--- trunk/lib/WebPAC/Validate.pm	2006/09/06 22:07:21	645
+++ trunk/lib/WebPAC/Validate.pm	2006/09/11 11:57:18	664
@@ -8,7 +8,7 @@
 use base 'WebPAC::Common';
 use File::Slurp;
 use List::Util qw/first/;
-use Data::Dumper;
+use Data::Dump qw/dump/;
 use WebPAC::Normalize qw/_pack_subfields_hash/;
 use Storable qw/dclone/;
 
@@ -18,11 +18,11 @@
 
 =head1 VERSION
 
-Version 0.03
+Version 0.07
 
 =cut
 
-our $VERSION = '0.03';
+our $VERSION = '0.07';
 
 =head1 SYNOPSIS
 
@@ -91,14 +91,20 @@
 		$log->logdie("need field name in line $curr_line: $l") unless (defined($fld));
 
 		if (@d) {
-			$v->{$fld} = \@d;
+			$v->{$fld} = [ map {
+				my $sf = $_;
+				if ( $sf =~ s/!(\*)?$/$1/ ) {
+					$self->{must_exist_sf}->{ $fld }->{ $sf }++;
+				};
+				$sf;
+			} @d ];
 		} else {
 			$v->{$fld} = 1;
 		}
 
 	}
 
-	$log->debug("current validation rules: ", Dumper($v));
+	$log->debug("current validation rules: ", dump($v));
 
 	$self->{rules} = $v;
 
@@ -111,7 +117,7 @@
 
 Validate record and return errors
 
-  my @errors = $validate->validate_errors( $rec );
+  my @errors = $validate->validate_errors( $rec, $rec_dump );
 
 =cut
 
@@ -121,13 +127,14 @@
 	my $log = $self->_get_logger();
 
 	my $rec = shift || $log->logdie("validate_errors need record");
+	my $rec_dump = shift;
 
 	$log->logdie("rec isn't HASH") unless (ref($rec) eq 'HASH');
 	$log->logdie("can't find validation rules") unless (my $r = $self->{rules});
 
-	my @errors;
+	my $errors;
 
-	$log->debug("rec = ", sub { Dumper($rec) }, "keys = ", keys %{ $rec });
+	$log->debug("rec = ", sub { dump($rec) }, "keys = ", keys %{ $rec });
 
 	my $fields;
 
@@ -138,13 +145,13 @@
 		$fields->{$f}++;
 
 		if ( ! defined($r->{$f}) ) {
-			push @errors, "field '$f' shouldn't exists";
+			$errors->{field}->{ $f }->{unexpected} = "this field is not expected";
 			next;
 		}
 
 
 		if (ref($rec->{$f}) ne 'ARRAY') {
-			push @errors, "field '$f' isn't repetable, probably bug in parsing input data";
+			$errors->{field}->{ $f }->{not_repeatable} = "probably bug in parsing input data";
 			next;
 		}
 
@@ -152,8 +159,11 @@
 			# can we have subfields?
 			if (ref($r->{$f}) eq 'ARRAY') {
 				# are values hashes? (has subfields)
-				if (ref($v) ne 'HASH') {
-					push @errors, "$f has value without subfields: $v";
+				if (! defined($v)) {
+#					$errors->{field}->{$f}->{empty} = undef;
+#					$errors->{dump} = $rec_dump if ($rec_dump);
+				} elsif (ref($v) ne 'HASH') {
+					$errors->{field}->{$f}->{missing_subfield} = join(",", @{ $r->{$f} }) . " required";
 					next;
 				} else {
 
@@ -163,47 +173,141 @@
 
 					delete($v->{subfields}) if (defined($v->{subfields}));
 
+					my $subfields;
+
 					foreach my $sf (keys %{ $v }) {
 
+						$subfields->{ $sf }++;
+
 						# is non-repeatable but with multiple values?
 						if ( ! first { $_ eq $sf.'*' } @{$r->{$f}} ) {
 							if ( ref($v->{$sf}) eq 'ARRAY' ) {
 								$sf_repeatable->{$sf}++;
 							};
 							if (! first { $_ eq $sf } @{ $r->{$f} }) {
-								push @errors, "$f has unknown subfield: $sf";
+								$errors->{field}->{ $f }->{subfield}->{extra}->{$sf}++;
 							}
 						}
 
 					}
 					if (my @r_sf = sort keys( %$sf_repeatable )) {
-						my $plural = $#r_sf > 0 ? 1 : 0;
 
-						push @errors, "$f subfield" .
-						( $plural ? 's ' : ' ' ) .
-						join(', ', @r_sf) .
-						( $plural ? ' are ' : ' is ' ) .
-						'repeatable in: ' .
-						join('', _pack_subfields_hash( $h, 1) );
+						foreach my $sf (@r_sf) {
+							$errors->{field}->{$f}->{subfield}->{extra_repeatable}->{$sf}++;
+							$errors->{field}->{$f}->{dump} =
+								join('', _pack_subfields_hash( $h, 1 ) );
+						}
+
+					}
+
+					if ( defined( $self->{must_exist_sf}->{$f} ) ) {
+						foreach my $sf (sort keys %{ $self->{must_exist_sf}->{$f} }) {
+#warn "====> $f $sf must exist\n";
+							$errors->{field}->{$f}->{subfield}->{missing}->{$sf}++
+								unless defined( $subfields->{$sf} );
+						}
 					}
+
 				}
 			} elsif (ref($v) eq 'HASH') {
-				push @errors, "$f has subfields which is not valid";
+				$errors->{field}->{$f}->{unexpected_subfields}++;
+				$errors->{field}->{$f}->{dump} =
+					join('', _pack_subfields_hash( $v, 1 ) );
 			}
 		}
 	}
 
 	foreach my $must (sort keys %{ $self->{must_exist} }) {
 		next if ($fields->{$must});
-		push @errors,
-			"field $must should exist, but it doesn't";
+		$errors->{field}->{$must}->{missing}++;
+		$errors->{dump} = $rec_dump if ($rec_dump);
+	}
+
+	if ($errors) {
+		$log->debug("errors: ", sub { dump( $errors ) } );
+
+		my $mfn = $rec->{'000'}->[0] || $log->logconfess("record ", dump( $rec ), " doesn't have MFN");
+		$self->{errors}->{$mfn} = $errors;
 	}
 
 	#$log->logcluck("return from this function is ARRAY") unless wantarray;
 
-	$log->debug("errors: ", join(", ", @errors)) if (@errors);
+	return $errors;
+}
+
+=head2 reset_errors
+
+Clean all accumulated errors for this input
+
+  $validate->reset_errors;
+
+=cut
+
+sub reset_errors {
+	my $self = shift;
+	delete ($self->{errors});
+}
+
+=head2 all_errors
+
+Return hash with all errors
+
+  print dump( $validate->all_errors );
+
+=cut
+
+sub all_errors {
+	my $self = shift;
+	return $self->{errors};
+}
+
+=head2 report
+
+Produce nice humanly readable report of errors
+
+  print $validate->report;
+
+=cut
+
+sub report {
+	my $self = shift;
+
+	sub unroll {
+		my ($rest,$o, $dump) = @_;
+
+#warn "# rest: $rest o: $o\n";
+
+		return unless ($rest);
+
+		if (ref($rest) ne 'HASH') {
+			$o .= "($rest)";
+			return ($o,$dump);
+		}
+
+		foreach my $k (sort keys %{ $rest }) {
+
+			if ($k eq 'dump') {
+				$dump = $rest->{dump};
+				warn "## dump: $dump\n";
+				next;
+			}
+			my $u;
+			($u, $dump) = unroll($rest->{$k}, $o, $dump);
+			$o .= "$k $u";
+
+
+		}
+		return ($o,$dump);
+	}
+
+	my $out = '';
+
+	foreach my $mfn (sort keys %{ $self->{errors} }) {
+		my ($msg,$dump) = unroll( $self->{errors}->{$mfn}, '', '' );
+		$out .= "MFN $mfn\n$msg\t$dump\n\n";
+	}
 
-	return @errors;
+	return $out;
 }
 
 =head1 AUTHOR