| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
# $Id: XWI.pm 226 2006-12-13 15:06:48Z anders $ |
|
2
|
|
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
# Copyright (c) 1996-1998 LUB NetLab, 2002-2006 Anders Ardö |
|
4
|
|
|
|
|
|
|
# |
|
5
|
|
|
|
|
|
|
# See the file LICENCE included in the distribution. |
|
6
|
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
package Combine::XWI; |
|
8
|
|
|
|
|
|
|
|
|
9
|
4
|
|
|
4
|
|
10472
|
use strict; |
|
|
4
|
|
|
|
|
9
|
|
|
|
4
|
|
|
|
|
136
|
|
|
10
|
4
|
|
|
4
|
|
4080
|
use HTML::Entities; |
|
|
4
|
|
|
|
|
29657
|
|
|
|
4
|
|
|
|
|
9287
|
|
|
11
|
|
|
|
|
|
|
|
|
12
|
|
|
|
|
|
|
sub new { |
|
13
|
3
|
|
|
3
|
1
|
87
|
my ($class) = @_; |
|
14
|
3
|
|
|
|
|
20
|
my $self = {}; |
|
15
|
3
|
|
|
|
|
24
|
bless $self, $class; |
|
16
|
3
|
|
|
|
|
42
|
$self->url_reset; |
|
17
|
3
|
|
|
|
|
34
|
$self->heading_reset; |
|
18
|
3
|
|
|
|
|
21
|
$self->link_reset; |
|
19
|
3
|
|
|
|
|
53
|
$self->meta_reset; |
|
20
|
3
|
|
|
|
|
22
|
$self->robot_reset; |
|
21
|
3
|
|
|
|
|
26
|
$self->topic_reset; |
|
22
|
3
|
|
|
|
|
17
|
return $self; |
|
23
|
|
|
|
|
|
|
} |
|
24
|
|
|
|
|
|
|
|
|
25
|
0
|
|
|
0
|
|
0
|
sub DESTROY { |
|
26
|
|
|
|
|
|
|
# print "an XWI object is destroyed" if Combine::Config::GetLoglev() > 2; |
|
27
|
|
|
|
|
|
|
} |
|
28
|
|
|
|
|
|
|
|
|
29
|
|
|
|
|
|
|
sub AUTOLOAD { |
|
30
|
22
|
|
|
22
|
|
1467
|
my ($self, $value) = @_; |
|
31
|
22
|
|
|
|
|
36
|
my $name = $Combine::XWI::AUTOLOAD; |
|
32
|
22
|
|
|
|
|
119
|
$name =~ s/.*://; |
|
33
|
22
|
100
|
|
|
|
60
|
if ($value) { |
|
34
|
17
|
|
|
|
|
38
|
$self->{$name} = $value; |
|
35
|
17
|
|
|
|
|
36
|
return undef; |
|
36
|
|
|
|
|
|
|
} |
|
37
|
|
|
|
|
|
|
else { |
|
38
|
5
|
|
|
|
|
20
|
return $self->{$name}; |
|
39
|
|
|
|
|
|
|
} |
|
40
|
|
|
|
|
|
|
} |
|
41
|
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
sub url_remove { |
|
43
|
0
|
|
|
0
|
1
|
0
|
my ($self,$url) = @_; |
|
44
|
0
|
|
|
|
|
0
|
my ($i,$next); |
|
45
|
0
|
|
|
|
|
0
|
my $index = 1; |
|
46
|
0
|
|
|
|
|
0
|
my $count = $self->{'url_count'}; |
|
47
|
0
|
|
0
|
|
|
0
|
while ( $index <= $count and $self->{"url_$index"} ne $url ) { |
|
48
|
0
|
|
|
|
|
0
|
$index++; |
|
49
|
|
|
|
|
|
|
} |
|
50
|
0
|
0
|
|
|
|
0
|
return undef if $index > $count; |
|
51
|
0
|
|
|
|
|
0
|
for ($i=$index; $i<$count; $i++) { |
|
52
|
0
|
|
|
|
|
0
|
$next = $i + 1; |
|
53
|
0
|
|
|
|
|
0
|
