add news item for ikiwiki 2.49

[ikiwiki.git] / IkiWiki / Plugin / search.pm

diff --git a/IkiWiki/Plugin/search.pm b/IkiWiki/Plugin/search.pm
index 3dc70f0b792bd89cb8dc668463dbb66f4ba5df99..3ac4351979e7b96bd81c334d7d873cb40d612498 100644 (file)
--- a/IkiWiki/Plugin/search.pm
+++ b/IkiWiki/Plugin/search.pm
@@ -20,19 +20,10 @@ sub checkconfig () { #{{{
                        error(sprintf(gettext("Must specify %s when using the search plugin"), $required));
                }
        }
-
+       
        if (! exists $config{omega_cgi}) {
                $config{omega_cgi}="/usr/lib/cgi-bin/omega/omega";
        }
-       
-       if (! -e $config{wikistatedir}."/xapian" || $config{rebuild}) {
-               writefile("omega.conf", $config{wikistatedir}."/xapian",
-                       "database_dir .\n".
-                       "template_dir ./templates\n");
-               writefile("query", $config{wikistatedir}."/xapian/templates",
-                       IkiWiki::misctemplate(gettext("search"),
-                               readfile(IkiWiki::template_file("searchquery.tmpl"))));
-       }
 } #}}}
 
 my $form;
@@ -54,23 +45,31 @@ sub pagetemplate (@) { #{{{
 } #}}}
 
 my $scrubber;
+my $stemmer;
 sub index (@) { #{{{
        my %params=@_;
        
-       return $params{content} if %IkiWiki::preprocessing;
+       return $params{content} if $IkiWiki::preprocessing{$params{destpage}};
+
+       setupfiles();
+
+       # A unique pageterm is used to identify the document for a page.
+       my $pageterm=pageterm($params{page});
+       return $params{content} unless defined $pageterm;
        
        my $db=xapiandb();
        my $doc=Search::Xapian::Document->new();
+       my $caption=IkiWiki::pagetitle($params{page});
        my $title;
        if (exists $pagestate{$params{page}}{meta} &&
-           exists $pagestate{$params{page}}{meta}{title}) {
+               exists $pagestate{$params{page}}{meta}{title}) {
                $title=$pagestate{$params{page}}{meta}{title};
        }
        else {
-               $title=IkiWiki::pagetitle($params{page});
+               $title=$caption;
        }
 
-       # Remove any html from text to be indexed.
+       # Remove html from text to be indexed.
        if (! defined $scrubber) {
                eval q{use HTML::Scrubber};
                if (! $@) {
@@ -94,22 +93,37 @@ sub index (@) { #{{{
        $sample=~s/\n/ /g;
        
        # data used by omega
+       # Decode html entities in it, since omega re-encodes them.
+       eval q{use HTML::Entities};
        $doc->set_data(
                "url=".urlto($params{page}, "")."\n".
-               "sample=$sample\n".
-               "caption=$title\n".
+               "sample=".decode_entities($sample)."\n".
+               "caption=".decode_entities($caption)."\n".
                "modtime=$IkiWiki::pagemtime{$params{page}}\n".
                "size=".length($params{content})."\n"
        );
 
+       # Index document and add terms for other metadata.
        my $tg = Search::Xapian::TermGenerator->new();
-       $tg->set_stemmer(new Search::Xapian::Stem("english"));
+       if (! $stemmer) {
+               my $langcode=$ENV{LANG} || "en";
+               $langcode=~s/_.*//;
+               eval { $stemmer=Search::Xapian::Stem->new($langcode) };
+               if ($@) {
+                       $stemmer=Search::Xapian::Stem->new("english");
+               }
+       }
+       $tg->set_stemmer($stemmer);
        $tg->set_document($doc);
        $tg->index_text($params{page}, 2);
-       $tg->index_text($title, 2);
+       $tg->index_text($caption, 2);
+       $tg->index_text($title, 2) if $title ne $caption;
        $tg->index_text($toindex);
+       $tg->index_text(lc($title), 1, "ZS"); # for title:foo
+       foreach my $link (@{$links{$params{page}}}) {
+               $tg->index_text(lc($link), 1, "ZLINK"); # for link:bar
+       }
 
-       my $pageterm=pageterm($params{page});
        $doc->add_term($pageterm);
        $db->replace_document_by_term($pageterm, $doc);
 
@@ -119,7 +133,8 @@ sub index (@) { #{{{
 sub delete (@) { #{{{
        my $db=xapiandb();
        foreach my $page (@_) {
-               $db->delete_document_by_term(pageterm($page));
+               my $pageterm=pageterm(pagename($page));
+               $db->delete_document_by_term($pageterm) if defined $pageterm;
        }
 } #}}}
 
@@ -131,6 +146,9 @@ sub cgi ($) { #{{{
                chdir("$config{wikistatedir}/xapian") || error("chdir: $!");
                $ENV{OMEGA_CONFIG_FILE}="./omega.conf";
                $ENV{CGIURL}=$config{cgiurl},
+               IkiWiki::loadindex();
+               $ENV{HELPLINK}=htmllink("", "", "ikiwiki/searching",
+                       noimageinline => 1, linktext => "Help");
                exec($config{omega_cgi}) || error("$config{omega_cgi} failed: $!");
        }
 } #}}}
@@ -138,9 +156,22 @@ sub cgi ($) { #{{{
 sub pageterm ($) { #{{{
        my $page=shift;
 
-       # TODO: check if > 255 char page names overflow term
-       # length; use sha1 if so?
-       return "U:".$page;
+       # 240 is the number used by omindex to decide when to hash an
+       # overlong term. This does not use a compatible hash method though.
+       if (length $page > 240) {
+               eval q{use Digest::SHA1};
+               if ($@) {
+                       debug("search: ".sprintf(gettext("need Digest::SHA1 to index %s"), $page)) if $@;
+                       return undef;
+               }
+
+               # Note no colon, therefore it's guaranteed to not overlap
+               # with a page with the same name as the hash..
+               return "U".lc(Digest::SHA1::sha1_hex($page));
+       }
+       else {
+               return "U:".$page;
+       }
 } #}}}
 
 my $db;
@@ -157,4 +188,15 @@ sub xapiandb () { #{{{
        return $db;
 } #}}}
 
+sub setupfiles () { #{{{
+       if (! -e $config{wikistatedir}."/xapian" || $config{rebuild}) {
+               writefile("omega.conf", $config{wikistatedir}."/xapian",
+                       "database_dir .\n".
+                       "template_dir ./templates\n");
+               writefile("query", $config{wikistatedir}."/xapian/templates",
+                       IkiWiki::misctemplate(gettext("search"),
+                               readfile(IkiWiki::template_file("searchquery.tmpl"))));
+       }
+} #}}}
+
 1