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 |
my $self = $config->{self} || $q->url(); |
29 |
|
30 |
sub debug { |
31 |
my ($text,$var) = @_; |
32 |
print "<pre>$text = ", dump($var), "</pre>"; |
33 |
} |
34 |
|
35 |
#debug('config', $config); |
36 |
|
37 |
sub json { |
38 |
return |
39 |
'<textarea id="json" style="display:none">' . |
40 |
$q->escapeHTML( JSON::Syck::Dump( $v ) ) . |
41 |
'</textarea>'; |
42 |
} |
43 |
|
44 |
sub sort_order { |
45 |
my $out; |
46 |
|
47 |
my $sort = $q->param('sort'); |
48 |
|
49 |
$out .= '<select name="sort" id="sort">'; |
50 |
|
51 |
foreach my $s (@{ $config->{estraier}->{order} }) { |
52 |
my ($text,$value) = %{$s}; |
53 |
$out .= qq{<option value="$value"} . |
54 |
( $sort eq $value ? ' selected' : '' ) . |
55 |
qq{>$text</option>}; |
56 |
} |
57 |
|
58 |
$out .= '</select>'; |
59 |
} |
60 |
|
61 |
sub get_results { |
62 |
my $p = {@_}; |
63 |
|
64 |
my ($search,$page) = ( $p->{search} , $p->{page}); |
65 |
|
66 |
sub next_page { |
67 |
return '<div id="next_page">' . |
68 |
join("\n", @_) . json() . '</div>'; |
69 |
} |
70 |
|
71 |
if (! $search || $search =~ m/^\s*$/) { |
72 |
$v->{status} = 'Enter search query'; |
73 |
return next_page(); |
74 |
} |
75 |
|
76 |
if (! $page) { |
77 |
$v->{status} = 'Error: no page number?'; |
78 |
return next_page(); |
79 |
} |
80 |
|
81 |
$search = join(" AND ", split(/\s+/, $search)) unless ($search =~ m/(?:AND|OR|\[|\])/); |
82 |
$v->{search} = $search; |
83 |
|
84 |
$v->{page} = $page; |
85 |
|
86 |
my $node = new Search::Estraier::Node(%{ $config->{estraier} }); |
87 |
|
88 |
my $on_page = 30; |
89 |
my $skip = ( $page - 1 ) * $on_page; |
90 |
|
91 |
my $cond = new Search::Estraier::Condition; |
92 |
$cond->set_phrase( $search ); |
93 |
$cond->set_max( $on_page * $page ); ## FIXME * $page is needed by hest 1.3.8 |
94 |
$cond->set_skip( $skip ); |
95 |
$cond->set_order( $p->{sort} ) if ($p->{sort}); |
96 |
|
97 |
my $nres = $node->search($cond, ( $config->{estraier}->{depth} || 0 ) ); |
98 |
|
99 |
my $out; |
100 |
|
101 |
if (defined($nres)) { |
102 |
|
103 |
$v->{hits} = $nres->hits; |
104 |
$v->{time} = $nres->hint('TIME'); |
105 |
|
106 |
if ($v->{hits} == 0) { |
107 |
$v->{status} = qq{<strong>No results for your search.</strong>}; |
108 |
return next_page(); |
109 |
} elsif ($nres->doc_num == 0) { |
110 |
$v->{status} = qq{<strong>Error getting results for page $page.</strong>}; |
111 |
return next_page('<strong>No results found.</strong>'); |
112 |
} |
113 |
|
114 |
$v->{max_page} = int( ($nres->hits + $on_page - 1) / $on_page ); |
115 |
|
116 |
$v->{status} = qq{ |
117 |
Got <b>$v->{hits}</b> results for <tt>$v->{search}</tt> |
118 |
in <em>$v->{time} s</em> |
119 |
}; |
120 |
|
121 |
sub html_snippet { |
122 |
my $text = shift || return; |
123 |
my $out = ''; |
124 |
foreach my $s (split(/[\n\r]{2}/, $text)) { |
125 |
$out .= ' ... ' if ($out); |
126 |
my ($pre,$hit,$post) = split(/\n/,$s,3); |
127 |
$hit =~ s/\t.*$//; |
128 |
$out .= |
129 |
$q->escapeHTML( $pre || '' ) . '<b>' . |
130 |
$q->escapeHTML( $hit || '' ) . '</b>' . |
131 |
$q->escapeHTML( $post || ''); |
132 |
} |
133 |
return $out; |
134 |
} |
135 |
|
136 |
sub attr_regex { |
137 |
my ($rdoc,$attr) = @_; |
138 |
my $text = $rdoc->attr( $attr ); |
139 |
return unless defined($text); |
140 |
|
141 |
if (my $r = $config->{estraier}->{attr_regex}->{$attr} ) { |
142 |
my $do = '$text =~ ' . $r . ';'; |
143 |
eval $do; |
144 |
if ($@) { |
145 |
warn "eval $do failed: $@\n"; |
146 |
} |
147 |
} |
148 |
return $text; |
149 |
} |
150 |
|
151 |
my @template; |
152 |
open(my $t, 'result.html') || die "result.html: $!"; |
153 |
while(<$t>) { |
154 |
push @template, $_; |
155 |
} |
156 |
close($t); |
157 |
|
158 |
# for each document in results |
159 |
for my $i ( 0 ... $nres->doc_num - 1 ) { |
160 |
|
161 |
my $rdoc = $nres->get_doc($i); |
162 |
my $uri = attr_regex( $rdoc, '@uri' ); |
163 |
my $nr = $skip + $i + 1; |
164 |
|
165 |
map { |
166 |
my $l = $_; |
167 |
$l =~ s/<%(.+?)%>/eval "$1"/ge; |
168 |
$out .= $l; |
169 |
} @template; |
170 |
|
171 |
} |
172 |
|
173 |
} else { |
174 |
$out .= 'error: ' . $node->status; |
175 |
} |
176 |
|
177 |
if ($v->{page} == $v->{max_page}) { |
178 |
$out .= next_page('<br/><strong>All results shown</strong>'); |
179 |
} else { |
180 |
$out .= next_page( |
181 |
'<br/><strong>Loading results...</strong><br/>', |
182 |
'If you are using the scroll bar, release the mouse to see more results.' |
183 |
); |
184 |
} |
185 |
|
186 |
return $out; |
187 |
|
188 |
} |
189 |
|
190 |
if ($q->path_info() eq '/snippet') { |
191 |
|
192 |
print get_results( |
193 |
search => $q->param('search') || '', |
194 |
page => $q->param('page') || 0, |
195 |
sort => $q->param('sort') || undef, |
196 |
); |
197 |
|
198 |
|
199 |
} else { |
200 |
|
201 |
my $get_results = get_results( |
202 |
search => $q->param('search') || '', |
203 |
page => 1, |
204 |
sort => $q->param('sort') || undef, |
205 |
); |
206 |
|
207 |
my $f = $q->path_info; |
208 |
$f =~ s/\W+//g; |
209 |
$f ||= 'search'; |
210 |
$f .= '.html'; |
211 |
open(my $s, $f) || die "$f: $!"; |
212 |
while(<$s>) { |
213 |
s/<%(.+?)%>/eval "$1"/ge; |
214 |
print; |
215 |
} |
216 |
close($s); |
217 |
|
218 |
} |