X-Git-Url: http://git.rot13.org/?p=MARC-Fast;a=blobdiff_plain;f=Fast.pm;h=aef3d5b5baead4b41eb5ec183f6196696f87c52f;hp=f944654aa902d6e30a897618a30238f334ecb4cd;hb=cabfd93a9ff6f82e8b67e3050ca3c2ea1ef0af37;hpb=5fc18887422d2dd62c716b73d81a226f3a295e5b

diff --git a/Fast.pm b/Fast.pm
index f944654..aef3d5b 100644
--- a/Fast.pm
+++ b/Fast.pm
@@ -7,7 +7,7 @@ use Data::Dump qw/dump/;
 BEGIN {
 	use Exporter ();
 	use vars qw ($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS);
-	$VERSION     = 0.09;
+	$VERSION     = 0.10;
 	@ISA         = qw (Exporter);
 	#Give a hoot don't pollute, do not export more than needed by default
 	@EXPORT      = qw ();
@@ -87,7 +87,8 @@ sub new {
 		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;
 		}
 
@@ -310,14 +311,14 @@ sub to_hash {
 
 	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 = $self->{'hash_filter'}->($l, $tag) if ($self->{'hash_filter'});
 
 			my $val;
 
@@ -350,7 +351,7 @@ sub to_hash {
 				$val = $l;
 			}
 
-			push @{$rec->{$rec_nr}}, $val;
+			push @{$rec->{$tag}}, $val;
 		}
 	}
 
@@ -384,6 +385,27 @@ sub to_ascii {
 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