1 |
dpavlin |
2 |
#!/usr/bin/perl -w |
2 |
|
|
|
3 |
|
|
use strict; |
4 |
|
|
|
5 |
|
|
use CGI::Simple; |
6 |
dpavlin |
5 |
use CGI::Carp qw(fatalsToBrowser warningsToBrowser); |
7 |
dpavlin |
2 |
use Search::Estraier; |
8 |
dpavlin |
5 |
use YAML::Syck; |
9 |
dpavlin |
15 |
use JSON::Syck; |
10 |
dpavlin |
5 |
use Data::Dump qw/dump/; |
11 |
dpavlin |
2 |
|
12 |
|
|
my $q = new CGI::Simple; |
13 |
|
|
print qq{Content-type: text/html\n\r\n\r}; |
14 |
|
|
|
15 |
dpavlin |
5 |
my $config = LoadFile('config.yml'); |
16 |
|
|
|
17 |
dpavlin |
15 |
my $v = { |
18 |
|
|
search => '', |
19 |
|
|
hits => 0, |
20 |
|
|
page => 0, |
21 |
|
|
max_page => 0, |
22 |
|
|
time => '', |
23 |
|
|
id => time() . rand(99), |
24 |
|
|
}; |
25 |
|
|
|
26 |
|
|
my $json; |
27 |
|
|
|
28 |
dpavlin |
5 |
#warn "config = ", dump($config); |
29 |
|
|
|
30 |
dpavlin |
15 |
sub get_results { |
31 |
|
|
my ($search, $page) = @_; |
32 |
dpavlin |
2 |
|
33 |
dpavlin |
15 |
if (! $search) { |
34 |
|
|
$v->{status} = 'Enter search query'; |
35 |
|
|
return; |
36 |
|
|
} |
37 |
dpavlin |
2 |
|
38 |
dpavlin |
15 |
if (! $page) { |
39 |
|
|
$v->{status} = 'Error: no page number?'; |
40 |
|
|
return; |
41 |
|
|
} |
42 |
dpavlin |
2 |
|
43 |
dpavlin |
15 |
$search = join(" AND ", split(/\s+/, $search)) unless ($search =~ m/(?:AND|OR|\[|\])/); |
44 |
|
|
$v->{search} = $search || ''; |
45 |
dpavlin |
2 |
|
46 |
dpavlin |
15 |
$v->{page} = $page; |
47 |
dpavlin |
2 |
|
48 |
dpavlin |
15 |
my $node = new Search::Estraier::Node(%{ $config->{estraier} }); |
49 |
dpavlin |
2 |
|
50 |
dpavlin |
15 |
my $on_page = 30; |
51 |
|
|
my $skip = ( $page - 1 ) * $on_page; |
52 |
dpavlin |
2 |
|
53 |
dpavlin |
15 |
my $cond = new Search::Estraier::Condition; |
54 |
|
|
$cond->set_phrase( $search ); |
55 |
|
|
$cond->set_max( $on_page ); |
56 |
|
|
$cond->set_skip( $skip ); |
57 |
dpavlin |
2 |
|
58 |
dpavlin |
15 |
my $nres = $node->search($cond, ( $config->{estraier}->{depth} || 0 ) ); |
59 |
dpavlin |
2 |
|
60 |
dpavlin |
15 |
my $out; |
61 |
dpavlin |
2 |
|
62 |
dpavlin |
15 |
if (defined($nres)) { |
63 |
|
|
$v->{hits} = $nres->hits; |
64 |
|
|
$v->{time} = $nres->hint('TIME'); |
65 |
|
|
$v->{max_page} = int( ($nres->hits + $on_page - 1) / $on_page ); |
66 |
|
|
|
67 |
|
|
$v->{status} = qq{ |
68 |
|
|
Got <b>$v->{hits}</b> results for <tt>$v->{search}</tt> |
69 |
|
|
in <em>$v->{time} s</em> |
70 |
|
|
}; |
71 |
|
|
|
72 |
|
|
sub html_snippet { |
73 |
|
|
my $text = shift || return; |
74 |
|
|
my $out = ''; |
75 |
|
|
foreach my $s (split(/[\n\r]{2}/, $text)) { |
76 |
|
|
$out .= ' ... ' if ($out); |
77 |
|
|
