--- trunk/all2xml.pl	2003/11/29 19:07:00	188
+++ trunk/all2xml.pl	2004/01/31 21:03:06	207
@@ -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});
@@ -195,10 +210,23 @@
 
 			# placeholder for all repeatable entries for index
 
-			sub mkformat {
+			sub chk_eval($) {
+				my $data = shift;
+				if ($data =~ s/eval{([^}]+)}//) {
+					if (eval "$1") {
+						return $data;
+					} else {
+						return undef;
+					}
+				} else {
+					return $data;
+				}
+			}
+
+			sub mkformat($$) {
 				my $x = shift || die "mkformat needs tag reference";
 				my $data = shift || return;
-				my $format_name = x($x->{format_name}) || return $data;
+				my $format_name = x($x->{format_name}) || return chk_eval($data);
 				my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!";
 				my $format_delimiter = x($x->{format_delimiter});
 				my @data;
@@ -211,10 +239,10 @@
 				if ($fmt) {
 					my $nr = scalar $fmt =~ s/%s/%s/g;
 					if (($#data+1) == $nr) {
-						return sprintf($fmt,@data);
+						return chk_eval(sprintf($fmt,@data));
 					} else {
-						print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n";
-						return $data;
+						#print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n";
+						return chk_eval($data);
 					}
 				} else {
 					print STDERR "usage of link '$format_name' without defined format (<link> tag)\n";
@@ -267,32 +295,33 @@
 						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 "");
 					}
 				}
 
 				# type="display" ; field for display
 				if ($d && $display) {
+					my $ldel = $delimiter;
 					if ($line_delimiter && $display_data) {
-						$display_data .= $line_delimiter;
+						$ldel = $line_delimiter;
 					}
 					if ($filter) {
 						no strict 'refs';
-						if ($display_data) {
-							$display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display)));
-						} else {
-							$display_data = join($delimiter,mkformat($x,&$filter($display)));
+						my @arr;
+						foreach my $tmp (&$filter($display)) {
+							my $tmp2 = mkformat($x,$tmp);
+							push @arr,$tmp2 if ($tmp2);
 						}
+						$display_data .= $ldel if ($display_data && @arr);
+						$display_data .= join($delimiter,@arr);
 					} else {
-						if ($display_data) {
-							$display_data .= $delimiter.mkformat($x,$display);
-						} else {
-							$display_data = mkformat($x,$display);
-						}
+						$display_data .= $ldel if ($display_data);
+						my $tmp = mkformat($x,$display);
+						$display_data = $tmp if ($tmp);
 					}
 				}
 						
@@ -362,7 +391,11 @@
 							@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, $idisplay;
 						}
@@ -430,8 +463,13 @@
 #print STDERR "field '$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}) {
+					my $tmp = mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]);
+					$display_data=$tmp if ($tmp);
+				} else {
+					$display_data = $cache->{display_data}->{$field}->[$page];
+				}
 				if ($display_data) { # default
 					if ($field eq "headline") {
 						$xml .= xmlify("headline", $display_data);
@@ -459,7 +497,7 @@
 
 					# 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 +539,7 @@
 
 				# 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));
 			}
 		}
 	}