--- trunk/lib/WebPAC/Normalize.pm	2005/12/15 17:01:10	253
+++ trunk/lib/WebPAC/Normalize.pm	2006/01/07 03:28:10	346
@@ -11,11 +11,11 @@
 
 =head1 VERSION
 
-Version 0.05
+Version 0.08
 
 =cut
 
-our $VERSION = '0.05';
+our $VERSION = '0.08';
 
 =head1 SYNOPSIS
 
@@ -47,6 +47,10 @@
 code defined as code ref on format after field substitution to producing
 output
 
+There is one built-in filter called C<regex> which can be use like this:
+
+  filter{regex(s/foo/bar/)}
+
 =item *
 
 optional C<lookup{...}> will be then performed. See C<WebPAC::Lookups>.
@@ -119,6 +123,15 @@
 
 	$log->debug("using lookup regex: ", $self->{lookup_regex}) if ($r && $l);
 
+	if (! $self->{filter} || ! $self->{filter}->{regex}) {
+		$log->debug("adding built-in filter regex");
+		$self->{filter}->{regex} = sub {
+			my ($val, $regex) = @_;
+			eval "\$val =~ $regex";
+			return $val;
+		};
+	}
+
 	$self ? return $self : return undef;
 }
 
@@ -144,7 +157,7 @@
 
 	$log->debug("data_structure rec = ", sub { Dumper($rec) });
 
-	$log->logdie("need unique ID (mfn) in field 000 of record ", sub { Dumper($rec) } ) unless (defined($rec->{'000'}));
+	$log->logdie("need unique ID (mfn) in field 000 of record " . Dumper($rec) ) unless (defined($rec->{'000'}));
 
 	my $id = $rec->{'000'}->[0] || $log->logdie("field 000 isn't array!");
 
@@ -157,9 +170,6 @@
 		$log->debug("cache miss, creating");
 	}
 
-	undef $self->{'currnet_filename'};
-	undef $self->{'headline'};
-
 	my @sorted_tags;
 	if ($self->{tags_by_order}) {
 		@sorted_tags = @{$self->{tags_by_order}};
@@ -278,8 +288,18 @@
 
  my $text = $webpac->parse($rec,'eval{"v901^a" eq "Deskriptor"}descriptor: v250^a', $i);
 
+Filters are implemented here. While simple form of filters looks like this:
+
+  filter{name_of_filter}
+
+but, filters can also have variable number of parametars like this:
+
+  filter{name_of_filter(param,param,param)}
+
 =cut
 
+my $warn_once;
+
 sub parse {
 	my $self = shift;
 
@@ -297,7 +317,7 @@
 
 	my @out;
 
-	$log->debug("format: $format");
+	$log->debug("format: $format [$i]");
 
 	my $eval_code;
 	# remove eval{...} from beginning
@@ -307,29 +327,45 @@
 	# remove filter{...} from beginning
 	$filter_name = $1 if ($format =~ s/^filter{([^}]+)}//s);
 
+	# did we found any (att all) field from format in row?
+	my $found_any;
+	# prefix before first field which we preserve it $found_any
 	my $prefix;
-	my $all_found=0;
+
+	my $f_step = 1;
 
 	while ($format =~ s/^(.*?)(v|s)(\d+)(?:\^(\w))?//s) {
 
 		my $del = $1 || '';
-		$prefix ||= $del if ($all_found == 0);
+		$prefix = $del if ($f_step == 1);
+
+		my $fld_type = lc($2);
 
 		# repeatable index
 		my $r = $i;
-		$r = 0 if (lc("$2") eq 's');
+		if ($fld_type eq 's') {
+			if ($found_any->{'v'}) {
+				$r = 0;
+			} else {
+				return;
+			}
+		}
 
 		my $found = 0;
 		my $tmp = $self->get_data(\$rec,$3,$4,$r,\$found);
 
 		if ($found) {
-			push @out, $del;
+			$found_any->{$fld_type} += $found;
+
+			# we will skip delimiter before first occurence of field!
+			push @out, $del unless($found_any->{$fld_type} == 1);
 			push @out, $tmp;
-			$all_found += $found;
 		}
+		$f_step++;
 	}
 
-	return if (! $all_found);
+	# test if any fields found?
+	return if (! $found_any->{'v'} && ! $found_any->{'s'});
 
 	my $out = join('',@out);
 
@@ -349,11 +385,21 @@
 		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: $out");
+	if ($filter_name) {
+		my @filter_args;
+		if ($filter_name =~ s/(\w+)\((.*)\)/$1/) {
+			@filter_args = split(/,/, $2);
+		}
+		if ($self->{'filter'}->{$filter_name}) {
+			$log->debug("about to filter{$filter_name} format: $out with arguments: ", join(",", @filter_args));
+			unshift @filter_args, $out;
+			$out = $self->{'filter'}->{$filter_name}->(@filter_args);
+			return unless(defined($out));
+			$log->debug("filter result: $out");
+		} elsif (! $warn_once->{$filter_name}) {
+			$log->warn("trying to use undefined filter $filter_name");
+			$warn_once->{$filter_name}++;
+		}
 	}
 
 	return $out;
@@ -433,6 +479,7 @@
 	}
 
 	my $found = 0;
+	my $just_single = 1;
 
 	my $eval_code;
 	# remove eval{...} from beginning
@@ -444,9 +491,14 @@
 
 	# do actual replacement of placeholders
 	# repeatable fields
-	$format =~ s/v(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,$i,\$found)/ges;
+	if ($format =~ s/v(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,$i,\$found)/ges) {
+		$just_single = 0;
+	}
+	
 	# non-repeatable fields
-	$format =~ s/s(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,0,\$found)/ges;
+	if ($format =~ s/s(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,0,\$found)/ges) {
+		return if ($i > 0 && $just_single);
+	}
 
 	if ($found) {
 		$log->debug("format: $format");
@@ -548,7 +600,8 @@
 			if ($$rec->{$f}->[$i] =~ /HASH/o) {
 				my $out;
 				foreach my $k (keys %{$$rec->{$f}->[$i]}) {
-					$out .= $$rec->{$f}->[$i]->{$k}." ";
+					my $v = $$rec->{$f}->[$i]->{$k};
+					$out .= "$v " if ($v);
 				}
 				return $out;
 			} else {