Merge branch 'master' of ssh://git.ikiwiki.info
[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 3.00;
8
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 => "indexhtml", id => "search", call => \&indexhtml);
14         hook(type => "delete", id => "search", call => \&delete);
15         hook(type => "cgi", id => "search", call => \&cgi);
16         hook(type => "disable", id => "search", call => \&disable);
17         hook(type => "needsbuild", id => "search", call => \&needsbuild);
18                 
19         eval q{ use Search::Xapian }; # load early to work around #622591
20 }
21
22 sub getsetup () {
23         return
24                 plugin => {
25                         safe => 1,
26                         rebuild => 1,
27                         section => "web",
28                 },
29                 omega_cgi => {
30                         type => "string",
31                         example => "/usr/lib/cgi-bin/omega/omega",
32                         description => "path to the omega cgi program",
33                         safe => 0, # external program
34                         rebuild => 0,
35                 },
36 }
37
38 sub checkconfig () {
39         foreach my $required (qw(url cgiurl)) {
40                 if (! length $config{$required}) {
41                         error(sprintf(gettext("Must specify %s when using the %s plugin"), $required, 'search'));
42                 }
43         }
44         
45         if (! defined $config{omega_cgi}) {
46                 $config{omega_cgi}="/usr/lib/cgi-bin/omega/omega";
47         }
48
49         # This is a mass dependency, so if the search form template
50         # changes, every page is rebuilt.
51         add_depends("", "templates/searchform.tmpl");
52 }
53
54 my $form;
55 sub pagetemplate (@) {
56         my %params=@_;
57         my $page=$params{page};
58         my $template=$params{template};
59
60         # Add search box to page header.
61         if ($template->query(name => "searchform")) {
62                 if (! defined $form) {
63                         my $searchform = template("searchform.tmpl", blind_cache => 1);
64                         $searchform->param(searchaction => IkiWiki::cgiurl());
65                         $searchform->param(html5 => $config{html5});
66                         $form=$searchform->output;
67                 }
68
69                 $template->param(searchform => $form);
70         }
71 }
72
73 my $scrubber;
74 my $stemmer;
75 sub indexhtml (@) {
76         my %params=@_;
77
78         setupfiles();
79
80         # A unique pageterm is used to identify the document for a page.
81         my $pageterm=pageterm($params{page});
82         return unless defined $pageterm;
83         
84         my $db=xapiandb();
85         my $doc=Search::Xapian::Document->new();
86         my $caption=pagetitle($params{page});
87         my $title;
88         if (exists $pagestate{$params{page}}{meta} &&
89                 exists $pagestate{$params{page}}{meta}{title}) {
90                 $title=$pagestate{$params{page}}{meta}{title};
91         }
92         else {
93                 $title=$caption;
94         }
95
96         # Remove html from text to be indexed.
97         if (! defined $scrubber) {
98                 eval q{use HTML::Scrubber};
99                 if (! $@) {
100                         $scrubber=HTML::Scrubber->new(allow => []);
101                 }
102         }
103         my $toindex = defined $scrubber ? $scrubber->scrub($params{content}) : $params{content};
104         
105         # Take 512 characters for a sample, then extend it out
106         # if it stopped in the middle of a word.
107         my $size=512;
108         my ($sample)=substr($toindex, 0, $size);
109         if (length($sample) == $size) {
110                 my $max=length($toindex);
111                 my $next;
112                 while ($size < $max &&
113                        ($next=substr($toindex, $size++, 1)) !~ /\s/) {
114                         $sample.=$next;
115                 }
116         }
117         $sample=~s/\n/ /g;
118         
119         my $url=urlto($params{destpage}, "");
120         if (defined $pagestate{$params{page}}{meta}{permalink}) {
121                 $url=$pagestate{$params{page}}{meta}{permalink}
122         }
123
124         # data used by omega
125         # Decode html entities in it, since omega re-encodes them.
126         eval q{use HTML::Entities};
127         error $@ if $@;
128         $doc->set_data(
129                 "url=".$url."\n".
130                 "sample=".decode_entities($sample)."\n".
131                 "caption=".decode_entities($caption)."\n".
132                 "modtime=$IkiWiki::pagemtime{$params{page}}\n".
133                 "size=".length($params{content})."\n"
134         );
135
136         # Index document and add terms for other metadata.
