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.
  84                 if (! defined $scrubber) {
  85                         eval q{use HTML::Scrubber};
  86                         error($@) if $@;
  87                         $scrubber=HTML::Scrubber->new(allow => []);
  88                 }
  89                 my $toindex=$scrubber->scrub($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 $next=substr($toindex, $size++, 1);
  97                         while (defined $next && $next !~ /\s/) {
  98                                 $sample.=$next;
  99                                 $next=substr($toindex, $size++, 1);
 100                         }
 101                 }
 102                 $sample=~s/\n/ /g;
 103
 104                 # data used by omega
 105                 $doc->set_data(
 106                         "url=".urlto($params{page}, "")."\n".
 107                         "sample=$sample\n".
 108                         "caption=$title\n".
 109                         "modtime=$IkiWiki::pagemtime{$params{page}}\n".
 110                         "size=".length($params{content})."\n"
 111                 );
 112
 113                 my $tg = Search::Xapian::TermGenerator->new();
 114                 $tg->set_stemmer(new Search::Xapian::Stem("english"));
 115                 $tg->set_document($doc);
 116                 $tg->index_text($params{page}, 2);
 117                 $tg->index_text($title, 2);
 118                 $tg->index_text($toindex);
 119
 120                 my $pageterm=pageterm($params{page});
 121                 $doc->add_term($pageterm);
 122                 $db->replace_document_by_term($pageterm, $doc);
 123         }
 124
 125         return $params{content};
 126 } #}}}
 127
 128 sub delete (@) { #{{{
 129         my $db=xapiandb();
 130         foreach my $page (@_) {
 131                 $db->delete_document_by_term(pageterm($page));
 132         }
 133 } #}}}
 134
 135 sub cgi ($) { #{{{
 136         my $cgi=shift;
 137
 138         if (defined $cgi->param('P')) {
 139                 # only works for GET requests
 140                 chdir("$config{wikistatedir}/xapian") || error("chdir: $!");
 141                 $ENV{OMEGA_CONFIG_FILE}="./omega.conf";
 142                 $ENV{CGIURL}=$config{cgiurl},
 143                 exec($config{omega_cgi}) || error("$config{omega_cgi} failed: $!");
 144         }
 145 } #}}}
 146
 147 sub pageterm ($) { #{{{
 148         my $page=shift;
 149
 150         # TODO: check if > 255 char page names overflow term
 151         # length; use sha1 if so?
 152         return "U".$page;
 153 } #}}}
 154
 155 my $db;
 156 sub xapiandb () { #{{{
 157         if (! defined $db) {
 158                 eval q{
 159                         use Search::Xapian;
 160                         use Search::Xapian::WritableDatabase;
 161                 };
 162                 error($@) if $@;
 163                 $db=Search::Xapian::WritableDatabase->new($config{wikistatedir}."/xapian/default",
 164                         Search::Xapian::DB_CREATE_OR_OPEN());
 165         }
 166         return $db;
 167 } #}}}
 168
 169 1