| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package Lingua::YaTeA::TermCandidate; |
|
2
|
5
|
|
|
5
|
|
25
|
use strict; |
|
|
5
|
|
|
|
|
10
|
|
|
|
5
|
|
|
|
|
112
|
|
|
3
|
5
|
|
|
5
|
|
20
|
use warnings; |
|
|
5
|
|
|
|
|
9
|
|
|
|
5
|
|
|
|
|
6329
|
|
|
4
|
|
|
|
|
|
|
# use UNIVERSAL; |
|
5
|
|
|
|
|
|
|
# use Scalar::Util qw(blessed); |
|
6
|
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
our $id = 0; |
|
8
|
|
|
|
|
|
|
our $VERSION=$Lingua::YaTeA::VERSION; |
|
9
|
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
sub new |
|
11
|
|
|
|
|
|
|
{ |
|
12
|
306
|
|
|
306
|
1
|
395
|
my ($class) = @_; |
|
13
|
306
|
|
|
|
|
335
|
my $this; |
|
14
|
306
|
|
|
|
|
668
|
$this->{ID} = $id++; |
|
15
|
306
|
|
|
|
|
450
|
$this->{KEY} = ""; |
|
16
|
306
|
|
|
|
|
444
|
$this->{HEAD} = (); |
|
17
|
306
|
|
|
|
|
430
|
$this->{WORDS} = []; |
|
18
|
306
|
|
|
|
|
469
|
$this->{OCCURRENCES} = []; |
|
19
|
306
|
|
|
|
|
408
|
$this->{RELIABILITY} = (); |
|
20
|
306
|
|
|
|
|
411
|
$this->{TERM_STATUS} = 1; |
|
21
|
306
|
|
|
|
|
547
|
$this->{ORIGINAL_PHRASE} = (); |
|
22
|
|
|
|
|
|
|
# $this->{WEIGHT} = 0; |
|
23
|
306
|
|
|
|
|
492
|
$this->{WEIGHTS} = {}; |
|
24
|
306
|
|
|
|
|
463
|
$this->{ROOT} = (); |
|
25
|
306
|
|
|
|
|
399
|
$this->{MNP_STATUS} = 0; # added by SA 13/02/2009 |
|
26
|
306
|
|
|
|
|
410
|
bless ($this,$class); |
|
27
|
306
|
|
|
|
|
562
|
return $this; |
|
28
|
|
|
|
|
|
|
} |
|
29
|
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
sub setROOT { |
|
31
|
301
|
|
|
301
|
0
|
415
|
my ($this, $ROOT) = @_; |
|
32
|
301
|
|
|
|
|
329
|
push @{$this->{ROOT}}, $ROOT; |
|
|
301
|
|
|
|
|
507
|
|
|
33
|
301
|
|
|
|
|
522
|
return($this->{ROOT}); |
|
34
|
|
|
|
|
|
|
} |
|
35
|
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
sub getROOT { |
|
37
|
0
|
|
|
0
|
0
|
0
|
my ($this) = @_; |
|
38
|
0
|
|
|
|
|
0
|
return($this->{ROOT}); |
|
39
|
|
|
|
|
|
|
} |
|
40
|
|
|
|
|
|
|
|
|
41
|
|
|
|
|
|
|
sub setTermStatus { |
|
42
|
0
|
|
|
0
|
1
|
0
|
my ($this, $status) = @_; |
|
43
|
0
|
|
|
|
|
0
|
$this->{TERM_STATUS} = $status; |
|
44
|
0
|
|
|
|
|
0
|
return($this->{TERM_STATUS}); |
|
45
|
|
|
|
|
|
|
} |
|
46
|
|
|
|
|
|
|
|
|
47
|
|
|
|
|
|
|
sub getTermStatus { |
|
48
|
0
|
|
|
0
|
1
|
0
|
my ($this) = @_; |
|
49
|
0
|
|
|
|
|
0
|
return($this->{TERM_STATUS}); |
|
50
|
|
|
|
|
|
|
} |
|
51
|
|
|
|
|
|
|
|
|
52
|
|
|
|
|
|
|
sub isTerm { |
|
53
|
0
|
|
|
0
|
1
|
0
|
my ($this) = @_; |
|
54
|
|
|
|
|
|
|
|
|
55
|
0
|
|
|
|
|
0
|
return($this->getTermStatus != 0); |
|
56
|
|
|
|
|
|
|
} |
|
57
|
|
|
|
|
|
|
|
|
58
|
|
|
|
|
|
|
sub getLength |
|
59
|
|
|
|
|
|
|
{ |
|
60
|
1210
|
|
|
1210
|
1
|
1470
|
my ($this) = @_; |
|
61
|
1210
|
|
|
|
|
1202
|
return scalar @{$this->getWords}; |
|
|
1210
|
|
|
|
|
1457
|
|
|
62
|
|
|
|
|
|
|
} |
|
63
|
|
|
|
|
|
|
|
|
64
|
|
|
|
|
|
|
sub addWord |
|
65
|
|
|
|
|
|
|
{ |
|
66
|
408
|
|
|
408
|
1
|
599
|
my ($this,$leaf,$words_a) = @_; |
|
67
|
408
|
|
|
|
|
456
|
push @{$this->{WORDS}}, $words_a->[$leaf->getIndex]; |
|
|
408
|
|
|
|
|
742
|
|
|
68
|
|
|
|
|
|
|
} |
|
69
|
|
|
|
|
|
|
|
|
70
|
|
|
|
|
|
|
sub addOccurrence |
|
71
|
|
|
|
|
|
|
{ |
|
72
|
320
|
|
|
320
|
1
|
429
|
my ($this,$occurrence) = @_; |
|
73
|
320
|
100
|
|
|
|
548
|
if($occurrence->isMaximal) |
|
74
|
|
|
|
|
|
|
{ |
|
75
|
1
|
|
|
|
|
2
|
$this->{MNP_STATUS} = 1; # added by SA 13/02/2009:: if at least one occurrence is a MNP, TC is a MNP |
|
76
|
|
|
|
|
|
|
} |
|
77
|
320
|
|
|
|
|
406
|
push @{$this->{OCCURRENCES}}, $occurrence; |
|
|
320
|
|
|
|
|
842
|
|
|
78
|
|
|
|
|
|
|
} |
|
79
|
|
|
|
|
|
|
|
|
80
|
|
|
|
|
|
|
sub addOccurrences |
|
81
|
|
|
|
|
|
|
{ |
|
82
|
66
|
|
|
66
|
1
|
98
|
my ($this,$occurrences_a) = @_; |
|
83
|
66
|
|
|
|
|
90
|
my $occurrence; |
|
84
|
66
|
|
|
|
|
94
|
foreach $occurrence (@$occurrences_a) |
|
85
|
|
|
|
|
|
|
{ |
|
86
|
68
|
|
|
|
|
95
|
$this->addOccurrence($occurrence); |
|
87
|
|
|
|
|
|
|
} |
|
88
|
|
|
|
|
|
|
} |
|
89
|
|
|
|
|
|
|
|
|
90
|
|
|
|
|
|
|
sub getKey |
|
91
|
|
|
|
|
|
|
{ |
|
92
|
1140
|
|
|
1140
|
1
|
1454
|
my ($this) = @_; |
|
93
|
1140
|
|
|
|
|
2955
|
return $this->{KEY}; |
|
94
|
|
|
|
|
|
|
} |
|
95
|
|
|
|
|
|
|
|
|
96
|
|
|
|
|
|
|
sub getID |
|
97
|
|
|
|
|
|
|
{ |
|
98
|
2303
|
|
|
2303
|
1
|
2825
|
my ($this) = @_; |
|
99
|
2303
|
|
|
|
|
4384
|
return $this->{ID}; |
|
100
|
|
|
|
|
|
|
} |
|
101
|
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
sub getMNPStatus |
|
103
|
|
|
|
|
|
|
{ |
|
104
|
240
|
|
|
240
|
1
|
319
|
my ($this) = @_; |
|
105
|
240
|
|
|
|
|
645
|
return $this->{MNP_STATUS}; |
|
106
|
|
|
|
|
|
|
} |
|
107
|
|
|
|
|
|
|
|
|
108
|
|
|
|
|
|
|
sub editKey |
|
109
|
|
|
|
|
|
|
{ |
|
110
|
692
|
|
|
692
|
1
|
969
|
my ($this,$string) = @_; |
|
111
|
692
|
|
|
|
|
1407
|
$this->{KEY} .= $string; |
|
112
|
|
|
|
|
|
|
} |
|
113
|
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
sub setHead |
|
115
|
|
|
|
|
|
|
{ |
|
116
|
240
|
|
|
240
|
1
|
292
|
my ($this) = @_; |
|
117
|
240
|
|
|
|
|
440
|
$this->{HEAD} = $this->searchHead(0); |
|
118
|
|
|
|
|
|
|
} |
|
119
|
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
sub getHead |
|
121
|
|
|
|
|
|
|
{ |
|
122
|
352
|
|
|
352
|
1
|
464
|
my ($this) = @_; |
|
123
|
352
|
|
|
|
|
582
|
return $this->{HEAD}; |
|
124
|
|
|
|
|
|
|
} |
|
125
|
|
|
|
|
|
|
|
|
126
|
|
|
|
|
|
|
sub setWeight |
|
127
|
|
|
|
|
|
|
{ |
|
128
|
240
|
|
|
240
|
1
|
265
|
my $this = shift; |
|
129
|
240
|
|
|
|
|
255
|
my $weight; |
|
130
|
|
|
|
|
|
|
my $weight_name; |
|
131
|
240
|
50
|
|
|
|
297
|
if (scalar(@_) == 2) { |
|
132
|
0
|
|
|
|
|
0
|
$weight_name = shift; |
|
133
|
|
|
|
|
|
|
} else { |
|
134
|
|
|
|
|
|
|
# default weight because it's the first |
|
135
|
240
|
|
|
|
|
266
|
$weight_name = "DDW"; |
|
136
|
|
|
|
|
|
|
} |
|
137
|
240
|
|
|
|
|
260
|
$weight = shift; |
|
138
|
240
|
|
|
|
|
302
|
$this->getWeights->{$weight_name} = $weight; |
|
139
|
|
|
|
|
|
|
|
|
140
|
|
|
|
|
|
|
# $this->{WEIGHT} = $weight; |
|
141
|
|
|
|
|
|
|
} |
|
142
|
|
|
|
|
|
|
|
|
143
|
|
|
|
|
|
|
sub getWeight |
|
144
|
|
|
|
|
|
|
{ |
|
145
|
1324
|
|
|
1324
|
1
|
1518
|
my $this = shift; |
|
146
|
|
|
|
|
|
|
|
|
147
|
1324
|
|
|
|
|
1341
|
my $weight_name; |
|
148
|
1324
|
50
|
|
|
|
1739
|
if (@_) { |
|
149
|
1324
|
|
|
|
|
1477
|
$weight_name = shift; |
|
150
|
|
|
|
|
|
|
} else { |
|
151
|
|
|
|
|
|
|
# default wieght because it's the first |
|
152
|
0
|
|
|
|
|
0
|
$weight_name = "DDW"; |
|
153
|
|
|
|
|
|
|
} |
|
154
|
1324
|
|
|
|
|
1636
|
return($this->getWeights->{$weight_name}); |
|
155
|
|
|
|
|
|
|
|
|
156
|
|
|
|
|
|
|
# return $this->{WEIGHT}; |
|
157
|
|
|
|
|
|
|
} |
|
158
|
|
|
|
|
|
|
|
|
159
|
|
|
|
|
|
|
sub setWeights |
|
160
|
|
|
|
|
|
|
{ |
|
161
|
0
|
|
|
0
|
1
|
0
|
my ($this,$weight) = @_; |
|
162
|
0
|
|
|
|
|
0
|
$this->{WEIGHTS} = $weight; |
|
163
|
|
|
|
|
|
|
} |
|
164
|
|
|
|
|
|
|
|
|
165
|
|
|
|
|
|
|
sub getWeights |
|
166
|
|
|
|
|
|
|
{ |
|
167
|
1566
|
|
|
1566
|
1
|
1937
|
my ($this) = @_; |
|
168
|
1566
|
|
|
|
|
3500
|
return($this->{WEIGHTS}); |
|
169
|
|
|
|
|
|
|
} |
|
170
|
|
|
|
|
|
|
|
|
171
|
|
|
|
|
|
|
sub getWeightNames |
|
172
|
|
|
|
|
|
|
{ |
|
173
|