137         my $tg = Search::Xapian::TermGenerator->new();
138         if (! $stemmer) {
139                 my $langcode=$ENV{LANG} || "en";
140                 $langcode=~s/_.*//;
141
142                 # This whitelist is here to work around a xapian bug (#486138)
143                 my @whitelist=qw{da de en es fi fr hu it no pt ru ro sv tr};
144
145                 if (grep { $_ eq $langcode } @whitelist) {
146                         $stemmer=Search::Xapian::Stem->new($langcode);
147                 }
148                 else {
149                         $stemmer=Search::Xapian::Stem->new("english");
150                 }
151         }
152         $tg->set_stemmer($stemmer);
153         $tg->set_document($doc);
154         $tg->index_text($params{page}, 2);
155         $tg->index_text($caption, 2);
156         $tg->index_text($title, 2) if $title ne $caption;
157         $tg->index_text($toindex);
158         $tg->index_text(lc($title), 1, "S"); # for title:foo
159         foreach my $link (@{$links{$params{page}}}) {
160                 $tg->index_text(lc($link), 1, "XLINK"); # for link:bar
161         }
162
163         $doc->add_term($pageterm);
164         $db->replace_document_by_term($pageterm, $doc);
165 }
166
167 sub delete (@) {
168         my $db=xapiandb();
169         foreach my $page (@_) {
170                 my $pageterm=pageterm(pagename($page));
171                 $db->delete_document_by_term($pageterm) if defined $pageterm;
172         }
173 }
174
175 sub cgi ($) {
176         my $cgi=shift;
177
178         if (defined $cgi->param('P')) {
179                 # only works for GET requests
180                 chdir("$config{wikistatedir}/xapian") || error("chdir: $!");
181                 $ENV{OMEGA_CONFIG_FILE}="./omega.conf";
182                 $ENV{CGIURL}=IkiWiki::cgiurl();
183                 IkiWiki::loadindex();
184                 $ENV{HELPLINK}=htmllink("", "", "ikiwiki/searching",
185                         noimageinline => 1, linktext => "Help");
186                 exec($config{omega_cgi}) || error("$config{omega_cgi} failed: $!");
187         }
188 }
189
190 sub pageterm ($) {
191         my $page=shift;
192
193         # 240 is the number used by omindex to decide when to hash an
194         # overlong term. This does not use a compatible hash method though.
195         eval q{use Encode};
196         if (length encode_utf8($page) > 240) {
197                 eval q{use Digest::SHA};
198                 if ($@) {
199                         debug("search: ".sprintf(gettext("need Digest::SHA to index %s"), $page)) if $@;
200                         return undef;
201                 }
202
203                 # Note no colon, therefore it's guaranteed to not overlap
204                 # with a page with the same name as the hash..
205                 return "U".lc(Digest::SHA::sha1_hex($page));
206         }
207         else {
208                 return "U:".$page;
209         }
210 }
211
212 my $db;
213 sub xapiandb () {
214         if (! defined $db) {
215                 eval q{
216                         use Search::Xapian;
217                         use Search::Xapian::WritableDatabase;
218                 };
219                 error($@) if $@;
220                 $db=Search::Xapian::WritableDatabase->new($config{wikistatedir}."/xapian/default",
221                         Search::Xapian::DB_CREATE_OR_OPEN());
222         }
223         return $db;
224 }
225
226 {
227 my $setup=0;
228 sub setupfiles () {
229         if (! $setup and (! -e $config{wikistatedir}."/xapian" || $config{rebuild})) {
230                 writefile("omega.conf", $config{wikistatedir}."/xapian",
231                         "database_dir .\n".
232                         "template_dir ./templates\n");
233                 omega_template();       
234                 $setup=1;
235         }
236 }
237 }
238
239 sub needsbuild {
240         my $list=shift;
241         if (grep {
242                 $_ eq "templates/page.tmpl" ||
243                 $_ eq "templates/searchquery.tmpl"
244         } @$list) {
245                 omega_template();
246         }
247 }
248
249 sub omega_template {
250         # Avoid omega interpreting anything in the cgitemplate
251         # as an omegascript command.
252         eval q{use IkiWiki::CGI};
253         my $template=IkiWiki::cgitemplate(undef, gettext("search"), "\0",
254                 searchform => "", # avoid showing the small search form
255         );
256         eval q{use HTML::Entities};
257         error $@ if $@;
258         $template=encode_entities($template, '\$');
259
260         my $querytemplate=readfile(IkiWiki::template_file("searchquery.tmpl"));
261         $template=~s/\0/$querytemplate/;
262         writefile("query", $config{wikistatedir}."/xapian/templates",
263                 $template);
264 }
265
266 sub disable () {
267         if (-d $config{wikistatedir}."/xapian") {
268                 system("rm", "-rf", $config{wikistatedir}."/xapian");
269         }
270 }
271
272 1