my ($pre,$hit,$post) = split(/\n/,$s,3); |
78 |
|
|
$hit =~ s/\t.*$//; |
79 |
|
|
$out .= |
80 |
|
|
$q->escapeHTML( $pre || '' ) . '<b>' . |
81 |
|
|
$q->escapeHTML( $hit || '' ) . '</b>' . |
82 |
|
|
$q->escapeHTML( $post || ''); |
83 |
|
|
} |
84 |
|
|
return $out; |
85 |
dpavlin |
2 |
} |
86 |
|
|
|
87 |
dpavlin |
15 |
sub attr_regex { |
88 |
|
|
my ($rdoc,$attr) = @_; |
89 |
|
|
my $text = $rdoc->attr( $attr ); |
90 |
|
|
return unless defined($text); |
91 |
dpavlin |
2 |
|
92 |
dpavlin |
15 |
if (my $r = $config->{estraier}->{attr_regex}->{$attr} ) { |
93 |
|
|
my $do = '$text =~ ' . $r . ';'; |
94 |
|
|
eval $do; |
95 |
|
|
if ($@) { |
96 |
|
|
warn "eval $do failed: $@\n"; |
97 |
|
|
} |
98 |
dpavlin |
14 |
} |
99 |
dpavlin |
15 |
return $text; |
100 |
dpavlin |
14 |
} |
101 |
|
|
|
102 |
dpavlin |
15 |
# for each document in results |
103 |
|
|
for my $i ( 0 ... $nres->doc_num - 1 ) { |
104 |
dpavlin |
2 |
|
105 |
dpavlin |
15 |
my $rdoc = $nres->get_doc($i); |
106 |
dpavlin |
2 |
|
107 |
dpavlin |
15 |
$out .= '<div class="item">' . |
108 |
|
|
'<h1>' . $rdoc->attr('@title') . '</h1>' . |
109 |
|
|
'<p>' . html_snippet( $rdoc->snippet ) . '</p>' . |
110 |
|
|
'<h2>' . attr_regex( $rdoc, 'source' ) . '</h2>'; |
111 |
|
|
my $uri = attr_regex( $rdoc, '@uri' ); |
112 |
|
|
$out .= |
113 |
|
|
qq{<a href="$uri"><tt>$uri</tt></a> } . |
114 |
|
|
attr_regex( $rdoc, '@mdate' ) . |
115 |
|
|
' [' . ( $skip + $i + 1 ) . ']'; |
116 |
|
|
} |
117 |
dpavlin |
2 |
|
118 |
dpavlin |
15 |
} else { |
119 |
|
|
$out .= 'error: ' . $node->status; |
120 |
dpavlin |
2 |
} |
121 |
dpavlin |
15 |
|
122 |
|
|
$json = '<textarea id="json" style="display:none">' . |
123 |
|
|
$q->escapeHTML( JSON::Syck::Dump( $v ) ) . |
124 |
|
|
'</textarea>'; |
125 |
|
|
|
126 |
|
|
return ($out,$json); |
127 |
|
|
|
128 |
dpavlin |
2 |
} |
129 |
|
|
|
130 |
dpavlin |
15 |
if ($q->path_info() eq '/snippet') { |
131 |
dpavlin |
2 |
|
132 |
dpavlin |
15 |
print join("\n<!-- json data -->", |
133 |
|
|
get_results( |
134 |
|
|
$q->param('search'), |
135 |
|
|
$q->param('page'), |
136 |
|
|
) |
137 |
|
|
); |
138 |
dpavlin |
2 |
|
139 |
|
|
|
140 |
|
|
} else { |
141 |
|
|
|
142 |
dpavlin |
15 |
my ($get_results, $json) = get_results( $q->param('search'), 1 ); |
143 |
dpavlin |
2 |
|
144 |
|
|
my $f = $q->path_info; |
145 |
|
|
$f =~ s/\W+//g; |
146 |
|
|
$f ||= 'search'; |
147 |
|
|
$f .= '.html'; |
148 |
|
|
open(my $s, $f) || die "$f: $!"; |
149 |
|
|
while(<$s>) { |
150 |
dpavlin |
4 |
s/<%(.+?)%>/eval "$1"/ge; |
151 |
dpavlin |
2 |
print; |
152 |
|
|
} |
153 |
|
|
close($f); |
154 |
|
|
|
155 |
|
|
} |