more substr fixes

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

diff --git a/IkiWiki/Plugin/search.pm b/IkiWiki/Plugin/search.pm
index a28d58a1d4b5403583d0cd5207f6c7bf2ada0183..2de1cf7ac886710b896b12864d6ae76728cb0ada 100644 (file)
--- a/IkiWiki/Plugin/search.pm
+++ b/IkiWiki/Plugin/search.pm
@@ -58,9 +58,10 @@ sub pagetemplate (@) { #{{{
 
 my %toindex;
 sub needsbuild ($) { #{{{
-       %toindex = map { $_ => 1 } @{shift()};
+       %toindex = map { pagename($_) => 1 } @{shift()};
 } #}}}
 
+my $scrubber;
 sub filter (@) { #{{{
        my %params=@_;
        
@@ -68,16 +69,43 @@ sub filter (@) { #{{{
                # index page
                my $db=xapiandb();
                my $doc=Search::Xapian::Document->new();
-               my $title=$params{page};
+               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=\n". # TODO
+                       "sample=$sample\n".
                        "caption=$title\n".
                        "modtime=$IkiWiki::pagemtime{$params{page}}\n".
                        "size=".length($params{content})."\n"
@@ -88,7 +116,7 @@ sub filter (@) { #{{{
                $tg->set_document($doc);
                $tg->index_text($params{page}, 2);
                $tg->index_text($title, 2);
-               $tg->index_text($params{content}); # TODO html strip; preprocessor too
+               $tg->index_text($toindex);
 
                my $pageterm=pageterm($params{page});
                $doc->add_term($pageterm);
@@ -122,7 +150,7 @@ sub pageterm ($) { #{{{
 
        # TODO: check if > 255 char page names overflow term
        # length; use sha1 if so?
-       return "P".$page;
+       return "U:".$page;
 } #}}}
 
 my $db;