--- trunk/all2xml.pl	2002/11/24 20:52:11	1
+++ trunk/all2xml.pl	2003/01/11 19:55:30	9
@@ -5,20 +5,145 @@
 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;
+use DBI;
+
+my $config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1);
+my $dbh = DBI->connect("DBI:Pg:dbname=webpac","","") || die $DBI::errstr; # FIX
+# FIX; select relname from pg_class where relname like 'index_%' ;
+$dbh->begin_work || die $dbh->errstr();
 
-my $config=XMLin();
-
-print Dumper($config);
+$dbh->do("delete from index_author") || die $dbh->errstr();
+$dbh->do("delete from index_title") || die $dbh->errstr();
 
 my %opts;
 
-getopts('d:m:q', \%opts);
+# usage:
+# 	-d directory name
+# 	-m multiple directories
+# 	-q quiet
+# 	-s run swish
+
+getopts('d:m:qs', \%opts);
+
+my $db_dir = $opts{d} || "ps";	# FIX
+
+#die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts);
+
+#print Dumper($config->{indexer});
+#print "-" x 70,"\n";
+
+# how to convert isis code page to UTF8?
+my $isis_map = Unicode::Map8->new($config->{isis_codepage}) || die;
+
+sub isis2xml {
+
+	my $row = shift @_;
+
+	my $xml;
+	$xml->{db_dir} = [ $db_dir ];	# FIX remove?
+
+	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};
+			}
+		}
+	}
+
+	foreach my $field (keys %{$config->{indexer}}) {
 
-my $db_dir = $opts{d};
+		my $display_data = "";
+		my $swish_data = "";
+		my $index_data = "";
+
+		foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
+
+			my $display_tmp = "";
+			my $swish_tmp = "";
+			my $index_tmp = "";
+
+			my $format = $x->{content};
+			my $s = 1;	# swish only
+			my $d = 1;	# display only
+			my $i = 0;	# index only
+			$s = 0 if (lc($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";	
+			# 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);
+						$swish_tmp .= $isis_tmp." " if ($s);
+						$index_tmp .= $prefix . $isis_tmp if ($i);
+#print STDERR " $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);
+			$index_tmp .= $prefix if ($index_tmp);
+
+#			$display_data .= $display_tmp if ($display_tmp ne "");
+#			$swish_data .= $swish_tmp if ($swish_tmp ne "");
+			$display_data .= $display_tmp;
+			$swish_data .= $swish_tmp;
+			$index_data .= $index_tmp;
+
+		}
+#print "--display:$display_data\n--swish:$swish_data\n";
+		#$xml->{$field."_display"} = $isis_map->tou($display_data)->utf8 if ($display_data);
+		#$xml->{$field."_swish"} = unac_string($config->{isis_codepage},$swish_data) if ($swish_data);
+		$xml->{$field."_display" } = [ $isis_map->tou($display_data)->utf8 ] if ($display_data);
+		$xml->{$field."_swish"} = [ unac_string($config->{isis_codepage},$swish_data) ] if ($swish_data);
+
+		# index
+		if ($index_data && $index_data ne "") {
+			my $sql = "select $field from index_$field where upper($field)=upper(?)";
+			my $sth = $dbh->prepare($sql) || die $dbh->errstr();
+			$sth->execute($index_data) || die "SQL: $sql; ".$dbh->errstr();
+#print STDERR "--->$index_data<---\n";
+			if (! $sth->fetchrow_hashref) {
+				my $sql = "insert into index_$field values (?)";
+				my $sth = $dbh->prepare($sql) || die $dbh->errstr();
+#print STDERR "$sql: $index_data<!----\n";
+				$sth->execute($index_data) || die "SQL: $sql; ".$dbh->errstr();
+			}
+		}
 
-die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts);
+	}
+	if ($xml) {
+		return XMLout($xml, rootname => 'xml', keeproot => 0, noattr => 0 );
+	} else {
+		return;
+	}
+}
 
-#--------------------------------------------------------------------
+##########################################################################
 
 my $last_tell=0;
 
@@ -31,36 +156,95 @@
 my @isis_dbs;
 
 foreach (@isis_dirs) {
-	if (-e "$common::isis_data/$db_dir/$_/LIBRI") {
-		push @isis_dbs,"$common::isis_data/$db_dir/$_/LIBRI/LIBRI";
+	if (-e $config->{isis_data}."/$db_dir/$_/LIBRI") {
+		push @isis_dbs,$config->{isis_data}."/$db_dir/$_/LIBRI/LIBRI";
 	}
-	if (-e "$common::isis_data/$db_dir/$_/PERI") {
-		push @isis_dbs,"$common::isis_data/$db_dir/$_/PERI/PERI";
+	if (-e $config->{isis_data}."/$db_dir/$_/PERI") {
+		push @isis_dbs,$config->{isis_data}."/$db_dir/$_/PERI/PERI";
 	}
-	if (-e "$common::isis_data/$db_dir/$_/AMS") {
-		push @isis_dbs,"$common::isis_data/$db_dir/$_/AMS/AMS";
+	if (-e $config->{isis_data}."/$db_dir/$_/AMS") {
+		push @isis_dbs,$config->{isis_data}."/$db_dir/$_/AMS/AMS";
 	}
-	if (-e "$common::isis_data/$db_dir/$_/ARTI") {
-#		push @isis_dbs,"$common::isis_data/$db_dir/$_/ARTI/ARTI";
+	if (-e $config->{isis_data}."/$db_dir/$_/ARTI") {
+#		push @isis_dbs,$config->{isis_data}."/$db_dir/$_/ARTI/ARTI";
 	}
 }
 
+print STDERR "FATAL: Can't find isis database.\nPerhaps isis_data (".$config->{isis_data}.") has wrong value?\n" if (! @isis_dbs);
+
+my $db;
+
 foreach my $isis_db (@isis_dbs) {
 
-	my $db = OpenIsis::open( "$isis_db" ) || warn "can't open '$isis_db'";
+
+	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 );
 
-	my $last_pcnt = 0;
+	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 $pcnt = int($row->{mfn} * 100 / $max_rowid);
-		if ($pcnt != $last_pcnt) {
-			printf STDERR ("%5d / %5d -- %-2d %%\n",$row->{mfn},$max_rowid,$pcnt) if (! $opts{q});
-			$last_pcnt = $pcnt;
+			# 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;
+			}
+
+			if (my $xml = isis2xml($row)) {
+				my $path = $isis_db;
+				$path =~ s#$config->{isis_data}/*##g;
+				my $out = "Path-Name: $path#".$row->{mfn}."\n";
+				$out .= "Content-Length: ".(length($xml)+1)."\n";
+				$out .= "Document-Type: XML\n\n$xml\n";
+				print $out;
+			}
 		}
 	}
+	print STDERR "\n";
 }
+
+$dbh->commit || die $dbh->errstr();
+
+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
+
+=cut