--- trunk/lib/WebPAC/Normalize.pm	2006/06/30 18:48:33	550
+++ trunk/lib/WebPAC/Normalize.pm	2006/09/11 12:56:05	667
@@ -3,13 +3,20 @@
 @EXPORT = qw/
 	_set_rec _set_lookup
 	_get_ds _clean_ds
+	_debug
+	_pack_subfields_hash
 
 	tag search display
 	marc marc_indicators marc_repeatable_subfield
+	marc_compose marc_leader
+	marc_duplicate marc_remove
+	marc_original_order
 
 	rec1 rec2 rec
 	regex prefix suffix surround
 	first lookup join_with
+
+	split_rec_on
 /;
 
 use warnings;
@@ -17,7 +24,7 @@
 
 #use base qw/WebPAC::Common/;
 use Data::Dump qw/dump/;
-use Encode qw/from_to/;
+use Storable qw/dclone/;
 
 # debugging warn(s)
 my $debug = 0;
@@ -29,11 +36,11 @@
 
 =head1 VERSION
 
-Version 0.06
+Version 0.19
 
 =cut
 
-our $VERSION = '0.06';
+our $VERSION = '0.19';
 
 =head1 SYNOPSIS
 
@@ -63,6 +70,7 @@
 	row => $row,
 	rules => $normalize_pl_config,
 	marc_encoding => 'utf-8',
+	config => $config,
   );
 
 Options C<lookup>, C<row>, C<rules> and C<log> are mandatory while all
@@ -84,6 +92,7 @@
 	no strict 'subs';
 	_set_lookup( $arg->{lookup} );
 	_set_rec( $arg->{row} );
+	_set_config( $arg->{config} );
 	_clean_ds( %{ $arg } );
 	eval "$arg->{rules}";
 	die "error evaling $arg->{rules}: $@\n" if ($@);
@@ -105,6 +114,34 @@
 	$rec = shift or die "no record hash";
 }
 
+=head2 _set_config
+
+Set current config hash
+
+  _set_config( $config );
+
+Magic keys are:
+
+=over 4
+
+=item _
+
+Code of current database
+
+=item _mfn
+
+Current MFN
+
+=back
+
+=cut
+
+my $config;
+
+sub _set_config {
+	$config = shift;
+}
+
 =head2 _get_ds
 
 Return hash formatted as data structure
@@ -113,7 +150,8 @@
 
 =cut
 
-my ($out,$marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators);
+my ($out, $marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators);
+my ($marc_record_offset, $marc_fetch_offset) = (0, 0);
 
 sub _get_ds {
 	return $out;
@@ -130,6 +168,7 @@
 sub _clean_ds {
 	my $a = {@_};
 	($out,$marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators) = ();
+	($marc_record_offset, $marc_fetch_offset) = (0,0);
 	$marc_encoding = $a->{marc_encoding};
 }
 
@@ -153,40 +192,96 @@
 
 	$marc->add_fields( WebPAC::Normalize:_get_marc_fields() );
 
-
-
 We are using I<magic> which detect repeatable fields only from
 sequence of field/subfield data generated by normalization.
 
-Repeatable field is created if there is second occurence of same subfield or
-if any of indicators are different. This is sane for most cases except for
-non-repeatable fields with repeatable subfields.
+Repeatable field is created when there is second occurence of same subfield or
+if any of indicators are different.
+
+This is sane for most cases. Something like:
+
+  900a-1 900b-1 900c-1
+  900a-2 900b-2
+  900a-3
+
+will be created from any combination of:
+
+  900a-1 900a-2 900a-3 900b-1 900b-2 900c-1
+
+and following rules:
+
+  marc('900','a', rec('200','a') );
+  marc('900','b', rec('200','b') );
+  marc('900','c', rec('200','c') );
+
+which might not be what you have in mind. If you need repeatable subfield,
+define it using C<marc_repeatable_subfield> like this:
+
+  marc_repeatable_subfield('900','a');
+  marc('900','a', rec('200','a') );
+  marc('900','b', rec('200','b') );
+  marc('900','c', rec('200','c') );
+
+will create:
+
+  900a-1 900a-2 900a-3 900b-1 900c-1
+  900b-2
+
+There is also support for returning next or specific using:
+
+  while (my $mf = WebPAC::Normalize:_get_marc_fields( fetch_next => 1 ) ) {
+  	# do something with $mf
+  }
+
+will always return fields from next MARC record or
+
+  my $mf = WebPAC::Normalize::_get_marc_fields( offset => 42 );
 
-You can change behaviour of that using C<marc_repeatable_subfield>.
+will return 42th copy record (if it exists).
 