240
|
|
|
240
|
1
|
337
|
my ($this) = @_; |
|
174
|
240
|
|
|
|
|
291
|
return(keys %{$this->{WEIGHTS}}); |
|
|
240
|
|
|
|
|
828
|
|
|
175
|
|
|
|
|
|
|
} |
|
176
|
|
|
|
|
|
|
|
|
177
|
|
|
|
|
|
|
sub getWords |
|
178
|
|
|
|
|
|
|
{ |
|
179
|
1893
|
|
|
1893
|
1
|
2219
|
my ($this) = @_; |
|
180
|
1893
|
|
|
|
|
3295
|
return $this->{WORDS}; |
|
181
|
|
|
|
|
|
|
} |
|
182
|
|
|
|
|
|
|
|
|
183
|
|
|
|
|
|
|
sub getWord |
|
184
|
|
|
|
|
|
|
{ |
|
185
|
0
|
|
|
0
|
0
|
0
|
my ($this,$index) = @_; |
|
186
|
0
|
|
|
|
|
0
|
return $this->getWords->[$index]; |
|
187
|
|
|
|
|
|
|
} |
|
188
|
|
|
|
|
|
|
|
|
189
|
|
|
|
|
|
|
sub getOccurrences |
|
190
|
|
|
|
|
|
|
{ |
|
191
|
2976
|
|
|
2976
|
1
|
3479
|
my ($this) = @_; |
|
192
|
2976
|
|
|
|
|
5613
|
return $this->{OCCURRENCES}; |
|
193
|
|
|
|
|
|
|
} |
|
194
|
|
|
|
|
|
|
|
|
195
|
|
|
|
|
|
|
sub getOccurrencesNumber |
|
196
|
|
|
|
|
|
|
{ |
|
197
|
0
|
|
|
0
|
0
|
0
|
my ($this) = @_; |
|
198
|
0
|
|
|
|
|
0
|
return scalar @{$this->getOccurrences}; |
|
|
0
|
|
|
|
|
0
|
|
|
199
|
|
|
|
|
|
|
} |
|
200
|
|
|
|
|
|
|
|
|
201
|
|
|
|
|
|
|
sub buildLinguisticInfos |
|
202
|
|
|
|
|
|
|
{ |
|
203
|
240
|
|
|
240
|
1
|
365
|
my ($this,$tagset) = @_; |
|
204
|
240
|
|
|
|
|
524
|
my $if; |
|
205
|
|
|
|
|
|
|
my $pos; |
|
206
|
240
|
|
|
|
|
0
|
my $lf; |
|
207
|
240
|
|
|
|
|
0
|
my $word; |
|
208
|
|
|
|
|
|
|
|
|
209
|
240
|
|
|
|
|
283
|
foreach $word (@{$this->getWords}) |
|
|
240
|
|
|
|
|
365
|
|
|
210
|
|
|
|
|
|
|
{ |
|
211
|
489
|
|
|
|
|
903
|
$if .= $word->getIF . " " ; |
|
212
|
489
|
100
|
|
|
|
832
|
if ($tagset->existTag('PREPOSITIONS',$word->getIF)) |
|
213
|
|
|
|
|
|
|
{ |
|
214
|
45
|
|
|
|
|
72
|
$pos .= $word->getLF . " "; |
|
215
|
|
|
|
|
|
|
} |
|
216
|
|
|
|
|
|
|
else |
|
217
|
|
|
|
|
|
|
{ |
|
218
|
444
|
|
|
|
|
724
|
$pos .= $word->getPOS . " "; |
|
219
|
|
|
|
|
|
|
} |
|
220
|
489
|
|
|
|
|
897
|
$lf .= $word->getLF . " " ; |
|
221
|
|
|
|
|
|
|
} |
|
222
|
240
|
|
|
|
|
1178
|
$if =~ s/\s+$//; |
|
223
|
240
|
|
|
|
|
679
|
$pos =~ s/\s+$//; |
|
224
|
240
|
|
|
|
|
668
|
$lf =~ s/\s+$//; |
|
225
|
240
|
|
|
|
|
743
|
return ($if,$pos,$lf); |
|
226
|
|
|
|
|
|
|
|
|
227
|
|
|
|
|
|
|
} |
|
228
|
|
|
|
|
|
|
|
|
229
|
|
|
|
|
|
|
sub getIF |
|
230
|
|
|
|
|
|
|
{ |
|
231
|
123
|
|
|
123
|
1
|
165
|
my ($this) = @_; |
|
232
|
123
|
|
|
|
|
137
|
my $word; |
|
233
|
|
|
|
|
|
|
