--- trunk2/lib/WebPAC.pm	2004/09/12 20:31:34	431
+++ trunk2/lib/WebPAC.pm	2004/09/15 21:21:36	453
@@ -34,8 +34,8 @@
 
  my $webpac = new WebPAC(
  	config_file => 'name.conf',
-	[code_page => 'ISO-8859-2',]
-	[low_mem => 1,]
+	code_page => 'ISO-8859-2',
+	low_mem => 1,
  );
 
 Default C<code_page> is C<ISO-8859-2>.
@@ -123,7 +123,7 @@
 			$log->debug("removed '$db_file' from last run");
 		}
 
-		use DBM::Deep;
+		require DBM::Deep;
 
 		my $db = new DBM::Deep $db_file;
 
@@ -185,6 +185,8 @@
 	$log->logcroak("need filename") if (! $arg->{'filename'});
 	my $code_page = $arg->{'code_page'} || '852';
 
+	$log->logdie("can't find database ",$arg->{'filename'}) unless (glob($arg->{'filename'}.'.*'));
+
 	# store data in object
 	$self->{'isis_filename'} = $arg->{'filename'};
 	$self->{'isis_code_page'} = $code_page;
@@ -207,6 +209,8 @@
 	if (my $s = $self->{'start_mfn'}) {
 		$log->info("skipping to MFN $s");
 		$startmfn = $s;
+	} else {
+		$self->{'start_mfn'} = $startmfn;
 	}
 
 	$maxmfn = $startmfn + $self->{limit_mfn} if ($self->{limit_mfn});
@@ -262,7 +266,7 @@
 
 	}
 
-	$self->{'current_mfn'} = $startmfn;
+	$self->{'current_mfn'} = -1;
 	$self->{'last_pcnt'} = 0;
 
 	$log->debug("max mfn: $maxmfn");
@@ -285,7 +289,15 @@
 
 	my $log = $self->_get_logger();
 
-	my $mfn = $self->{'current_mfn'}++ || $log->logconfess("it seems that you didn't load database!");
+	$log->logconfess("it seems that you didn't load database!") unless ($self->{'current_mfn'});
+
+	if ($self->{'current_mfn'} == -1) {
+		$self->{'current_mfn'} = $self->{'start_mfn'};
+	} else {
+		$self->{'current_mfn'}++;
+	}
+
+	my $mfn = $self->{'current_mfn'};
 
 	if ($mfn > $self->{'max_mfn'}) {
 		$self->{'current_mfn'} = $self->{'max_mfn'};
@@ -302,6 +314,19 @@
 	}
 }
 
+=head2 mfn
+
+Returns current record number (MFN).
+
+ print $webpac->mfn;
+
+=cut
+
+sub mfn {
+	my $self = shift;
+	return $self->{'current_mfn'};
+}
+
 =head2 progress_bar
 
 Draw progress bar on STDERR.
@@ -333,7 +358,7 @@
 	if ($p < $self->{'last_pcnt'}) {
 		$self->{'last_pcnt'} = $p;
 		$self->{'last_t'} = time();
-		$self->{'last_curr'} = 1;
+		$self->{'last_curr'} = undef;
 	}
 
 	if ($p != $self->{'last_pcnt'}) {
@@ -752,6 +777,31 @@
 	return @arr;
 }
 
+=head2 sort_arr
+
+Sort array ignoring case and html in data
+
+ my @sorted = $webpac->sort_arr(@unsorted);
+
+=cut
+
+sub sort_arr {
+	my $self = shift;
+
+	my $log = $self->_get_logger();
+
+	# FIXME add Schwartzian Transformation?
+
+	my @sorted = sort {
+		$a =~ s#<[^>]+/*>##;
+		$b =~ s#<[^>]+/*>##;
+		lc($b) cmp lc($a)
+	} @_;
+	$log->debug("sorted values: ",sub { join(", ",@sorted) });
+
+	return @sorted;
+}
+
 
 =head2 data_structure
 
@@ -808,6 +858,11 @@
 			}
 			next if (! @v);
 
+			if ($tag->{'sort'}) {
+				@v = $self->sort_arr(@v);
+				$log->warn("sort within tag is usually not what you want!");
+			}
+
 			# use format?
 			if ($tag->{'format_name'}) {
 				@v = map { $self->apply_format($tag->{'format_name'},$tag->{'format_delimiter'},$_) } @v;
@@ -822,12 +877,35 @@
 				next; # don't return headline in data_structure!
 			}
 
-			# does tag have type?
-			if ($tag->{'type'}) {
-				push @{$row->{$tag->{'type'}}}, @v;
-			} else {
-				push @{$row->{'display'}}, @v;
-				push @{$row->{'swish'}}, @v;
+			# delimiter will join repeatable fields
+			if ($tag->{'delimiter'}) {
+				@v = ( join($tag->{'delimiter'}, @v) );
+			}
+
+			# 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}}, $last;
+
+				} else {
+					push @{$row->{$type}}, @v;
+				}
 			}
 
 
@@ -840,6 +918,11 @@
 			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");
+			}
+
 			push @ds, $row;
 
 			$log->debug("row $field: ",sub { Dumper($row) });