj.save()
skladnica_frame_obj.arguments.add(*skladnica_args_ls)
for lemma in lemmas.all():
lemma.skladnica_frames.add(skladnica_frame_obj)
for lemma in Lemma.objects.filter(old=False, entry=u'być'):
mark_sub_frames(lemma)
for old_lemma in Lemma.objects.filter(entry=lemma.entry, old=True):
old_lemma.skladnica_frames.clear()
old_lemma.skladnica_frames.add(*lemma.skladnica_frames.all())
finally:
f.close()
nlemmafile.close()
underfile.close()
except IOError:
return 'Error: Can not work on file %s, check if it exists!' % SKL_FRAMES_PATH