line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package WWW::Mixi::Scraper::Plugin::ViewBBS;
|
2
|
|
|
|
|
|
|
|
3
|
1
|
|
|
1
|
|
928
|
use strict;
|
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
36
|
|
4
|
1
|
|
|
1
|
|
5
|
use warnings;
|
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
24
|
|
5
|
1
|
|
|
1
|
|
5
|
use WWW::Mixi::Scraper::Plugin;
|
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
8
|
|
6
|
1
|
|
|
1
|
|
6
|
use WWW::Mixi::Scraper::Utils qw( _datetime _uri );
|
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
637
|
|
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
validator {qw(
|
9
|
|
|
|
|
|
|
id is_number
|
10
|
|
|
|
|
|
|
comm_id is_number
|
11
|
|
|
|
|
|
|
comment_count is_number
|
12
|
|
|
|
|
|
|
page is_number_or_all
|
13
|
|
|
|
|
|
|
)};
|
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
sub scrape {
|
16
|
0
|
|
|
0
|
1
|
|
my ($self, $html) = @_;
|
17
|
|
|
|
|
|
|
|
18
|
0
|
|
|
|
|
|
my %scraper;
|
19
|
|
|
|
|
|
|
$scraper{images} = scraper {
|
20
|
0
|
|
|
0
|
|
|
process 'a',
|
21
|
|
|
|
|
|
|
link => '@onClick';
|
22
|
0
|
|
|
|
|
|
process 'a>img',
|
23
|
|
|
|
|
|
|
thumb_link => '@src';
|
24
|
0
|
|
|
|
|
|
result qw( link thumb_link );
|
25
|
0
|
|
|
|
|
|
};
|
26
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
$scraper{topic} = scraper {
|
28
|
0
|
|
|
0
|
|
|
process 'dl[class="bbsList01 bbsDetail"]>dt>span.date',
|
29
|
|
|
|
|
|
|
time => 'TEXT';
|
30
|
0
|
|
|
|
|
|
process 'dl[class="bbsList01 bbsDetail"]>dt>span.titleSpan',
|
31
|
|
|
|
|
|
|
subject => 'TEXT';
|
32
|
0
|
|
|
|
|
|
process 'dd.bbsContent>dl>dt>a',
|
33
|
|
|
|
|
|
|
name => 'TEXT',
|
34
|
|
|
|
|
|
|
name_link => '@href';
|
35
|
0
|
|
|
|
|
|
process 'dd.bbsContent>dl>dd',
|
36
|
|
|
|
|
|
|
description => $self->html_or_text;
|
37
|
0
|
|
|
|
|
|
process 'dd.bbsContent>dl>dd>div.communityPhoto>table>tr>td',
|
38
|
|
|
|
|
|
|
'images[]' => $scraper{images};
|
39
|
0
|
|
|
|
|
|
result qw( time subject description name name_link images link );
|
40
|
0
|
|
|
|
|
|
};
|
41
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
# bbs topic is not an array
|
43
|
0
|
|
|
|
|
|
my $stash = $self->post_process($scraper{topic}->scrape(\$html))->[0];
|
44
|
|
|
|
|
|
|
|
45
|
|
|
|
|
|
|
# XXX: this fails when you test with local files.
|
46
|
|
|
|
|
|
|
# However, this link cannot be extracted from the html,
|
47
|
|
|
|
|
|
|
# at least as of writing this. ugh.
|
48
|
0
|
|
|
|
|
|
$stash->{link} = $self->{uri};
|
49
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
$scraper{comments} = scraper {
|
51
|
0
|
|
|
0
|
|
|
process 'dt>a',
|
52
|
|
|
|
|
|
|
link => '@href',
|
53
|
|
|
|
|
|
|
name => 'TEXT';
|
54
|
0
|
|
|
|
|
|
process 'dd',
|
55
|
|
|
|
|
|
|
description => $self->html_or_text;
|
56
|
0
|
|
|
|
|
|
result qw( link name description );
|
57
|
0
|
|
|
|
|
|
};
|
58
|
|
|
|
|
|
|
|
59
|
|
|
|
|
|
|
$scraper{list} = scraper {
|
60
|
0
|
|
|
0
|
|
|
process 'dl.commentList01>dt[class="commentDate clearfix"]>span.date',
|
61
|
|
|
|
|
|
|
'times[]' => 'TEXT';
|
62
|
0
|
|
|
|
|
|
process 'dl.commentList01>dt[class="commentDate clearfix"]>span.senderId',
|
63
|
|
|
|
|
|
|
'sender_ids[]' => 'TEXT';
|
64
|
0
|
|
|
|
|
|
process 'dl.commentList01>dd>dl.commentContent01',
|
65
|
|
|
|
|
|
|
'comments[]' => $scraper{comments};
|
66
|
0
|
|
|
|
|
|
result qw( times sender_ids comments );
|
67
|
0
|
|
|
|
|
|
};
|
68
|
|
|
|
|
|
|
|
69
|
0
|
|
|
|
|
|
my $stash_c = $self->post_process($scraper{list}->scrape(\$html))->[0];
|
70
|
|
|
|
|
|
|
|
71
|
0
|
0
|
|
|
|
|
my @comments = @{ $stash_c->{comments} || [] };
|
|
0
|
|
|
|
|
|
|
72
|
0
|
0
|
|
|
|
|
my @times = @{ $stash_c->{times} || [] };
|
|
0
|
|
|
|
|
|
|
73
|
0
|
0
|
|
|
|
|
my @sender_ids = @{ $stash_c->{sender_ids} || [] };
|
|
0
|
|
|
|
|
|
|
74
|
0
|
|
|
|
|
|
foreach my $comment ( @comments ) {
|
75
|
0
|
|
|
|
|
|
$comment->{time} = _datetime( shift @times );
|
76
|
0
|
|
|
|
|
|
$comment->{subject} = shift @sender_ids;
|
77
|
|
|
|
|
|
|
|
78
|
|
|
|
|
|
|
# incompatible with WWW::Mixi to let comment links
|
79
|
|
|
|
|
|
|
# look more 'permanent' to make plagger/rss readers happier
|
80
|
0
|
|
|
|
|
|
$comment->{name_link} = _uri( $comment->{link} );
|
81
|
0
|
0
|
|
|
|
|
$comment->{link} = $stash->{link}
|
82
|
|
|
|
|
|
|
? _uri( $stash->{link} . '#' . $comment->{subject} )
|
83
|
|
|
|
|
|
|
: undef;
|
84
|
|
|
|
|
|
|
}
|
85
|
0
|
|
|
|
|
|
$stash->{comments} = \@comments;
|
86
|
|
|
|
|
|
|
|
87
|
0
|
|
|
|
|
|
return $stash;
|
88
|
|
|
|
|
|
|
}
|
89
|
|
|
|
|
|
|
|
90
|
|
|
|
|
|
|
1;
|
91
|
|
|
|
|
|
|
|
92
|
|
|
|
|
|
|
__END__
|