--- trunk/Fast.pm	2007/11/06 20:06:07	24
+++ trunk/lib/MARC/Fast.pm	2013/08/22 11:24:36	47
@@ -2,12 +2,12 @@
 
 use strict;
 use Carp;
-use Data::Dumper;
+use Data::Dump qw/dump/;
 
 BEGIN {
 	use Exporter ();
 	use vars qw ($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS);
-	$VERSION     = 0.09;
+	$VERSION     = 0.12;
 	@ISA         = qw (Exporter);
 	#Give a hoot don't pollute, do not export more than needed by default
 	@EXPORT      = qw ();
@@ -87,7 +87,8 @@
 		my $len = read($self->{fh}, $leader, 24);
 
 		if ($len < 24) {
-			carp "short read of leader, aborting\n";
+			warn "short read of leader, aborting\n";
+			$self->{count}--;
 			last;
 		}
 
@@ -124,6 +125,7 @@
 
 		# skip to next record
 		my $o = substr($leader,0,5);
+		warn "# in record ", $self->{count}," record length isn't number but: ",dump($o),"\n" unless $o =~ m/^\d+$/;
 		if ($o > 24) {
 			seek($self->{fh},$o-24,1) if ($o);
 		} else {
@@ -275,11 +277,14 @@
 
 Read record with specified MFN and convert it to hash
 
-  my $hash = $marc->to_hash( $mfn, include_subfields => 1, );
+  my $hash = $marc->to_hash( $mfn, include_subfields => 1,
+	hash_filter => sub { my ($l,$tag) = @_; return $l; }
+  );
 
 It has ability to convert characters (using C<hash_filter>) from MARC
 database before creating structures enabling character re-mapping or quick
-fix-up of data.
+fix-up of data. If you specified C<hash_filter> both in C<new> and C<to_hash>
+only the one from C<to_hash> will be used.
 
 This function returns hash which is like this:
 
@@ -303,20 +308,21 @@
 	my $mfn = shift || confess "need mfn!";
 
 	my $args = {@_};
+	my $filter_coderef = $args->{'hash_filter'} || $self->{'hash_filter'};
 
 	# init record to include MFN as field 000
 	my $rec = { '000' => [ $mfn ] };
 
 	my $row = $self->fetch($mfn) || return;
 
-	foreach my $rec_nr (keys %{$row}) {
-		foreach my $l (@{$row->{$rec_nr}}) {
+	foreach my $tag (keys %{$row}) {
+		foreach my $l (@{$row->{$tag}}) {
 
 			# remove end marker
 			$l =~ s/\x1E$//;
 
 			# filter output
-			$l = $self->{'hash_filter'}->($l, $rec_nr) if ($self->{'hash_filter'});
+			$l = $filter_coderef->($l, $tag) if $filter_coderef;
 
 			my $val;
 
@@ -331,25 +337,27 @@
 				foreach my $t (split(/\x1F/,$l)) {
 					next if (! $t);
 					my $f = substr($t,0,1);
+					my $v = substr($t,1);
 
 					push @subfields, ( $f, $sf_usage->{$f}++ || 0 );
 
 					# repeatable subfiled -- convert it to array
-					if ($val->{$f}) {
+					if ( defined $val->{$f} ) {
 						if ( ref($val->{$f}) ne 'ARRAY' ) {
-							$val->{$f} = [ $val->{$f}, $val ];
+							$val->{$f} = [ $val->{$f}, $v ];
 						} else {
-							push @{$val->{$f}}, $val;
+							push @{$val->{$f}}, $v;
 						}
+					} else {
+						$val->{$f} = $v;
 					}
-					$val->{substr($t,0,1)} = substr($t,1);
 				}
 				$val->{subfields} = [ @subfields ] if $args->{include_subfields};
 			} else {
 				$val = $l;
 			}
 
-			push @{$rec->{$rec_nr}}, $val;
+			push @{$rec->{$tag}}, $val;
 		}
 	}
 
@@ -383,6 +391,27 @@
 1;
 __END__
 
+=head1 UTF-8 ENCODING
+
+This module does nothing with encoding. But, since MARC format is byte
+oriented even when using UTF-8 which has variable number of bytes for each
+character, file is opened in binary mode.
+
+As a result, all scalars recturned to perl don't have utf-8 flag. Solution is
+to use C<hash_filter> and L<Encode> to decode utf-8 encoding like this:
+
+  use Encode;
+
+  my $marc = new MARC::Fast(
+  	marcdb => 'utf8.marc',
+	hash_filter => sub {
+		Encode::decode( 'utf-8', $_[0] );
+	},
+  );
+
+This will affect C<to_hash>, but C<fetch> will still return binary representation
+since it doesn't support C<hash_filter>.
+
 =head1 AUTHOR
 
 	Dobrica Pavlinusic