--- trunk/lib/WebPAC/Normalize.pm	2005/07/17 22:48:25	22
+++ trunk/lib/WebPAC/Normalize.pm	2009/05/30 15:26:25	1210
@@ -1,650 +1,960 @@
 package WebPAC::Normalize;
+use Exporter 'import';
+our @EXPORT = qw/
+	_set_ds _set_lookup
+	_set_load_row
+	_get_ds _clean_ds
+	_debug
+	_pack_subfields_hash
+
+	to
+	search_display search display sorted
+
+	rec1 rec2 rec
+	frec frec_eq frec_ne
+	regex prefix suffix surround
+	first lookup join_with
+	save_into_lookup
+
+	split_rec_on
+
+	get set
+	count
+
+	row
+	rec_array
+
+/;
 
 use warnings;
 use strict;
-use Data::Dumper;
 
-=head1 NAME
+#use base qw/WebPAC::Common/;
+use Data::Dump qw/dump/;
+use Carp qw/confess/;
+
+# debugging warn(s)
+my $debug = 0;
+_debug( $debug );
+
+# FIXME
+use WebPAC::Normalize::ISBN;
+push @EXPORT, ( 'isbn_10', 'isbn_13' );
+
+use WebPAC::Normalize::MARC;
+push @EXPORT, ( qw/
+	marc marc_indicators marc_repeatable_subfield
+	marc_compose marc_leader marc_fixed
+	marc_duplicate marc_remove marc_count
+	marc_original_order
+	marc_template
+/);
 
-WebPAC::Normalize - data mungling for normalisation
+use Storable qw/dclone/;
 
-=head1 VERSION
+=head1 NAME
 
-Version 0.01
+WebPAC::Normalize - describe normalisaton rules using sets
 
 =cut
 
-our $VERSION = '0.01';
+our $VERSION = '0.36';
 
 =head1 SYNOPSIS
 
-This package contains code that mungle data to produce normalized format.
-
-It contains several assumptions:
+This module uses C<conf/normalize/*.pl> files to perform normalisation
+from input records using perl functions which are specialized for set
+processing.
+
+Sets are implemented as arrays, and normalisation file is valid perl, which
+means that you check it's validity before running WebPAC using
+C<perl -c normalize.pl>.
+
+Normalisation can generate multiple output normalized data. For now, supported output
+types (on the left side of definition) are: C<search_display>, C<display>, C<search> and
+C<marc>.
 
-=over
+=head1 FUNCTIONS
 
-=item *
+Functions which start with C<_> are private and used by WebPAC internally.
+All other functions are available for use within normalisation rules.
 
-format of fields is defined using C<v123^a> notation for repeatable fields
-or C<s123^a> for single (or first) value, where C<123> is field number and
-C<a> is subfield.
+=head2 data_structure
 
-=item *
+Return data structure
 
-source data records (C<$rec>) have unique identifiers in field C<000>
+  my $ds = WebPAC::Normalize::data_structure(
+	lookup => $lookup_hash,
+	row => $row,
+	rules => $normalize_pl_config,
+	marc_encoding => 'utf-8',
+	config => $config,
+	load_row_coderef => sub {
+		my ($database,$input,$mfn) = @_;
+		$store->load_row( database => $database, input => $input, id => $mfn );
+	},
+  );
 
-=item *
+Options C<row>, C<rules> and C<log> are mandatory while all
+other are optional.
 
-optional C<eval{length('v123^a') == 3}> tag at B<beginning of format> will be
-perl code that is evaluated before producing output (value of field will be
-interpolated before that)
+C<load_row_coderef> is closure only used when executing lookups, so they will
+die if it's not defined.
 
-=item *
+This function will B<die> if normalizastion can't be evaled.
 
-optional C<filter{filter_name}> at B<begining of format> will apply perl
-code defined as code ref on format after field substitution to producing
-output
+Since this function isn't exported you have to call it with 
+C<WebPAC::Normalize::data_structure>.
 
-=item *
+=cut
 
-optional C<lookup{...}> will be then performed. See C<WebPAC::Lookups>.
+my $load_row_coderef;
 
-=item *
+sub data_structure {
+	my $arg = {@_};
 
-at end, optional C<format>s rules are resolved. Format rules are similar to
-C<sprintf> and can also contain C<lookup{...}> which is performed after
-values are inserted in format.
+	die "need row argument" unless ($arg->{row});
+	die "need normalisation argument" unless ($arg->{rules});
 
-=back
+	_set_lookup( $arg->{lookup} ) if defined($arg->{lookup});
+	_set_ds( $arg->{row} );
+	_set_config( $arg->{config} ) if defined($arg->{config});
+	_clean_ds( %{ $arg } );
+	$load_row_coderef = $arg->{load_row_coderef};
 