$self->{"url_$i"} = $self->{"url_$next"}; |
|
54
|
|
|
|
|
|
|
} |
|
55
|
0
|
|
|
|
|
0
|
$self->{'url_count'}--; |
|
56
|
|
|
|
|
|
|
} |
|
57
|
|
|
|
|
|
|
|
|
58
|
|
|
|
|
|
|
sub url_reset { |
|
59
|
3
|
|
|
3
|
1
|
9
|
my ($self) = @_; |
|
60
|
3
|
|
|
|
|
68
|
$self->{'url_point'} = 1; |
|
61
|
3
|
|
|
|
|
13
|
$self->{'url_count'} = 0; |
|
62
|
|
|
|
|
|
|
} |
|
63
|
|
|
|
|
|
|
|
|
64
|
|
|
|
|
|
|
sub url_rewind { |
|
65
|
0
|
|
|
0
|
1
|
0
|
my ($self) = @_; |
|
66
|
0
|
|
|
|
|
0
|
$self->{'url_point'} = 1; |
|
67
|
|
|
|
|
|
|
} |
|
68
|
|
|
|
|
|
|
|
|
69
|
|
|
|
|
|
|
sub url_add { |
|
70
|
5
|
|
|
5
|
1
|
35
|
my ($self, $url) = @_; |
|
71
|
5
|
|
|
|
|
17
|
$self->{'url_count'}++; |
|
72
|
5
|
|
|
|
|
19
|
my $point = $self->{'url_count'}; |
|
73
|
5
|
|
|
|
|
16
|
$self->{"url_$point"} = $url; |
|
74
|
5
|
|
|
|
|
15
|
return $self->{'url_count'}; |
|
75
|
|
|
|
|
|
|
} |
|
76
|
|
|
|
|
|
|
|
|
77
|
|
|
|
|
|
|
sub url_get { |
|
78
|
0
|
|
|
0
|
1
|
0
|
my ($self) = @_; |
|
79
|
0
|
|
|
|
|
0
|
my $point = $self->{'url_point'}; |
|
80
|
0
|
0
|
|
|
|
0
|
return undef unless $point <= $self->{'url_count'}; |
|
81
|
0
|
|
|
|
|
0
|
$self->{'url_point'}++; |
|
82
|
0
|
|
|
|
|
0
|
return $self->{"url_$point"}; |
|
83
|
|
|
|
|
|
|
} |
|
84
|
|
|
|
|
|
|
|
|
85
|
|
|
|
|
|
|
sub meta_reset { |
|
86
|
3
|
|
|
3
|
1
|
12
|
my ($self) = @_; |
|
87
|
3
|
|
|
|
|
15
|
$self->{'meta_point'} = 1; |
|
88
|
3
|
|
|
|
|
16
|
$self->{'meta_count'} = 0; |
|
89
|
|
|
|
|
|
|
} |
|
90
|
|
|
|
|
|
|
|
|
91
|
|
|
|
|
|
|
sub meta_rewind { |
|
92
|
1
|
|
|
1
|
1
|
503
|
my ($self) = @_; |
|
93
|
1
|
|
|
|
|
3
|
$self->{'meta_point'} = 1; |
|
94
|
|
|
|
|
|
|
} |
|
95
|
|
|
|
|
|
|
|
|
96
|
|
|
|
|
|
|
sub meta_add { |
|
97
|
9
|
|
|
9
|
1
|
37
|
my ($self, $meta_name, $meta_content) = @_; |
|
98
|
9
|
|
|
|
|
16
|
$self->{'meta_count'}++; |
|
99
|
9
|
|
|
|
|
15
|
my $point = $self->{'meta_count'}; |
|
100
|
9
|
|
|
|
|
30
|
$self->{"meta_" . $point . "_name"} = $meta_name; |
|
101
|
9
|
|
|
|
|
44
|
$self->{"meta_" . $point . "_content" } = |
|
102
|
|
|
|
|
|
|
HTML::Entities::decode_entities($meta_content); |
|
103
|
|
|
|
|
|
|
# special for robots meta-tag |
|
104
|
9
|
50
|
|
|
|
48
|
if ( $meta_name eq "robots" ) { |
|
105
|
0
|
|
|
|
|
0
|
$self->{metarobots} = $meta_content; |
|
106
|
|
|
|
|
|
|
} |
|
107
|
9
|
|
|
|
|
27
|
return $self->{'meta_count'}; |
|
108
|
|
|
|
|
|
|
} |
|
109
|
