Merge branch 'master' into xapian
[ikiwiki] / IkiWiki / Plugin / search.pm
1 #!/usr/bin/perl
2 # xapian-omega search engine plugin
3 package IkiWiki::Plugin::search;
4
5 use warnings;
6 use strict;
7 use IkiWiki 2.00;
8
9 sub import { #{{{
10         hook(type => "checkconfig", id => "search", call => \&checkconfig);
11         hook(type => "pagetemplate", id => "search", call => \&pagetemplate);
12         # run last so other needsbuild hooks can modify the list
13         hook(type => "needsbuild", id => "search", call => \&needsbuild,
14                 last => 1);
15         hook(type => "filter", id => "search", call => \&filter);
16         hook(type => "delete", id => "search", call => \&delete);
17         hook(type => "cgi", id => "search", call => \&cgi);
18 } # }}}
19
20 sub checkconfig () { #{{{
21         foreach my $required (qw(url cgiurl)) {
22                 if (! length $config{$required}) {
23                         error(sprintf(gettext("Must specify %s when using the search plugin"), $required));
24                 }
25         }
26
27         if (! exists $config{omega_cgi}) {
28                 $config{omega_cgi}="/usr/lib/cgi-bin/omega/omega";
29         }
30         
31         if (! -e $config{wikistatedir}."/xapian" || $config{rebuild}) {
32                 writefile("omega.conf", $config{wikistatedir}."/xapian",
33                         "database_dir .\n".
34                         "template_dir ./templates\n");
35                 writefile("query", $config{wikistatedir}."/xapian/templates",
36                         IkiWiki::misctemplate(gettext("search"),
37                                 readfile(IkiWiki::template_file("searchquery.tmpl"))));
38         }
39 } #}}}
40
41 my $form;
42 sub pagetemplate (@) { #{{{
43         my %params=@_;
44         my $page=$params{page};
45         my $template=$params{template};
46
47         # Add search box to page header.
48         if ($template->query(name => "searchform")) {
49                 if (! defined $form) {
50                         my $searchform = template("searchform.tmpl", blind_cache => 1);
51                         $searchform->param(searchaction => $config{cgiurl});
52                         $form=$searchform->output;
53                 }
54
55                 $template->param(searchform => $form);
56         }
57 } #}}}
58
59 my %toindex;
60 sub needsbuild ($) { #{{{
61         %toindex = map { pagename($_) => 1 } @{shift()};
62 } #}}}
63
64 my $scrubber;
65 sub filter (@) { #{{{
66         my %params=@_;
67         
68         if ($params{page} eq $params{destpage} && $toindex{$params{page}}) {
69                 # index page
70                 my $db=xapiandb();
71                 my $doc=Search::Xapian::Document->new();
72                 my $title;
73                 if (exists $pagestate{$params{page}}{meta} &&
74                     exists $pagestate{$params{page}}{meta}{title}) {
75                         $title=$pagestate{$params{page}}{meta}{title};
76                 }
77                 else {
78                         $title=IkiWiki::pagetitle($params{page});
79                 }
80
81                 # Remove any html from text to be indexed.
82                 # TODO: This removes html that is in eg, a markdown pre,
83                 # which should not be removed, really.
84                 if (! defined $scrubber) {
85                         eval q{use HTML::Scrubber};
86                         if (! $@) {
87                                 $scrubber=HTML::Scrubber->new(allow => []);
88                         }
89                 }
90                 my $toindex = defined $scrubber ? $scrubber->scrub($params{content}) : $params{content};
91                 
92                 # Take 512 characters for a sample, then extend it out
93                 # if it stopped in the middle of a word.
94                 my $size=512;
95                 my ($sample)=substr($toindex, 0, $size);
96                 if (length($sample) == $size) {
97                         my $max=length($toindex);
98                         my $next;
99                         while ($size < $max &&
100                                ($next=substr($toindex, $size++, 1)) !~ /\s/) {
101                                 $sample.=$next;
102                         }
103                 }
104                 $sample=~s/\n/ /g;
105                 
106                 # data used by omega
107                 $doc->set_data(
108                         "url=".urlto($params{page}, "")."\n".
109                         "sample=$sample\n".
110                         "caption=$title\n".
111                         "modtime=$IkiWiki::pagemtime{$params{page}}\n".
112                         "size=".length($params{content})."\n"
113                 );
114
115                 my $tg = Search::Xapian::TermGenerator->new();
116                 $tg->set_stemmer(new Search::Xapian::Stem("english"));
117                 $tg->set_document($doc);
118                 $tg->index_text($params{page}, 2);
119                 $tg->index_text($title, 2);
120                 $tg->index_text($toindex);
121
122                 my $pageterm=pageterm($params{page});
123                 $doc->add_term($pageterm);
124                 $db->replace_document_by_term($pageterm, $doc);
125         }
126
127         return $params{content};
128 } #}}}
129
130 sub delete (@) { #{{{
131         my $db=xapiandb();
132         foreach my $page (@_) {
133                 $db->delete_document_by_term(pageterm($page));
134         }
135 } #}}}
136
137 sub cgi ($) { #{{{
138         my $cgi=shift;
139
140         if (defined $cgi->param('P')) {
141                 # only works for GET requests
142                 chdir("$config{wikistatedir}/xapian") || error("chdir: $!");
143                 $ENV{OMEGA_CONFIG_FILE}="./omega.conf";
144                 $ENV{CGIURL}=$config{cgiurl},
145                 exec($config{omega_cgi}) || error("$config{omega_cgi} failed: $!");
146         }
147 } #}}}
148
149 sub pageterm ($) { #{{{
150         my $page=shift;
151
152         # TODO: check if > 255 char page names overflow term
153         # length; use sha1 if so?
154         return "U:".$page;
155 } #}}}
156
157 my $db;
158 sub xapiandb () { #{{{
159         if (! defined $db) {
160                 eval q{
161                         use Search::Xapian;
162                         use Search::Xapian::WritableDatabase;
163                 };
164                 error($@) if $@;
165                 $db=Search::Xapian::WritableDatabase->new($config{wikistatedir}."/xapian/default",
166                         Search::Xapian::DB_CREATE_OR_OPEN());
167         }
168         return $db;
169 } #}}}
170
171 1