lib/WebPAC/Validate.pm

package WebPAC::Validate;

use warnings;
use strict;

use blib;

use base 'WebPAC::Common';
use File::Slurp;
use List::Util qw/first/;
use Data::Dump qw/dump/;
use WebPAC::Normalize qw/_pack_subfields_hash/;
use Storable qw/dclone/;

=head1 NAME

WebPAC::Validate - provide simple validation for records

=head1 VERSION

Version 0.12

=cut

our $VERSION = '0.12';

=head1 SYNOPSIS

This module provide a simple way to validate your file against a simple
configuration file in following format:

  # field 10 doesn't have any subfields
  10
  # same with 101
  101
  # field 200 have valid subfields a-g
  # and field e is repeatable
  200 a b c d e* f g
  # field 205 can have only subfield a
  # and must exists
  205! a
  # while 210 can have a c or d
  210 a c d
  # field which is ignored in validation
  999-

=head1 FUNCTIONS

=head2 new

Create new validation object

  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 {
        my $class = shift;
        my $self = {@_};
        bless($self, $class);

        my $log = $self->_get_logger();

        $log->logdie("need path or delimiters_path") unless ( $self->{path} || $self->{delimiters_path} );

        if ( $self->{path} ) {

                my $v_file = read_file( $self->{path} ) ||
                        $log->logdie("can't open validate path $self->{path}: $!");

                my $v;
                my $curr_line = 1;

                foreach my $l (split(/[\n\r]+/, $v_file)) {
                        $curr_line++;

                        # skip comments and whitespaces
                        next if ($l =~ /^#/ || $l =~ /^\s*$/);

                        $l =~ s/^\s+//;
                        $l =~ s/\s+$//;

                        my @d = split(/\s+/, $l);

                        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));

                        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));

                $self->{rules} = $v;

                $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_rec

Validate record and return errors

  my @errors = $validate->validate_rec( $rec, $rec_dump );

=cut

sub validate_rec {
        my $self = shift;

        my $log = $self->_get_logger();

        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});
        my $r = $self->{rules};

        my $errors;

        $log->debug("rec = ", sub { dump($rec) }, "keys = ", keys %{ $rec });

        my $fields;

        foreach my $f (keys %{ $rec }) {

                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}) ) {
                        $errors->{ $f }->{unexpected} = "this field is not expected";
                        next;
                }


                if (ref($rec->{$f}) ne 'ARRAY') {
                        $errors->{ $f }->{not_repeatable} = "probably bug in parsing input data";
                        next;
                }

                foreach my $v (@{ $rec->{$f} }) {
                        # can we have subfields?
                        if (ref($r->{$f}) eq 'ARRAY') {
                                # are values hashes? (has subfields)
                                if (! defined($v)) {
#                                       $errors->{$f}->{empty} = undef;
#                                       $errors->{dump} = $rec_dump if ($rec_dump);
                                } elsif (ref($v) ne 'HASH') {
                                        $errors->{$f}->{missing_subfield} = join(",", @{ $r->{$f} }) . " required";
                                        next;
                                } else {

                                        my $h = dclone( $v );

                                        my $sf_repeatable;

                                        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} }) {
                                                                $errors->{ $f }->{subfield}->{extra}->{$sf}++;
                                                        }
                                                }

                                        }
                                        if (my @r_sf = sort keys( %$sf_repeatable )) {

                                                foreach my $sf (@r_sf) {
                                                        $errors->{$f}->{subfield}->{extra_repeatable}->{$sf}++;
                                                        $errors->{$f}->{dump} = _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";
                                                        $errors->{$f}->{subfield}->{missing}->{$sf}++
                                                                unless defined( $subfields->{$sf} );
                                                }
                                        }

                                }
                        } elsif (ref($v) eq 'HASH') {
                                $errors->{$f}->{unexpected_subfields}++;
                                $errors->{$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}++;
                $errors->{dump} = $rec_dump if ($rec_dump);
        }

        if ($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;
        }

        #$log->logcluck("return from this function is ARRAY") unless wantarray;

        return $errors;
}

=head2 reset

Clean all accumulated errors for this input and remember delimiter templates
for L<save_delimiters_templates>

  $validate->reset;

This function B<must> be called after each input to provide accurate statistics.