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
sub meta_get { |
|
111
|
4
|
|
|
4
|
1
|
3262
|
my ($self) = @_; |
|
112
|
4
|
|
|
|
|
11
|
my $point = $self->{'meta_point'}; |
|
113
|
4
|
100
|
|
|
|
21
|
return undef unless $point <= $self->{'meta_count'}; |
|
114
|
3
|
|
|
|
|
6
|
$self->{'meta_point'}++; |
|
115
|
3
|
|
|
|
|
13
|
my $meta_name = $self->{"meta_" . $point . "_name"}; |
|
116
|
3
|
|
|
|
|
12
|
my $meta_content = $self->{"meta_" . $point . "_content"}; |
|
117
|
3
|
|
|
|
|
12
|
return($meta_name, $meta_content); |
|
118
|
|
|
|
|
|
|
} |
|
119
|
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
sub robot_reset { |
|
121
|
3
|
|
|
3
|
1
|
6
|
my ($self) = @_; |
|
122
|
3
|
|
|
|
|
15
|
$self->{'robot_point'} = 1; |
|
123
|
3
|
|
|
|
|
17
|
$self->{'robot_count'} = 0; |
|
124
|
|
|
|
|
|
|
} |
|
125
|
|
|
|
|
|
|
|
|
126
|
|
|
|
|
|
|
sub robot_rewind { |
|
127
|
1
|
|
|
1
|
1
|
13
|
my ($self) = @_; |
|
128
|
1
|
|
|
|
|
4
|
$self->{'robot_point'} = 1; |
|
129
|
|
|
|
|
|
|
} |
|
130
|
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
sub robot_add { |
|
132
|
6
|
|
|
6
|
1
|
29
|
my ($self, $robot_name, $robot_content) = @_; |
|
133
|
6
|
|
|
|
|
11
|
$self->{'robot_count'}++; |
|
134
|
6
|
|
|
|
|
10
|
my $point = $self->{'robot_count'}; |
|
135
|
6
|
|
|
|
|
19
|
$self->{"robot_" . $point . "_name"} = $robot_name; |
|
136
|
6
|
|
|
|
|
14
|
$self->{"robot_" . $point . "_content" } = $robot_content; |
|
137
|
6
|
|
|
|
|
12
|
return $self->{'robot_count'}; |
|
138
|
|
|
|
|
|
|
} |
|
139
|
|
|
|
|
|
|
|
|
140
|
|
|
|
|
|
|
sub robot_get { |
|
141
|
3
|
|
|
3
|
1
|
1893
|
my ($self) = @_; |
|
142
|
3
|
|
|
|
|
8
|
my $point = $self->{'robot_point'}; |
|
143
|
3
|
100
|
|
|
|
205
|
return undef unless $point <= $self->{'robot_count'}; |
|
144
|
2
|
|
|
|
|
3
|
$self->{'robot_point'}++; |
|
145
|
2
|
|
|
|
|
9
|
my $robot_name = $self->{"robot_" . $point . "_name"}; |
|
146
|
2
|
|
|
|
|
6
|
my $robot_content = $self->{"robot_" . $point . "_content"}; |
|
147
|
2
|
|
|
|
|
9
|
return($robot_name, $robot_content); |
|
148
|
|
|
|
|
|
|
} |
|
149
|
|
|
|
|
|
|
|
|
150
|
|
|
|
|
|
|
|
|
151
|
|
|
|
|
|
|
sub topic_reset { |
|
152
|
3
|
|
|
3
|
1
|
15
|
my ($self) = @_; |
|
153
|
3
|
|
|
|
|
16
|
$self->{'topic_point'} = 1; |
|
154
|
3
|
|
|
|
|
12
|
$self->{'topic_count'} = 0; |
|
155
|
|
|
|
|
|
|
} |
|
156
|
|
|
|
|
|
|
|
|
157
|
|
|
|
|
|
|
sub topic_rewind { |
|
158
|
1