 =cut
 
 sub _get_marc_fields {
 
+	my $arg = {@_};
+	warn "### _get_marc_fields arg: ", dump($arg), $/ if ($debug > 2);
+	my $offset = $marc_fetch_offset;
+	if ($arg->{offset}) {
+		$offset = $arg->{offset};
+	} elsif($arg->{fetch_next}) {
+		$marc_fetch_offset++;
+	}
+
+	return if (! $marc_record || ref($marc_record) ne 'ARRAY');
+
+	warn "### full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 2);
+
+	my $marc_rec = $marc_record->[ $offset ];
+
+	warn "## _get_marc_fields (at offset: $offset) -- marc_record = ", dump( @$marc_rec ), $/ if ($debug > 1);
+
+	return if (! $marc_rec || ref($marc_rec) ne 'ARRAY' || $#{ $marc_rec } < 0);
 
 	# first, sort all existing fields 
 	# XXX might not be needed, but modern perl might randomize elements in hash
 	my @sorted_marc_record = sort {
-		$a->[0] . $a->[3] cmp $b->[0] . $b->[3]	
-	} @{ $marc_record };
+		$a->[0] . ( $a->[3] || '' ) cmp $b->[0] . ( $b->[3] || '')
+	} @{ $marc_rec };
 
+	@sorted_marc_record = @{ $marc_rec };	### FIXME disable sorting
+	
 	# output marc fields
 	my @m;
 
 	# count unique field-subfields (used for offset when walking to next subfield)
 	my $u;
-	map { $u->{ $_->[0] . $_->[3]  }++ } @sorted_marc_record;
+	map { $u->{ $_->[0] . ( $_->[3] || '')  }++ } @sorted_marc_record;
 
 	if ($debug) {
-		warn "## marc_repeatable_subfield ", dump( $marc_repeatable_subfield ), $/;
-		warn "## marc_record ", dump( $marc_record ), $/;
-		warn "## sorted_marc_record ", dump( \@sorted_marc_record ), $/;
-		warn "## subfield count ", dump( $u ), $/;
+		warn "## marc_repeatable_subfield = ", dump( $marc_repeatable_subfield ), $/ if ( $marc_repeatable_subfield );
+		warn "## marc_record[$offset] = ", dump( $marc_rec ), $/;
+		warn "## sorted_marc_record = ", dump( \@sorted_marc_record ), $/;
+		warn "## subfield count = ", dump( $u ), $/;
 	}
 
 	my $len = $#sorted_marc_record;
@@ -204,11 +299,11 @@
 		# mark it visited
 		$visited->{$i}++;
 
-		my $row = $sorted_marc_record[$i];
+		my $row = dclone( $sorted_marc_record[$i] );
 
 		# field and subfield which is key for
 		# marc_repeatable_subfield and u
-		my $fsf = $row->[0] . $row->[3];
+		my $fsf = $row->[0] . ( $row->[3] || '' );
 
 		if ($debug > 1) {
 
@@ -262,7 +357,22 @@
 		warn "## saved/3 ", dump( $field ),$/ if ($debug);
 	}
 
-	return @m;
+	return \@m;
+}
+
+=head2 _debug
+
+Change level of debug warnings
+
+  _debug( 2 );
+
+=cut
+
+sub _debug {
+	my $l = shift;
+	return $debug unless defined($l);
+	warn "debug level $l",$/ if ($l > 0);
+	$debug = $l;
 }
 
 =head1 Functions to create C<data_structure>
@@ -319,11 +429,31 @@
 	$out->{$name}->{search} = \@o;
 }
 
