--- trunk/all2xml.pl	2003/01/11 06:14:48	5
+++ trunk2/all2all.pl	2004/06/13 20:39:44	349
@@ -1,209 +1,126 @@
 #!/usr/bin/perl -w
 
-use strict;
-use OpenIsis;
-use Getopt::Std;
-use Data::Dumper;
-use XML::Simple;
-use Text::Unaccent 1.02;	# 1.01 won't compile on my platform,
-require Unicode::Map8;
-
-my $config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1);
-
-my %opts;
-
-getopts('d:m:q', \%opts);
+=head1 NAME
 
-my $db_dir = $opts{d} || "ps";	# FIX
+webpac.pl - basic script for all WebPac needs
 
-#die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts);
+=cut
 
-#print Dumper($config->{indexer});
-#print "-" x 70,"\n";
+use strict;
+use OpenIsis;
+use Text::Iconv;
+use Data::Dumper;
 
-# how to convert isis code page to UTF8?
-my $isis_map = Unicode::Map8->new($config->{isis_codepage}) || die;
+my $cp = Text::Iconv->new("852", "iso-8859-2");
 
-sub isis2xml {
+$|=1;
 
-	my $row = shift @_;
+my $db = OpenIsis::open( shift @ARGV || '/data/hidra/THS-500/THS' );
+my $maxmfn = OpenIsis::maxRowid( $db ) || 1;
 
-	my $xml;
+print "rows: $maxmfn\n\n";
 
-	sub isis_sf {
-		my $row = shift @_;
-		my $isis_id = shift @_;
-		my $subfield = shift @_;
-		if ($row->{$isis_id}->[0]) {
-			my $sf = OpenIsis::subfields($row->{$isis_id}->[0]);
-			if (! defined $subfield || length($subfield) == 0) {
-				# subfield list undef, empty or no defined subfields for this record
-				my $all_sf = $row->{$isis_id}->[0];
-				$all_sf =~ s/\^./ /g;	nuke definirions
-				return $all_sf;
-			} elsif ($sf->{$subfield}) {
-				return $sf->{$subfield};
-			}
+my $data;
+
+my @lookup = (
+	{ 'key' => 'd:v900', 'val' => 'v250^a' },
+	{ 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' },
+	{ 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'ma:v561^4:v562^4:v461^1', 'val' => 'v900' },
+	{ 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'da:v561^4:v562^4:v461^1', 'val' => 'v900' },
+);
+
+sub fill_in {
+	my $tmp = shift || return;
+	my $data = shift || return;
+	my $i = shift || 0;
+
+	my $found = 0;
+
+	sub get_sf {
+		my ($found,$data,$f,$sf,$i) = @_;
+		if ($$data->{$f} && $$data->{$f}->[$i]->{$sf}) {
+			$$found++;
+			return $$data->{$f}->[$i]->{$sf};
+		} else {
+			return '';
 		}
 	}
 
-	foreach my $field (keys %{$config->{indexer}}) {
-
-		my $display_data = "";
-		my $index_data = "";
-
-		foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
-
-			my $display_tmp = "";
-			my $index_tmp = "";
-
-			my $format = $x->{content};
-			my $i = 1;	# index only
-			my $d = 1;	# display only
-			$i = 0 if (lc($x->{type}) eq "display");
-			$d = 0 if (lc($x->{type}) eq "index");
-#print "## i: $i d: $d ## $format ##";	
-			# parse format
-			my $prefix = "";
-			if ($format =~ s/^([^\d]+)//) {
-				$prefix = $1;
-			}
-			while ($format) {
-				if ($format =~ s/^(\d\d\d)(\w?)//) {
-					my $isis_tmp = isis_sf($row,$1,$2);
-					if ($isis_tmp) {
-#						$display_tmp .= $prefix . "/$1/$2/".$isis_tmp if ($d);
-						$display_tmp .= $prefix . $isis_tmp if ($d);
-						$index_tmp .= $isis_tmp." " if ($i);
-#print " $isis_tmp <--\n";
-					}
-					$prefix = "";
-				} elsif ($format =~ s/^([^\d]+)//) {
-					$prefix = $1;
-				} else {
-					print STDERR "WARNING: unparsed format '$format'\n";
-					last;
-				};
-			}
-			# add suffix
-			$display_tmp .= $prefix if ($display_tmp);
-
-#			$display_data .= $display_tmp if ($display_tmp ne "");
-#			$index_data .= $index_tmp if ($index_tmp ne "");
-			$display_data .= $display_tmp;
-			$index_data .= $index_tmp;
-
+	sub get_nosf {
+		my ($found,$data,$f,$i) = @_;
+		if ($$data->{$f} && $$data->{$f}->[$i]) {
+			$$found++;
+			return $$data->{$f}->[$i];
+		} else {
+			return '';
 		}
-#print "--display:$display_data\n--index:$index_data\n";
-		$xml->{$field}->{display} .= $isis_map->tou($display_data)->utf8 if ($display_data);
-		$xml->{$field}->{index} .= unac_string($config->{isis_codepage},$index_data) if ($index_data);
-	
-	}
-	if ($xml) {
-		return XMLout($xml, rootname => 'xml', noattr => 1 );
-	} else {
-		return;
 	}
-}
-
-##########################################################################
 
