| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package Lingua::YaTeA::WordOccurrence; |
|
2
|
3
|
|
|
3
|
|
19
|
use strict; |
|
|
3
|
|
|
|
|
8
|
|
|
|
3
|
|
|
|
|
80
|
|
|
3
|
3
|
|
|
3
|
|
16
|
use warnings; |
|
|
3
|
|
|
|
|
6
|
|
|
|
3
|
|
|
|
|
73
|
|
|
4
|
3
|
|
|
3
|
|
401
|
use Lingua::YaTeA::ChunkingDataSet; |
|
|
3
|
|
|
|
|
8
|
|
|
|
3
|
|
|
|
|
26
|
|
|
5
|
3
|
|
|
3
|
|
67
|
use UNIVERSAL; |
|
|
3
|
|
|
|
|
16
|
|
|
|
3
|
|
|
|
|
12
|
|
|
6
|
3
|
|
|
3
|
|
103
|
use Scalar::Util qw(blessed); |
|
|
3
|
|
|
|
|
7
|
|
|
|
3
|
|
|
|
|
1836
|
|
|
7
|
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
our $VERSION=$Lingua::YaTeA::VERSION; |
|
9
|
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
sub new |
|
11
|
|
|
|
|
|
|
{ |
|
12
|
616
|
|
|
616
|
1
|
948
|
my ($class,$form) = @_; |
|
13
|
616
|
|
|
|
|
1062
|
my $this = {}; |
|
14
|
616
|
|
|
|
|
1086
|
bless ($this,$class); |
|
15
|
616
|
|
|
|
|
1109
|
$this->{FORM} = $this->setForm($form); |
|
16
|
616
|
|
|
|
|
1350
|
return $this; |
|
17
|
|
|
|
|
|
|
} |
|
18
|
|
|
|
|
|
|
|
|
19
|
|
|
|
|
|
|
sub setForm |
|
20
|
|
|
|
|
|
|
{ |
|
21
|
616
|
|
|
616
|
1
|
1060
|
my ($this,$form) = @_; |
|
22
|
616
|
|
|
|
|
1579
|
$this->{FORM} = $form; |
|
23
|
|
|
|
|
|
|
} |
|
24
|
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
sub getForm |
|
26
|
|
|
|
|
|
|
{ |
|
27
|
612
|
|
|
612
|
1
|
1070
|
my ($this,$form) = @_; |
|
28
|
612
|
|
|
|
|
2612
|
return $this->{FORM}; |
|
29
|
|
|
|
|
|
|
} |
|
30
|
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
sub isChunkEnd |
|
32
|
|
|
|
|
|
|
{ |
|
33
|
612
|
|
|
612
|
1
|
1359
|
my ($this,$action,$split_after,$sentence_boundary,$document_boundary,$chunking_data) = @_; |
|
34
|
|
|
|
|
|
|
|
|
35
|
612
|
100
|
66
|
|
|
2622
|
if ((blessed($this)) && ($this->isa("Lingua::YaTeA::WordFromCorpus"))) |
|
36
|
|
|
|
|
|
|
{ |
|
37
|
594
|
100
|
|
|
|
1234
|
if ($Lingua::YaTeA::Corpus::tt_counter != 0){ # word is between testified term frontiers: not end |
|
38
|
6
|
|
|
|
|
20
|
return 0; |
|
39
|
|
|
|
|
|
|
} |
|
40
|
588
|
|
|
|
|
863
|
$$split_after--; |
|
41
|
588
|
100
|
|
|
|
1092
|
if($$split_after == 0) |
|
42
|
|
|
|
|
|
|
{ |
|
43
|
2
|
|
|
|
|
9
|
return 1; |
|
44
|
|
|
|
|
|
|
} |
|
45
|
586
|
100
|
66
|
|
|
1327
|
if ( # word is a sentence or document boundary : end |
|
46
|
|
|
|
|
|
|
($this->isSentenceBoundary($sentence_boundary) == 1) |
|
47
|
|
|
|
|
|
|
|| |
|
48
|
|
|
|
|
|
|
($this->isDocumentBoundary($document_boundary) == 1) |
|
49
|
|
|
|
|
|
|
) |
|
50
|
|
|
|
|
|
|
{ |
|
51
|
18
|
|
|
|
|
69
|
return 1; |
|
52
|
|
|
|
|
|
|
} |
|
53
|
|
|
|
|
|
|
# Test if word is a chunking frontier |
|
54
|
568
|
100
|
|
|
|
1327
|
if ($this->isChunkingFrontier($chunking_data)) |
|
55
|
|
|
|
|
|
|
{ |
|
56
|
214
|
50
|
|
|
|
421
|
if($Lingua::YaTeA::Corpus::tt_counter == 0) |
|
57
|
|
|
|
|
|
|
{ |
|
58
|
214
|
|
|
|
|
782
|
return 1; |
|
59
|
|
|
|
|
|
|
} |
|
60
|
|
|
|
|
|
|
} |
|
61
|
|
|
|
|
|
|
|
|
62
|
|
|
|
|
|
|
} |
|
63
|
|
|
|
|
|
|
else{ |
|
64
|
18
|
100
|
66
|
|
|
114
|
if ((blessed($this)) && ($this->isa("Lingua::YaTeA::ForbiddenStructureMark"))) |
|
65