-This also describes order in which transformations are applied (eval,
-filter, lookup, format) which is important to undestand when deciding how to
-solve your data mungling and normalisation process.
+	no strict 'subs';
+	no warnings 'redefine';
+	eval "$arg->{rules};";
+	die "error evaling $arg->{rules}: $@\n" if ($@);
 
+	return _get_ds();
+}
 
+=head2 _set_ds
 
+Set current record hash
 
-=head1 FUNCTIONS
+  _set_ds( $rec );
 
-=head2 new
+=cut
 
-Create new normalisation object
+my $rec;
 
-  my $n = new WebPAC::Normalize::Something(
- 	filter => {
-		'filter_name_1' => sub {
-			# filter code
-			return length($_);
-		}, ...
-	},
-	db => $webpac_db_obj,
-	lookup_regex => $lookup->regex,
-  );
+sub _set_ds {
+	$rec = shift or die "no record hash";
+	$WebPAC::Normalize::MARC::rec = $rec;
+}
 
-Parametar C<filter> defines user supplied snippets of perl code which can
-be use with C<filter{...}> notation.
+=head2
 
-Recommended parametar C<lookup_regex> is used to enable parsing of lookups
-in structures.
+  my $rec = _get_rec();
 
 =cut
 
-sub new {
-	my $class = shift;
-        my $self = {@_};
-        bless($self, $class);
+sub _get_rec { $rec };
 
-	$self ? return $self : return undef;
+sub rec_array {
+	my $d = $rec->{ $_[0] };
+	return @$d if ref($d) eq 'ARRAY';
+	die "field $_[0] not array: ",dump( $d );
 }
 
+=head2 _set_config
 
-=head2 data_structure
+Set current config hash
 
-Create in-memory data structure which represents normalized layout from
-C<conf/normalize/*.xml>.
+  _set_config( $config );
 
-This structures are used to produce output.
+Magic keys are:
 
- my @ds = $webpac->data_structure($rec);
+=over 4
 
-B<Note: historical oddity follows>
+=item _
 
-This method will also set C<< $webpac->{'currnet_filename'} >> if there is
-C<< <filename> >> tag and C<< $webpac->{'headline'} >> if there is
-C<< <headline> >> tag.
+Code of current database
 
-=cut
+=item _mfn
 
-sub data_structure {
-	my $self = shift;
+Current MFN
 
-	my $log = $self->_get_logger();
+=back
 
-	my $rec = shift;
-	$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o);
+=cut
 
-	my $cache_file;
+my $config;
 
-	if ($self->{'db'}) {
-		my @ds = $self->{'db'}->load_ds($rec);
-		return @ds if (@ds);
-	}
+sub _set_config {
+	$config = shift;
+}
 
-	undef $self->{'currnet_filename'};
-	undef $self->{'headline'};
+=head2 _get_ds
 
-	my @sorted_tags;
-	if ($self->{tags_by_order}) {
-		@sorted_tags = @{$self->{tags_by_order}};
-	} else {
-		@sorted_tags = sort { $self->_sort_by_order } keys %{$self->{'import_xml'}->{'indexer'}};
-		$self->{tags_by_order} = \@sorted_tags;
-	}
+Return hash formatted as data structure
 
-	my @ds;
+  my $ds = _get_ds();
 
-	$log->debug("tags: ",sub { join(", ",@sorted_tags) });
+=cut
 
-	foreach my $field (@sorted_tags) {
+my $out;
 
-		my $row;
+sub _get_ds {
+#warn "## out = ",dump($out);
+	return $out;
+}
 
-#print "field $field [",$self->{'tag'},"] = ",Dumper($self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}});
+=head2 _clean_ds
 
-		foreach my $tag (@{$self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}}) {
-			my $format = $tag->{'value'} || $tag->{'content'};
+Clean data structure hash for next record
 
-			$log->debug("format: $format");
+  _clean_ds();
 
-			my @v;
-			if ($self->{'lookup_regex'} && $format =~ $self->{'lookup_regex'}) {
-				@v = $self->fill_in_to_arr($rec,$format);
-			} else {
-				@v = $self->parse_to_arr($rec,$format);
-			}
-			next if (! @v);
+=cut
 
-			if ($tag->{'sort'}) {
-				@v = $self->sort_arr(@v);
-			}
+sub _clean_ds {
+	my $a = {@_};
+	$out = undef;
+	WebPAC::Normalize::MARC::_clean();
+}
 
-			# use format?
-			if ($tag->{'format_name'}) {
-				@v = map { $self->apply_format($tag->{'format_name'},$tag->{'format_delimiter'},$_) } @v;
-			}
+=head2 _set_lookup
 
