]> git.vanrenterghem.biz Git - git.ikiwiki.info.git/blob - IkiWiki/Plugin/search.pm
add section information
[git.ikiwiki.info.git] / IkiWiki / Plugin / search.pm
1 #!/usr/bin/perl
2 # xapian-omega search engine plugin
3 package IkiWiki::Plugin::search;
5 use warnings;
6 use strict;
7 use IkiWiki 3.00;
9 sub import {
10         hook(type => "getsetup", id => "search", call => \&getsetup);
11         hook(type => "checkconfig", id => "search", call => \&checkconfig);
12         hook(type => "pagetemplate", id => "search", call => \&pagetemplate);
13         hook(type => "postscan", id => "search", call => \&index);
14         hook(type => "delete", id => "search", call => \&delete);
15         hook(type => "cgi", id => "search", call => \&cgi);
16 }
18 sub getsetup () {
19         return
20                 plugin => {
21                         safe => 1,
22                         rebuild => 1,
23                         section => "web",
24                 },
25                 omega_cgi => {
26                         type => "string",
27                         example => "/usr/lib/cgi-bin/omega/omega",
28                         description => "path to the omega cgi program",
29                         safe => 0, # external program
30                         rebuild => 0,
31                 },
32 }
34 sub checkconfig () {
35         foreach my $required (qw(url cgiurl)) {
36                 if (! length $config{$required}) {
37                         error(sprintf(gettext("Must specify %s when using the %s plugin"), $required, 'search'));
38                 }
39         }
40         
41         if (! defined $config{omega_cgi}) {
42                 $config{omega_cgi}="/usr/lib/cgi-bin/omega/omega";
43         }
44 }
46 my $form;
47 sub pagetemplate (@) {
48         my %params=@_;
49         my $page=$params{page};
50         my $template=$params{template};
52         # Add search box to page header.
53         if ($template->query(name => "searchform")) {
54                 if (! defined $form) {
55                         my $searchform = template("searchform.tmpl", blind_cache => 1);
56                         $searchform->param(searchaction => $config{cgiurl});
57                         $form=$searchform->output;
58                 }
60                 $template->param(searchform => $form);
61         }
62 }
64 my $scrubber;
65 my $stemmer;
66 sub index (@) {
67         my %params=@_;
69         setupfiles();
71         # A unique pageterm is used to identify the document for a page.
72         my $pageterm=pageterm($params{page});
73         return $params{content} unless defined $pageterm;
74         
75         my $db=xapiandb();
76         my $doc=Search::Xapian::Document->new();
77         my $caption=pagetitle($params{page});
78         my $title;
79         if (exists $pagestate{$params{page}}{meta} &&
80                 exists $pagestate{$params{page}}{meta}{title}) {
81                 $title=$pagestate{$params{page}}{meta}{title};
82         }
83         else {
84                 $title=$caption;
85         }
87         # Remove html from text to be indexed.
88         if (! defined $scrubber) {
89                 eval q{use HTML::Scrubber};
90                 if (! $@) {
91                         $scrubber=HTML::Scrubber->new(allow => []);
92                 }
93         }
94         my $toindex = defined $scrubber ? $scrubber->scrub($params{content}) : $params{content};
95         
96         # Take 512 characters for a sample, then extend it out
97         # if it stopped in the middle of a word.
98         my $size=512;
99         my ($sample)=substr($toindex, 0, $size);
100         if (length($sample) == $size) {
101                 my $max=length($toindex);
102                 my $next;
103                 while ($size < $max &&
104                        ($next=substr($toindex, $size++, 1)) !~ /\s/) {
105                         $sample.=$next;
106                 }
107         }
108         $sample=~s/\n/ /g;
109         
110         # data used by omega
111         # Decode html entities in it, since omega re-encodes them.
112         eval q{use HTML::Entities};
113         $doc->set_data(
114                 "url=".urlto($params{page}, "")."\n".
115                 "sample=".decode_entities($sample)."\n".