my $if; |
|
234
|
123
|
|
|
|
|
132
|
foreach $word (@{$this->getWords}) |
|
|
123
|
|
|
|
|
177
|
|
|
235
|
|
|
|
|
|
|
{ |
|
236
|
386
|
|
|
|
|
641
|
$if .= $word->getIF . " " ; |
|
237
|
|
|
|
|
|
|
} |
|
238
|
123
|
|
|
|
|
528
|
$if =~ s/\s+$//; |
|
239
|
123
|
|
|
|
|
328
|
return $if; |
|
240
|
|
|
|
|
|
|
} |
|
241
|
|
|
|
|
|
|
|
|
242
|
|
|
|
|
|
|
sub getLF |
|
243
|
|
|
|
|
|
|
{ |
|
244
|
112
|
|
|
112
|
1
|
150
|
my ($this) = @_; |
|
245
|
112
|
|
|
|
|
127
|
my $word; |
|
246
|
|
|
|
|
|
|
my $lf; |
|
247
|
112
|
|
|
|
|
129
|
foreach $word (@{$this->getWords}) |
|
|
112
|
|
|
|
|
156
|
|
|
248
|
|
|
|
|
|
|
{ |
|
249
|
361
|
|
|
|
|
548
|
$lf .= $word->getLF . " " ; |
|
250
|
|
|
|
|
|
|
} |
|
251
|
112
|
|
|
|
|
383
|
$lf =~ s/\s+$//; |
|
252
|
112
|
|
|
|
|
275
|
return $lf; |
|
253
|
|
|
|
|
|
|
} |
|
254
|
|
|
|
|
|
|
|
|
255
|
|
|
|
|
|
|
sub getPOS |
|
256
|
|
|
|
|
|
|
{ |
|
257
|
0
|
|
|
0
|
1
|
0
|
my ($this) = @_; |
|
258
|
0
|
|
|
|
|
0
|
my $word; |
|
259
|
|
|
|
|
|
|
my $pos; |
|
260
|
0
|
|
|
|
|
0
|
foreach $word (@{$this->getWords}) |
|
|
0
|
|
|
|
|
0
|
|
|
261
|
|
|
|
|
|
|
{ |
|
262
|
0
|
|
|
|
|
0
|
$pos .= $word->getPOS . " " ; |
|
263
|
|
|
|
|
|
|
} |
|
264
|
0
|
|
|
|
|
0
|
$pos =~ s/\s+$//; |
|
265
|
0
|
|
|
|
|
0
|
return $pos; |
|
266
|
|
|
|
|
|
|
} |
|
267
|
|
|
|
|
|
|
|
|
268
|
|
|
|
|
|
|
sub getFrequency |
|
269
|
|
|
|
|
|
|
{ |
|
270
|
2312
|
|
|
2312
|
1
|
2704
|
my ($this) = @_; |
|
271
|
2312
|
|
|
|
|
2501
|
return scalar @{$this->getOccurrences}; |
|
|
2312
|
|
|
|
|
2887
|
|
|
272
|
|
|
|
|
|
|
} |
|
273
|
|
|
|
|
|
|
|
|
274
|
|
|
|
|
|
|
sub setReliability |
|
275
|
|
|
|
|
|
|
{ |
|
276
|
240
|
|
|
240
|
1
|
360
|
my ($this,$reliability) = @_; |
|
277
|
240
|
|
|
|
|
371
|
$this->{RELIABILITY} = $reliability; |
|
278
|
|
|
|
|
|
|
} |
|
279
|
|
|
|
|
|
|
|
|
280
|
|
|
|
|
|
|
sub getReliability |
|
281
|
|
|
|
|
|
|
{ |
|
282
|
240
|
|
|
240
|
1
|
312
|
my ($this) = @_; |
|
283
|
240
|
|
|
|
|
1073
|
return $this->{RELIABILITY}; |
|
284
|
|
|
|
|
|
|
} |
|
285
|
|
|
|
|
|
|
|
|
286
|
|
|
|
|
|
|
sub getOriginalPhrase |
|
287
|
|
|
|
|
|
|
{ |
|
288
|
0
|
|
|
0
|
1
|
|
my ($this) = @_; |
|
289
|
0
|
|
|
|
|
|
return $this->{ORIGINAL_PHRASE}; |
|
290
|
|
|
|
|
|
|
} |
|
291
|
|
|
|
|
|
|
|
|
292
|
|
|
|
|
|
|
|
|
293
|
|
|
|
|
|
|
|
|
294
|
|
|
|
|
|
|
1; |
|
295
|
|
|
|
|
|
|
|
|
296
|
|
|
|
|
|
|
__END__ |