--- trunk/all2xml.pl	2003/11/29 15:59:19	182
+++ trunk/all2xml.pl	2004/03/09 12:17:05	255
@@ -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;
@@ -117,6 +119,7 @@
 	delete $cache->{swish_data};
 	delete $cache->{swish_exact_data};
 	delete $cache->{index_data};
+	delete $cache->{index_delimiter};
 	my @page_fields;	# names of fields
 
 
@@ -137,7 +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;
+			}
 		}
 	}
 
@@ -164,6 +173,14 @@
 		my %page_max = ();
 		# default line_delimiter if using
 		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}}) {
 
@@ -177,7 +194,7 @@
 			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) = (0,0,0,1) if (lc($x->{type}) eq "index");
+			($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/);
 
@@ -193,10 +210,24 @@
 
 			# placeholder for all repeatable entries for index
 
-			sub mkformat {
+			sub chk_eval($) {
+				my $data = shift;
+				return if (! defined($data));
+				if ($data && $data =~ s/\s*eval{([^}]+)}\s*//) {
+					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;
@@ -209,10 +240,10 @@
 				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";
@@ -240,7 +271,7 @@
 								$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;
 							}
@@ -265,32 +296,37 @@
 						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 "");
 					}
+
+					$tmp = chk_eval($tmp);
+					$swish_data .= $tmp if ($s && $tmp);
+					$swish_exact_data .= "xxbxx $tmp xxexx " if ($se && $tmp 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);
 					}
 				}
 						
@@ -302,7 +338,8 @@
 						no strict 'refs';
 						$idisplay = &$filter($idisplay);
 					}
-					push @index_data, $idisplay if (! $iterate_by_page);
+					$idisplay = chk_eval($idisplay);
+					push @index_data, $idisplay if ($idisplay && !$iterate_by_page);
 				}
 
 				# store fields in lookup
@@ -311,11 +348,21 @@
 						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'?";
 						}
@@ -360,7 +407,11 @@
 							@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;
 						}
@@ -371,9 +422,14 @@
 			}
 
 			if (! $iterate_by_page) {
+				my $idel = $x->{index_delimiter};
 				# fill data in index
-				foreach my $d (@index_data) {
-					$index->insert($field, $d, $path);
+				foreach my $tmp (@index_data) {
+					my $i = $d = $tmp;
+					if ($idel && $tmp =~ m/$idel/) {
+						($i,$d) = split(/$idel/,$tmp);
+					}
+					$index->insert($field, $i, $d, $path);
 				}
 				@index_data = ();
 			}
@@ -423,8 +479,13 @@
 #print STDERR "field '$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);
@@ -452,12 +513,18 @@
 
 					# 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));
 				}
 				
-				foreach my $d (@{$cache->{index_data}->{$field}->[$page]}) {
-					$index->insert($field, $d, $path);
-#print STDERR "index $field: $d [$path]\n";
+				my $idel = $cache->{index_delimiter}->{$field};
+				foreach my $tmp (@{$cache->{index_data}->{$field}->[$page]}) {
+					my $i = $tmp;
+					my $d = $tmp;
+					if ($idel && $tmp =~ m/$idel/) {
+						($i,$d) = split(/$idel/,$tmp);
+					}
+					$index->insert($field, $i, $d, $path);
+#print STDERR "index [$idel] $field: $i --> $d [$path]\n";
 				}
 			}
 	
@@ -488,7 +555,7 @@
 
 				# 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));
 			}
 		}
 	}
@@ -541,6 +608,8 @@
 		#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
@@ -677,8 +746,9 @@
 		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 "$database doesn't have 'excel_file' defined!";
@@ -732,29 +802,32 @@
 		}
 	} elsif ($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'";
+
+		# read MARC file in memory
+		$marc->nextmarc(-1);
 
-		my $i=0;	# record nr.
+		my $max_rec = $marc->marc_count();
 
-		my $rec;
+		for(my $i=1; $i<=$max_rec; $i++) {
 
-		while ($marc->nextmarc(1)) {
+			progress($i,$max_rec);
 
-			# XXX
-			fakeprogress($i++);
+			# store value for marc_sf.pm
+			$main::cache->{marc_record} = $i;
 
 			my $swishpath = $database."#".$i;
 
@@ -766,6 +839,9 @@
 				print "Document-Type: XML\n\n$xml\n";
 			}
 		}
+
+		print STDERR "\n";
+
 	} elsif ($type_base eq "feed") {
 
 		$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage);