--- trunk/all2xml.pl	2003/12/21 03:27:02	197
+++ trunk/all2xml.pl	2004/03/14 11:50:29	274
@@ -150,6 +150,25 @@
 		}
 	}
 
+	sub init_visible_type($) {
+		my $type = shift;
+
+		# swish, swish_exact, display, index, index_lookup
+		# swish and display defaults
+		my ($s,$se,$d,$i,$il) = (1,0,1,0,0);
+		if (lc($type) eq "display") {
+			$s = 0;
+		} elsif (lc($type) eq "swish") {
+			$d = 0;
+		} elsif (lc($type) eq "index") {
+			($s,$se,$d,$i) = (0,1,0,1);
+		} elsif (lc($type) eq "swish_exact") {
+			($s,$se,$d,$i) = (0,1,0,0);
+		} elsif (lc($type) =~ /^lookup/) {
+			($s,$se,$d,$i,$il) = (0,1,0,0,1);
+		}
+		return ($s,$se,$d,$i,$il);
+	}
 
 	# begin real work: go field by field
 	foreach my $field (@sorted_tags) {
@@ -189,14 +208,7 @@
 
 			my $repeat_off = 0;	# init repeatable offset
 
-			# swish, swish_exact, display, index, index_lookup
-			# swish and display defaults
-			my ($s,$se,$d,$i,$il) = (1,0,1,0,0);
-			$s = 0 if (lc($x->{type}) eq "display");
-			$d = 0 if (lc($x->{type}) eq "swish");
-			($s,$se,$d,$i) = (0,1,0,1) if (lc($x->{type}) eq "index");
-			($s,$se,$d,$i) = (0,1,0,0) if (lc($x->{type}) eq "swish_exact");
-			($s,$se,$d,$i,$il) = (0,1,0,0,1) if (lc($x->{type}) =~ /^lookup/);
+			my ($s,$se,$d,$i,$il) = init_visible_type($x->{type});
 
 			# what will separate last line from this one?
 			if ($display_data && $x->{append}) {
@@ -228,7 +240,7 @@
 					if (($#data+1) == $nr) {
 						return sprintf($fmt,@data);
 					} else {
-						print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n";
+						#print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n";
 						return $data;
 					}
 				} else {
@@ -257,7 +269,7 @@
 								$display = $new_display;
 								$cache->{lhash}->{$display} = $new_display;
 							} else {
-								print STDERR "WARNING: lookup for '$display' didn't find anything.\n";
+#								print STDERR "WARNING: lookup for '$display' didn't find anything.\n";
 								$display = "";
 								$cache->{lhash}->{$display} = $null;
 							}
@@ -278,36 +290,35 @@
 				}
 				# type="swish" ; field for swish
 				if ($swish) {
+					my $tmp = $swish;
 					if ($filter && ($s || $se)) {
 						no strict 'refs';
-						my $tmp = join(" ",&$filter($swish)) if ($s || $se);
-						$swish_data .= $tmp if ($s);
-						$swish_exact_data .= "xxbxx $tmp xxexx " if ($se && $tmp ne "");
-
-					} else {
-						$swish_data .= $swish if ($s);
-						$swish_exact_data .= "xxbxx $swish xxexx " if ($se && $swish ne "");
+						$tmp = join(" ",&$filter($tmp)) if ($s || $se);
 					}
+
+					$swish_data .= $tmp if ($s && $tmp);
+					$swish_exact_data .= "xxbxx $tmp xxexx " if ($tmp && $tmp ne "" && $se);
 				}
 
 				# 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);
 					}
 				}
 						
@@ -319,7 +330,7 @@
 						no strict 'refs';
 						$idisplay = &$filter($idisplay);
 					}
-					push @index_data, $idisplay if (! $iterate_by_page);
+					push @index_data, $idisplay if ($idisplay && !$iterate_by_page);
 				}
 
