[mb-commits] r9211 - mb_server/branches/TaggingAgain-BRANCH/cgi-bin/MusicBrainz/Server

Author: luks
Date: 2007-06-24 08:07:21 +0000 (Sun, 24 Jun 2007)
New Revision: 9211

Modified:
   mb_server/branches/TaggingAgain-BRANCH/cgi-bin/MusicBrainz/Server/Tag.pm
Log:
Don't replace non-ascii 'word' characters in tag names. Allow '-' in tag names.

Modified: 
mb_server/branches/TaggingAgain-BRANCH/cgi-bin/MusicBrainz/Server/Tag.pm
===================================================================
--- mb_server/branches/TaggingAgain-BRANCH/cgi-bin/MusicBrainz/Server/Tag.pm    
2007-06-23 22:26:05 UTC (rev 9210)
+++ mb_server/branches/TaggingAgain-BRANCH/cgi-bin/MusicBrainz/Server/Tag.pm    
2007-06-24 08:07:21 UTC (rev 9211)
@@ -32,6 +32,7 @@
 use List::Util qw( min max );
 use URI::Escape qw( uri_escape );
 use MusicBrainz::Server::Validation qw( encode_entities );
+use Encode qw( decode encode );
 
 sub Update
 {
@@ -41,13 +42,14 @@
 
        @new_tags = grep {
                # remove non-word characters
-               $_ =~ s/\W+/ /sg;
+               $_ =~ s/[^\p{IsWord}-]+/ /sg;
                # combine multiple spaces into one
                $_ =~ s/\s+/ /sg;
                # remove leading and trailing whitespace
                $_ =~ s/^\s*(.*?)\s*$/$1/;
+               $_ = encode "utf-8", $_;
                $_;
-       } split ',', lc($input);
+       } split ',', lc(decode "utf-8", $input);
        # make sure the list contains only unique tags
        @new_tags = keys %{{ map { $_ => 1 } @new_tags }};