X-Git-Url: http://koha-dev.rot13.org:8081/gitweb/?a=blobdiff_plain;f=C4%2FCharset.pm;h=8b69848d5dd5f8bdd6f25aad5c7b969821ac9ad0;hb=1e7437bbae653840136fd1c2faed80a83aa08d2b;hp=a4e6b716f8e5b44d251c0cebb1475e827ec9ac88;hpb=c548761bc190d9b6a5f2989c0ab01b54f4d61879;p=srvgit

diff --git a/C4/Charset.pm b/C4/Charset.pm
index a4e6b716f8..8b69848d5d 100644
--- a/C4/Charset.pm
+++ b/C4/Charset.pm
@@ -39,6 +39,7 @@ BEGIN {
         SetUTF8Flag
         SetMarcUnicodeFlag
         StripNonXmlChars
+        nsb_clean
     );
 }
 
@@ -382,6 +383,40 @@ sub StripNonXmlChars {
     return $str;
 }
 
+
+
+=head2 nsb_clean
+
+=over 4
+
+nsb_clean($string);
+
+=back
+
+Removes Non Sorting Block characters
+
+=cut
+sub nsb_clean {
+    my $NSB  = '\x88' ;        # NSB : begin Non Sorting Block
+    my $NSE  = '\x89' ;        # NSE : Non Sorting Block end
+    my $NSB2 = '\x98' ;        # NSB : begin Non Sorting Block
+    my $NSE2 = '\x9C' ;        # NSE : Non Sorting Block end
+    my $C2   = '\xC2' ;        # What is this char ? It is sometimes left by the regexp after removing NSB / NSE
+
+    # handles non sorting blocks
+    my ($string) = @_ ;
+    $_ = $string ;
+    s/$NSB//g ;
+    s/$NSE//g ;
+    s/$NSB2//g ;
+    s/$NSE2//g ;
+    s/$C2//g ;
+    $string = $_ ;
+
+    return($string) ;
+}
+
+
 =head1 INTERNAL FUNCTIONS
 
 =head2 _default_marc21_charconv_to_utf8