=cut

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

Return hash with all errors

  print dump( $validate->all_errors );

=cut

sub all_errors {
        my $self = shift;
        return $self->{errors};
}

=head2 report_error

Produce nice humanly readable report of single error

  print $validate->report_error( $error_hash );

=cut

sub report_error {
        my $self = shift;

        my $h = shift || die "no hash?";

        sub _unroll {
                my ($self, $tree, $accumulated) = @_;

                my $log = $self->_get_logger();

                $log->debug("# ",
                        ( $tree                 ? "tree: $tree "                                        : '' ),
                        ( $accumulated  ? "accumulated: $accumulated "          : '' ),
                );

                my $results;

                if (ref($tree) ne 'HASH') {
                        return ("$accumulated\t($tree)", undef);
                }

                my $dump;

                foreach my $k (sort keys %{ $tree }) {

                        if ($k eq 'dump') {
                                $dump = $tree->{dump};
                                #warn "## dump ",dump($dump),"\n";
                                next;
                        }

                        $log->debug("current: $k");

                        my ($new_results, $new_dump) = $self->_unroll($tree->{$k},
                                $accumulated ? "$accumulated\t$k" : $k
                        );

                        $log->debug(
                                ( $new_results          ? "new_results: " . dump($new_results) ." "     : '' ),
                        );

                        push @$results, $new_results if ($new_results);
                        $dump = $new_dump if ($new_dump);

                }

                $log->debug(
                        ( $results              ? "results: " . dump($results) ." "     : '' ),
                );

                if ($#$results == 0) {
                        return ($results->[0], $dump);
                } else {
                        return ($results, $dump);
                }
        }


        sub _reformat {
                my $l = shift;
                $l =~ s/\t/ /g;
                $l =~ s/_/ /g;
                return $l;
        }

        my $out = '';

        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

Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>

=head1 COPYRIGHT & LICENSE

Copyright 2006 Dobrica Pavlinusic, All Rights Reserved.

This program is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.

