--- branches/tehnika/all2xml.pl	2005/02/17 18:18:49	674
+++ branches/tehnika/all2xml.pl	2005/02/28 10:43:38	684
@@ -11,6 +11,7 @@
 #use GDBM_File;
 use Fcntl;	# for O_RDWR
 use TDB_File;
+use Carp;
 
 $|=1;
 
@@ -62,7 +63,7 @@
 	'isis' => 'isis',
 	'excel' => 'column',
 	'marc' => 'marc',
-	'feed' => 'feed'
+	'feed' => 'feed',
 );
 
 my $cache;	# for cacheing
@@ -112,6 +113,10 @@
 		$cache->{tags_by_order} = \@sorted_tags;
 	}
 
+	if (! @sorted_tags) {
+		print STDERR "WARNING: no tags for this type found in import_xml file!\n";
+	}
+
 	# lookup key
 	my $lookup_key;
 
@@ -220,7 +225,7 @@
 
 		my ($swish,$display);
 
-		my $tag = $type2tag{$type} || die "can't find which tag to use for type $type";
+		my $tag = $cfg->val($database, 'import_xml_tag') || $type2tag{$type} || die "can't find which tag to use for type $type";
 
 		# is this field page-by-page?
 		my $iterate_by_page = $config->{indexer}->{$field}->{iterate_by_page};
@@ -662,13 +667,22 @@
 		print STDERR "opening lookup file '$lookup_file'\n";
 	}
 
-print STDERR "reading ./import_xml/$type.xml\n";
+	my $import_xml_type = $cfg->val($database, 'import_xml_file') || $type;
+	my $import_xml_file = "./import_xml/$import_xml_type.xml";
+
+	if (! -r $import_xml_file) {
+		print STDERR "ERROR: file $import_xml_file not readable skipping!\n";
+		next;
+	}
+
+	print STDERR "reading $import_xml_file\n";
 
 	# extract just type basic
 	my $type_base = $type;
 	$type_base =~ s/_.+$//g;
 
-	$config=XMLin("./import_xml/$type.xml", ForceArray => [ $type2tag{$type_base}, 'config', 'format' ], ForceContent => 1 );
+	my $tag = $cfg->val($database, 'import_xml_tag') || $type2tag{$type_base} || die "can't find which tag to use for type $type";
+	$config=XMLin($import_xml_file, ForceArray => [ $tag, 'config', 'format' ], ForceContent => 1 );
 
 	# helper for progress bar
 	sub fmt_time {
@@ -730,7 +744,6 @@
 	}
 
 	# now read database
-print STDERR "using: $type...\n";
 
 	# erase cache for tags by order in this database
 	delete $cache->{tags_by_order};
@@ -927,6 +940,72 @@
 		}
 		# close lookup
 		untie %lhash if (%lhash);
+
+	} elsif ($type_base eq "dbf") {
+
+		my $dbf_file = $cfg -> val($database, 'dbf_file') || die "$database doesn't have 'dbf_file' defined!";
+		my $dbf_codepage = $cfg -> val($database, 'dbf_codepage') || die "$database doesn't have 'dbf_codepage' defined!";
+		my $dbf_mapping = $cfg -> val($database, 'dbf_mapping') || die "$database doesn't have 'dbf_mapping' defined!";
+
+		$import2cp = Text::Iconv->new($dbf_codepage,$codepage);
+		require XBase;
+		my $db = new XBase $dbf_file;
+
+		if (! $db) {
+			print STDERR "ERROR: can't read DBF database: $dbf_file, skipping...\n";
+			next;
+		}
+
+		my $max_rowid = $db->last_record;
+
+		print STDERR "Reading database: $dbf_file [$max_rowid rows]\n";
+
+		my %dbf2iso;
+		foreach my $m (split(/[\n\r]+/,$dbf_mapping)) {
+			my ($col,$fld) = split(/\s+/,$m,2);
+			$dbf2iso{$col} = $fld;
+		}
+
+#print STDERR "## dbf2iso: ",Dumper(\%dbf2iso),"\n## /dbf2iso\n";
+
+		# bad, bad...
+		require "to_hash.pm";
+
+		foreach my $row_id (0 .. $max_rowid) {
+			my $dbf_row = $db->get_record_as_hash($row_id);
+			if ($dbf_row) {
+
+#print STDERR "## dbf_row: ",Dumper($dbf_row),"\n## /dbf_row\n";
+				# apply mapping from config file
+				# all unspecified records will get _ in
+				# front of them - _DELETE will be __DELETE
+				my $rec;
+				map { 
+					my $new_fld = $dbf2iso{$_} || '_'.$_;
+					my $data = $dbf_row->{$_};
+					push @{ $rec->{$new_fld} }, $data if ($data && $data !~ /^(?:\s+|\$a\.|)$/);
+				} keys %{$dbf_row};
+#print STDERR "## rec: ",Dumper($rec),"\n## /rec\n";
+				my $row = to_hash($row_id+1, $rec);
+
+				$row->{mfn} = $row_id+1;
+				$row->{record} = $rec;
+
+#print STDERR "## row: ",Dumper($row),"\n## /row\n";
+				progress($row->{mfn}, $max_rowid);
+
+				my $swishpath = $path."#".int($row->{mfn});
+
+				if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
+					$xml = $cp2utf->convert($xml);
+					use bytes;	# as opposed to chars
+					print "Path-Name: $swishpath\n";
+					print "Content-Length: ".(length($xml)+1)."\n";
+					print "Document-Type: XML\n\n$xml\n";
+				}
+			}
+		}
+		print STDERR "\n";
 	}
 }