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 * $page ); ## FIXME * $page is needed by hest 1.3.8 |
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 |
|
101 |
$v->{hits} = $nres->hits; |
102 |
$v->{time} = $nres->hint('TIME'); |
103 |
|
104 |
if ($v->{hits} == 0) { |
105 |
$v->{status} = qq{<strong>No results for your search.</strong>}; |
106 |
return next_page(); |
107 |
} elsif ($nres->doc_num == 0) { |
108 |
$v->{status} = qq{<strong>Error getting results for page $page.</strong>}; |
109 |
return next_page('<strong>No results found.</strong>'); |
110 |
} |
111 |
|
112 |
$v->{max_page} = int( ($nres->hits + $on_page - 1) / $on_page ); |
113 |
|
114 |
$v->{status} = qq{ |
115 |
Got <b>$v->{hits}</b> results for <tt>$v->{search}</tt> |
116 |
in <em>$v->{time} s</em> |
117 |
}; |
118 |
|
119 |
sub html_snippet { |
120 |
my $text = shift || return; |
121 |
my $out = ''; |
122 |
foreach my $s (split(/[\n\r]{2}/, $text)) { |
123 |
$out .= ' ... ' if ($out); |
124 |
my ($pre,$hit,$post) = split(/\n/,$s,3); |
125 |
$hit =~ s/\t.*$//; |
126 |
$out .= |
127 |
$q->escapeHTML( $pre || '' ) . '<b>' . |
128 |
$q->escapeHTML( $hit || '' ) . '</b>' . |
129 |
$q->escapeHTML( $post || ''); |
130 |
} |
131 |
return $out; |
132 |
} |
133 |
|
134 |
sub attr_regex { |
135 |
my ($rdoc,$attr) = @_; |
136 |
my $text = $rdoc->attr( $attr ); |
137 |
return unless defined($text); |
138 |
|
139 |
if (my $r = $config->{estraier}->{attr_regex}->{$attr} ) { |
140 |
my $do = '$text =~ ' . $r . ';'; |
141 |
eval $do; |
142 |
if ($@) { |
143 |
warn "eval $do failed: $@\n"; |
144 |
} |
145 |
} |
146 |
return $text; |
147 |
} |
148 |
|
149 |
my @template; |
150 |
open(my $t, 'result.html') || die "result.html: $!"; |
151 |
while(<$t>) { |
152 |
push @template, $_; |
153 |
} |
154 |
close($t); |
155 |
|
156 |
# for each document in results |
157 |
for my $i ( 0 ... $nres->doc_num - 1 ) { |
158 |
|
159 |
my $rdoc = $nres->get_doc($i); |
160 |
my $uri = attr_regex( $rdoc, '@uri' ); |
161 |
my $nr = $skip + $i + 1; |
162 |
|
163 |
map { |
164 |
my $l = $_; |
165 |
$l =~ s/<%(.+?)%>/eval "$1"/ge; |
166 |
$out .= $l; |
167 |
} @template; |
168 |
|
169 |
} |
170 |
|
171 |
} else { |
172 |
$out .= 'error: ' . $node->status; |
173 |
} |
174 |
|
175 |
if ($v->{page} == $v->{max_page}) { |
176 |
$out .= next_page('<br/><strong>All results shown</strong>'); |
177 |
} else { |
178 |
$out .= next_page( |
179 |
'<br/><strong>Loading results...</strong><br/>', |
180 |
'If you are using the scroll bar, release the mouse to see more results.' |
181 |
); |
182 |
} |
183 |
|
184 |
return $out; |
185 |
|
186 |
} |
187 |
|
188 |
if ($q->path_info() eq '/snippet') { |
189 |
|
190 |
print get_results( |
191 |
search => $q->param('search') || '', |
192 |
page => $q->param('page') || 0, |
193 |
sort => $q->param('sort') || undef, |
194 |
); |
195 |
|
196 |
|
197 |
} else { |
198 |
|
199 |
my $get_results = get_results( |
200 |
search => $q->param('search') || '', |
201 |
page => 1, |
202 |
sort => $q->param('sort') || undef, |
203 |
); |
204 |
|
205 |
my $f = $q->path_info; |
206 |
$f =~ s/\W+//g; |
207 |
$f ||= 'search'; |
208 |
$f .= '.html'; |
209 |
open(my $s, $f) || die "$f: $!"; |
210 |
while(<$s>) { |
211 |
s/<%(.+?)%>/eval "$1"/ge; |
212 |
print; |
213 |
} |
214 |
close($s); |
215 |
|
216 |
} |