| 1 |
#!/usr/bin/perl -w |
| 2 |
|
| 3 |
use strict; |
| 4 |
|
| 5 |
use CGI::Simple; |
| 6 |
use CGI::Carp qw(fatalsToBrowser warningsToBrowser); |
| 7 |
use Search::Estraier; |
| 8 |
use YAML::Syck; |
| 9 |
use JSON::Syck; |
| 10 |
use Data::Dump qw/dump/; |
| 11 |
|
| 12 |
my $q = new CGI::Simple; |
| 13 |
print qq{Content-type: text/html\n\r\n\r}; |
| 14 |
|
| 15 |
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 $json; |
| 27 |
|
| 28 |
sub debug { |
| 29 |
my ($text,$var) = @_; |
| 30 |
print "<pre>$text = ", dump($var), "</pre>"; |
| 31 |
} |
| 32 |
|
| 33 |
#debug('config', $config); |
| 34 |
|
| 35 |
sub json { |
| 36 |
return |
| 37 |
'<textarea id="json" style="display:none">' . |
| 38 |
$q->escapeHTML( JSON::Syck::Dump( $v ) ) . |
| 39 |
'</textarea>'; |
| 40 |
} |
| 41 |
|
| 42 |
sub sort_order { |
| 43 |
my $out; |
| 44 |
|
| 45 |
my $sort = $q->param('sort'); |
| 46 |
|
| 47 |
$out .= '<select name="sort" id="sort">'; |
| 48 |
|
| 49 |
foreach my $s (@{ $config->{estraier}->{order} }) { |
| 50 |
my ($text,$value) = %{$s}; |
| 51 |
$out .= qq{<option value="$value"} . |
| 52 |
( $sort eq $value ? ' selected' : '' ) . |
| 53 |
qq{>$text</option>}; |
| 54 |
} |
| 55 |
|
| 56 |
$out .= '</select>'; |
| 57 |
} |
| 58 |
|
| 59 |
sub get_results { |
| 60 |
my $p = {@_}; |
| 61 |
|
| 62 |
my ($search,$page) = ( $p->{search} , $p->{page}); |
| 63 |
|
| 64 |
sub next_page { |
| 65 |
return '<div id="next_page">' . |
| 66 |
join("\n", @_) . json() . '</div>'; |
| 67 |
} |
| 68 |
|
| 69 |
if (! $search || $search =~ m/^\s*$/) { |
| 70 |
$v->{status} = 'Enter search query'; |
| 71 |
return next_page(); |
| 72 |
} |
| 73 |
|
| 74 |
if (! $page) { |
| 75 |
$v->{status} = 'Error: no page number?'; |
| 76 |
return next_page(); |
| 77 |
} |
| 78 |
|
| 79 |
$search = join(" AND ", split(/\s+/, $search)) unless ($search =~ m/(?:AND|OR|\[|\])/); |
| 80 |
$v->{search} = $search; |
| 81 |
|
| 82 |
$v->{page} = $page; |
| 83 |
|
| 84 |
my $node = new Search::Estraier::Node(%{ $config->{estraier} }); |
| 85 |
|
| 86 |
my $on_page = 30; |
| 87 |
my $skip = ( $page - 1 ) * $on_page; |
| 88 |
|
| 89 |
my $cond = new Search::Estraier::Condition; |
| 90 |
$cond->set_phrase( $search ); |
| 91 |
$cond->set_max( $on_page ); |
| 92 |
$cond->set_skip( $skip ); |
| 93 |
$cond->set_order( $p->{sort} ) if ($p->{sort}); |
| 94 |
|
| 95 |
my $nres = $node->search($cond, ( $config->{estraier}->{depth} || 0 ) ); |
| 96 |
|
| 97 |
my $out; |
| 98 |
|
| 99 |
if (defined($nres)) { |
| 100 |
$v->{hits} = $nres->hits; |
| 101 |
$v->{time} = $nres->hint('TIME'); |
| 102 |
$v->{max_page} = int( ($nres->hits + $on_page - 1) / $on_page ); |
| 103 |
|
| 104 |
$v->{status} = qq{ |
| 105 |
Got <b>$v->{hits}</b> results for <tt>$v->{search}</tt> |
| 106 |
in <em>$v->{time} s</em> |
| 107 |
}; |
| 108 |
|
| 109 |
sub html_snippet { |
| 110 |
my $text = shift || return; |
| 111 |
my $out = ''; |
| 112 |
foreach my $s (split(/[\n\r]{2}/, $text)) { |
| 113 |
$out .= ' ... ' if ($out); |
| 114 |
my ($pre,$hit,$post) = split(/\n/,$s,3); |
| 115 |
$hit =~ s/\t.*$//; |
| 116 |
$out .= |
| 117 |
$q->escapeHTML( $pre || '' ) . '<b>' . |
| 118 |
$q->escapeHTML( $hit || '' ) . '</b>' . |
| 119 |
$q->escapeHTML( $post || ''); |
| 120 |
} |
| 121 |
return $out; |
| 122 |
} |
| 123 |
|
| 124 |
sub attr_regex { |
| 125 |
my ($rdoc,$attr) = @_; |
| 126 |
my $text = $rdoc->attr( $attr ); |
| 127 |
return unless defined($text); |
| 128 |
|
| 129 |
if (my $r = $config->{estraier}->{attr_regex}->{$attr} ) { |
| 130 |
my $do = '$text =~ ' . $r . ';'; |
| 131 |
eval $do; |
| 132 |
if ($@) { |
| 133 |
warn "eval $do failed: $@\n"; |
| 134 |
} |
| 135 |
} |
| 136 |
return $text; |
| 137 |
} |
| 138 |
|
| 139 |
# for each document in results |
| 140 |
for my $i ( 0 ... $nres->doc_num - 1 ) { |
| 141 |
|
| 142 |
my $rdoc = $nres->get_doc($i); |
| 143 |
|
| 144 |
$out .= '<div class="item">' . |
| 145 |
'<h1>' . attr_regex( $rdoc, '@title' ) . '</h1>' . |
| 146 |
'<p>' . html_snippet( $rdoc->snippet ) . '</p>' . |
| 147 |
'<strong>' . attr_regex( $rdoc, 'source' ) . '</strong>' . |
| 148 |
' [' . attr_regex( $rdoc, '@size' ) . ' bytes]<br/>'; |
| 149 |
my $uri = attr_regex( $rdoc, '@uri' ); |
| 150 |
$out .= |
| 151 |
qq{<a href="$uri"><tt>$uri</tt></a> } . |
| 152 |
attr_regex( $rdoc, '@mdate' ) . |
| 153 |
' [' . ( $skip + $i + 1 ) . ']'; |
| 154 |
} |
| 155 |
|
| 156 |
} else { |
| 157 |
$out .= 'error: ' . $node->status; |
| 158 |
} |
| 159 |
|
| 160 |
if ($v->{page} == $v->{max_page}) { |
| 161 |
$out .= next_page('<strong>All results shown</strong>'); |
| 162 |
} else { |
| 163 |
$out .= next_page( |
| 164 |
'<strong>Loading results...</strong><br/>', |
| 165 |
'If you are using the scroll bar, release the mouse to see more results.' |
| 166 |
); |
| 167 |
} |
| 168 |
|
| 169 |
return $out; |
| 170 |
|
| 171 |
} |
| 172 |
|
| 173 |
if ($q->path_info() eq '/snippet') { |
| 174 |
|
| 175 |
print get_results( |
| 176 |
search => $q->param('search') || '', |
| 177 |
page => $q->param('page') || 0, |
| 178 |
sort => $q->param('sort') || undef, |
| 179 |
); |
| 180 |
|
| 181 |
|
| 182 |
} else { |
| 183 |
|
| 184 |
my $get_results = get_results( |
| 185 |
search => $q->param('search') || '', |
| 186 |
page => 1, |
| 187 |
sort => $q->param('sort') || undef, |
| 188 |
); |
| 189 |
|
| 190 |
my $f = $q->path_info; |
| 191 |
$f =~ s/\W+//g; |
| 192 |
$f ||= 'search'; |
| 193 |
$f .= '.html'; |
| 194 |
open(my $s, $f) || die "$f: $!"; |
| 195 |
while(<$s>) { |
| 196 |
s/<%(.+?)%>/eval "$1"/ge; |
| 197 |
print; |
| 198 |
} |
| 199 |
close($f); |
| 200 |
|
| 201 |
} |
Parent Directory
| 
Revision Log