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 }
  17
  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 }
  33
  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
  45         # This is a mass dependency, so if the search form template
  46         # changes, every page is rebuilt.
  47         add_depends("", "templates/searchform.tmpl");
  48 }
  49
  50 my $form;
  51 sub pagetemplate (@) {
  52         my %params=@_;
  53         my $page=$params{page};
  54         my $template=$params{template};
  55
  56         # Add search box to page header.
  57         if ($template->query(name => "searchform")) {
  58                 if (! defined $form) {
  59                         my $searchform = template("searchform.tmpl", blind_cache => 1);
  60                         $searchform->param(searchaction => $config{cgiurl});
  61                         $searchform->param(html5 => $config{html5});
  62                         $form=$searchform->output;
  63                 }
  64
  65                 $template->param(searchform => $form);
  66         }
  67 }
  68
  69 my $scrubber;
  70 my $stemmer;
  71 sub indexhtml (@) {
  72         my %params=@_;
  73
  74         setupfiles();
  75
  76         # A unique pageterm is used to identify the document for a page.
  77         my $pageterm=pageterm($params{page});
  78         return unless defined $pageterm;
  79
  80         my $db=xapiandb();
  81         my $doc=Search::Xapian::Document->new();
  82         my $caption=pagetitle($params{page});
  83         my $title;
  84         if (exists $pagestate{$params{page}}{meta} &&
  85                 exists $pagestate{$params{page}}{meta}{title}) {
  86                 $title=$pagestate{$params{page}}{meta}{title};
  87         }
  88         else {
  89                 $title=$caption;
  90         }
  91
  92         # Remove html from text to be indexed.
  93         if (! defined $scrubber) {
  94                 eval q{use HTML::Scrubber};
  95                 if (! $@) {
  96                         $scrubber=HTML::Scrubber->new(allow => []);
  97                 }
  98         }
  99         my $toindex = defined $scrubber ? $scrubber->scrub($params{content}) : $params{content};
 100
 101         # Take 512 characters for a sample, then extend it out
 102         # if it stopped in the middle of a word.
 103         my $size=512;
 104         my ($sample)=substr($toindex, 0, $size);
 105         if (length($sample) == $size) {
 106                 my $max=length($toindex);
 107                 my $next;
 108                 while ($size < $max &&
 109                        ($next=substr($toindex, $size++, 1)) !~ /\s/) {
 110                         $sample.=$next;
 111                 }
 112         }
 113         $sample=~s/\n/ /g;
 114
 115         my $url=urlto($params{destpage}, "");
 116         if (defined $pagestate{$params{page}}{meta}{permalink}) {
 117                 $url=$pagestate{$params{page}}{meta}{permalink}
 118         }
 119
 120         # data used by omega
 121         # Decode html entities in it, since omega re-encodes them.
 122         eval q{use HTML::Entities};
 123         error $@ if $@;
 124         $doc->set_data(
 125                 "url=".$url."\n".
 126                 "sample=".decode_entities($sample)."\n".
 127                 "caption=".decode_entities($caption)."\n".
 128                 "modtime=$IkiWiki::pagemtime{$params{page}}\n".
 129                 "size=".length($params{content})."\n"
 130         );
 131
 132         # Index document and add terms for other metadata.
