--- trunk/parse_format.pm	2003/11/24 01:16:04	176
+++ trunk/parse_format.pm	2005/02/27 23:07:35	678
@@ -10,13 +10,17 @@
 	my $i = shift || 0;	# isis repeatable number
 	my $codepage = shift || die "parse_format must be called with codepage!";
 	if ($type eq "isis") {
-		return parse_iso_format($format,$row,$i,$codepage,'isis_sf');
+		return parse_iso_format($format,$row,$i,$codepage,'hash_sf');
 	} elsif ($type eq "excel") {
 		return parse_excel_format($format,$row,$i,$codepage);
 	} elsif ($type eq "marc") {
 		return parse_iso_format($format,$row,$i,$codepage,'marc_sf');
 	} elsif ($type eq "feed") {
 		return parse_feed_format($format,$row,$i,$codepage);
+	} elsif ($type eq "dbf") {
+		return parse_iso_format($format,$row,$i,$codepage,'hash_sf');
+	} else {
+		confess "FATAL: unknown type '$type'";
 	}
 }
 
@@ -55,6 +59,9 @@
 
 		my $f = $format;
 
+		my $eval;
+		$eval = $1 if ($f =~ s/^eval{([^}]+?)}//);
+
 		if ($f =~ s/^([^\d]+)//) {
 			if ($f) {	# there is more to parse
 				push @fmt,$1;
@@ -80,7 +87,7 @@
 			# this might be our local scpeciality -- fields 10 and 11
 			# (as opposed to 010 and 011) so they are strictly listed
 			# here
-			} elsif ($f =~ s/^(1[01])//) {
+			} elsif ($f =~ s/^(1[01]\w?)//) {
 				push @fmt,$1;
 				push @fmt,undef;
 			} elsif ($f =~ s/^mfn//i) {
@@ -111,9 +118,12 @@
 			}
 		}
 		push @fmt,'' if ($#fmt % 3 != 0);	# add empty suffix
+
+		$cache->{format_eval}->{$format} = $eval; # store eval string (if any)
+
 		$cache->{format}->{$format} = \@fmt;
 		
-		print STDERR "storing format for '$format': [",join("|",@fmt),"]\n";
+#		print STDERR "storing format for '$format': [",join("|",@fmt),"]\n";
 #		print STDERR "storing format for '$format':",Dumper(@fmt),"\n";
 #		print STDERR Dumper($cache->{format}->{$format});
 	}
@@ -151,11 +161,33 @@
 		$sufix = shift @fmt;
 	}
 	$display = $prefix.$display.$sufix if ($display);
-	print STDERR "format left unused: [",join("|",@fmt),"]\n" if (@fmt);
 
-	print STDERR "format: [",join("|",@{$tmp}),"]\n" if (@fmt);
+	my $eval = $cache->{format_eval}->{$format};
+	if ($eval) {
+		sub fld2str {
+			my ($func,$row,$f,$sf,$i) = @_;
+#print STDERR "## in fld2str\n";
+			my $tmp = $codepage->convert(&$func($row,$f,$sf,$i)) ||  $codepage->convert(&$func($row,$f,$sf,0)) || '';
+			return "'$tmp'";
+		}
+
+		$eval =~ s/v(\d+)\^(\w*)/fld2str($func,$row,$1,$2,$i)/eg;
+#print STDERR "## eval: $eval\n";
+		if (eval "$eval") {
+			die "eval error: eval{$eval}: $@" if ($@);
+			return ($swish,$display);
+		} else {
+			die "eval error: eval{$eval}: $@" if ($@);
+			return (undef,undef);
+		}
+	}
+
+	if (@fmt) {
+		print STDERR "format left unused: [",join("|",@fmt),"]\n";
+		print STDERR "format: [",join("|",@{$tmp}),"]\n";
+	}
 
-#	print STDERR "display: $display swish: $swish\n";
+#	print STDERR "format: {",$format || '',"} display: {",$display || '',"} swish: {",$swish || '',"}\n";
 
 	return ($swish,$display);
 }
@@ -166,7 +198,14 @@
 	my $format = shift;
 	my $row = shift;
 	my $i = shift;
-	my $codepage = shift;
+	#my $codepage = shift;
+	#
+	# data allready comes in utf-8 due to change in
+	# SpreadSheet::ParseExcel::FmtDefault line 69 from
+	#	return pack('C*', unpack('n*', $sTxt));
+	# to following which returns utf-8:
+	#	return pack('U*', unpack('n*', $sTxt));
+	#
 
 	return if ($i > 0);	# Excel doesn't support repeatable fields
 
@@ -187,9 +226,6 @@
 #print STDERR "--$1-> $format -[",length($format),"] ";
 			if ($row->{$1}) {
 				my $tmp = $row->{$1};
-				if ($codepage) {
-					$tmp = $codepage->convert($tmp) || warn "excel: $1 '$tmp' can't convert";
-				}
 				$display .= $prefix . $tmp;
 				$swish .= $tmp." ";
 #print STDERR " == $tmp";
@@ -198,7 +234,7 @@
 		} elsif ($format =~ s/^([^A-Z\|]+)(\|[A-Z]{1,2}\|)/$2/) {
 			$prefix .= $1 if ($display);
 		} else {
-			print STDERR "unparsed format: $format\n";
+			#print STDERR "unparsed format: $format\n";
 			$prefix .= $format;
 			$format = "";
 		}