FuzzyIndex

git-svn-id: http://opensvn.csie.org/pttbbs/pttbbs/trunk/pttbbs@966 63ad8ddf-47c3-0310-b6dd-a9e9d9715204

2 files changed, 28 insertions, 5 deletions

diff --git a/blog/blog.pl b/blog/blog.pl
index 96c929af..515c1e65 100755
--- a/blog/blog.pl
+++ b/blog/blog.pl
@@ -1,5 +1,5 @@
 #!/usr/bin/perl
-# $Id: blog.pl,v 1.23 2003/06/19 13:06:58 in2 Exp $
+# $Id: blog.pl,v 1.24 2003/06/25 08:00:32 in2 Exp $
 use CGI qw/:standard/;
 use lib qw/./;
 use LocalVars;
@@ -9,6 +9,7 @@ use Data::Dumper;
 use Date::Calc qw(:all);
 use Template;
 use HTML::Calendar::Simple;
+use OurNet::FuzzyIndex;
 
 use vars qw/@emonth @cnumber %config %attr %article %th/;
 
@@ -69,6 +70,15 @@ sub main
 		       $ptr);
 	}
     }
+    elsif( $attr{"$fn.loadBlog"} =~ /FuzzySearch/i ){
+	my $idx = OurNet::FuzzyIndex->new("$BLOGDATA/$brdname.idx");
+	my %result = $idx->query($th{SearchKey} = param('SearchKey'),
+				 MATCH_FUZZY);
+	foreach my $t (sort { $result{$b} <=> $result{$a} } keys(%result)) {
+	    AddArticle('blog', $attr{"$fn.loadBlogFields"},
+		       $idx->getkey($t), sprintf("%5.1f", $result{$t} / 10));
+	}
+    }
 
     if( $attr{"$fn.loadBlogPrevNext"} ){
 	my $s = packdate($y, $m, $d);
@@ -225,9 +235,9 @@ sub main
 	print "<pre>template error: ". $tmpl->error();
 }
 
-sub AddArticle($$$)
+sub AddArticle($$$;$)
 {
-    my($cl, $fields, $s) = @_;
+    my($cl, $fields, $s, $score) = @_;
     my($content, $short) = ();
     $content = applyfilter($article{"$s.content"}, $config{outputfilter})
 	if( $fields =~ /content/i );
@@ -248,6 +258,7 @@ sub AddArticle($$$)
 		       author => (($fields !~ /author/i) ? '' :
 				  $article{"$s.author"}),
 		       short  => $short,
+		       score  => $score,
 		   }
         if( $article{"$s.title"} );
 }
diff --git a/blog/builddb.pl b/blog/builddb.pl
index b519587b..1cec2615 100755
--- a/blog/builddb.pl
+++ b/blog/builddb.pl
@@ -7,6 +7,7 @@ use IO::Handle;
 use Data::Dumper;
 use BBSFileHeader;
 use DB_File;
+use OurNet::FuzzyIndex;
 
 sub main
 {
@@ -99,12 +100,17 @@ sub buildconfigure($$)
 sub builddata($$$$$$)
 {
     my($board, $rbh, $rebuild, $contentonly, $number, $force) = @_;
-    my(%dat, $dbfn, $y, $m, $d, $t, $currid);
+    my(%dat, $dbfn, $idxfn, $y, $m, $d, $t, $currid, $idx);
 
     $dbfn = "$BLOGROOT/$board.db";
-    unlink $dbfn if( $rebuild );
+    $idxfn = "$BLOGROOT/$board.idx";
+    if( $rebuild ){
+	unlink $dbfn;
+	unlink $idxfn;
+    }
 
     tie %dat, 'DB_File', $dbfn, O_CREAT | O_RDWR, 0666, $DB_HASH;
+    $idx = OurNet::FuzzyIndex->new($idxfn);
     foreach( $number ? $number : (1..($rbh->{num} - 1)) ){
 	if( !(($y, $m, $d, $t) =
 	      $rbh->{"$_.title"} =~ /(\d+)\.(\d+).(\d+),(.*)/) ){
@@ -127,6 +133,9 @@ sub builddata($$$$$$)
 			  $dat{"$currid.content"} = $rbh->{"$_.content"});
 	    $dat{"$currid.short"} = ("$c[0]\n$c[1]\n$c[2]\n$c[3]\n");
 
+	    $idx->delete($currid) if( $idx->findkey($currid) );
+	    $idx->insert($currid, $rbh->{"$_.content"});
+
 	    if( !$contentonly ){
 		debugmsg("\tbuilding $currid linking... ");
 		if( $dat{$currid} ){
@@ -163,6 +172,9 @@ sub builddata($$$$$$)
 	}
     }
     untie %dat;
+    $idx->sync();
+    undef $idx;
+    chmod 0666, $idxfn;
 }
 
 sub getdir($$$$$)