-			if ($field eq 'filename') {
-				$self->{'current_filename'} = join('',@v);
-				$log->debug("filename: ",$self->{'current_filename'});
-			} elsif ($field eq 'headline') {
-				$self->{'headline'} .= join('',@v);
-				$log->debug("headline: ",$self->{'headline'});
-				next; # don't return headline in data_structure!
-			}
+Set current lookup hash
 
-			# delimiter will join repeatable fields
-			if ($tag->{'delimiter'}) {
-				@v = ( join($tag->{'delimiter'}, @v) );
-			}
+  _set_lookup( $lookup );
 
-			# default types 
-			my @types = qw(display swish);
-			# override by type attribute
-			@types = ( $tag->{'type'} ) if ($tag->{'type'});
-
-			foreach my $type (@types) {
-				# append to previous line?
-				$log->debug("type: $type ",sub { join(" ",@v) }, $row->{'append'} || 'no append');
-				if ($tag->{'append'}) {
-
-					# I will delimit appended part with
-					# delimiter (or ,)
-					my $d = $tag->{'delimiter'};
-					# default delimiter
-					$d ||= " ";
-
-					my $last = pop @{$row->{$type}};
-					$d = "" if (! $last);
-					$last .= $d . join($d, @v);
-					push @{$row->{$type}}, $last;
+=cut
 
-				} else {
-					push @{$row->{$type}}, @v;
-				}
-			}
+my $lookup;
 
+sub _set_lookup {
+	$lookup = shift;
+}
 
-		}
+=head2 _get_lookup
 
-		if ($row) {
-			$row->{'tag'} = $field;
+Get current lookup hash
 
-			# TODO: name_sigular, name_plural
-			my $name = $self->{'import_xml'}->{'indexer'}->{$field}->{'name'};
-			$row->{'name'} = $name ? $self->_x($name) : $field;
-
-			# post-sort all values in field
-			if ($self->{'import_xml'}->{'indexer'}->{$field}->{'sort'}) {
-				$log->warn("sort at field tag not implemented");
-			}
+  my $lookup = _get_lookup();
 
-			push @ds, $row;
+=cut
 
-			$log->debug("row $field: ",sub { Dumper($row) });
-		}
+sub _get_lookup {
+	return $lookup;
+}
 
-	}
+=head2 _set_load_row
+
+Setup code reference which will return L<data_structure> from
+L<WebPAC::Store>
+
+  _set_load_row(sub {
+		my ($database,$input,$mfn) = @_;
+		$store->load_row( database => $database, input => $input, id => $mfn );
+  });
 
-	$self->{'db'}->save_ds(
-		ds => \@ds,
-		current_filename => $self->{'current_filename'},
-		headline => $self->{'headline'},
-	) if ($self->{'db'});
+=cut
 
-	return @ds;
+sub _set_load_row {
+	my $coderef = shift;
+	confess "argument isn't CODE" unless ref($coderef) eq 'CODE';
 
+	$load_row_coderef = $coderef;
 }
 
-=head2 parse
+=head2 _debug
 
-Perform smart parsing of string, skipping delimiters for fields which aren't
-defined. It can also eval code in format starting with C<eval{...}> and
-return output or nothing depending on eval code.
+Change level of debug warnings
 
- my $text = $webpac->parse($rec,'eval{"v901^a" eq "Deskriptor"}descriptor: v250^a', $i);
+  _debug( 2 );
 
 =cut
 
-sub parse {
-	my $self = shift;
+sub _debug {
+	my $l = shift;
+	return $debug unless defined($l);
+	warn "debug level $l",$/ if ($l > 0);
+	$debug = $l;
+	$WebPAC::Normalize::MARC::debug = $debug;
+}
 
-	my ($rec, $format_utf8, $i) = @_;
+=head1 Functions to create C<data_structure>
 
-	return if (! $format_utf8);
+Those functions generally have to first in your normalization file.
 
-	my $log = $self->_get_logger();
+=head2 to
 
-	$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o);
+Generic way to set values for some name
 
-	$i = 0 if (! $i);
+  to('field-name', 'name-value' => rec('200','a') );
 
-	my $format = $self->_x($format_utf8) || $log->logconfess("can't convert '$format_utf8' from UTF-8 to ",$self->{'code_page'});
+There are many helpers defined below which might be easier to use.
 