=cut

1; # End of WebPAC::Validate
1	dpavlin	514	package WebPAC::Validate;
2
3			use warnings;
4			use strict;
5
6			use blib;
7
8			use base 'WebPAC::Common';
9			use File::Slurp;
10	dpavlin	515	use List::Util qw/first/;
11	dpavlin	652	use Data::Dump qw/dump/;
12	dpavlin	643	use WebPAC::Normalize qw/_pack_subfields_hash/;
13			use Storable qw/dclone/;
14	dpavlin	514
15			=head1 NAME
16
17			WebPAC::Validate - provide simple validation for records
18
19			=head1 VERSION
20
21	dpavlin	849	Version 0.12
22	dpavlin	514
23			=cut
24
25	dpavlin	849	our $VERSION = '0.12';
26	dpavlin	514
27			=head1 SYNOPSIS
28
29			This module provide a simple way to validate your file against a simple
30			configuration file in following format:
31
32			# field 10 doesn't have any subfields
33			10
34			# same with 101
35			101
36			# field 200 have valid subfields a-g
37	dpavlin	645	# and field e is repeatable
38			200 a b c d e* f g
39	dpavlin	514	# field 205 can have only subfield a
40	dpavlin	645	# and must exists
41			205! a
42	dpavlin	514	# while 210 can have a c or d
43			210 a c d
44	dpavlin	677	# field which is ignored in validation
45			999-
46	dpavlin	514
47			=head1 FUNCTIONS
48
49			=head2 new
50
51			Create new validation object
52
53			my $validate = new WebPAC::Validate(
54	dpavlin	643	path => 'conf/validate/file',
55	dpavlin	827	delimiters => [ ' : ', ' / ', ' ; ', ' , ' ],
56	dpavlin	837	delimiters_path => 'conf/validate/delimiters/file',
57	dpavlin	514	);
58
59	dpavlin	827	Optional parametar C<delimiters> will turn on validating of delimiters. Be
60			careful here, those delimiters are just stuck into regex, so they can
61			contain L<perlre> regexpes.
62
63	dpavlin	514	=cut
64
65			sub new {
66			my $class = shift;
67			my $self = {@_};
68			bless($self, $class);
69
70			my $log = $self->_get_logger();
71
72	dpavlin	839	$log->logdie("need path or delimiters_path") unless ( $self->{path} \|\| $self->{delimiters_path} );
73
74	dpavlin	837	if ( $self->{path} ) {
75	dpavlin	514
76	dpavlin	837	my $v_file = read_file( $self->{path} ) \|\|
77			$log->logdie("can't open validate path $self->{path}: $!");
78	dpavlin	514
79	dpavlin	837	my $v;
80			my $curr_line = 1;
81	dpavlin	514
82	dpavlin	837	foreach my $l (split(/[\n\r]+/, $v_file)) {
83			$curr_line++;
84	dpavlin	514
85	dpavlin	837	# skip comments and whitespaces
86			next if ($l =~ /^#/ \|\| $l =~ /^\s*$/);
87	dpavlin	516
88	dpavlin	837	$l =~ s/^\s+//;
89			$l =~ s/\s+$//;
90	dpavlin	514
91	dpavlin	837	my @d = split(/\s+/, $l);
92	dpavlin	514
93	dpavlin	837	my $fld = shift @d;
94	dpavlin	514
95	dpavlin	837	if ($fld =~ s/!$//) {
96			$self->{must_exist}->{$fld}++;
97			} elsif ($fld =~ s/-$//) {
98			$self->{dont_validate}->{$fld}++;
99			}
100	dpavlin	645
101	dpavlin	837	$log->logdie("need field name in line $curr_line: $l") unless (defined($fld));
102	dpavlin	516
103	dpavlin	837	if (@d) {
104			$v->{$fld} = [ map {
105			my $sf = $_;
106			if ( $sf =~ s/!(\*)?$/$1/ ) {
107			$self->{must_exist_sf}->{ $fld }->{ $sf }++;
108			};
109			$sf;
110			} @d ];
111			} else {
112			$v->{$fld} = 1;
113			}
114
115	dpavlin	514	}
116
117	dpavlin	837	$log->debug("current validation rules: ", dump($v));
118	dpavlin	514
119	dpavlin	837	$self->{rules} = $v;
120	dpavlin	514
121	dpavlin	837	$log->info("validation uses rules from $self->{path}");
122			}
123	dpavlin	514
124	dpavlin	827	if ( $self->{delimiters} ) {
125			$self->{delimiters_regex} = '(\^[a-z0-9]\|' . join('\|', @{ $self->{delimiters} }) . ')';
126			$log->info("validation check delimiters with regex $self->{delimiters_regex}");
127			}
128
129	dpavlin	837	if ( my $path = $self->{delimiters_path} ) {
130			if ( -e $path ) {
131			$log->info("using delimiter validation rules from $path");