+=head2 marc_leader
+
+Setup fields within MARC leader or get leader
+
+  marc_leader('05','c');
+  my $leader = marc_leader();
+
+=cut
+
+sub marc_leader {
+	my ($offset,$value) = @_;
+
+	if ($offset) {
+		$out->{' leader'}->{ $offset } = $value;
+	} else {
+		return $out->{' leader'};
+	}
+}
+
 =head2 marc
 
 Save value for MARC field
 
   marc('900','a', rec('200','a') );
+  marc('001', rec('000') );
 
 =cut
 
@@ -331,14 +461,20 @@
 	my $f = shift or die "marc needs field";
 	die "marc field must be numer" unless ($f =~ /^\d+$/);
 
-	my $sf = shift or die "marc needs subfield";
+	my $sf;
+	if ($f >= 10) {
+		$sf = shift or die "marc needs subfield";
+	}
 
 	foreach (@_) {
 		my $v = $_;		# make var read-write for Encode
 		next unless (defined($v) && $v !~ /^\s*$/);
-		from_to($v, 'iso-8859-2', $marc_encoding) if ($marc_encoding);
 		my ($i1,$i2) = defined($marc_indicators->{$f}) ? @{ $marc_indicators->{$f} } : (' ',' ');
-		push @{ $marc_record }, [ $f, $i1, $i2, $sf => $v ];
+		if (defined $sf) {
+			push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $i1, $i2, $sf => $v ];
+		} else {
+			push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $v ];
+		}
 	}
 }
 