-	my @out;
+=cut
 
-	$log->debug("format: $format");
+sub to {
+	my $type = shift or confess "need type -- BUG?";
+	my $name = shift or confess "needs name as first argument";
+	my @o = grep { defined($_) && $_ ne '' } @_;
+	return unless (@o);
+	$out->{$name}->{$type} = \@o;
+}
 
-	my $eval_code;
-	# remove eval{...} from beginning
-	$eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s);
+=head2 search_display
 
-	my $filter_name;
-	# remove filter{...} from beginning
-	$filter_name = $1 if ($format =~ s/^filter{([^}]+)}//s);
+Define output for L<search> and L<display> at the same time
 
-	my $prefix;
-	my $all_found=0;
+  search_display('Title', rec('200','a') );
 
-	while ($format =~ s/^(.*?)(v|s)(\d+)(?:\^(\w))?//s) {
+=cut
 
-		my $del = $1 || '';
-		$prefix ||= $del if ($all_found == 0);
+sub search_display {
+	my $name = shift or die "search_display needs name as first argument";
+	my @o = grep { defined($_) && $_ ne '' } @_;
+	return unless (@o);
+	$out->{$name}->{search} = \@o;
+	$out->{$name}->{display} = \@o;
+}
 
-		# repeatable index
-		my $r = $i;
-		$r = 0 if (lc("$2") eq 's');
+=head2 tag
 
-		my $found = 0;
-		my $tmp = $self->get_data(\$rec,$3,$4,$r,\$found);
+Old name for L<search_display>, it will probably be removed at one point.
 
-		if ($found) {
-			push @out, $del;
-			push @out, $tmp;
-			$all_found += $found;
-		}
-	}
+=cut
 
-	return if (! $all_found);
+sub tag {
+	search_display( @_ );
+}
 
-	my $out = join('',@out);
+=head2 display
 
-	if ($out) {
-		# add rest of format (suffix)
-		$out .= $format;
+Define output just for I<display>
 
-		# add prefix if not there
-		$out = $prefix . $out if ($out !~ m/^\Q$prefix\E/);
+  @v = display('Title', rec('200','a') );
 
-		$log->debug("result: $out");
-	}
+=cut
 
-	if ($eval_code) {
-		my $eval = $self->fill_in($rec,$eval_code,$i) || return;
-		$log->debug("about to eval{$eval} format: $out");
-		return if (! $self->_eval($eval));
-	}
-	
-	if ($filter_name && $self->{'filter'}->{$filter_name}) {
-		$log->debug("about to filter{$filter_name} format: $out");
-		$out = $self->{'filter'}->{$filter_name}->($out);
-		return unless(defined($out));
-		$log->debug("filter result: $out");
-	}
+sub display { to( 'display', @_ ) }
 
-	return $out;
-}
+=head2 search
 
-=head2 parse_to_arr
+Prepare values just for I<search>
 
-Similar to C<parse>, but returns array of all repeatable fields
-
- my @arr = $webpac->parse_to_arr($rec,'v250^a');
+  @v = search('Title', rec('200','a') );
 
 =cut
 
-sub parse_to_arr {
-	my $self = shift;
+sub search { to( 'search', @_ ) }
 
-	my ($rec, $format_utf8) = @_;
+=head2 sorted
 
-	my $log = $self->_get_logger();
+Insert into lists which will be automatically sorted
 
-	$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o);
-	return if (! $format_utf8);
+ sorted('Title', rec('200','a') );
 
-	my $i = 0;
-	my @arr;
+=cut
 
-	while (my $v = $self->parse($rec,$format_utf8,$i++)) {
-		push @arr, $v;
-	}
+sub sorted { to( 'sorted', @_ ) }
 
-	$log->debug("format '$format_utf8' returned ",--$i," elements: ", sub { join(" | ",@arr) }) if (@arr);
+=head2 row
 
-	return @arr;
-}
+Insert new row of data into output module
 
+  row( column => 'foo', column2 => 'bar' );
 
-=head2 fill_in
+=cut
+
+use Data::Dump qw/dump/;
 
-Workhourse of all: takes record from in-memory structure of database and
-strings with placeholders and returns string or array of with substituted
-values from record.
+sub row {
+	die "array doesn't have odd number of elements but $#_: ",dump( @_ ) if $#_ % 2 == 1;
+	my $table = shift @_;
+	push @{ $out->{'_rows'}->{$table} }, {@_};
+}
 
- my $text = $webpac->fill_in($rec,'v250^a');
 
-Optional argument is ordinal number for repeatable fields. By default,
-it's assume to be first repeatable field (fields are perl array, so first
-element is 0).
-Following example will read second value from repeatable field.
+=head1 Functions to extract data from input
 
- my $text = $webpac->fill_in($rec,'Title: v250^a',1);
+This function should be used inside functions to create C<data_structure> described
+above.
 
-This function B<does not> perform parsing of format to inteligenty skip
-delimiters before fields which aren't used.
+=head2 _pack_subfields_hash
 
-This method will automatically decode UTF-8 string to local code page
-if needed.
+ @subfields = _pack_subfields_hash( $h );
+ $subfields = _pack_subfields_hash( $h, 1 );
+
+Return each subfield value in array or pack them all together and return scalar
+with subfields (denoted by C<^>) and values.
 
 =cut
 
-sub fill_in {
-	my $self = shift;
+sub _pack_subfields_hash {
+
+	warn "## _pack_subfields_hash( ",dump(@_), " )\n" if ($debug > 1);
 
-	my $log = $self->_get_logger();
+	my ($hash,$include_subfields) = @_;
 
-	my $rec = shift || $log->logconfess("need data record");
-	my $format = shift || $log->logconfess("need format to parse");
-	# iteration (for repeatable fields)
-	my $i = shift || 0;
+	# sanity and ease of use
+	return $hash if (ref($hash) ne 'HASH');
 
-	$log->logdie("infitite loop in format $format") if ($i > ($self->{'max_mfn'} || 9999));
+	my $h = dclone( $hash );
 
-	# FIXME remove for speedup?
-	$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o);
+	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";
 
-	if (utf8::is_utf8($format)) {
-		$format = $self->_x($format);
+				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 $sf (sort keys %$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};
+		}
 	}
+}
 
-	my $found = 0;
+=head2 rec1
 
-	my $eval_code;
-	# remove eval{...} from beginning
-	$eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s);
+Return all values in some field
 
-	my $filter_name;
-	# remove filter{...} from beginning
-	$filter_name = $1 if ($format =~ s/^filter{([^}]+)}//s);
+  @v = rec1('200')
 
-	# do actual replacement of placeholders
-	# repeatable fields
-	$format =~ s/v(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,$i,\$found)/ges;
-	# non-repeatable fields
-	$format =~ s/s(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,0,\$found)/ges;
+TODO: order of values is probably same as in source data, need to investigate that
 
-	if ($found) {
-		$log->debug("format: $format");
-		if ($eval_code) {
-			my $eval = $self->fill_in($rec,$eval_code,$i);
-			return if (! $self->_eval($eval));
-		}
-		if ($filter_name && $self->{'filter'}->{$filter_name}) {
-			$log->debug("filter '$filter_name' for $format");
-			$format = $self->{'filter'}->{$filter_name}->($format);
-			return unless(defined($format));
-			$log->debug("filter result: $format");
+=cut
+
+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') {
+		my @out;
+		foreach my $h ( @{ $rec->{$f} } ) {
+			if (ref($h) eq 'HASH') {
+				push @out, ( _pack_subfields_hash( $h ) );
+			} else {
+				push @out, $h;
+			}
 		}
-		# do we have lookups?
-		if ($self->{'lookup'}) {
-			return $self->lookup($format);
+		return @out;
+	} elsif( defined($rec->{$f}) ) {
+		return $rec->{$f};
+	}
+}
+
+=head2 rec2
+
+Return all values in specific field and subfield
+
+  @v = rec2('200','a')
+
+=cut
+
+sub rec2 {
+	my $f = shift;
+	return unless (defined($rec && $rec->{$f}));
+	my $sf = shift;
+	warn "rec2($f,$sf) = ", dump( $rec->{$f} ), $/ if ($debug > 1);
+	return map {
+		if (ref($_->{$sf}) eq 'ARRAY') {
+			@{ $_->{$sf} };
 		} else {
-			return $format;
+			$_->{$sf};
 		}
+	} grep { ref($_) eq 'HASH' && defined $_->{$sf} } @{ $rec->{$f} };
+}
+
+=head2 rec
+
+syntaxtic sugar for
+
+  @v = rec('200')
+  @v = rec('200','a')
+
+If rec() returns just single value, it will
+return scalar, not array.
+
+=cut
+
+sub rec {
+	my @out;
+	if ($#_ == 0) {
+		@out = rec1(@_);
+	} elsif ($#_ == 1) {
+		@out = rec2(@_);
+	}
+	if ($#out == 0 && ! wantarray) {
+		return $out[0];
+	} elsif (@out) {
+		return @out;
 	} else {
-		return;
+		return '';
 	}
 }
 
+=head2 frec
+
+Returns first value from field
+
+  $v = frec('200');
+  $v = frec('200','a');
+
+=cut
+
+sub frec {
+	my @out = rec(@_);
+	warn "rec(",dump(@_),") has more than one return value, ignoring\n" if $#out > 0;
+	return shift @out;
+}
+
+=head2 frec_eq
+
+=head2 frec_ne
+
+Check if first values from two fields are same or different
+
+  if ( frec_eq( 900 => 'a', 910 => 'c' ) ) {
+  	# values are same
+  } else {
+    # values are different
+  }
 
-=head2 fill_in_to_arr
+Strictly speaking C<frec_eq> and C<frec_ne> wouldn't be needed if you
+could write something like:
 
-Similar to C<fill_in>, but returns array of all repeatable fields. Usable
-for fields which have lookups, so they shouldn't be parsed but rather
-C<fill_id>ed.
+  if ( frec( '900','a' ) eq frec( '910','c' ) ) {
+  	# yada tada
+  }
 
- my @arr = $webpac->fill_in_to_arr($rec,'[v900];;[v250^a]');
+but you can't since our parser L<WebPAC::Parser> will remove all whitespaces
+in order to parse text and create invalid function C<eqfrec>.
 