|
|
|
1
|
1
|
10
|
my ($self) = @_; |
|
159
|
1
|
|
|
|
|
4
|
$self->{'topic_point'} = 1; |
|
160
|
|
|
|
|
|
|
} |
|
161
|
|
|
|
|
|
|
|
|
162
|
|
|
|
|
|
|
sub topic_add { |
|
163
|
6
|
|
|
6
|
1
|
41
|
my ($self, $topic_cls, $topic_absscore, $topic_relscore, $terms, $algorithm) = @_; |
|
164
|
6
|
|
|
|
|
11
|
$self->{'topic_count'}++; |
|
165
|
6
|
|
|
|
|
9
|
my $point = $self->{'topic_count'}; |
|
166
|
6
|
|
|
|
|
17
|
$self->{"topic_" . $point . "_cls"} = $topic_cls; |
|
167
|
6
|
|
|
|
|
21
|
$self->{"topic_" . $point . "_absscore" } = $topic_absscore; |
|
168
|
6
|
|
|
|
|
18
|
$self->{"topic_" . $point . "_relscore" } = $topic_relscore; |
|
169
|
6
|
|
|
|
|
18
|
$self->{"topic_" . $point . "_terms" } = $terms; |
|
170
|
6
|
|
|
|
|
17
|
$self->{"topic_" . $point . "_algorithm" } = $algorithm; |
|
171
|
6
|
|
|
|
|
13
|
return $self->{'topic_count'}; |
|
172
|
|
|
|
|
|
|
} |
|
173
|
|
|
|
|
|
|
|
|
174
|
|
|
|
|
|
|
sub topic_get { |
|
175
|
3
|
|
|
3
|
1
|
2246
|
my ($self) = @_; |
|
176
|
3
|
|
|
|
|
11
|
my $point = $self->{'topic_point'}; |
|
177
|
3
|
100
|
|
|
|
14
|
return undef unless $point <= $self->{'topic_count'}; |
|
178
|
2
|
|
|
|
|
3
|
$self->{'topic_point'}++; |
|
179
|
2
|
|
|
|
|
10
|
my $topic_cls = $self->{"topic_" . $point . "_cls"}; |
|
180
|
2
|
|
|
|
|
7
|
my $topic_absscore = $self->{"topic_" . $point . "_absscore"}; |
|
181
|
2
|
|
|
|
|
7
|
my $topic_relscore = $self->{"topic_" . $point . "_relscore"}; |
|
182
|
2
|
|
|
|
|
6
|
my $terms = $self->{"topic_" . $point . "_terms"}; |
|
183
|
2
|
|
|
|
|
7
|
my $algorithm = $self->{"topic_" . $point . "_algorithm"}; |
|
184
|
2
|
|
|
|
|
10
|
return($topic_cls, $topic_absscore, $topic_relscore, $terms, $algorithm); |
|
185
|
|
|
|
|
|
|
} |
|
186
|
|
|
|
|
|
|
|
|
187
|
|
|
|
|
|
|
sub xmeta_reset { |
|
188
|
0
|
|
|
0
|
1
|
0
|
my ($self) = @_; |
|
189
|
0
|
|
|
|
|
0
|
$self->{'xmeta_point'} = 1; |
|
190
|
0
|
|
|
|
|
0
|
$self->{'xmeta_count'} = 0; |
|
191
|
|
|
|
|
|
|
} |
|
192
|
|
|
|
|
|
|
|
|
193
|
|
|
|
|
|
|
sub xmeta_rewind { |
|
194
|
0
|
|
|
0
|
1
|
0
|
my ($self) = @_; |
|
195
|
0
|
|
|
|
|
0
|
$self->{'xmeta_point'} = 1; |
|
196
|
|
|
|
|
|
|
} |
|
197
|
|
|
|
|
|
|
|
|
198
|
|
|
|
|
|
|
sub xmeta_add { |
|
199
|
0
|
|
|
0
|
1
|
0
|
my ($self, $meta_name, $meta_content, $meta_scheme, |
|
200
|
|
|
|
|
|
|
$meta_lang, $meta_group) = @_; |
|
201
|
0
|
|
|
|
|
0
|
$self->{'xmeta_count'}++; |
|
202
|
