--- trunk/all2xml.pl	2003/11/29 19:07:00	188
+++ trunk/all2xml.pl	2003/12/21 03:27:02	197
@@ -72,6 +72,8 @@
 # if you are tight on memory, turn this off
 my $use_lhash_cache = 1;
 
+my $last_field_name;	# cache to prevent repeated fields
+
 sub data2xml {
 
 	use xmlify;
@@ -138,7 +140,13 @@
 			print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
 		}
 		if ($field_name) {
-			return x($field_name);
+			if (! $last_field_name) {
+				$last_field_name = x($field_name);
+				return $last_field_name;
+			} elsif ($field_name ne $last_field_name) {
+				$last_field_name = x($field_name);
+				return $last_field_name;
+			}
 		}
 	}
 
@@ -167,6 +175,13 @@
 		my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '<br/>';
 		$cache->{index_delimiter}->{$field} = $config->{indexer}->{$field}->{index_delimiter};
 
+		my $format_name = $config->{indexer}->{$field}->{format_name};
+		my $format_delimiter = $config->{indexer}->{$field}->{format_delimiter};
+		if ($format_name && $format_delimiter) {
+			$cache->{format}->{$field}->{format_name} = $format_name;
+			$cache->{format}->{$field}->{format_delimiter} = $format_delimiter;
+		}
+
 		foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) {
 
 			my $format = x($x->{content});
@@ -195,7 +210,7 @@
 
 			# placeholder for all repeatable entries for index
 
-			sub mkformat {
+			sub mkformat($$) {
 				my $x = shift || die "mkformat needs tag reference";
 				my $data = shift || return;
 				my $format_name = x($x->{format_name}) || return $data;
@@ -267,11 +282,11 @@
 						no strict 'refs';
 						my $tmp = join(" ",&$filter($swish)) if ($s || $se);
 						$swish_data .= $tmp if ($s);
-						$swish_exact_data .= $tmp if ($se);
+						$swish_exact_data .= "xxbxx $tmp xxexx " if ($se && $tmp ne "");
 
 					} else {
 						$swish_data .= $swish if ($s);
-						$swish_exact_data .= $swish if ($se);
+						$swish_exact_data .= "xxbxx $swish xxexx " if ($se && $swish ne "");
 					}
 				}
 
@@ -362,7 +377,11 @@
 							@index_data = @{$cache->{index_data}->{$field}->[$page]};
 						}
 						if ($x->{append}) {
-							$index_data[$#index_data] .= $idisplay;
+							if (@index_data) {
+								$index_data[$#index_data] .= $idisplay;
+							} else {
+								push @index_data, $idisplay;
+							}
 						} else {
 							push @index_data, $idisplay;
 						}
@@ -430,8 +449,12 @@
 #print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n";
 #print STDERR Dumper($cache->{display_data});
 			for (my $page=0; $page <= $nr_pages; $page++) {
-
-				my $display_data = $cache->{display_data}->{$field}->[$page];
+				my $display_data;
+				if ($cache->{format}->{$field}) {
+					$display_data=mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]);
+				} else {
+					$display_data = $cache->{display_data}->{$field}->[$page];
+				}
 				if ($display_data) { # default
 					if ($field eq "headline") {
 						$xml .= xmlify("headline", $display_data);
@@ -459,7 +482,7 @@
 
 					# add delimiters before and after word.
 					# That is required to produce exact match
-					$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx'));
+					$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data));
 				}
 				
 				my $idel = $cache->{index_delimiter}->{$field};
@@ -501,7 +524,7 @@
 
 				# add delimiters before and after word.
 				# That is required to produce exact match
-				$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx'));
+				$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data));
 			}
 		}
 	}