--- trunk/lib/WebPAC/Normalize.pm	2005/11/20 20:13:39	74
+++ trunk/lib/WebPAC/Normalize.pm	2005/12/16 16:00:18	261
@@ -11,11 +11,11 @@
 
 =head1 VERSION
 
-Version 0.02
+Version 0.06
 
 =cut
 
-our $VERSION = '0.02';
+our $VERSION = '0.06';
 
 =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>.
@@ -82,11 +86,15 @@
 	db => $db_obj,
 	lookup_regex => $lookup->regex,
 	lookup => $lookup_obj,
+	prefix => 'foobar',
   );
 
 Parametar C<filter> defines user supplied snippets of perl code which can
 be use with C<filter{...}> notation.
 
+C<prefix> is used to form filename for database record (to support multiple
+source files which are joined in one database).
+
 Recommended parametar C<lookup_regex> is used to enable parsing of lookups
 in structures. If you pass this parametar, you must also pass C<lookup>
 which is C<WebPAC::Lookup> object.
@@ -111,6 +119,19 @@
 
 	$log->logdie("lookup must be WebPAC::Lookup object") if ($self->{'lookup'} && ! $self->{'lookup'}->isa('WebPAC::Lookup'));
 
+	$log->warn("no prefix defined. please check that!") unless ($self->{'prefix'});
+
+	$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;
 }
 
@@ -124,12 +145,6 @@
 
  my $ds = $webpac->data_structure($rec);
 
-B<Note: historical oddity follows>
-
-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.
-
 =cut
 
 sub data_structure {
@@ -140,10 +155,16 @@
 	my $rec = shift;
 	$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o);
 
+	$log->debug("data_structure rec = ", sub { Dumper($rec) });
+
+	$log->logdie("need unique ID (mfn) in field 000 of record ", sub { Dumper($rec) } ) unless (defined($rec->{'000'}));
+
+	my $id = $rec->{'000'}->[0] || $log->logdie("field 000 isn't array!");
+
 	my $cache_file;
 
 	if ($self->{'db'}) {
-		my $ds = $self->{'db'}->load_ds($rec);
+		my $ds = $self->{'db'}->load_ds( id => $id, prefix => $self->{prefix} );
 		$log->debug("load_ds( rec = ", sub { Dumper($rec) }, ") = ", sub { Dumper($ds) });
 		return $ds if ($ds);
 		$log->debug("cache miss, creating");
@@ -195,15 +216,6 @@
 				@v = map { $self->apply_format($tag->{'format_name'},$tag->{'format_delimiter'},$_) } @v;
 			}
 
-			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!
-			}
-
 			# delimiter will join repeatable fields
 			if ($tag->{'delimiter'}) {
 				@v = ( join($tag->{'delimiter'}, @v) );
@@ -216,7 +228,7 @@
 
 			foreach my $type (@types) {
 				# append to previous line?
-				$log->debug("type: $type ",sub { join(" ",@v) }, $row->{'append'} || 'no append');
+				$log->debug("type: $type ",sub { join(" ",@v) }, " ", $row->{'append'} || 'no append');
 				if ($tag->{'append'}) {
 
 					# I will delimit appended part with
@@ -257,12 +269,10 @@
 
 	}
 
-	$log->logdie("there is no current_filename defined! Do you have filename tag in conf/normalize/?.xml") unless ($self->{'current_filename'});
-
 	$self->{'db'}->save_ds(
+		id => $id,
 		ds => $ds,
-		current_filename => $self->{'current_filename'},
-		headline => $self->{'headline'},
+		prefix => $self->{prefix},
 	) if ($self->{'db'});
 
 	$log->debug("ds: ", sub { Dumper($ds) });
@@ -281,8 +291,18 @@
 
  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;
 
@@ -352,11 +372,21 @@
 		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 $out;
@@ -466,7 +496,9 @@
 		# do we have lookups?
 		if ($self->{'lookup'}) {
 			if ($self->{'lookup'}->can('lookup')) {
-				return $self->{'lookup'}->lookup($format);
+				my @lookup = $self->{lookup}->lookup($format);
+				$log->debug("lookup $format", join(", ", @lookup));
+				return @lookup;
 			} else {
 				$log->warn("Have lookup object but can't invoke lookup method");
 			}
@@ -680,4 +712,4 @@
 
 =cut
 
-1; # End of WebPAC::DB
+1; # End of WebPAC::Normalize