--- trunk/lib/WebPAC/Normalize.pm	2005/07/17 10:42:23	15
+++ trunk/lib/WebPAC/Normalize.pm	2005/12/19 15:34:47	295
@@ -2,8 +2,8 @@
 
 use warnings;
 use strict;
+use base 'WebPAC::Common';
 use Data::Dumper;
-use Storable;
 
 =head1 NAME
 
@@ -11,11 +11,11 @@
 
 =head1 VERSION
 
-Version 0.01
+Version 0.07
 
 =cut
 
-our $VERSION = '0.01';
+our $VERSION = '0.07';
 
 =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>.
@@ -79,18 +83,21 @@
 			return length($_);
 		}, ...
 	},
-	cache_data_structure => './cache/ds/',
+	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.
 
-Optional parameter C<cache_data_structure> defines path to directory
-in which cache file for C<data_structure> call will be created.
+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.
+in structures. If you pass this parametar, you must also pass C<lookup>
+which is C<WebPAC::Lookup> object.
 
 =cut
 
@@ -99,49 +106,33 @@
         my $self = {@_};
         bless($self, $class);
 
-	$self->setup_cache_dir( $self->{'cache_data_structure'} );
-
-	$self ? return $self : return undef;
-}
-
-=head2 setup_cache_dir
+	my $r = $self->{'lookup_regex'} ? 1 : 0;
+	my $l = $self->{'lookup'} ? 1 : 0;
 
-Check if specified cache directory exist, and if not, disable caching.
-
- $setup_cache_dir('./cache/ds/');
-
-If you pass false or zero value to this function, it will disable
-cacheing.
-
-=cut
+	my $log = $self->_get_logger();
 
-sub setup_cache_dir {
-	my $self = shift;
+	# those two must be in pair
+	if ( ($r & $l) != ($r || $l) ) {
+		my $log = $self->_get_logger();
+		$log->logdie("lookup_regex and lookup must be in pair");
+	}
 
-	my $dir = shift;
+	$log->logdie("lookup must be WebPAC::Lookup object") if ($self->{'lookup'} && ! $self->{'lookup'}->isa('WebPAC::Lookup'));
 
-	my $log = $self->_get_logger();
+	$log->warn("no prefix defined. please check that!") unless ($self->{'prefix'});
 
-	if ($dir) {
-		my $msg;
-		if (! -e $dir) {
-			$msg = "doesn't exist";
-		} elsif (! -d $dir) {
-			$msg = "is not directory";
-		} elsif (! -w $dir) {
-			$msg = "not writable";
-		}
+	$log->debug("using lookup regex: ", $self->{lookup_regex}) if ($r && $l);
 
-		if ($msg) {
-			undef $self->{'cache_data_structure'};
-			$log->warn("cache_data_structure $dir $msg, disabling...");
-		} else {
-			$log->debug("using cache dir $dir");
-		}
-	} else {
-		$log->debug("disabling cache");
-		undef $self->{'cache_data_structure'};
+	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;
 }
 
 
@@ -152,13 +143,7 @@
 
 This structures are used to produce output.
 
- 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.
+ my $ds = $webpac->data_structure($rec);
 
 =cut
 
@@ -170,42 +155,21 @@
 	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 (my $cache_path = $self->{'cache_data_structure'}) {
-		my $id = $rec->{'000'};
-		$id = $rec->{'000'}->[0] if ($id =~ m/^ARRAY/o);
-		unless (defined($id)) {
-			$log->warn("Can't use cache_data_structure on records without unique identifier in field 000");
-			undef $self->{'cache_data_structure'};
-		} else {
-			$cache_file = "$cache_path/$id";
-			if (-r $cache_file) {
-				my $ds_ref = retrieve($cache_file);
-				if ($ds_ref) {
-					$log->debug("cache hit: $cache_file");
-					my $ok = 1;
-					foreach my $f (qw(current_filename headline)) {
-						if ($ds_ref->{$f}) {
-							$self->{$f} = $ds_ref->{$f};
-						} else {
-							$ok = 0;
-						}
-					};
-					if ($ok && $ds_ref->{'ds'}) {
-						return @{ $ds_ref->{'ds'} };
-					} else {
-						$log->warn("cache_data_structure $cache_path corrupt. Use rm $cache_path/* to re-create it on next run!");
-						undef $self->{'cache_data_structure'};
-					}
-				}
-			}
-		}
+	if ($self->{'db'}) {
+		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");
 	}
 