 				# store fields in lookup
@@ -328,11 +339,21 @@
 						if ($lookup_key) {
 							print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)";
 						} else {
-							$lookup_key = $display;
+							if ($filter) {
+								no strict 'refs';
+								$lookup_key = &$filter($display);
+							} else {
+								$lookup_key = $display;
+							}
 						}
 					} elsif (lc($x->{type}) eq "lookup_val") {
 						if ($lookup_key) {
-							$lhash{$lookup_key} = $display;
+							if ($filter) {
+								no strict 'refs';
+								$lhash{$lookup_key} = &$filter($display);
+							} else {
+								$lhash{$lookup_key} = $display;
+							}
 						} else {
 							print STDERR "WARNING: no lookup_key defined for  '$display'?";
 						}
@@ -411,16 +432,12 @@
 			my $delimiter = x($x->{delimiter}) || ' ';
 			my $val = $cfg->val($database, x($x->{content}));
 
-			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");
-			# no support for swish exact in config.
-			# IMHO, it's useless
-			($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
+			# FIXME index_lookup is not supported!
+			my ($s,$se,$d,$i,$il) = init_visible_type($x->{type});
 
 			if ($val) {
 				$display_data .= $delimiter.$val if ($d);
-				$swish_data .= $val if ($s);
+				$swish_data .= " ".$val if ($s);
 				$index->insert($field, $val, $path) if ($i);
 			}
 
@@ -451,7 +468,8 @@
 			for (my $page=0; $page <= $nr_pages; $page++) {
 				my $display_data;
 				if ($cache->{format}->{$field}) {
-					$display_data=mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]);
+					my $tmp = mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]);
+					$display_data=$tmp if ($tmp);
 				} else {
 					$display_data = $cache->{display_data}->{$field}->[$page];
 				}
@@ -577,6 +595,8 @@
 		#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644;
 		tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644;
 		print STDERR "creating lookup file '$lookup_file'\n";
+		# delete memory cache for lookup file
+		delete $cache->{lhash};
 	}
 
 	# open existing lookup file
@@ -713,8 +733,9 @@
 		print STDERR "\n";
 
 	} elsif ($type_base eq "excel") {
-		use Spreadsheet::ParseExcel;
-		use Spreadsheet::ParseExcel::Utility qw(int2col);
+		require Spreadsheet::ParseExcel;
+		require Spreadsheet::ParseExcel::Utility;
+		import Spreadsheet::ParseExcel::Utility qw(int2col);
 		
 		$import2cp = Text::Iconv->new($config->{excel_codepage},$codepage);
 		my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!";
@@ -768,29 +789,32 @@
 		}
 	} elsif ($type_base eq "marc") {
 
-		use MARC;
+		require MARC;
 		
 		$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage);
 		my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!";
 
 		# optional argument is format
-		my $format = x($config->{format}) || 'usmarc';
+		my $format = x($config->{marc_format}) || 'usmarc';
 
 		print STDERR "Reading MARC file '$marc_file'\n";
 
 		my $marc = new MARC;
 		my $nr = $marc->openmarc({
 				file=>$marc_file, format=>$format
-			}) || die "Can't open MARC file '$marc_file'";
+			}) || die "Can't open MARC file '$marc_file' with format '$format'";
 
-		my $i=0;	# record nr.
+		# read MARC file in memory
+		$marc->nextmarc(-1);
 
-		my $rec;
+		my $max_rec = $marc->marc_count();
 
-		while ($marc->nextmarc(1)) {
+		for(my $i=1; $i<=$max_rec; $i++) {
 
-			# XXX
-			fakeprogress($i++);
+			progress($i,$max_rec);
+
+			# store value for marc_sf.pm
+			$main::cache->{marc_record} = $i;
 
 			my $swishpath = $database."#".$i;
 
@@ -802,6 +826,9 @@
 				print "Document-Type: XML\n\n$xml\n";
 			}
 		}
+
+		print STDERR "\n";
+
 	} elsif ($type_base eq "feed") {
 
 		$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage);