--- trunk/tools/phpmylib2marc.pl	2004/02/01 15:57:17	208
+++ trunk/tools/phpmylib2marc.pl	2004/02/01 22:18:05	216
@@ -2,7 +2,7 @@
 #
 # This script will try (hard) to convert database from
 # PhpMyLibrary (http://phpmylibrary.sourceforge.net/) back
-# to MARC format
+# to MARC format (ISO 2709)
 #
 # 2003-01-31 Dobrica Pavlinusic <dpavlin@rot13.org>
 #
@@ -24,12 +24,14 @@
 my $passwd = "";
 
 my $usage = 0;
+my $debug = 0;
 
 my $result = GetOptions(
 	"database=s" => \$database,
 	"host=s" => \$host,
 	"user=s" => \$user,
 	"password=s" => \$passwd,
+	"debug!" => \$debug,
 	"help!" => \$usage,
 	);
 
@@ -43,16 +45,25 @@
 my $dbh = DBI->connect($dsn, $user, $passwd, {'RaiseError' => 1});
 
 # UNIMARC leader format
-my $leader_fmt = qq{%05diam0 22%05d   45  };
+#my $leader_fmt = qq{%05diam0 22%05d   45  };
+# MARC leader format
+my $leader_fmt = qq{%05dcas  22%05d a 4500};
 
 
 my $sth = $dbh->prepare("SELECT marc FROM tblbib");
 $sth->execute();
+
+my $count = 0;
+my $rec_nr = 0;
+
 while (my $row = $sth->fetchrow_hashref()) {
 	my $marc = $row->{'marc'};
+	$rec_nr++;
 
 	my $real_len = length($marc);
 
+	my $skip = 0;	# skip this record?
+
 	# fix PhpMyLibrary MARC (why do I have to do this? It's MARC,
 	# for gaddem sake!!!
 	
@@ -83,9 +94,77 @@
 	#           (xxxxx)nam  22(.....)   450 <---
 	$marc =~ m/^(.....)......(.....)polerioj/ || die "record: '$marc' unparsable!";
 	my ($reclen,$base_addr) = ($1,$2);
-	my $leader = sprintf($leader_fmt,length($marc)+1,$base_addr);
 
-	print $leader . substr($marc,24).chr(29);
+	my $directory = substr($marc,24,$base_addr-24);
+	my $fields = substr($marc,$base_addr-1);
+
+	print STDERR "# $rec_nr fields: '$fields'\n" if ($debug);
+	print STDERR "# $rec_nr directory: [",length($directory),"]\n" if ($debug);
+
+	# PhpMyLibrary MARC records don't have indicators, so we'll add them
+
+	my $o = 0;	# offset
+	my $new_dictionary;
+	my $new_fields;
+
+	while (!$skip && $directory =~ s/(\d{3})(\d{4})(\d{5})//) {
+		my ($tag,$len,$addr) = ($1,$2,$3);
+
+		print STDERR "tag/len/addr: $tag $len $addr\n" if ($debug);
+
+		sub check_field($) {
+			my $f = shift;
+			my $del = substr($f,0,1);
+		
+			die "expected 0x1e, got '$del' (".ord($del)."): '$f'" if (ord($del) != 30);
+		}
+
+		if (($addr+$len) > length($fields)) {
+			print STDERR "WARNING: error in dictionary on record $rec_nr skipping...\n" if ($debug);
+			$skip = 1;
+			next;
+		}
+
+		# take field
+		my $f = substr($fields,$addr,$len);
+		print STDERR "data $tag [$len] $addr: '$f'\n" if ($debug);
+
+		my $del = substr($fields,$addr+$len,1);
+
+		# check field delimiters...
+		if ($del ne chr(30)) {
+			print STDERR "WARNING: skipping record $rec_nr, can't find delimiters got: '$del'\n" if ($debug);
+			$skip = 1;
+			next;
+		}
+
+		check_field($f);
+			
+		if ($tag =~ m/^00/) {
+			# fields 001-008 doesn't have indicators
+			$new_dictionary .= sprintf("%03d%04d%05d",$tag,$len,$addr);
+			$new_fields.=$f;
+		} else {
+			$new_dictionary .= sprintf("%03d%04d%05d",$tag,($len+2),($addr+$o));
+			$new_fields.=chr(30)."  ".substr($f,1);
+			$o += 2;
+		}
+	}
+
+	if (! $skip) {
+		my $new_leader = sprintf($leader_fmt,24+length($new_dictionary.$new_fields)+2,$base_addr);
+		my $new_marc = $new_leader . $new_dictionary . $new_fields . chr(30);
+		$new_marc .= chr(29);	# end of record
+
+		print STDERR "original and new marc: [$rec_nr]\n$marc\n$new_marc\n\n" if ($debug);
+		print "$new_marc";
+		$count++;
+	}
+
+#	last if ($count > 100);
+
 }
 $sth->finish();
 $dbh->disconnect();
+
+print STDERR "$count records from database $database converted...\n";