-	undef $self->{'currnet_filename'};
-	undef $self->{'headline'};
-
 	my @sorted_tags;
 	if ($self->{tags_by_order}) {
 		@sorted_tags = @{$self->{tags_by_order}};
@@ -214,7 +178,7 @@
 		$self->{tags_by_order} = \@sorted_tags;
 	}
 
-	my @ds;
+	my $ds;
 
 	$log->debug("tags: ",sub { join(", ",@sorted_tags) });
 
@@ -225,7 +189,10 @@
 #print "field $field [",$self->{'tag'},"] = ",Dumper($self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}});
 
 		foreach my $tag (@{$self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}}) {
-			my $format = $tag->{'value'} || $tag->{'content'};
+			my $format;
+
+			$log->logdie("expected tag HASH and got $tag") unless (ref($tag) eq 'HASH');
+			$format = $tag->{'value'} || $tag->{'content'};
 
 			$log->debug("format: $format");
 
@@ -246,28 +213,19 @@
 				@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) );
 			}
 
 			# default types 
-			my @types = qw(display swish);
+			my @types = qw(display search);
 			# 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');
+				$log->debug("type: $type ",sub { join(" ",@v) }, " ", $row->{'append'} || 'no append');
 				if ($tag->{'append'}) {
 
 					# I will delimit appended part with
@@ -294,30 +252,31 @@
 
 			# TODO: name_sigular, name_plural
 			my $name = $self->{'import_xml'}->{'indexer'}->{$field}->{'name'};
-			$row->{'name'} = $name ? $self->_x($name) : $field;
+			my $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");
 			}
 
-			push @ds, $row;
+			$ds->{$row_name} = $row;
 
 			$log->debug("row $field: ",sub { Dumper($row) });
 		}
 
 	}
 
-	if ($cache_file) {
-		store {
-			ds => \@ds,
-			current_filename => $self->{'current_filename'},
-			headline => $self->{'headline'},
-		}, $cache_file;
-		$log->debug("created storable cache file $cache_file");
-	}
+	$self->{'db'}->save_ds(
+		id => $id,
+		ds => $ds,
+		prefix => $self->{prefix},
+	) if ($self->{'db'});
+
+	$log->debug("ds: ", sub { Dumper($ds) });
 
-	return @ds;
+	$log->logconfess("data structure returned is not array any more!") if wantarray;
+
+	return $ds;
 
 }
 
@@ -329,8 +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;
 
@@ -400,11 +369,21 @@
 		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 $out;
@@ -513,7 +492,13 @@
 		}
 		# do we have lookups?
 		if ($self->{'lookup'}) {
-			return $self->lookup($format);
+			if ($self->{'lookup'}->can('lookup')) {
+				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");
+			}
 		} else {
 			return $format;
 		}
@@ -586,7 +571,7 @@
 		if ($sf && $$rec->{$f}->[$i]->{$sf}) {
 			$$found++ if (defined($$found));
 			return $$rec->{$f}->[$i]->{$sf};
-		} elsif ($$rec->{$f}->[$i]) {
+		} elsif (! $sf && $$rec->{$f}->[$i]) {
 			$$found++ if (defined($$found));
 			# it still might have subfield, just
 			# not specified, so we'll dump all
@@ -599,6 +584,8 @@
 			} else {
 				return $$rec->{$f}->[$i];
 			}
+		} else {
+			return '';
 		}
 	} else {
 		return '';
@@ -639,7 +626,7 @@
 	$log->debug("using format $name [$format] on $data to produce: $out");
 
 	if ($self->{'lookup_regex'} && $out =~ $self->{'lookup_regex'}) {
-		return $self->lookup($out);
+		return $self->{'lookup'}->lookup($out);
 	} else {
 		return $out;
 	}
@@ -722,4 +709,4 @@
 
 =cut
 
-1; # End of WebPAC::DB
+1; # End of WebPAC::Normalize