Item13405: Fix normalization of filenames on NFD filesystems

RCSStoreContrib/lib/Foswiki/Store/Rcs/Store.pm

@@ -38,6 +38,7 @@ our @ISA = ('Foswiki::Store');
 
 use Assert;
 use Error qw( :try );
+use Encode;
 
 use Foswiki          ();
 use Foswiki::Meta    ();
@@ -113,7 +114,16 @@ sub readTopic {
 
     ( my $text, $isLatest ) = $handler->getRevision($version);
     $text = '' unless defined $text;
-    $text = _decode($text);
+
+    # Use Encode::decode directly. Don't NFC normalize the topic text.
+    # Only need to normalize filenames.
+    $text = Encode::decode(
+        $Foswiki::cfg{Store}{Encoding} || 'utf-8',
+        $text,
+
+        #Encode::FB_CROAK # DEBUG
+        Encode::FB_PERLQQ
+    );
     $text =~ s/\r//g;    # Remove carriage returns
     Foswiki::Serialise::deserialise( $text, 'Embedded', $topicObject );
 

core/lib/Foswiki.spec

@@ -1357,6 +1357,15 @@ $Foswiki::cfg{PluralToSingular} = $TRUE;
 # Leave this undefined to use the default of utf-8 encoding.
 # $Foswiki::cfg{Store}{Encoding} = undef;
 
+# **BOOLEAN LABEL="NFC Normalize Filenames" **
+# Foswiki uses NFC normalization for all network operations, but assumes
+# that the file system is also NFC normalized.  Some systems such as OSx
+# enforce NFD normalization for filenames.  If Foswiki is installed on one
+# of these sysetms, or accesses such a system via a remote file system
+# like NFS, then all directory / filename read operations must be NFC
+# normalized.
+$Foswiki::cfg{NFCNormalizeFilenames} = $FALSE;
+
 # **PERL LABEL="Implementation Classes" EXPERT**
 # Customisation of the Foswiki Store implementation. This allows
 # extension modules to hook into the store implementation at a very low level.

core/lib/Foswiki/Store.pm

@@ -61,6 +61,7 @@ use Assert;
 use Foswiki          ();
 use Foswiki::Sandbox ();
 use HTML::Entities   ();
+use Unicode::Normalize qw(NFC);
 
 BEGIN {
     if ( $Foswiki::cfg{UseLocale} ) {
@@ -251,13 +252,18 @@ currently selected {Store}{Encoding} (or UTF-8 if none is set)
 into perl characters (unicode). May die if $octets contains
 an invalid byte sequence for the encoding.
 
+This utility function should not be used to decode topic text.
+For performance reasons, we don't need to NFC normalize file contents,
+only the filenames themselves.
+
 =cut
 
 sub decode {
     return $_[0] unless defined $_[0];
     my $s = $_[0];
-    return Encode::decode( $Foswiki::cfg{Store}{Encoding} || 'utf-8',
+    my $decoded = Encode::decode( $Foswiki::cfg{Store}{Encoding} || 'utf-8',
         $s, Encode::FB_CROAK );
+    return ( $Foswiki::cfg{NFCNormalizeFilenames} ) ? NFC($decoded) : $decoded;
 }
 
 =begin TML