6 |
use CGI::Carp qw(fatalsToBrowser warningsToBrowser); |
use CGI::Carp qw(fatalsToBrowser warningsToBrowser); |
7 |
use Search::Estraier; |
use Search::Estraier; |
8 |
use YAML::Syck; |
use YAML::Syck; |
9 |
|
use JSON::Syck; |
10 |
use Data::Dump qw/dump/; |
use Data::Dump qw/dump/; |
11 |
|
|
12 |
my $q = new CGI::Simple; |
my $q = new CGI::Simple; |
14 |
|
|
15 |
my $config = LoadFile('config.yml'); |
my $config = LoadFile('config.yml'); |
16 |
|
|
17 |
|
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 $json; |
27 |
|
|
28 |
#warn "config = ", dump($config); |
#warn "config = ", dump($config); |
29 |
|
|
30 |
if ($q->path_info() eq '/snippet') { |
sub json { |
31 |
|
return |
32 |
|
'<textarea id="json" style="display:none">' . |
33 |
|
$q->escapeHTML( JSON::Syck::Dump( $v ) ) . |
34 |
|
'</textarea>'; |
35 |
|
} |
36 |
|
|
37 |
print qq{ |
sub get_results { |
38 |
<HTML> |
my ($search, $page) = @_; |
|
<HEAD> |
|
|
<META CONTENT="text/html; charset=utf-8" HTTP-EQUIV="Content-Type"> |
|
|
<META CONTENT="no-cache" HTTP-EQUIV="Pragma"> |
|
|
<META CONTENT="-1" HTTP-EQUIV="Expires"> |
|
|
</HEAD> |
|
|
|
|
|
<div> |
|
|
<div class="post"> |
|
|
<ul> |
|
|
}; |
|
39 |
|
|
40 |
my $node = new Search::Estraier::Node(%{ $config->{estraier} }); |
warn "get_results( $search , $page )\n"; |
41 |
|
|
42 |
my $o = $q->param('index') || 0; |
if (! $search || $search =~ m/^\s*$/) { |
43 |
my $search = $q->param('q'); |
$v->{status} = 'Enter search query'; |
44 |
$search = join(" AND ", split(/\s+/, $search)) unless ($search =~ m/(?:AND|OR|\[|\])/); |
return('', json); |
45 |
|
} |
46 |
my $on_page = 30; |
|
47 |
my $skip = $o * $on_page; |
if (! $page) { |
48 |
|
$v->{status} = 'Error: no page number?'; |
49 |
my $cond = new Search::Estraier::Condition; |
return('', json); |
|
$cond->set_phrase( $search ); |
|
|
$cond->set_max( $on_page ); |
|
|
$cond->set_skip( $skip ); |
|
|
|
|
|
my $nres = $node->search($cond, ( $config->{estraier}->{depth} || 0 ) ); |
|
|
|
|
|
my $max = 0; |
|
|
|
|
|
if (defined($nres)) { |
|
|
$max = $nres->hits; |
|
|
my $time = $nres->hint('TIME'); |
|
|
print qq{ |
|
|
<div id="status_update" style="display:none;"> |
|
|
Got <b>$max</b> results for <tt>$search</tt> in <em>$time s</em> |
|
|
</div> |
|
|
}; |
|
|
|
|
|
sub html_snippet { |
|
|
my $text = shift || return; |
|
|
my $out = ''; |
|
|
foreach my $s (split(/[\n\r]{2}/, $text)) { |
|
|
$out .= ' ... ' if ($out); |
|
|
my ($pre,$hit,$post) = split(/\n/,$s,3); |
|
|
$hit =~ s/\t.*$//; |
|
|
$out .= |
|
|
$q->escapeHTML( $pre || '' ) . '<b>' . |
|
|
$q->escapeHTML( $hit || '' ) . '</b>' . |
|
|
$q->escapeHTML( $post || ''); |
|
|
} |
|
|
return $out; |
|
50 |
} |
} |
51 |
|
|
52 |
|
$search = join(" AND ", split(/\s+/, $search)) unless ($search =~ m/(?:AND|OR|\[|\])/); |
53 |
|
$v->{search} = $search || ''; |
54 |
|
|
55 |
|
$v->{page} = $page; |
56 |
|
|
57 |
|
my $node = new Search::Estraier::Node(%{ $config->{estraier} }); |
58 |
|
|
59 |
|
my $on_page = 30; |
60 |
|
my $skip = ( $page - 1 ) * $on_page; |
61 |
|
|
62 |
|
my $cond = new Search::Estraier::Condition; |
63 |
|
$cond->set_phrase( $search ); |
64 |
|
$cond->set_max( $on_page ); |
65 |
|
$cond->set_skip( $skip ); |
66 |
|
|
67 |
|
my $nres = $node->search($cond, ( $config->{estraier}->{depth} || 0 ) ); |
68 |
|
|
69 |
|
my $out; |
70 |
|
|
71 |
|
if (defined($nres)) { |
72 |
|
$v->{hits} = $nres->hits; |
73 |
|
$v->{time} = $nres->hint('TIME'); |
