--- trunk/Fast.pm	2005/12/28 22:16:39	8
+++ trunk/lib/MARC/Fast.pm	2013/08/22 11:24:36	47
@@ -1,13 +1,13 @@
-
 package MARC::Fast;
+
 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     = 0.03;
+	$VERSION     = 0.12;
 	@ISA         = qw (Exporter);
 	#Give a hoot don't pollute, do not export more than needed by default
 	@EXPORT      = qw ();
@@ -23,12 +23,21 @@
 
   use MARC::Fast;
 
+  my $marc = new MARC::Fast(
+  	marcdb => 'unimarc.iso',
+  );
+
+  foreach my $mfn ( 1 .. $marc->count ) {
+  	print $marc->to_ascii( $mfn );
+  }
+
+For longer example with command line options look at L<scripts/dump_fastmarc.pl>
 
 =head1 DESCRIPTION
 
 This is very fast alternative to C<MARC> and C<MARC::Record> modules.
 
-It's is also very sutable for random access to MARC records (as opposed to
+It's is also very subtable for random access to MARC records (as opposed to
 sequential one).
 
 =head1 METHODS
@@ -43,7 +52,7 @@
 	debug => 0,
 	assert => 0,
 	hash_filter => sub {
-		my $t = shift;
+		my ($t, $record_number) = @_;
 		$t =~ s/foo/bar/;
 		return $t;
 	},
@@ -78,7 +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;
 		}
 
@@ -111,10 +121,11 @@
 		print STDERR "REC ",$self->{count},": $leader\n" if ($self->{debug});
 
 		# store leader for later
-		push @{$self->{leaders}}, $leader;
+		push @{$self->{leader}}, $leader;
 
 		# 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 {
@@ -145,14 +156,22 @@
 
   my $hash = $marc->fetch(42);
 
+First record number is C<1>
+
 =cut
 
 sub fetch {
 	my $self = shift;
 
-	my $rec_nr = shift || return;
+	my $rec_nr = shift;
+
+	if ( ! $rec_nr ) {
+		$self->{last_leader} = undef;
+		return;
+	}
 
-	my $leader = $self->{leaders}->[$rec_nr - 1];
+	my $leader = $self->{leader}->[$rec_nr - 1];
+	$self->{last_leader} = $leader;
 	unless ($leader) {
 		carp "can't find record $rec_nr";
 		return;
@@ -234,15 +253,38 @@
 }
 
 
+=head2 last_leader
+
+Returns leader of last record L<fetch>ed
+
+  print $marc->last_leader;
+
+Added in version 0.08 of this module, so if you need it use:
+
+  use MARC::Fast 0.08;
+
+to be sure that it's supported.
+
+=cut
+
+sub last_leader {
+	my $self = shift;
+	return $self->{last_leader};
+}
+
+
 =head2 to_hash
 
 Read record with specified MFN and convert it to hash
 
-  my $hash = $marc->to_hash($mfn);
+  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:
 
@@ -265,65 +307,110 @@
 
 	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 $k (keys %{$row}) {
-		foreach my $l (@{$row->{$k}}) {
+	foreach my $tag (keys %{$row}) {
+		foreach my $l (@{$row->{$tag}}) {
 
 			# remove end marker
 			$l =~ s/\x1E$//;
 
 			# filter output
-			$l = $self->{'hash_filter'}->($l) if ($self->{'hash_filter'});
+			$l = $filter_coderef->($l, $tag) if $filter_coderef;
 
 			my $val;
 
 			# has identifiers?
 			($val->{'i1'},$val->{'i2'}) = ($1,$2) if ($l =~ s/^([01 #])([01 #])\x1F/\x1F/);
 
+			my $sf_usage;
+			my @subfields;
+
 			# has subfields?
 			if ($l =~ m/\x1F/) {
 				foreach my $t (split(/\x1F/,$l)) {
 					next if (! $t);
 					my $f = substr($t,0,1);
-					# repeatable subfileds. When we hit first one,
-					# store CURRENT (up to that) in first repetition
-					# of this record. Then, new record with same
-					# identifiers will be created.
-					if ($val->{$f}) {
-						push @{$rec->{$k}}, $val;
-						$val = {
-							i1 => $val->{i1},
-							i2 => $val->{i2},
-						};
+					my $v = substr($t,1);
+
+					push @subfields, ( $f, $sf_usage->{$f}++ || 0 );
+
+					# repeatable subfiled -- convert it to array
+					if ( defined $val->{$f} ) {
+						if ( ref($val->{$f}) ne 'ARRAY' ) {
+							$val->{$f} = [ $val->{$f}, $v ];
+						} else {
+							push @{$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 @{$rec->{$k}}, $val;
+			push @{$rec->{$tag}}, $val;
 		}
 	}
 
 	return $rec;
 }
 
+=head2 to_ascii
+
+  print $marc->to_ascii( 42 );
+
+=cut
+
+sub to_ascii {
+	my $self = shift;
+
+	my $mfn = shift || confess "need mfn";
+	my $row = $self->fetch($mfn) || return;
+
+	my $out;
+
+	foreach my $f (sort keys %{$row}) {
+		my $dump = join('', @{ $row->{$f} });
+		$dump =~ s/\x1e$//;
+		$dump =~ s/\x1f/\$/g;
+		$out .= "$f\t$dump\n";
+	}
+
+	return $out;
+}
 
 1;
 __END__
 
-=head1 BUGS
+=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:
 
-=head1 SUPPORT
+  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
 
@@ -343,6 +430,6 @@
 
 =head1 SEE ALSO
 
-perl(1).
+L<Biblio::Isis>, perl(1).
 
 =cut