@@ -378,12 +514,264 @@
 	@{ $marc_indicators->{$f} } = ($i1,$i2);
 }
 
+=head2 marc_compose
+
+Save values for each MARC subfield explicitly
+
+  marc_compose('900',
+  	'a', rec('200','a')
+  	'b', rec('201','a')
+  	'a', rec('200','b')
+  	'c', rec('200','c')
+  );
+
+If you specify C<+> for subfield, value will be appended
+to previous defined subfield.
+
+=cut
+
+sub marc_compose {
+	my $f = shift or die "marc_compose needs field";
+	die "marc_compose field must be numer" unless ($f =~ /^\d+$/);
+
+	my ($i1,$i2) = defined($marc_indicators->{$f}) ? @{ $marc_indicators->{$f} } : (' ',' ');
+	my $m = [ $f, $i1, $i2 ];
+
+	warn "### marc_compose input subfields = ", dump(@_),$/ if ($debug > 2);
+
+	if ($#_ % 2 != 1) {
+		die "ERROR: marc_compose",dump($f,@_)," not valid (must be even).\nDo you need to add first() or join() around some argument?\n";
+	}
+
+	while (@_) {
+		my $sf = shift;
+		my $v = shift;
+
+		next unless (defined($v) && $v !~ /^\s*$/);
+		warn "## ++ marc_compose($f,$sf,$v) ", dump( $m ),$/ if ($debug > 1);
+		if ($sf ne '+') {
+			push @$m, ( $sf, $v );
+		} else {
+			$m->[ $#$m ] .= $v;
+		}
+	}
+
+	warn "## marc_compose current marc = ", dump( $m ),$/ if ($debug > 1);
+
+	push @{ $marc_record->[ $marc_record_offset ] }, $m if ($#{$m} > 2);
+}
+
+=head2 marc_duplicate
+
+Generate copy of current MARC record and continue working on copy
+
+  marc_duplicate();
+
+Copies can be accessed using C<< _get_marc_fields( fetch_next => 1 ) >> or
+C<< _get_marc_fields( offset => 42 ) >>.
+
+=cut
+
+sub marc_duplicate {
+	 my $m = $marc_record->[ -1 ];
+	 die "can't duplicate record which isn't defined" unless ($m);
+	 push @{ $marc_record }, dclone( $m );
+	 warn "## marc_duplicate = ", dump(@$marc_record), $/ if ($debug > 1);
+	 $marc_record_offset = $#{ $marc_record };
+	 warn "## marc_record_offset = $marc_record_offset", $/ if ($debug > 1);
+}
+
+=head2 marc_remove
+
+Remove some field or subfield from MARC record.
+
+  marc_remove('200');
+  marc_remove('200','a');
+
+This will erase field C<200> or C<200^a> from current MARC record.
+
+This is useful after calling C<marc_duplicate> or on it's own (but, you
+should probably just remove that subfield definition if you are not
+using C<marc_duplicate>).
+
+FIXME: support fields < 10.
+
+=cut
+
+sub marc_remove {
+	my ($f, $sf) = @_;
+
+	die "marc_remove needs record number" unless defined($f);
+
+	my $marc = $marc_record->[ $marc_record_offset ];
+
+	warn "### marc_remove before = ", dump( $marc ), $/ if ($debug > 2);
+
+	my $i = 0;
+	foreach ( 0 .. $#{ $marc } ) {
+		last unless (defined $marc->[$i]);
+		warn "#### working on ",dump( @{ $marc->[$i] }), $/ if ($debug > 3);
+		if ($marc->[$i]->[0] eq $f) {
+			if (! defined $sf) {
+				# remove whole field
+				splice @$marc, $i, 1;
+				warn "#### slice \@\$marc, $i, 1 = ",dump( @{ $marc }), $/ if ($debug > 3);
+				$i--;
+			} else {
+				foreach my $j ( 0 .. (( $#{ $marc->[$i] } - 3 ) / 2) ) {
+					my $o = ($j * 2) + 3;
+					if ($marc->[$i]->[$o] eq $sf) {
+						# remove subfield
+						splice @{$marc->[$i]}, $o, 2;
+						warn "#### slice \@{\$marc->[$i]}, $o, 2 = ", dump( @{ $marc }), $/ if ($debug > 3);
+						# is record now empty?
+						if ($#{ $marc->[$i] } == 2) {
+							splice @$marc, $i, 1;
+							warn "#### slice \@\$marc, $i, 1 = ", dump( @{ $marc }), $/ if ($debug > 3);
+							$i--;
+						};
+					}
+				}
+			}
+		}
+		$i++;
+	}
+
+	warn "### marc_remove($f", $sf ? ",$sf" : "", ") after = ", dump( $marc ), $/ if ($debug > 2);
+
+	$marc_record->[ $marc_record_offset ] = $marc;
+
+	warn "## full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 1);
+}
+
+=head2 marc_original_order
+
+Copy all subfields preserving original order to marc field.
+
+  marc_original_order( marc_field_number, original_input_field_number );
+
+Please note that field numbers are consistent with other commands (marc
+field number first), but somewhat counter-intuitive (destination and then
+source).
+
+You might want to use this command if you are just renaming subfields or
+using pre-processing modify_record in C<config.yml> and don't need any
+post-processing or want to preserve order of original subfields.
+
+
+=cut
+
+sub marc_original_order {
+
+	my ($to, $from) = @_;
+	die "marc_original_order needs from and to fields\n" unless ($from && $to);
+
+	return unless defined($rec->{$from});
+
+	my $r = $rec->{$from};
+	die "record field $from isn't array\n" unless (ref($r) eq 'ARRAY');
+
+	my ($i1,$i2) = defined($marc_indicators->{$to}) ? @{ $marc_indicators->{$to} } : (' ',' ');
+	warn "## marc_original_order($to,$from) source = ", dump( $r ),$/ if ($debug > 1);
+
+	foreach my $d (@$r) {
+
+		if (! defined($d->{subfields}) && ref($d->{subfields}) ne 'ARRAY') {
+			warn "# marc_original_order($to,$from): field $from doesn't have subfields specification\n";
+			next;
+		}
+	
+		my @sfs = @{ $d->{subfields} };
+
+		die "field $from doesn't have even number of subfields specifications\n" unless($#sfs % 2 == 1);
+
+		warn "#--> d: ",dump($d), "\n#--> sfs: ",dump(@sfs),$/ if ($debug > 2);
+
+		my $m = [ $to, $i1, $i2 ];
+
+		while (my $sf = shift @sfs) {
+
+			warn "#--> sf: ",dump($sf), $/ if ($debug > 2);
+			my $offset = shift @sfs;
+			die "corrupted sufields specification for field $from\n" unless defined($offset);
+
+			my $v;
+			if (ref($d->{$sf}) eq 'ARRAY') {
+				$v = $d->{$sf}->[$offset] if (defined($d->{$sf}->[$offset]));
+			} elsif ($offset == 0) {
+				$v = $d->{$sf};
+			} else {
+				die "field $from subfield '$sf' need occurence $offset which doesn't exist", dump($d->{$sf});
+			}
+			push @$m, ( $sf, $v ) if (defined($v));
+		}
+
+		if ($#{$m} > 2) {
+			push @{ $marc_record->[ $marc_record_offset ] }, $m;
+		}
+	}
+
+	warn "## marc_record = ", dump( $marc_record ),$/ if ($debug > 1);
+}
+
 
