--- trunk/all2xml.pl	2003/02/22 21:33:04	17
+++ trunk/all2xml.pl	2003/02/23 07:08:54	29
@@ -40,13 +40,14 @@
 
 Text::Iconv->raise_error(1);     # Conversion errors raise exceptions
 
-#my $isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8');
-#my $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage});
 my $isis_codepage;
 my $index_codepage;
 my $cludge_codepage = Text::Iconv->new('UTF8','ISO8859-1');
 my $xml_codepage;
 
+my $XML_CHARSET = 'UTF8';
+
+
 sub isis2xml {
 
 	use xmlify;
@@ -62,47 +63,104 @@
 
 	my %field_usage;	# counter for usage of each field
 
+
 	foreach my $field (keys %{$config->{indexer}}) {
 
 		$field_usage{$field}++;
 
 		my $swish_data = "";
 		my $display_data = "";
+		my $line_delimiter = "";
 
 		foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
 
 			my $format = $x->{content};
+			my $delimiter = $x->{delimiter} || ' ';
+
+			# FIX: this is ugly, UGLY, cludge string is returned
+			# in UTF8 encoding , but as if source charset
+			# is ISO8859-1 and not some other. This break other
+			# character encodings, so we convert it first
+			# back to ISO8859-1
+			$format = $xml_codepage->convert($format);
+			$delimiter = $xml_codepage->convert($delimiter) if ($delimiter);
+
+			my $isis_i = 0;		# isis repeatable offset
+
 			my ($s,$d,$i) = (1,1,0);	# swish, display default
 			$s = 0 if (lc($x->{type}) eq "display");
 			$d = 0 if (lc($x->{type}) eq "swish");
 			($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
 #print STDERR "## s: $s d: $d i: $i ## $format ##\n";	
 
-			$format = $cludge_codepage->convert($format);
-			my ($swish,$display) = parse_format($format,$row);
+			$display_data .= $line_delimiter if ($display_data && $display_data !~ /$line_delimiter$/);
+
+			my ($swish,$display) = (1,1);
+
+			while ($swish || $display) {
+				($swish,$display) = parse_format($format,$row,$isis_i++);
 #print STDERR "s: $swish\nd: $display\n" if ($swish);
 
 #print STDERR "swish: $swish<-- display: $display<--\n";
-			# FIX: this is ugly, UGLY, cludge: OpenIsis return
-			# UTF8 encoding of strings, but as if source charset
-			# is ISO8859-1 and not some other. This breaks our
-			# isis character encoding, so we convert it first
-			# back to ISO8859-1 (which can actually be different
-			# encoding in isis)
-
-			$swish_data .= $swish if ($s && $swish);
-			$display_data .= $display if ($d && $display);
-
-			# insert into index
-			if ($i && $display) {
-				my $index_data = $index_codepage->convert($display) || $display;
-				$index->insert($field, $index_data, $db_dir);
+
+				# filter="name" ; filter this field through
+				# filter/[name].pm
+				my $filter = $x->{filter};
+				if ($filter) {
+					require "filter/".$filter.".pm";
+				}
+				# type="swish" ; field for swish
+				if ($s && $swish) {
+					if ($filter) {
+print STDERR "using filter '$filter'\n";
+						no strict 'refs';
+						$swish_data .= join(" ",&$filter($swish));
+					} else {
+						$swish_data .= $swish;
+					}
+				}
+
+				# type="display" ; field for display
+				if ($d && $display) {
+					if ($filter) {
+						no strict 'refs';
+						$display_data .= join($delimiter,&$filter($display));
+					} else {
+						if ($display_data) {
+							$display_data .= $delimiter.$display;
+						} else {
+							$display_data .= $display;
+						}
+					}
+				}
+						
+				# type="index" ; insert into index
+				if ($i && $display) {
+					my $index_data = $index_codepage->convert($display) || $display;
+					if ($filter) {
+						no strict 'refs';
+						foreach my $d (&$filter($index_data)) {
+							$index->insert($field, $d, $db_dir);
+						}
+					} else {
+						$index->insert($field, $index_data, $db_dir);
+					}
+				}
+			}
+			if ($x->{append}) {
+				$line_delimiter = ' ';
+			} else {
+				$line_delimiter = '<br/>';
 			}
 		}
 
 
 #print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data);
 		if ($display_data) {
+
+			# remove last <br>
+			$display_data =~ s/$line_delimiter$//;
+
 			$display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !";
 			# FIX: this is removed and replaced by html tag.
 			#$xml .= xmlify($field."_display", $display_data);
@@ -117,17 +175,24 @@
 					$field_name = $config->{indexer}->{$field}->{name_singular}."#-#";
 				} elsif ($config->{indexer}->{$field}->{name_plural}) {
 					$field_name = $config->{indexer}->{$field}->{name_plural}."#-#";
-				} else {
+				} elsif ($config->{indexer}->{$field}->{name}) {
 					$field_name = $config->{indexer}->{$field}->{name}."#-#";
+				} else {
+					print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
 				}
 				if ($field_name) {
 					$html .= $xml_codepage->convert($field_name);
+#					$html .= "-->".$field_name."<--";
 				}
 				$html .= $display_data."###\n";
 			}
 		}
 		if ($swish_data) {
 			my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2');
+			# remove extra spaces
+			$swish_data =~ s/ +/ /g;
+			$swish_data =~ s/ +$//g;
+
 			$swish_data = $i->convert($swish_data);
 			$xml .= xmlify($field."_swish",unac_string('ISO8859-2',$swish_data));
 			#$swish_data = $isis_codepage->convert($swish_data)."##" || $swish_data;
@@ -163,9 +228,9 @@
 
 	# read configuration for this type
 	$config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1);
-	$isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8');
+	$isis_codepage = Text::Iconv->new($config->{isis_codepage},$XML_CHARSET);
 	$index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage});
-	$xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),'UTF8');
+	$xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),$XML_CHARSET);
 
 	my $db = OpenIsis::open( $isis_db );
 	if (0) {