132			open(my $d, $path) \|\| $log->fatal("can't open $path: $!");
133			while(<$d>) {
134			chomp($d);
135			if (/^\s(#)\s(\d+)\t+(\d+)\t+(.)$/) {
136			my ($comment,$field,$count,$template) = ($1,$2,$3,$4);
137			$self->{_validate_delimiters_templates}->{$field}->{$template} = $count unless ($comment);
138			} else {
139			warn "## ignored $d\n";
140			}
141			}
142			close($d);
143	dpavlin	841	#warn "_validate_delimiters_templates = ",dump( $self->{_validate_delimiters_templates} );
144	dpavlin	837	} else {
145			$log->warn("delimiters path $path doesn't exist, it will be created after this run");
146			}
147			}
148
149	dpavlin	514	$self ? return $self : return undef;
150			}
151
152	dpavlin	768	=head2 validate_rec
153	dpavlin	515
154			Validate record and return errors
155
156	dpavlin	768	my @errors = $validate->validate_rec( $rec, $rec_dump );
157	dpavlin	515
158			=cut
159
160	dpavlin	768	sub validate_rec {
161	dpavlin	515	my $self = shift;
162
163			my $log = $self->_get_logger();
164
165	dpavlin	768	my $rec = shift \|\| $log->logdie("validate_rec need record");
166	dpavlin	653	my $rec_dump = shift;
167	dpavlin	515
168			$log->logdie("rec isn't HASH") unless (ref($rec) eq 'HASH');
169	dpavlin	837	# $log->logdie("can't find validation rules") unless (my $r = $self->{rules});
170			my $r = $self->{rules};
171	dpavlin	515
172	dpavlin	652	my $errors;
173	dpavlin	515
174	dpavlin	652	$log->debug("rec = ", sub { dump($rec) }, "keys = ", keys %{ $rec });
175	dpavlin	515
176	dpavlin	645	my $fields;
177
178	dpavlin	515	foreach my $f (keys %{ $rec }) {
179
180	dpavlin	516	next if (!defined($f) \|\| $f eq '' \|\| $f eq '000');
181	dpavlin	515
182	dpavlin	827	# first check delimiters
183			if ( my $regex = $self->{delimiters_regex} ) {
184
185			foreach my $v (@{ $rec->{$f} }) {
186	dpavlin	835	my $l = _pack_subfields_hash( $v, 1 );
187			my $subfield_dump = $l;
188			my $template = '';
189			$l =~ s/$regex/$template.=$1/eg;
190			#warn "## template: $template\n";
191
192			if ( $template ) {
193			$self->{_delimiters_templates}->{$f}->{$template}++;
194
195			if ( my $v = $self->{_validate_delimiters_templates} ) {
196	dpavlin	838	if ( ! defined( $v->{$f}->{$template} ) ) {
197	dpavlin	843	$errors->{$f}->{potentially_invalid_combination} = $template;
198	dpavlin	835	$errors->{$f}->{dump} = $subfield_dump;
199	dpavlin	838	#} else {
200			# warn "## $f $template ok\n";
201	dpavlin	835	}
202			}
203			}
204	dpavlin	827	}
205			}
206
207	dpavlin	837	next unless ( $r ); # skip validation of no rules are specified
208
209	dpavlin	677	next if (defined( $self->{dont_validate}->{$f} ));
210
211			# track field usage
212	dpavlin	645	$fields->{$f}++;
213
214			if ( ! defined($r->{$f}) ) {
215	dpavlin	666	$errors->{ $f }->{unexpected} = "this field is not expected";
216	dpavlin	515	next;
217			}
218
219	dpavlin	645
220	dpavlin	515	if (ref($rec->{$f}) ne 'ARRAY') {
221	dpavlin	666	$errors->{ $f }->{not_repeatable} = "probably bug in parsing input data";
222	dpavlin	515	next;
223			}
224
225			foreach my $v (@{ $rec->{$f} }) {
226			# can we have subfields?
227			if (ref($r->{$f}) eq 'ARRAY') {
228			# are values hashes? (has subfields)
229	dpavlin	652	if (! defined($v)) {
230	dpavlin	666	# $errors->{$f}->{empty} = undef;
231	dpavlin	656	# $errors->{dump} = $rec_dump if ($rec_dump);
232	dpavlin	652	} elsif (ref($v) ne 'HASH') {
233	dpavlin	666	$errors->{$f}->{missing_subfield} = join(",", @{ $r->{$f} }) . " required";
234	dpavlin	515	next;
235			} else {
236	dpavlin	643
237			my $h = dclone( $v );
238
239	dpavlin	645	my $sf_repeatable;
240
241	dpavlin	643	delete($v->{subfields}) if (defined($v->{subfields}));
242
243	dpavlin	646	my $subfields;
244
245	dpavlin	515	foreach my $sf (keys %{ $v }) {
246	dpavlin	643
247	dpavlin	646	$subfields->{ $sf }++;