116                 "caption=".decode_entities($caption)."\n".
117                 "modtime=$IkiWiki::pagemtime{$params{page}}\n".
118                 "size=".length($params{content})."\n"
119         );
121         # Index document and add terms for other metadata.
122         my $tg = Search::Xapian::TermGenerator->new();
123         if (! $stemmer) {
124                 my $langcode=$ENV{LANG} || "en";
125                 $langcode=~s/_.*//;
127                 # This whitelist is here to work around a xapian bug (#486138)
128                 my @whitelist=qw{da de en es fi fr hu it no pt ru ro sv tr};
130                 if (grep { $_ eq $langcode } @whitelist) {
131                         $stemmer=Search::Xapian::Stem->new($langcode);
132                 }
133                 else {
134                         $stemmer=Search::Xapian::Stem->new("english");
135                 }
136         }
137         $tg->set_stemmer($stemmer);
138         $tg->set_document($doc);
139         $tg->index_text($params{page}, 2);
140         $tg->index_text($caption, 2);
141         $tg->index_text($title, 2) if $title ne $caption;
142         $tg->index_text($toindex);
143         $tg->index_text(lc($title), 1, "S"); # for title:foo
144         foreach my $link (@{$links{$params{page}}}) {
145                 $tg->index_text(lc($link), 1, "XLINK"); # for link:bar
146         }
148         $doc->add_term($pageterm);
149         $db->replace_document_by_term($pageterm, $doc);
152 sub delete (@) {
153         my $db=xapiandb();
154         foreach my $page (@_) {
155                 my $pageterm=pageterm(pagename($page));
156                 $db->delete_document_by_term($pageterm) if defined $pageterm;
157         }
160 sub cgi ($) {
161         my $cgi=shift;
163         if (defined $cgi->param('P')) {
164                 # only works for GET requests
165                 chdir("$config{wikistatedir}/xapian") || error("chdir: $!");
166                 $ENV{OMEGA_CONFIG_FILE}="./omega.conf";
167                 $ENV{CGIURL}=$config{cgiurl},
168                 IkiWiki::loadindex();
169                 $ENV{HELPLINK}=htmllink("", "", "ikiwiki/searching",
170                         noimageinline => 1, linktext => "Help");
171                 exec($config{omega_cgi}) || error("$config{omega_cgi} failed: $!");
172         }
175 sub pageterm ($) {
176         my $page=shift;
178         # 240 is the number used by omindex to decide when to hash an
179         # overlong term. This does not use a compatible hash method though.
180         if (length $page > 240) {
181                 eval q{use Digest::SHA1};
182                 if ($@) {
183                         debug("search: ".sprintf(gettext("need Digest::SHA1 to index %s"), $page)) if $@;
184                         return undef;
185                 }
187                 # Note no colon, therefore it's guaranteed to not overlap
188                 # with a page with the same name as the hash..
189                 return "U".lc(Digest::SHA1::sha1_hex($page));
190         }
191         else {
192                 return "U:".$page;
193         }
196 my $db;
197 sub xapiandb () {
198         if (! defined $db) {
199                 eval q{
200                         use Search::Xapian;
201                         use Search::Xapian::WritableDatabase;
202                 };
203                 error($@) if $@;
204                 $db=Search::Xapian::WritableDatabase->new($config{wikistatedir}."/xapian/default",
205                         Search::Xapian::DB_CREATE_OR_OPEN());
206         }
207         return $db;
211 my $setup=0;
212 sub setupfiles () {
213         if (! $setup and (! -e $config{wikistatedir}."/xapian" || $config{rebuild})) {
214                 writefile("omega.conf", $config{wikistatedir}."/xapian",
215                         "database_dir .\n".
216                         "template_dir ./templates\n");
217                 writefile("query", $config{wikistatedir}."/xapian/templates",
218                         IkiWiki::misctemplate(gettext("search"),
219                                 readfile(IkiWiki::template_file("searchquery.tmpl"))));
220                 $setup=1;
221         }