 =cut
 
-sub fill_in_to_arr {
-	my $self = shift;
+sub frec_eq {
+	my ( $f1,$sf1, $f2, $sf2 ) = @_;
+	return (rec( $f1, $sf1 ))[0] eq (rec( $f2, $sf2 ))[0];
+}
+
+sub frec_ne {
+	return ! frec_eq( @_ );
+}
 
-	my ($rec, $format_utf8) = @_;
+=head2 regex
 
-	my $log = $self->_get_logger();
+Apply regex to some or all values
 
-	$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o);
-	return if (! $format_utf8);
+  @v = regex( 's/foo/bar/g', @v );
 
-	my $i = 0;
-	my @arr;
+=cut
 
-	while (my @v = $self->fill_in($rec,$format_utf8,$i++)) {
-		push @arr, @v;
+sub regex {
+	my $r = shift;
+	my @out;
+	#warn "r: $r\n", dump(\@_);
+	foreach my $t (@_) {
+		next unless ($t);
+		eval "\$t =~ $r";
+		push @out, $t if ($t && $t ne '');
 	}
+	return @out;
+}
+
+=head2 prefix
+
+Prefix all values with a string
+
+  @v = prefix( 'my_', @v );
+
+=cut
+
+sub prefix {
+	my $p = shift;
+	return @_ unless defined( $p );
+	return map { $p . $_ } grep { defined($_) } @_;
+}
 
-	$log->debug("format '$format_utf8' returned ",--$i," elements: ", sub { join(" | ",@arr) }) if (@arr);
+=head2 suffix
 
-	return @arr;
+suffix all values with a string
+
+  @v = suffix( '_my', @v );
+
+=cut
+
+sub suffix {
+	my $s = shift;
+	return @_ unless defined( $s );
+	return map { $_ . $s } grep { defined($_) } @_;
 }
 
+=head2 surround
 
-=head2 get_data
+surround all values with a two strings
 
-Returns value from record.
+  @v = surround( 'prefix_', '_suffix', @v );
 
- my $text = $self->get_data(\$rec,$f,$sf,$i,\$found);
+=cut
+
+sub surround {
+	my $p = shift;
+	my $s = shift;
+	$p = '' unless defined( $p );
+	$s = '' unless defined( $s );
+	return map { $p . $_ . $s } grep { defined($_) } @_;
+}
 
-Arguments are:
-record reference C<$rec>,
-field C<$f>,
-optional subfiled C<$sf>,
-index for repeatable values C<$i>.
+=head2 first
 
-Optinal variable C<$found> will be incremeted if there
-is field.
+Return first element
 
-Returns value or empty string.
+  $v = first( @v );
 
 =cut
 
-sub get_data {
-	my $self = shift;
+sub first {
+	my $r = shift;
+	return $r;
+}
 