0
|
|
|
|
|
0
|
my $point = $self->{'xmeta_count'}; |
|
203
|
0
|
|
|
|
|
0
|
$self->{"xmeta_" . $point . "_name"} = $meta_name; |
|
204
|
0
|
|
|
|
|
0
|
$self->{"xmeta_" . $point . "_content" } = $meta_content; |
|
205
|
0
|
|
|
|
|
0
|
$self->{"xmeta_" . $point . "_scheme" } = $meta_scheme; |
|
206
|
0
|
|
|
|
|
0
|
$self->{"xmeta_" . $point . "_lang" } = $meta_lang; |
|
207
|
0
|
|
|
|
|
0
|
$self->{"xmeta_" . $point . "_group" } = $meta_group; |
|
208
|
0
|
|
|
|
|
0
|
return $self->{'xmeta_count'}; |
|
209
|
|
|
|
|
|
|
} |
|
210
|
|
|
|
|
|
|
|
|
211
|
|
|
|
|
|
|
sub xmeta_get { |
|
212
|
0
|
|
|
0
|
1
|
0
|
my ($self) = @_; |
|
213
|
0
|
|
|
|
|
0
|
my $point = $self->{'xmeta_point'}; |
|
214
|
0
|
0
|
|
|
|
0
|
return undef unless $point <= $self->{'xmeta_count'}; |
|
215
|
0
|
|
|
|
|
0
|
$self->{'xmeta_point'}++; |
|
216
|
0
|
|
|
|
|
0
|
my $meta_name = $self->{"meta_" . $point . "_name"}; |
|
217
|
0
|
|
|
|
|
0
|
my $meta_content = $self->{"meta_" . $point . "_content"}; |
|
218
|
0
|
|
|
|
|
0
|
my $meta_scheme = $self->{"xmeta_" . $point . "_scheme" }; |
|
219
|
0
|
|
|
|
|
0
|
my $meta_lang = $self->{"xmeta_" . $point . "_lang" }; |
|
220
|
0
|
|
|
|
|
0
|
my $meta_group = $self->{"xmeta_" . $point . "_group" }; |
|
221
|
|
|
|
|
|
|
|
|
222
|
0
|
|
|
|
|
0
|
return ($meta_name, $meta_content, $meta_scheme,$meta_lang, $meta_group); |
|
223
|
|
|
|
|
|
|
} |
|
224
|
|
|
|
|
|
|
|
|
225
|
|
|
|
|
|
|
|
|
226
|
|
|
|
|
|
|
sub heading_reset { |
|
227
|
3
|
|
|
3
|
1
|
10
|
my ($self) = @_; |
|
228
|
3
|
|
|
|
|
18
|
$self->{'heading_point'} = 1; |
|
229
|
3
|
|
|
|
|
17
|
$self->{'heading_count'} = 0; |
|
230
|
|
|
|
|
|
|
} |
|
231
|
|
|
|
|
|
|
|
|
232
|
|
|
|
|
|
|
sub heading_rewind { |
|
233
|
1
|
|
|
1
|
1
|
2
|
my ($self) = @_; |
|
234
|
1
|
|
|
|
|
3
|
$self->{'heading_point'} = 1; |
|
235
|
|
|
|
|
|
|
} |
|
236
|
|
|
|
|
|
|
|
|
237
|
|
|
|
|
|
|
sub heading_add { |
|
238
|
6
|
|
|
6
|
1
|
53
|
my ($self, $heading) = @_; |
|
239
|
6
|
|
|
|
|
12
|
$self->{'heading_count'}++; |
|
240
|
6
|
|
|
|
|
10
|
my $point = $self->{'heading_count'}; |
|
241
|
6
|
|
|
|
|
83
|
$self->{"heading_$point"} = HTML::Entities::decode_entities($heading); |
|
242
|
6
|
|
|
|
|
17
|
return $self->{'heading_count'}; |
|
243
|
|
|
|
|
|
|
} |
|
244
|
|
|
|
|
|
|
|
|
245
|
|
|
|
|
|
|
sub heading_get { |
|
246
|
3
|
|
|
3
|
1
|
1086
|
my ($self) = @_; |
|
247
|
3
|
|
|
|
|
16
|