 =head1 Functions to extract data from input
 
 This function should be used inside functions to create C<data_structure> described
 above.
 
+=head2 _pack_subfields_hash
+
+ @values = _pack_subfields_hash( $h, $include_subfields )
+
+=cut
+
+sub _pack_subfields_hash {
+
+	warn "## _pack_subfields_hash( ",dump(@_), " )\n" if ($debug > 1);
+
+	my ($h,$include_subfields) = @_;
+
+	if ( defined($h->{subfields}) ) {
+		my $sfs = delete $h->{subfields} || die "no subfields?";
+		my @out;
+		while (@$sfs) {
+			my $sf = shift @$sfs;
+			push @out, '^' . $sf if ($include_subfields);
+			my $o = shift @$sfs;
+			if ($o == 0 && ref( $h->{$sf} ) ne 'ARRAY' ) {
+				# single element subfields are not arrays
+warn "====> $sf $o / $#$sfs ", dump( $sfs, $h->{$sf} ), "\n";
+
+				push @out, $h->{$sf};
+			} else {
+warn "====> $sf $o / $#$sfs ", dump( $sfs, $h->{$sf} ), "\n";
+				push @out, $h->{$sf}->[$o];
+			}
+		}
+		if ($include_subfields) {
+			return join('', @out);
+		} else {
+			return @out;
+		}
+	} else {
+		if ($include_subfields) {
+			my $out = '';
+			foreach my $sf (keys %$h) {
+				if (ref($h->{$sf}) eq 'ARRAY') {
+					$out .= '^' . $sf . join('^' . $sf, @{ $h->{$sf} });
+				} else {
+					$out .= '^' . $sf . $h->{$sf};
+				}
+			}
+			return $out;
+		} else {
+			# FIXME this should probably be in alphabetical order instead of hash order
+			values %{$h};
+		}
+	}
+}
+
 =head2 rec1
 
 Return all values in some field
@@ -396,15 +784,19 @@
 
 sub rec1 {
 	my $f = shift;
+	warn "rec1($f) = ", dump( $rec->{$f} ), $/ if ($debug > 1);
 	return unless (defined($rec) && defined($rec->{$f}));
+	warn "rec1($f) = ", dump( $rec->{$f} ), $/ if ($debug > 1);
 	if (ref($rec->{$f}) eq 'ARRAY') {
-		return map { 
-			if (ref($_) eq 'HASH') {
-				values %{$_};
+		my @out;
+		foreach my $h ( @{ $rec->{$f} } ) {
+			if (ref($h) eq 'HASH') {
+				push @out, ( _pack_subfields_hash( $h ) );
 			} else {
-				$_;
+				push @out, $h;
 			}
-		} @{ $rec->{$f} };
+		}
+		return @out;
 	} elsif( defined($rec->{$f}) ) {
 		return $rec->{$f};
 	}
@@ -422,7 +814,14 @@
 	my $f = shift;
 	return unless (defined($rec && $rec->{$f}));
 	my $sf = shift;
-	return map { $_->{$sf} } grep { ref($_) eq 'HASH' && $_->{$sf} } @{ $rec->{$f} };
+	warn "rec2($f,$sf) = ", dump( $rec->{$f} ), $/ if ($debug > 1);
+	return map {
+		if (ref($_->{$sf}) eq 'ARRAY') {
+			@{ $_->{$sf} };
+		} else {
+			$_->{$sf};
+		}
+	} grep { ref($_) eq 'HASH' && $_->{$sf} } @{ $rec->{$f} };
 }
 
