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 => "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 } # }}}
  17
  18 sub getsetup () { #{{{
  19         return
  20                 omega_cgi => {
  21                         type => "string",
  22                         example => "/usr/lib/cgi-bin/omega/omega",
  23                         description => "path to the omega cgi program",
  24                         safe => 0, # external program
  25                         rebuild => 0,
  26                 },
  27 } #}}}
  28
  29 sub checkconfig () { #{{{
  30         foreach my $required (qw(url cgiurl)) {
  31                 if (! length $config{$required}) {
  32                         error(sprintf(gettext("Must specify %s when using the search plugin"), $required));
  33                 }
  34         }
  35
  36         if (! defined $config{omega_cgi}) {
  37                 $config{omega_cgi}="/usr/lib/cgi-bin/omega/omega";
  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 $scrubber;
  60 my $stemmer;
  61 sub index (@) { #{{{
  62         my %params=@_;
  63
  64         setupfiles();
  65
  66         # A unique pageterm is used to identify the document for a page.
  67         my $pageterm=pageterm($params{page});
  68         return $params{content} unless defined $pageterm;
  69
  70         my $db=xapiandb();
  71         my $doc=Search::Xapian::Document->new();
  72         my $caption=IkiWiki::pagetitle($params{page});
  73         my $title;
  74         if (exists $pagestate{$params{page}}{meta} &&
  75                 exists $pagestate{$params{page}}{meta}{title}) {
  76                 $title=$pagestate{$params{page}}{meta}{title};
  77         }
  78         else {
  79                 $title=$caption;
  80         }
  81
  82         # Remove html from text to be indexed.
  83         if (! defined $scrubber) {
  84                 eval q{use HTML::Scrubber};
  85                 if (! $@) {
  86                         $scrubber=HTML::Scrubber->new(allow => []);
  87                 }
  88         }
  89         my $toindex = defined $scrubber ? $scrubber->scrub($params{content}) : $params{content};
  90
  91         # Take 512 characters for a sample, then extend it out
  92         # if it stopped in the middle of a word.
  93         my $size=512;
  94         my ($sample)=substr($toindex, 0, $size);
  95         if (length($sample) == $size) {
  96                 my $max=length($toindex);
  97                 my $next;
  98                 while ($size < $max &&
  99                        ($next=substr($toindex, $size++, 1)) !~ /\s/) {
 100                         $sample.=$next;
 101                 }
 102         }
 103         $sample=~s/\n/ /g;
 104
 105         # data used by omega
 106         # Decode html entities in it, since omega re-encodes them.
 107         eval q{use HTML::Entities};
 108         $doc->set_data(
 109                 "url=".urlto($params{page}, "")."\n".
 110                 "sample=".decode_entities($sample)."\n".
 111                 "caption=".decode_entities($caption)."\n".
 112                 "modtime=$IkiWiki::pagemtime{$params{page}}\n".
 113                 "size=".length($params{content})."\n"
 114         );
 115
 116         # Index document and add terms for other metadata.
 117         my $tg = Search::Xapian::TermGenerator->new();
 118         if (! $stemmer) {
 119                 my $langcode=$ENV{LANG} || "en";
 120                 $langcode=~s/_.*//;
 121
 122                 # This whitelist is here to work around a xapian bug (#486138)
 123                 my @whitelist=qw{da de en es fi fr hu it no pt ru ro sv tr};
 124
 125                 if (grep { $_ eq $langcode } @whitelist) {
 126                         $stemmer=Search::Xapian::Stem->new($langcode);
 127                 }
 128                 else {
 129                         $stemmer=Search::Xapian::Stem->new("english");
 130                 }
 131         }
 132         $tg->set_stemmer($stemmer);
 133         $tg->set_document($doc);
 134         $tg->index_text($params{page}, 2);
 135         $tg->index_text($caption, 2);
 136         $tg->index_text($title, 2) if $title ne $caption;
 137         $tg->index_text($toindex);
 138         $tg->index_text(lc($title), 1, "S"); # for title:foo
 139         foreach my $link (@{$links{$params{page}}}) {
 140                 $tg->index_text(lc($link), 1, "XLINK"); # for link:bar
 141         }
 142
 143         $doc->add_term($pageterm);
 144         $db->replace_document_by_term($pageterm, $doc);
 145 } #}}}
 146
 147 sub delete (@) { #{{{
 148         my $db=xapiandb();
 149         foreach my $page (@_) {
 150                 my $pageterm=pageterm(pagename($page));
 151                 $db->delete_document_by_term($pageterm) if defined $pageterm;
 152         }
 153 } #}}}
 154
 155 sub cgi ($) { #{{{
 156         my $cgi=shift;
 157
 158         if (defined $cgi->param('P')) {
 159                 # only works for GET requests
 160                 chdir("$config{wikistatedir}/xapian") || error("chdir: $!");
 161                 $ENV{OMEGA_CONFIG_FILE}="./omega.conf";
 162                 $ENV{CGIURL}=$config{cgiurl},
 163                 IkiWiki::loadindex();
 164                 $ENV{HELPLINK}=htmllink("", "", "ikiwiki/searching",
 165                         noimageinline => 1, linktext => "Help");
 166                 exec($config{omega_cgi}) || error("$config{omega_cgi} failed: $!");
 167         }
 168 } #}}}
 169
 170 sub pageterm ($) { #{{{
 171         my $page=shift;
 172
 173         # 240 is the number used by omindex to decide when to hash an
 174         # overlong term. This does not use a compatible hash method though.
 175         if (length $page > 240) {
 176                 eval q{use Digest::SHA1};
 177                 if ($@) {
 178                         debug("search: ".sprintf(gettext("need Digest::SHA1 to index %s"), $page)) if $@;
 179                         return undef;
 180                 }
 181
 182                 # Note no colon, therefore it's guaranteed to not overlap
 183                 # with a page with the same name as the hash..
 184                 return "U".lc(Digest::SHA1::sha1_hex($page));
 185         }
 186         else {
 187                 return "U:".$page;
 188         }
 189 } #}}}
 190
 191 my $db;
 192 sub xapiandb () { #{{{
 193         if (! defined $db) {
 194                 eval q{
 195                         use Search::Xapian;
 196                         use Search::Xapian::WritableDatabase;
 197                 };
 198                 error($@) if $@;
 199                 $db=Search::Xapian::WritableDatabase->new($config{wikistatedir}."/xapian/default",
 200                         Search::Xapian::DB_CREATE_OR_OPEN());
 201         }
 202         return $db;
 203 } #}}}
 204
 205 {
 206 my $setup=0;
 207 sub setupfiles () { #{{{
 208         if (! $setup and (! -e $config{wikistatedir}."/xapian" || $config{rebuild})) {
 209                 writefile("omega.conf", $config{wikistatedir}."/xapian",
 210                         "database_dir .\n".
 211                         "template_dir ./templates\n");
 212                 writefile("query", $config{wikistatedir}."/xapian/templates",
 213                         IkiWiki::misctemplate(gettext("search"),
 214                                 readfile(IkiWiki::template_file("searchquery.tmpl"))));
 215                 $setup=1;
 216         }
 217 } #}}}
 218 }
 219
 220 1