-my $last_tell=0;
-
-my @isis_dirs = ( '.' );	# use dirname as database name
-
-if ($opts{m}) {
-	@isis_dirs = split(/,/,$opts{m});
-}
+	$tmp =~ s/v(\d+)\^(\w)/get_sf(\$found,\$data,$1,$2,$i)/ges;
+	$tmp =~ s/v(\d+)/get_nosf(\$found,\$data,$1,$i)/ges;
 
-my @isis_dbs;
-
-foreach (@isis_dirs) {
-	if (-e $config->{isis_data}."/$db_dir/$_/LIBRI") {
-		push @isis_dbs,$config->{isis_data}."/$db_dir/$_/LIBRI/LIBRI";
-	}
-	if (-e $config->{isis_data}."/$db_dir/$_/PERI") {
-		push @isis_dbs,$config->{isis_data}."/$db_dir/$_/PERI/PERI";
-	}
-	if (-e $config->{isis_data}."/$db_dir/$_/AMS") {
-		push @isis_dbs,$config->{isis_data}."/$db_dir/$_/AMS/AMS";
-	}
-	if (-e $config->{isis_data}."/$db_dir/$_/ARTI") {
-#		push @isis_dbs,$config->{isis_data}."/$db_dir/$_/ARTI/ARTI";
+	if ($found) {
+		return $tmp;
+	} else {
+		return;
 	}
 }
 
-print STDERR "FATAL: Can't find isis database.\nPerhaps isis_data (".$config->{isis_data}.") has wrong value?\n" if (! @isis_dbs);
+print "reading database...\n";
 
-my $db;
+for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) {
 
-foreach my $isis_db (@isis_dbs) {
+	my $row = OpenIsis::read( $db, $mfn );
+	foreach my $k (keys %{$row}) {
+		if ($k ne "mfn") {
+			foreach my $l (@{$row->{$k}}) {
+				$l = $cp->convert($l);
+				# has subfields?
+				my $val;
+				if ($l =~ m/\^/) {
+					foreach my $t (split(/\^/,$l)) {
+						next if (! $t);
+						$val->{substr($t,0,1)} = substr($t,1);
+					}
+				} else {
+					$val = $l;
+				}
 
+				push @{$data->{$mfn}->{$k}}, $val;
+			}
+		}
 
-	my $db = OpenIsis::open( $isis_db );
-	if (0) {
-#	# FIX
-#	if (! $db ) {
-		print STDERR "WARNING: can't open '$isis_db'\n";
-		next ;
 	}
 
-	my $max_rowid = OpenIsis::maxRowid( $db );
-
-	print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
 
-	my $last_p = 0;
-
-#	{ my $row_id = 1;
-# FIX
-	for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
-		my $row = OpenIsis::read( $db, $row_id );
-		if ($row && $row->{mfn}) {
-
-			# output current process indicator
-			my $p = int($row->{mfn} * 100 / $max_rowid);
-			if ($p != $last_p) {
-				printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q});
-				$last_p = $p;
+	foreach my $i (@lookup) {
+		my $d = $data->{$mfn};
+		if ($i->{'eval'}) {
+			my $eval = fill_in($i->{'eval'},$d);
+			my $key = fill_in($i->{'key'}, $d);
+			my $val = fill_in($i->{'val'}, $d);
+			if ($key && $val && eval $eval) {
+				push @{$data->{lookup}->{$key}}, $val;
 			}
-
-			if (my $xml = isis2xml($row)) {
-				print "Path-Name: $isis_db#".$row->{mfn}."\n";
-				print "Content-Length: ".(length($xml)+1)."\n";
-				print "Document-Type: XML\n\n$xml\n";
+		} else {
+			my $key = fill_in($i->{'key'}, $d);
+			my $val = fill_in($i->{'val'}, $d);
+			if ($key && $val) {
+				push @{$data->{lookup}->{$key}}, $val;
 			}
 		}
 	}
-	print STDERR "\n";
-}
 
+}
+print "\n";
 
-1;
-__END__
-##########################################################################
-
-=head1 NAME
-
-isis2xml.pl - read isis file and dump XML
-
-=head1 DESCRIPTION
-
-This command will read ISIS data file using OpenIsis perl module and
-create XML file for usage with I<SWISH-E>
-indexer. Dispite it's name, this script B<isn't general xml generator>
-from isis files (isis allready has something like that). Output of this
-script is tailor-made for SWISH-E.
-
-=head1 AUTHOR
-
-Dobrica Pavlinusic <dpavlin@rot13.org>
-
-=head1 COPYRIGHT
-
-GNU Public License (GPL) v2 or later
-
-=head1 SEE ALSO
-
-SWISH-E web site at http://www.swish-e.org
+for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) {
+	my $d = $data->{$mfn};
+	print $d->{'250'}->[0]->{'a'},"\n";
+	my $t = fill_in('v553^1;;v553^a', $d);
+	print " Uži pojam: $t\n" if ($t);
+	$t = fill_in('[a:v251];;[d:[a:v251]]', $d);
+	print " Uži pojam: $t\n" if ($t);
+}
 
-=cut
+print Dumper($data);