--- trunk/lib/WebPAC/Validate.pm	2006/09/11 12:32:51	666
+++ trunk/lib/WebPAC/Validate.pm	2007/05/27 10:50:37	849
@@ -18,11 +18,11 @@
 
 =head1 VERSION
 
-Version 0.07
+Version 0.12
 
 =cut
 
-our $VERSION = '0.07';
+our $VERSION = '0.12';
 
 =head1 SYNOPSIS
 
@@ -41,6 +41,8 @@
   205! a
   # while 210 can have a c or d
   210 a c d
+  # field which is ignored in validation
+  999-
 
 =head1 FUNCTIONS
 
@@ -50,8 +52,14 @@
 
   my $validate = new WebPAC::Validate(
   	path => 'conf/validate/file',
+	delimiters => [ ' : ', ' / ', ' ; ', ' , ' ],
+	delimiters_path => 'conf/validate/delimiters/file',
   );
 
+Optional parametar C<delimiters> will turn on validating of delimiters. Be
+careful here, those delimiters are just stuck into regex, so they can
+contain L<perlre> regexpes.
+
 =cut
 
 sub new {
@@ -61,76 +69,105 @@
 
 	my $log = $self->_get_logger();
 
-	foreach my $p (qw/path/) {
-		$log->logconfess("need $p") unless ($self->{$p});
-	}
+	$log->logdie("need path or delimiters_path") unless ( $self->{path} || $self->{delimiters_path} );
 
-	my $v_file = read_file( $self->{path} ) ||
-		$log->logdie("can't open validate path $self->{path}: $!");
+	if ( $self->{path} ) {
 
-	my $v;
-	my $curr_line = 1;
+		my $v_file = read_file( $self->{path} ) ||
+			$log->logdie("can't open validate path $self->{path}: $!");
 
-	foreach my $l (split(/[\n\r]+/, $v_file)) {
-		$curr_line++;
+		my $v;
+		my $curr_line = 1;
 
-		# skip comments and whitespaces
-		next if ($l =~ /^#/ || $l =~ /^\s*$/);
+		foreach my $l (split(/[\n\r]+/, $v_file)) {
+			$curr_line++;
 
-		$l =~ s/^\s+//;
-		$l =~ s/\s+$//;
+			# skip comments and whitespaces
+			next if ($l =~ /^#/ || $l =~ /^\s*$/);
 
-		my @d = split(/\s+/, $l);
+			$l =~ s/^\s+//;
+			$l =~ s/\s+$//;
 
-		my $fld = shift @d;
+			my @d = split(/\s+/, $l);
 
-		if ($fld =~ s/!$//) {
-			$self->{must_exist}->{$fld}++;
-		}
+			my $fld = shift @d;
+
+			if ($fld =~ s/!$//) {
+				$self->{must_exist}->{$fld}++;
+			} elsif ($fld =~ s/-$//) {
+				$self->{dont_validate}->{$fld}++;
+			}
 
-		$log->logdie("need field name in line $curr_line: $l") unless (defined($fld));
+			$log->logdie("need field name in line $curr_line: $l") unless (defined($fld));
+
+			if (@d) {
+				$v->{$fld} = [ map {
+					my $sf = $_;
+					if ( $sf =~ s/!(\*)?$/$1/ ) {
+						$self->{must_exist_sf}->{ $fld }->{ $sf }++;
+					};
+					$sf;
+				} @d ];
+			} else {
+				$v->{$fld} = 1;
+			}
 
-		if (@d) {
-			$v->{$fld} = [ map {
-				my $sf = $_;
-				if ( $sf =~ s/!(\*)?$/$1/ ) {
-					$self->{must_exist_sf}->{ $fld }->{ $sf }++;
-				};
-				$sf;
-			} @d ];
-		} else {
-			$v->{$fld} = 1;
 		}
 
-	}
+		$log->debug("current validation rules: ", dump($v));
 
-	$log->debug("current validation rules: ", dump($v));
+		$self->{rules} = $v;
 
-	$self->{rules} = $v;
+		$log->info("validation uses rules from $self->{path}");
+	}
 
-	$log->info("validation uses rules from $self->{path}");
+	if ( $self->{delimiters} ) {
+		$self->{delimiters_regex} = '(\^[a-z0-9]|' . join('|', @{ $self->{delimiters} }) . ')';
+		$log->info("validation check delimiters with regex $self->{delimiters_regex}");
+	}
+
+	if ( my $path = $self->{delimiters_path} ) {
+		if ( -e $path ) {
+			$log->info("using delimiter validation rules from $path");
+			open(my $d, $path) || $log->fatal("can't open $path: $!");
+			while(<$d>) {
+				chomp($d);
+				if (/^\s*(#*)\s*(\d+)\t+(\d+)\t+(.*)$/) {
+					my ($comment,$field,$count,$template) = ($1,$2,$3,$4);
+					$self->{_validate_delimiters_templates}->{$field}->{$template} = $count unless ($comment);
+				} else {
+					warn "## ignored $d\n";
+				}
+			}
+			close($d);
+			#warn "_validate_delimiters_templates = ",dump( $self->{_validate_delimiters_templates} );
+		} else {
+			$log->warn("delimiters path $path doesn't exist, it will be created after this run");
+		}
+	}
 
 	$self ? return $self : return undef;
 }
 
-=head2 validate_errors
+=head2 validate_rec
 
 Validate record and return errors
 
-  my @errors = $validate->validate_errors( $rec, $rec_dump );
+  my @errors = $validate->validate_rec( $rec, $rec_dump );
 
 =cut
 
-sub validate_errors {
+sub validate_rec {
 	my $self = shift;
 
 	my $log = $self->_get_logger();
 
-	my $rec = shift || $log->logdie("validate_errors need record");
+	my $rec = shift || $log->logdie("validate_rec 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});
+#	$log->logdie("can't find validation rules") unless (my $r = $self->{rules});
+	my $r = $self->{rules};
 
 	my $errors;
 
@@ -142,6 +179,36 @@
 
 		next if (!defined($f) || $f eq '' || $f eq '000');
 
+		# first check delimiters
+		if ( my $regex = $self->{delimiters_regex} ) {
+
+			foreach my $v (@{ $rec->{$f} }) {
+				my $l = _pack_subfields_hash( $v, 1 );
+				my $subfield_dump = $l;
+				my $template = '';
+				$l =~ s/$regex/$template.=$1/eg;
+				#warn "## template: $template\n";
+
+				if ( $template ) {
+					$self->{_delimiters_templates}->{$f}->{$template}++;
+
+					if ( my $v = $self->{_validate_delimiters_templates} ) {
+						if ( ! defined( $v->{$f}->{$template} ) ) {
+							$errors->{$f}->{potentially_invalid_combination} = $template;
+							$errors->{$f}->{dump} = $subfield_dump;
+						#} else {
+						#	warn "## $f $template ok\n";
+						}
+					}
+				}
+			}
+		}
+
+		next unless ( $r );	# skip validation of no rules are specified
+
+		next if (defined( $self->{dont_validate}->{$f} ));
+
+		# track field usage
 		$fields->{$f}++;
 
 		if ( ! defined($r->{$f}) ) {
@@ -194,8 +261,7 @@
 
 						foreach my $sf (@r_sf) {
 							$errors->{$f}->{subfield}->{extra_repeatable}->{$sf}++;
-							$errors->{$f}->{dump} =
-								join('', _pack_subfields_hash( $h, 1 ) );
+							$errors->{$f}->{dump} = _pack_subfields_hash( $h, 1 );
 						}
 
 					}
@@ -211,12 +277,13 @@
 				}
 			} elsif (ref($v) eq 'HASH') {
 				$errors->{$f}->{unexpected_subfields}++;
-				$errors->{$f}->{dump} =
-					join('', _pack_subfields_hash( $v, 1 ) );
+				$errors->{$f}->{dump} = _pack_subfields_hash( $v, 1 );
 			}
 		}
 	}
 
+	$log->debug("_delimiters_templates = ", dump( $self->{_delimiters_templates} ) );
+
 	foreach my $must (sort keys %{ $self->{must_exist} }) {
 		next if ($fields->{$must});
 		$errors->{$must}->{missing}++;
@@ -224,7 +291,7 @@
 	}
 
 	if ($errors) {
-		$log->debug("errors: ", sub { dump( $errors ) } );
+		$log->debug("errors: ", $self->report_error( $errors ) );
 
 		my $mfn = $rec->{'000'}->[0] || $log->logconfess("record ", dump( $rec ), " doesn't have MFN");
 		$self->{errors}->{$mfn} = $errors;
@@ -235,17 +302,37 @@
 	return $errors;
 }
 
-=head2 reset_errors
+=head2 reset
 
-Clean all accumulated errors for this input
+Clean all accumulated errors for this input and remember delimiter templates
+for L<save_delimiters_templates>
 
-  $validate->reset_errors;
+  $validate->reset;
+
+This function B<must> be called after each input to provide accurate statistics.
 
 =cut
 
-sub reset_errors {
+sub reset {
 	my $self = shift;
+
+	my $log = $self->_get_logger;
+
 	delete ($self->{errors});
+
+	if ( ! $self->{_delimiters_templates} ) {
+		$log->debug("called without _delimiters_templates?");
+		return;
+	}
+
+	foreach my $f ( keys %{ $self->{_delimiters_templates} } ) {
+		foreach my $t ( keys %{ $self->{_delimiters_templates}->{$f} } ) {
+			$self->{_accumulated_delimiters_templates}->{$f}->{$t} +=
+				$self->{_delimiters_templates}->{$f}->{$t};
+		}
+	}
+	$log->debug("_accumulated_delimiters_templates = ", dump( $self->{_accumulated_delimiter_templates} ));
+	delete ($self->{_delimiters_templates});
 }
 
 =head2 all_errors
@@ -261,18 +348,20 @@
 	return $self->{errors};
 }
 
-=head2 report
+=head2 report_error
 
-Produce nice humanly readable report of errors
+Produce nice humanly readable report of single error
 
-  print $validate->report;
+  print $validate->report_error( $error_hash );
 
 =cut
 
-sub report {
+sub report_error {
 	my $self = shift;
 
-	sub unroll {
+	my $h = shift || die "no hash?";
+
+	sub _unroll {
 		my ($self, $tree, $accumulated) = @_;
 
 		my $log = $self->_get_logger();
@@ -294,13 +383,13 @@
 
 			if ($k eq 'dump') {
 				$dump = $tree->{dump};
-				warn "## dump: ",dump($dump),"\n";
+				#warn "## dump ",dump($dump),"\n";
 				next;
 			}
 
 			$log->debug("current: $k");
 
-			my ($new_results, $new_dump) = $self->unroll($tree->{$k},
+			my ($new_results, $new_dump) = $self->_unroll($tree->{$k},
 				$accumulated ? "$accumulated\t$k" : $k
 			);
 
@@ -324,37 +413,149 @@
 		}
 	}
 
-	my $log = $self->_get_logger();
 
-	my $out = '';
-	my $e = $self->{errors} || return;
-
-	sub reformat {
+	sub _reformat {
 		my $l = shift;
 		$l =~ s/\t/ /g;
-		$l =~ s/_/ /;
+		$l =~ s/_/ /g;
 		return $l;
 	}
 
-	foreach my $mfn (sort keys %$e) {
-		$out .= "MFN $mfn\n";
+	my $out = '';
 
-		for my $f (sort keys %{ $e->{$mfn} }) {
-			my ($r, $d) = $self->unroll( $e->{$mfn}->{$f} );
-			my $e = $f . ': ';
-			if (ref($r) eq 'ARRAY') {
-				$e .= join(", ", map { reformat( $_ ) } @$r);
-			} else {
-				$e .= reformat( $r );
-			}
-			$e .= "\n\t$d" if ($d);
-			$e .= "\n";
-			$log->debug("MFN $mfn | $e");
-			$out .= $e;
+	for my $f (sort keys %{ $h }) {
+		$out .= "$f: ";
+		
+		my ($r, $d) = $self->_unroll( $h->{$f} );
+		my $e;
+		if (ref($r) eq 'ARRAY') {
+			$e .= join(", ", map { _reformat( $_ ) } @$r);
+		} else {
+			$e .= _reformat( $r );
 		}
+		$e .= "\n\t$d" if ($d);
+
+		$out .= $e . "\n";
+	}
+	return $out;
+}
+
+
+=head2 report
+
+Produce nice humanly readable report of errors
+
+  print $validate->report;
+
+=cut
+
+sub report {
+	my $self = shift;
+	my $e = $self->{errors} || return;
+
+	my $out;
+	foreach my $mfn (sort { $a <=> $b } keys %$e) {
+		$out .= "MFN $mfn\n" . $self->report_error( $e->{$mfn} ) . "\n";
 	}
 
 	return $out;
+
+}
+
+=head2 delimiters_templates
+
+Generate report of delimiter tamplates 
+
+  my $report = $validate->delimiter_teplates(
+	report => 1,
+	accumulated => 1,
+  );
+
+Options:
+
+=over 4
+
+=item report
+
+Generate humanly readable report with single fields
+
+=item accumulated
+
+Use accumulated data from all inputs
+
+=back
+
+=cut
+
+sub delimiters_templates {
+	my $self = shift;
+
+	my $args = {@_};
+
+	my $t;
+	if ( $args->{accumulated} ) {
+		$t = $self->{_accumulated_delimiters_templates};
+	} else {
+		$t = $self->{_delimiters_templates};
+	}
+
+	my $log = $self->_get_logger;
+
+	unless ($t) {
+		$log->error("called without delimiters");
+		return;
+	}
+
+	my $out;
+
+	foreach my $f (sort { $a <=> $b } keys %$t) {
+		$out .= "$f\n" if ( $args->{report} );
+		foreach my $template (sort { $a cmp $b } keys %{ $t->{$f} }) {
+			my $count = $t->{$f}->{$template};
+			$out .= 
+				( $count ? "" : "# " ) .
+				( $args->{report} ? "" : "$f" ) .
+				"\t$count\t$template\n";
+		}
+	}
+
+	return $out;
+}
+
+=head2 save_delimiters_templates
+
+Save accumulated delimiter templates
+
+  $validator->save_delimiters_template( '/path/to/validate/delimiters' );
+
+=cut
+
+sub save_delimiters_templates {
+	my $self = shift;
+
+	my $path = $self->{delimiters_path};
+
+	return unless ( $path );
+
+	my $log = $self->_get_logger;
+
+	if ( ! $self->{_accumulated_delimiters_templates} ) {
+		$log->error('no _accumulated_delimiters_templates found, reset');
+		$self->reset;
+	}
+
+	if ( ! $self->{_delimiters_templates} ) {
+		$log->error('found _delimiters_templates, calling reset');
+		$self->reset;
+	}
+
+	$path .= '.new' if ( -e $path );
+
+	open(my $d, '>', $path) || $log->fatal("can't open $path: $!");
+	print $d $self->delimiters_templates( accumulated => 1 );
+	close($d);
+
+	$log->info("new delimiters templates saved to $path");
 }
 
 =head1 AUTHOR