|
|
|
|
|
|
|
{ |
|
66
|
12
|
100
|
|
|
|
43
|
if ($this->isOpener) |
|
67
|
|
|
|
|
|
|
{ |
|
68
|
6
|
|
|
|
|
11
|
$Lingua::YaTeA::Corpus::forbidden_counter ++; |
|
69
|
6
|
|
|
|
|
19
|
$$action = $this->getAction; |
|
70
|
6
|
100
|
|
|
|
20
|
if($$action eq "split") |
|
71
|
|
|
|
|
|
|
{ |
|
72
|
|
|
|
|
|
|
|
|
73
|
2
|
|
|
|
|
12
|
$$split_after = $this->getSplitAfter +1; |
|
74
|
|
|
|
|
|
|
} |
|
75
|
|
|
|
|
|
|
} |
|
76
|
|
|
|
|
|
|
else{ |
|
77
|
6
|
50
|
|
|
|
21
|
if ($this->isCloser) |
|
78
|
|
|
|
|
|
|
{ |
|
79
|
6
|
|
|
|
|
13
|
$Lingua::YaTeA::Corpus::forbidden_counter --; |
|
80
|
6
|
100
|
66
|
|
|
26
|
if (($Lingua::YaTeA::Corpus::tt_counter == 0) && ($this->getAction ne "split")) { |
|
81
|
|
|
|
|
|
|
# warn "==>1\n"; |
|
82
|
4
|
|
|
|
|
20
|
return 1; |
|
83
|
|
|
|
|
|
|
} else { |
|
84
|
2
|
|
|
|
|
9
|
return(0); |
|
85
|
|
|
|
|
|
|
} |
|
86
|
|
|
|
|
|
|
} |
|
87
|
|
|
|
|
|
|
else{ |
|
88
|
0
|
|
|
|
|
0
|
die "erreur\n"; |
|
89
|
|
|
|
|
|
|
} |
|
90
|
|
|
|
|
|
|
} |
|
91
|
|
|
|
|
|
|
} |
|
92
|
|
|
|
|
|
|
else |
|
93
|
|
|
|
|
|
|
{ |
|
94
|
6
|
50
|
33
|
|
|
34
|
if ((blessed($this)) && ($this->isa('Lingua::YaTeA::TestifiedTermMark'))) { |
|
95
|
6
|
100
|
|
|
|
17
|
if ($this->isOpener) |
|
96
|
|
|
|
|
|
|
{ |
|
97
|
3
|
|
|
|
|
7
|
$Lingua::YaTeA::Corpus::tt_counter++; |
|
98
|
|
|
|
|
|
|
} |
|
99
|
|
|
|
|
|
|
else |
|
100
|
|
|
|
|
|
|
{ |
|
101
|
3
|
50
|
|
|
|
9
|
if ($this->isCloser) |
|
102
|
|
|
|
|
|
|
{ |
|
103
|
3
|
|
|
|
|
7
|
$Lingua::YaTeA::Corpus::tt_counter--; |
|
104
|
|
|
|
|
|
|
} |
|
105
|
|
|
|
|
|
|
} |
|
106
|
6
|
|
|
|
|
24
|
return 0; |
|
107
|
|
|
|
|
|
|
} |
|
108
|
|
|
|
|
|
|
} |
|
109
|
|
|
|
|
|
|
} |
|
110
|
|
|
|
|
|
|
|
|
111
|
360
|
100
|
66
|
|
|
1669
|
if( # word is in a forbidden structure but out of testified term frontiers: end |
|
112
|
|
|
|
|
|
|
($Lingua::YaTeA::Corpus::forbidden_counter != 0) |
|
113
|
|
|
|
|
|
|
&& |
|
114
|
|
|
|
|
|
|
($Lingua::YaTeA::Corpus::tt_counter == 0) |
|
115
|
|
|
|
|
|
|
) |
|
116
|
|
|
|
|
|
|
{ |
|
117
|
12
|
100
|
|
|
|
32
|
if ($$action eq "delete") |
|
118
|
|
|
|
|
|
|
{ |
|
119
|
8
|
|
|
|
|
39
|
return 1; |
|
120
|
|
|
|
|
|
|
} |
|
121
|
|
|
|
|
|
|
else{ |
|
122
|
|
|
|
|
|
|
|
|
123
|
|
|
|
|
|
|
# $Lingua::YaTeA::Corpus::split_counter--; |
|
124
|
|
|
|
|
|
|
# if($Lingua::YaTeA::Corpus::split_counter == 0) |
|
125
|
|
|
|
|
|
|
|
|
126
|
|
|
|
|
|
|
#return 0; |
|
127
|
|
|
|
|
|
|
} |
|
128
|
4
|
|
|
|
|
16
|
return 0; |
|
129
|
|
|
|
|
|
|
} |
|
130
|
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
|
|
132
|
|
|
|
|
|
|
} |
|
133
|
|
|
|
|
|
|
|
|
134
|
|
|
|
|
|
|
sub print |
|
135
|
|
|
|
|
|
|
{ |
|
136
|
612
|
|
|
612
|
1
|
1055
|
my ($this,$fh) = @_; |
|
137
|
612
|
50
|
|
|
|
1107
|
if(defined $fh) |
|
138
|
|
|
|
|
|
|
{ |
|
139
|
612
|
|
|
|
|
1270
|
print $fh $this->getForm . "\n"; |
|
140
|
|
|
|
|
|
|
} |
|
141
|
|
|
|
|
|
|
else |
|
142
|
|
|
|
|
|
|
{ |
|
143
|
0
|
|
|
|
|
|
print $this->getForm . "\n"; |
|
144
|
|
|
|
|
|
|
} |
|
145
|
|
|
|
|
|
|
} |
|
146
|
|
|
|
|
|
|
|
|
147
|
|
|
|
|
|
|
1; |
|
148
|
|
|
|
|
|
|
|
|
149
|
|
|
|
|
|
|
__END__ |