74 |
|
$v->{max_page} = int( ($nres->hits + $on_page - 1) / $on_page ); |
75 |
|
|
76 |
|
$v->{status} = qq{ |
77 |
|
Got <b>$v->{hits}</b> results for <tt>$v->{search}</tt> |
78 |
|
in <em>$v->{time} s</em> |
79 |
|
}; |
80 |
|
|
81 |
|
sub html_snippet { |
82 |
|
my $text = shift || return; |
83 |
|
my $out = ''; |
84 |
|
foreach my $s (split(/[\n\r]{2}/, $text)) { |
85 |
|
$out .= ' ... ' if ($out); |
86 |
|
my ($pre,$hit,$post) = split(/\n/,$s,3); |
87 |
|
$hit =~ s/\t.*$//; |
88 |
|
$out .= |
89 |
|
$q->escapeHTML( $pre || '' ) . '<b>' . |
90 |
|
$q->escapeHTML( $hit || '' ) . '</b>' . |
91 |
|
$q->escapeHTML( $post || ''); |
92 |
|
} |
93 |
|
return $out; |
94 |
|
} |
95 |
|
|
96 |
|
sub attr_regex { |
97 |
|
my ($rdoc,$attr) = @_; |
98 |
|
my $text = $rdoc->attr( $attr ); |
99 |
|
return unless defined($text); |
100 |
|
|
101 |
|
if (my $r = $config->{estraier}->{attr_regex}->{$attr} ) { |
102 |
|
my $do = '$text =~ ' . $r . ';'; |
103 |
|
eval $do; |
104 |
|
if ($@) { |
105 |
|
warn "eval $do failed: $@\n"; |
106 |
|
} |
107 |
|
} |
108 |
|
return $text; |
109 |
|
} |
110 |
|
|
111 |
# for each document in results |
# for each document in results |
112 |
for my $i ( 0 ... $nres->doc_num - 1 ) { |
for my $i ( 0 ... $nres->doc_num - 1 ) { |
113 |
|
|
114 |
my $rdoc = $nres->get_doc($i); |
my $rdoc = $nres->get_doc($i); |
115 |
|
|
116 |
print "<li>"; |
$out .= '<div class="item">' . |
117 |
|
'<h1>' . $rdoc->attr('@title') . '</h1>' . |
118 |
|
'<p>' . html_snippet( $rdoc->snippet ) . '</p>' . |
119 |
|
'<h2>' . attr_regex( $rdoc, 'source' ) . '</h2>'; |
120 |
|
my $uri = attr_regex( $rdoc, '@uri' ); |
121 |
|
$out .= |
122 |
|
qq{<a href="$uri"><tt>$uri</tt></a> } . |
123 |
|
attr_regex( $rdoc, '@mdate' ) . |
124 |
|
' [' . ( $skip + $i + 1 ) . ']'; |
125 |
|
} |
126 |
|
|
127 |
print "<h1>", $rdoc->attr('@title'),"</h1>\n"; |
} else { |
128 |
print "<h2>", $rdoc->attr('source'),"</h2>\n"; |
$out .= 'error: ' . $node->status; |
|
print "<h3>", $rdoc->attr('@mdate'),"</h3>\n"; |
|
|
print "<p>", html_snippet( $rdoc->snippet ),"</p>\n"; |
|
|
print "[", $skip + $i, "] "; |
|
|
print "<tt>", $rdoc->attr('@uri'),"</tt>"; |
|
|
print "</li>"; |
|
129 |
} |
} |
|
} else { |
|
|
die "error: ", $node->status,"\n"; |
|
|
} |
|
130 |
|
|
131 |
|
return ($out,json); |
132 |
|
|
133 |
print qq{ |
} |
|
</ul> |
|
|
</div> |
|
134 |
|
|
135 |
</div> |
if ($q->path_info() eq '/snippet') { |
136 |
|
|
137 |
|
print join("\n<!-- json data -->", |
138 |
|
get_results( |
139 |
|
$q->param('search'), |
140 |
|
$q->param('page'), |
141 |
|
) |
142 |
|
); |
143 |
|
|
|
</html> |
|
|
}; |
|
144 |
|
|
145 |
} else { |
} else { |
146 |
|
|
147 |
sub page_id { |
my ($get_results,$json); |
148 |
my $page_id = time() . rand(99); |
|
149 |
warn "page_id = $page_id\n"; |
if ($q->param('search')) { |
150 |
return $page_id; |
($get_results, $json) = get_results( $q->param('search'), 1 ); |
151 |
}; |
} else { |
152 |
|
$v->{status} = 'Enter search query'; |
153 |
|
($get_results, $json) = ('', json); |
154 |
|
} |
155 |
|
|
156 |
my $f = $q->path_info; |
my $f = $q->path_info; |
157 |
$f =~ s/\W+//g; |
$f =~ s/\W+//g; |
159 |
$f .= '.html'; |
$f .= '.html'; |
160 |
open(my $s, $f) || die "$f: $!"; |
open(my $s, $f) || die "$f: $!"; |
161 |
while(<$s>) { |
while(<$s>) { |
|
no strict 'vars'; |
|
162 |
s/<%(.+?)%>/eval "$1"/ge; |
s/<%(.+?)%>/eval "$1"/ge; |
163 |
print; |
print; |
164 |
} |
} |