 133         my $tg = Search::Xapian::TermGenerator->new();
 134         if (! $stemmer) {
 135                 my $langcode=$ENV{LANG} || "en";
 136                 $langcode=~s/_.*//;
 137
 138                 # This whitelist is here to work around a xapian bug (#486138)
 139                 my @whitelist=qw{da de en es fi fr hu it no pt ru ro sv tr};
 140
 141                 if (grep { $_ eq $langcode } @whitelist) {
 142                         $stemmer=Search::Xapian::Stem->new($langcode);
 143                 }
 144                 else {
 145                         $stemmer=Search::Xapian::Stem->new("english");
 146                 }
 147         }
 148         $tg->set_stemmer($stemmer);
 149         $tg->set_document($doc);
 150         $tg->index_text($params{page}, 2);
 151         $tg->index_text($caption, 2);
 152         $tg->index_text($title, 2) if $title ne $caption;
 153         $tg->index_text($toindex);
 154         $tg->index_text(lc($title), 1, "S"); # for title:foo
 155         foreach my $link (@{$links{$params{page}}}) {
 156                 $tg->index_text(lc($link), 1, "XLINK"); # for link:bar
 157         }
 158
 159         $doc->add_term($pageterm);
 160         $db->replace_document_by_term($pageterm, $doc);
 161 }
 162
 163 sub delete (@) {
 164         my $db=xapiandb();
 165         foreach my $page (@_) {
 166                 my $pageterm=pageterm(pagename($page));
 167                 $db->delete_document_by_term($pageterm) if defined $pageterm;
 168         }
 169 }
 170
 171 sub cgi ($) {
 172         my $cgi=shift;
 173
 174         if (defined $cgi->param('P')) {
 175                 # only works for GET requests
 176                 chdir("$config{wikistatedir}/xapian") || error("chdir: $!");
 177                 $ENV{OMEGA_CONFIG_FILE}="./omega.conf";
 178                 $ENV{CGIURL}=$config{cgiurl},
 179                 IkiWiki::loadindex();
 180                 $ENV{HELPLINK}=htmllink("", "", "ikiwiki/searching",
 181                         noimageinline => 1, linktext => "Help");
 182                 exec($config{omega_cgi}) || error("$config{omega_cgi} failed: $!");
 183         }
 184 }
 185
 186 sub pageterm ($) {
 187         my $page=shift;
 188
 189         # 240 is the number used by omindex to decide when to hash an
 190         # overlong term. This does not use a compatible hash method though.
 191         if (length $page > 240) {
 192                 eval q{use Digest::SHA1};
 193                 if ($@) {
 194                         debug("search: ".sprintf(gettext("need Digest::SHA1 to index %s"), $page)) if $@;
 195                         return undef;
 196                 }
 197
 198                 # Note no colon, therefore it's guaranteed to not overlap
 199                 # with a page with the same name as the hash..
 200                 return "U".lc(Digest::SHA1::sha1_hex($page));
 201         }
 202         else {
 203                 return "U:".$page;
 204         }
 205 }
 206
 207 my $db;
 208 sub xapiandb () {
 209         if (! defined $db) {
 210                 eval q{
 211                         use Search::Xapian;
 212                         use Search::Xapian::WritableDatabase;
 213                 };
 214                 error($@) if $@;
 215                 $db=Search::Xapian::WritableDatabase->new($config{wikistatedir}."/xapian/default",
 216                         Search::Xapian::DB_CREATE_OR_OPEN());
 217         }
 218         return $db;
 219 }
 220
 221 {
 222 my $setup=0;
 223 sub setupfiles () {
 224         if (! $setup and (! -e $config{wikistatedir}."/xapian" || $config{rebuild})) {
 225                 writefile("omega.conf", $config{wikistatedir}."/xapian",
 226                         "database_dir .\n".
 227                         "template_dir ./templates\n");
 228
 229                 # Avoid omega interpreting anything in the misctemplate
 230                 # as an omegascript command.
 231                 my $misctemplate=IkiWiki::misctemplate(gettext("search"), "\0");
 232                 eval q{use HTML::Entities};
 233                 error $@ if $@;
 234                 $misctemplate=encode_entities($misctemplate, '\$');
 235
 236                 my $querytemplate=readfile(IkiWiki::template_file("searchquery.tmpl"));
 237                 $misctemplate=~s/\0/$querytemplate/;
 238
 239                 writefile("query", $config{wikistatedir}."/xapian/templates",
 240                         $misctemplate);
 241                 $setup=1;
 242         }
 243 }
 244 }
 245
 246 1