 =head2 rec
@@ -435,10 +834,16 @@
 =cut
 
 sub rec {
+	my @out;
 	if ($#_ == 0) {
-		return rec1(@_);
+		@out = rec1(@_);
 	} elsif ($#_ == 1) {
-		return rec2(@_);
+		@out = rec2(@_);
+	}
+	if (@out) {
+		return @out;
+	} else {
+		return '';
 	}
 }
 
@@ -471,7 +876,7 @@
 =cut
 
 sub prefix {
-	my $p = shift or die "prefix needs string as first argument";
+	my $p = shift or return;
 	return map { $p . $_ } grep { defined($_) } @_;
 }
 
@@ -534,6 +939,78 @@
 	}
 }
 
+=head2 config
+
+Consult config values stored in C<config.yml>
+
+  # return database code (key under databases in yaml)
+  $database_code = config();	# use _ from hash
+  $database_name = config('name');
+  $database_input_name = config('input name');
+  $tag = config('input normalize tag');
+
+Up to three levels are supported.
+
+=cut
+
+sub config {
+	return unless ($config);
+
+	my $p = shift;
+
+	$p ||= '';
+
+	my $v;
+
+	warn "### getting config($p)\n" if ($debug > 1);
+
+	my @p = split(/\s+/,$p);
+	if ($#p < 0) {
+		$v = $config->{ '_' };	# special, database code
+	} else {
+
+		my $c = dclone( $config );
+
+		foreach my $k (@p) {
+			warn "### k: $k c = ",dump($c),$/ if ($debug > 1);
+			if (ref($c) eq 'ARRAY') {
+				$c = shift @$c;
+				warn "config($p) taking first occurence of '$k', probably not what you wanted!\n";
+				last;
+			}
+
+			if (! defined($c->{$k}) ) {
+				$c = undef;
+				last;
+			} else {
+				$c = $c->{$k};
+			}
+		}
+		$v = $c if ($c);
+
+	}
+
+	warn "## config( '$p' ) = ",dump( $v ),$/ if ($v && $debug);
+	warn "config( '$p' ) is empty\n" if (! $v);
+
+	return $v;
+}
+
+=head2 id
+
+Returns unique id of this record
+
+  $id = id();
+
+Returns C<42/2> for 2nd occurence of MFN 42.
+
+=cut
+
+sub id {
+	my $mfn = $config->{_mfn} || die "no _mfn in config data";
+	return $mfn . $#{$marc_record} ? $#{$marc_record} + 1 : '';
+}
+
 =head2 join_with
 
 Joins walues with some delimiter
@@ -544,7 +1021,46 @@
 
 sub join_with {
 	my $d = shift;
-	return join($d, grep { defined($_) && $_ ne '' } @_);
+	warn "### join_with('$d',",dump(@_),")\n" if ($debug > 2);
+	my $v = join($d, grep { defined($_) && $_ ne '' } @_);
+	return '' unless defined($v);
+	return $v;
+}
+
+=head2 split_rec_on
+
+Split record subfield on some regex and take one of parts out
+
+  $a_before_semi_column =
+  	split_rec_on('200','a', /\s*;\s*/, $part);
+
+C<$part> is optional number of element. First element is
+B<1>, not 0!
+
+If there is no C<$part> parameter or C<$part> is 0, this function will
+return all values produced by splitting.
+
+=cut
+
+sub split_rec_on {
+	die "split_rec_on need (fld,sf,regex[,part]" if ($#_ < 2);
+
+	my ($fld, $sf, $regex, $part) = @_;
+	warn "### regex ", ref($regex), $regex, $/ if ($debug > 2);
+
+	my @r = rec( $fld, $sf );
+	my $v = shift @r;
+	warn "### first rec($fld,$sf) = ",dump($v),$/ if ($debug > 2);
+
+	return '' if ( ! defined($v) || $v =~ /^\s*$/);
+
+	my @s = split( $regex, $v );
+	warn "## split_rec_on($fld,$sf,$regex,$part) = ",dump(@s),$/ if ($debug > 1);
+	if ($part && $part > 0) {
+		return $s[ $part - 1 ];
+	} else {
+		return @s;
+	}
 }
 
 # END