-	my ($rec,$f,$sf,$i,$found) = @_;
+=head2 lookup
 
-	if ($$rec->{$f}) {
-		return '' if (! $$rec->{$f}->[$i]);
-		no strict 'refs';
-		if ($sf && $$rec->{$f}->[$i]->{$sf}) {
-			$$found++ if (defined($$found));
-			return $$rec->{$f}->[$i]->{$sf};
-		} elsif ($$rec->{$f}->[$i]) {
-			$$found++ if (defined($$found));
-			# it still might have subfield, just
-			# not specified, so we'll dump all
-			if ($$rec->{$f}->[$i] =~ /HASH/o) {
-				my $out;
-				foreach my $k (keys %{$$rec->{$f}->[$i]}) {
-					$out .= $$rec->{$f}->[$i]->{$k}." ";
-				}
-				return $out;
-			} else {
-				return $$rec->{$f}->[$i];
-			}
+Consult lookup hashes for some value
+
+  @v = lookup(
+  	sub {
+		'ffkk/peri/mfn'.rec('000')
+	},
+	'ffkk','peri','200-a-200-e',
+	sub {
+		first(rec(200,'a')).' '.first(rec('200','e'))
+	}
+  );
+
+Code like above will be B<automatically generated> using L<WebPAC::Parse> from
+normal lookup definition in C<conf/lookup/something.pl> which looks like:
+
+  lookup(
+	# which results to return from record recorded in lookup
+	sub { 'ffkk/peri/mfn' . rec('000') },
+	# from which database and input
+	'ffkk','peri',
+	# such that following values match
+	sub { first(rec(200,'a')) . ' ' . first(rec('200','e')) },
+	# if this part is missing, we will try to match same fields
+	# from lookup record and current one, or you can override
+	# which records to use from current record using
+	sub { rec('900','x') . ' ' . rec('900','y') },
+  )
+
+You can think about this lookup as SQL (if that helps):
+
+  select
+  	sub { what }
+  from
+  	database, input
+  where
+    sub { filter from lookuped record }
+  having
+    sub { optional filter on current record }
+
+Easy as pie, right?
+
+=cut
+
+sub lookup {
+	my ($what, $database, $input, $key, $having) = @_;
+
+	confess "lookup needs 5 arguments: what, database, input, key, having\n" unless ($#_ == 4);
+
+	warn "## lookup ($database, $input, $key)", $/ if ($debug > 1);
+	return unless (defined($lookup->{$database}->{$input}->{$key}));
+
+	confess "lookup really need load_row_coderef added to data_structure\n" unless ($load_row_coderef);
+
+	my $mfns;
+	my @having = $having->();
+
+	warn "## having = ", dump( @having ) if ($debug > 2);
+
+	foreach my $h ( @having ) {
+		if (defined($lookup->{$database}->{$input}->{$key}->{$h})) {
+			warn "lookup for $database/$input/$key/$h return ",dump($lookup->{$database}->{$input}->{$key}->{$h}),"\n" if ($debug);
+			$mfns->{$_}++ foreach keys %{ $lookup->{$database}->{$input}->{$key}->{$h} };
 		}
+	}
+
+	return unless ($mfns);
+
+	my @mfns = sort keys %$mfns;
+
+	warn "# lookup loading $database/$input/$key mfn ", join(",",@mfns)," having ",dump(@having),"\n" if ($debug);
+
+	my $old_rec = $rec;
+	my @out;
+
+	foreach my $mfn (@mfns) {
+		$rec = $load_row_coderef->( $database, $input, $mfn );
+
+		warn "got $database/$input/$mfn = ", dump($rec), $/ if ($debug);
+
+		my @vals = $what->();
+
+		push @out, ( @vals );
+
+		warn "lookup for mfn $mfn returned ", dump(@vals), $/ if ($debug);
+	}
+
+#	if (ref($lookup->{$k}) eq 'ARRAY') {
+#		return @{ $lookup->{$k} };
+#	} else {
+#		return $lookup->{$k};
+#	}
+
+	$rec = $old_rec;
+
+	warn "## lookup returns = ", dump(@out), $/ if ($debug);
+
+	if ($#out == 0) {
+		return $out[0];
 	} else {
-		return '';
+		return @out;
 	}
 }
 
+=head2 save_into_lookup
+
+Save value into lookup. It associates current database, input
+and specific keys with one or more values which will be
+associated over MFN.
 
-=head2 apply_format
+MFN will be extracted from first occurence current of field 000
+in current record, or if it doesn't exist from L<_set_config> C<_mfn>.
 
-Apply format specified in tag with C<format_name="name"> and
-C<format_delimiter=";;">.
+  my $nr = save_into_lookup($database,$input,$key,sub {
+ 	# code which produce one or more values 
+  });
 
- my $text = $webpac->apply_format($format_name,$format_delimiter,$data);
+It returns number of items saved.
 
-Formats can contain C<lookup{...}> if you need them.
+This function shouldn't be called directly, it's called from code created by
+L<WebPAC::Parser>. 
 
 =cut
 
-sub apply_format {
-	my $self = shift;
+sub save_into_lookup {
+	my ($database,$input,$key,$coderef) = @_;
+	die "save_into_lookup needs database" unless defined($database);
+	die "save_into_lookup needs input" unless defined($input);
+	die "save_into_lookup needs key" unless defined($key);
+	die "save_into_lookup needs CODE" unless ( defined($coderef) && ref($coderef) eq 'CODE' );
+
+	warn "## save_into_lookup rec = ", dump($rec), " config = ", dump($config), $/ if ($debug > 2);
 
-	my ($name,$delimiter,$data) = @_;
+	my $mfn = 
+		defined($rec->{'000'}->[0])	?	$rec->{'000'}->[0]	:
+		defined($config->{_mfn})	?	$config->{_mfn}		:
+										die "mfn not defined or zero";
 
-	my $log = $self->_get_logger();
+	my $nr = 0;
 
-	if (! $self->{'import_xml'}->{'format'}->{$name}) {
-		$log->warn("<format name=\"$name\"> is not defined in ",$self->{'import_xml_file'});
-		return $data;
+	foreach my $v ( $coderef->() ) {
+		$lookup->{$database}->{$input}->{$key}->{$v}->{$mfn}++;
+		warn "# saved lookup $database/$input/$key [$v] $mfn\n" if ($debug > 1);
+		$nr++;
 	}
 
-	$log->warn("no delimiter for format $name") if (! $delimiter);
+	return $nr;
+}
+
+=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');
+  $database_input_name = config('input name');
+
+Up to three levels are supported.
+
+=cut
+
+sub config {
+	return unless ($config);
 
