]> sipb.mit.edu Git - ikiwiki.git/blobdiff - IkiWiki/Plugin/search.pm
more substr fixes
[ikiwiki.git] / IkiWiki / Plugin / search.pm
index e705d018af908d73b36269dd0894b350a31b90a2..2de1cf7ac886710b896b12864d6ae76728cb0ada 100644 (file)
@@ -9,8 +9,11 @@ use IkiWiki 2.00;
 sub import { #{{{
        hook(type => "checkconfig", id => "search", call => \&checkconfig);
        hook(type => "pagetemplate", id => "search", call => \&pagetemplate);
+       # run last so other needsbuild hooks can modify the list
+       hook(type => "needsbuild", id => "search", call => \&needsbuild,
+               last => 1);
+       hook(type => "filter", id => "search", call => \&filter);
        hook(type => "delete", id => "search", call => \&delete);
-       hook(type => "change", id => "search", call => \&change);
        hook(type => "cgi", id => "search", call => \&cgi);
 } # }}}
 
@@ -53,12 +56,81 @@ sub pagetemplate (@) { #{{{
        }
 } #}}}
 
-sub delete (@) { #{{{
-       debug(gettext("cleaning xapian search index"));
+my %toindex;
+sub needsbuild ($) { #{{{
+       %toindex = map { pagename($_) => 1 } @{shift()};
+} #}}}
+
+my $scrubber;
+sub filter (@) { #{{{
+       my %params=@_;
+       
+       if ($params{page} eq $params{destpage} && $toindex{$params{page}}) {
+               # index page
+               my $db=xapiandb();
+               my $doc=Search::Xapian::Document->new();
+               my $title;
+               if (exists $pagestate{$params{page}}{meta} &&
+                   exists $pagestate{$params{page}}{meta}{title}) {
+                       $title=$pagestate{$params{page}}{meta}{title};
+               }
+               else {
+                       $title=IkiWiki::pagetitle($params{page});
+               }
+
+               # Remove any html from text to be indexed.
+               # TODO: This removes html that is in eg, a markdown pre,
+               # which should not be removed.
+               if (! defined $scrubber) {
+                       eval q{use HTML::Scrubber};
+                       error($@) if $@;
+                       $scrubber=HTML::Scrubber->new(allow => []);
+               }
+               my $toindex=$scrubber->scrub($params{content});
+               
+               # Take 512 characters for a sample, then extend it out
+               # if it stopped in the middle of a word.
+               my $size=512;
+               my ($sample)=substr($toindex, 0, $size);
+               if (length($sample) == $size) {
+                       my $max=length($toindex);
+                       my $next;
+                       while ($size < $max &&
+                              ($next=substr($toindex, $size++, 1)) !~ /\s/) {
+                               $sample.=$next;
+                       }
+               }
+               $sample=~s/\n/ /g;
+               
+               # data used by omega
+               $doc->set_data(
+                       "url=".urlto($params{page}, "")."\n".
+                       "sample=$sample\n".
+                       "caption=$title\n".
+                       "modtime=$IkiWiki::pagemtime{$params{page}}\n".
+                       "size=".length($params{content})."\n"
+               );
+
+               my $tg = Search::Xapian::TermGenerator->new();
+               $tg->set_stemmer(new Search::Xapian::Stem("english"));
+               $tg->set_document($doc);
+               $tg->index_text($params{page}, 2);
+               $tg->index_text($title, 2);
+               $tg->index_text($toindex);
+
+               my $pageterm=pageterm($params{page});
+               $doc->add_term($pageterm);
+               $db->replace_document_by_term($pageterm, $doc);
+       }
+
+       return $params{content};
 } #}}}
 
-sub change (@) { #{{{
-       debug(gettext("updating xapian search index"));
+sub delete (@) { #{{{
+       my $db=xapiandb();
+       foreach my $page (@_) {
+               $db->delete_document_by_term(pageterm($page));
+       }
 } #}}}
 
 sub cgi ($) { #{{{
@@ -73,4 +145,26 @@ sub cgi ($) { #{{{
        }
 } #}}}
 
+sub pageterm ($) { #{{{
+       my $page=shift;
+
+       # TODO: check if > 255 char page names overflow term
+       # length; use sha1 if so?
+       return "U:".$page;
+} #}}}
+
+my $db;
+sub xapiandb () { #{{{
+       if (! defined $db) {
+               eval q{
+                       use Search::Xapian;
+                       use Search::Xapian::WritableDatabase;
+               };
+               error($@) if $@;
+               $db=Search::Xapian::WritableDatabase->new($config{wikistatedir}."/xapian/default",
+                       Search::Xapian::DB_CREATE_OR_OPEN());
+       }
+       return $db;
+} #}}}
+
 1