my $point = $self->{'heading_point'}; |
|
248
|
3
|
100
|
|
|
|
11
|
return undef unless $point <= $self->{'heading_count'}; |
|
249
|
2
|
|
|
|
|
4
|
$self->{'heading_point'}++; |
|
250
|
2
|
|
|
|
|
7
|
return $self->{"heading_$point"}; |
|
251
|
|
|
|
|
|
|
} |
|
252
|
|
|
|
|
|
|
|
|
253
|
|
|
|
|
|
|
sub link_reset { |
|
254
|
3
|
|
|
3
|
1
|
9
|
my ($self) = @_; |
|
255
|
3
|
|
|
|
|
13
|
$self->{'link_point'} = 1; |
|
256
|
3
|
|
|
|
|
14
|
$self->{'link_count'} = 0; |
|
257
|
|
|
|
|
|
|
} |
|
258
|
|
|
|
|
|
|
|
|
259
|
|
|
|
|
|
|
sub link_rewind { |
|
260
|
1
|
|
|
1
|
1
|
5
|
my ($self) = @_; |
|
261
|
1
|
|
|
|
|
3
|
$self->{'link_point'} = 1; |
|
262
|
|
|
|
|
|
|
} |
|
263
|
|
|
|
|
|
|
|
|
264
|
|
|
|
|
|
|
sub link_add { |
|
265
|
6
|
|
|
6
|
1
|
50
|
my ($self, $link_urlstr, $link_netlocid, $link_urlid, $link_text, $link_type) = @_; |
|
266
|
6
|
|
|
|
|
14
|
$self->{'link_count'}++; |
|
267
|
6
|
|
|
|
|
13
|
my $point = $self->{'link_count'}; |
|
268
|
6
|
|
|
|
|
45
|
$self->{"link_" . $point . "_text"} = |
|
269
|
|
|
|
|
|
|
HTML::Entities::decode_entities($link_text); |
|
270
|
6
|
|
|
|
|
34
|
$self->{"link_" . $point . "_urlstr" } = $link_urlstr; |
|
271
|
6
|
|
|
|
|
24
|
$self->{"link_" . $point . "_netlocid" } = $link_netlocid; |
|
272
|
6
|
|
|
|
|
25
|
$self->{"link_" . $point . "_urlid" } = $link_urlid; |
|
273
|
6
|
|
|
|
|
42
|
$self->{"link_" . $point . "_type" } = $link_type; |
|
274
|
6
|
|
|
|
|
18
|
return $self->{'link_count'}; |
|
275
|
|
|
|
|
|
|
} |
|
276
|
|
|
|
|
|
|
|
|
277
|
|
|
|
|
|
|
sub link_get { |
|
278
|
3
|
|
|
3
|
1
|
1445
|
my ($self) = @_; |
|
279
|
3
|
|
|
|
|
6
|
my $point = $self->{'link_point'}; |
|
280
|
3
|
100
|
|
|
|
12
|
return undef unless $point <= $self->{'link_count'}; |
|
281
|
2
|
|
|
|
|
3
|
$self->{'link_point'}++; |
|
282
|
2
|
|
|
|
|
6
|
my $link_text = $self->{"link_" . $point . "_text"}; |
|
283
|
2
|
|
|
|
|
3
|
my $link_urlstr = $self->{"link_" . $point . "_urlstr"}; |
|
284
|
2
|
|
|
|
|
4
|
my $link_netlocid = $self->{"link_" . $point . "_netlocid"}; |
|
285
|
2
|
|
|
|
|
7
|
my $link_urlid = $self->{"link_" . $point . "_urlid"}; |
|
286
|
2
|
|
|
|
|
5
|
my $link_type = $self->{"link_" . $point . "_type"}; |
|
287
|
2
|
|
|
|
|
7
|
return($link_urlstr, $link_netlocid, $link_urlid, $link_text, $link_type); |
|
288
|
|
|
|
|
|
|
} |
|
289
|
|
|
|
|
|
|
|
|
290
|
|
|
|
|
|
|
1; |
|
291
|
|
|
|
|
|
|
|
|
292
|
|
|
|
|
|
|
__END__ |