-	my $format = $self->_x($self->{'import_xml'}->{'format'}->{$name}->{'content'}) || $log->logdie("can't find format '$name'");
+	my $p = shift;
 
-	my @data = split(/\Q$delimiter\E/, $data);
+	$p ||= '';
 
-	my $out = sprintf($format, @data);
-	$log->debug("using format $name [$format] on $data to produce: $out");
+	my $v;
 
-	if ($self->{'lookup_regex'} && $out =~ $self->{'lookup_regex'}) {
-		return $self->lookup($out);
+	warn "### getting config($p)\n" if ($debug > 1);
+
+	my @p = split(/\s+/,$p);
+	if ($#p < 0) {
+		$v = $config->{ '_' };	# special, database code
 	} else {
-		return $out;
+
+		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 sort_arr
+=head2 id
+
+Returns unique id of this record
 
-Sort array ignoring case and html in data
+  $id = id();
 
- my @sorted = $webpac->sort_arr(@unsorted);
+Returns C<42/2> for 2nd occurence of MFN 42.
 
 =cut
 
-sub sort_arr {
-	my $self = shift;
+sub id {
+	my $mfn = $config->{_mfn} || die "no _mfn in config data";
+	return $mfn . ( WebPAC::Normalize::MARC::_created_marc_records() || '' );
+}
 
-	my $log = $self->_get_logger();
+=head2 join_with
 
-	# FIXME add Schwartzian Transformation?
+Joins walues with some delimiter
 
-	my @sorted = sort {
-		$a =~ s#<[^>]+/*>##;
-		$b =~ s#<[^>]+/*>##;
-		lc($b) cmp lc($a)
-	} @_;
-	$log->debug("sorted values: ",sub { join(", ",@sorted) });
+  $v = join_with(", ", @v);
 
-	return @sorted;
+=cut
+
+sub join_with {
+	my $d = shift;
+	warn "### join_with('$d',",dump(@_),")\n" if ($debug > 2);
+	my $v = join($d, grep { defined($_) && $_ ne '' } @_);
+	return '' unless defined($v);
+	return $v;
 }
 
+=head2 split_rec_on
 
-=head1 INTERNAL METHODS
+Split record subfield on some regex and take one of parts out
 
-=head2 _sort_by_order
+  $a_before_semi_column =
+  	split_rec_on('200','a', /\s*;\s*/, $part);
 
-Sort xml tags data structure accoding to C<order=""> attribute.
+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 _sort_by_order {
-	my $self = shift;
+sub split_rec_on {
+	die "split_rec_on need (fld,sf,regex[,part]" if ($#_ < 2);
 
-	my $va = $self->{'import_xml'}->{'indexer'}->{$a}->{'order'} ||
-		$self->{'import_xml'}->{'indexer'}->{$a};
-	my $vb = $self->{'import_xml'}->{'indexer'}->{$b}->{'order'} ||
-		$self->{'import_xml'}->{'indexer'}->{$b};
+	my ($fld, $sf, $regex, $part) = @_;
+	warn "### regex ", ref($regex), $regex, $/ if ($debug > 2);
 
-	return $va <=> $vb;
-}
+	my @r = rec( $fld, $sf );
+	my $v = shift @r;
+	warn "### first rec($fld,$sf) = ",dump($v),$/ if ($debug > 2);
 
-=head2 _x
+	return '' if ( ! defined($v) || $v =~ /^\s*$/);
 
-Convert strings from C<conf/normalize/*.xml> encoding into application
-specific encoding (optinally specified using C<code_page> to C<new>
-constructor).
+	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;
+	}
+}
 
- my $text = $n->_x('normalize text string');
+my $hash;
 
-This is a stub so that other modules doesn't have to implement it.
+=head2 set
+
+  set( key => 'value' );
 
 =cut
 
-sub _x {
-	my $self = shift;
-	return shift;
-}
+sub set {
+	my ($k,$v) = @_;
+	warn "## set ( $k => ", dump($v), " )", $/ if ( $debug );
+	$hash->{$k} = $v;
+};
 
+=head2 get
 
-=head1 AUTHOR
+  get( 'key' );
 
-Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
+=cut
 
-=head1 COPYRIGHT & LICENSE
+sub get {
+	my $k = shift || return;
+	my $v = $hash->{$k};
+	warn "## get $k = ", dump( $v ), $/ if ( $debug );
+	return $v;
+}
 
-Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
+=head2 count
 
-This program is free software; you can redistribute it and/or modify it
-under the same terms as Perl itself.
+  if ( count( @result ) == 1 ) {
+  	# do something if only 1 result is there
+  }
 
 =cut
 
-1; # End of WebPAC::DB
+sub count {
+	warn "## count ",dump(@_),$/ if ( $debug );
+	return @_ . '';
+}
+
+# END
+1;