248
249	dpavlin	645	# is non-repeatable but with multiple values?
250			if ( ! first { $_ eq $sf.'*' } @{$r->{$f}} ) {
251			if ( ref($v->{$sf}) eq 'ARRAY' ) {
252			$sf_repeatable->{$sf}++;
253			};
254			if (! first { $_ eq $sf } @{ $r->{$f} }) {
255	dpavlin	666	$errors->{ $f }->{subfield}->{extra}->{$sf}++;
256	dpavlin	645	}
257	dpavlin	515	}
258	dpavlin	643
259	dpavlin	515	}
260	dpavlin	645	if (my @r_sf = sort keys( %$sf_repeatable )) {
261
262	dpavlin	652	foreach my $sf (@r_sf) {
263	dpavlin	666	$errors->{$f}->{subfield}->{extra_repeatable}->{$sf}++;
264	dpavlin	670	$errors->{$f}->{dump} = _pack_subfields_hash( $h, 1 );
265	dpavlin	652	}
266
267	dpavlin	645	}
268	dpavlin	646
269			if ( defined( $self->{must_exist_sf}->{$f} ) ) {
270			foreach my $sf (sort keys %{ $self->{must_exist_sf}->{$f} }) {
271	dpavlin	647	#warn "====> $f $sf must exist\n";
272	dpavlin	666	$errors->{$f}->{subfield}->{missing}->{$sf}++
273	dpavlin	652	unless defined( $subfields->{$sf} );
274	dpavlin	646	}
275			}
276
277	dpavlin	515	}
278			} elsif (ref($v) eq 'HASH') {
279	dpavlin	666	$errors->{$f}->{unexpected_subfields}++;
280	dpavlin	670	$errors->{$f}->{dump} = _pack_subfields_hash( $v, 1 );
281	dpavlin	515	}
282			}
283			}
284
285	dpavlin	827	$log->debug("_delimiters_templates = ", dump( $self->{_delimiters_templates} ) );
286
287	dpavlin	645	foreach my $must (sort keys %{ $self->{must_exist} }) {
288			next if ($fields->{$must});
289	dpavlin	666	$errors->{$must}->{missing}++;
290	dpavlin	653	$errors->{dump} = $rec_dump if ($rec_dump);
291	dpavlin	645	}
292
293	dpavlin	652	if ($errors) {
294	dpavlin	671	$log->debug("errors: ", $self->report_error( $errors ) );
295	dpavlin	652
296	dpavlin	653	my $mfn = $rec->{'000'}->[0] \|\| $log->logconfess("record ", dump( $rec ), " doesn't have MFN");
297	dpavlin	652	$self->{errors}->{$mfn} = $errors;
298			}
299
300	dpavlin	515	#$log->logcluck("return from this function is ARRAY") unless wantarray;
301
302	dpavlin	652	return $errors;
303	dpavlin	515	}
304
305	dpavlin	849	=head2 reset
306	dpavlin	653
307	dpavlin	849	Clean all accumulated errors for this input and remember delimiter templates
308			for L<save_delimiters_templates>
309	dpavlin	653
310	dpavlin	849	$validate->reset;
311	dpavlin	653
312	dpavlin	849	This function B<must> be called after each input to provide accurate statistics.
313
314	dpavlin	653	=cut
315
316	dpavlin	849	sub reset {
317	dpavlin	653	my $self = shift;
318	dpavlin	849
319			my $log = $self->_get_logger;
320
321	dpavlin	653	delete ($self->{errors});
322	dpavlin	849
323			if ( ! $self->{_delimiters_templates} ) {
324			$log->debug("called without _delimiters_templates?");
325			return;
326			}
327
328			foreach my $f ( keys %{ $self->{_delimiters_templates} } ) {
329			foreach my $t ( keys %{ $self->{_delimiters_templates}->{$f} } ) {
330			$self->{_accumulated_delimiters_templates}->{$f}->{$t} +=
331			$self->{_delimiters_templates}->{$f}->{$t};
332			}
333			}
334			$log->debug("_accumulated_delimiters_templates = ", dump( $self->{_accumulated_delimiter_templates} ));
335			delete ($self->{_delimiters_templates});
336	dpavlin	653	}
337
338			=head2 all_errors
339
340			Return hash with all errors
341
342			print dump( $validate->all_errors );
343
344			=cut
345
346			sub all_errors {
347			my $self = shift;
348			return $self->{errors};
349			}
350
351	dpavlin	671	=head2 report_error
352	dpavlin	664
353	dpavlin	671	Produce nice humanly readable report of single error
354	dpavlin	664
355	dpavlin	671	print $validate->report_error( $error_hash );
356	dpavlin	664
357			=cut
358
359	dpavlin	671	sub report_error {
360	dpavlin	664	my $self = shift;
361
362	dpavlin	671	my $h = shift \|\| die "no hash?";
363
364			sub _unroll {
365	dpavlin	666	my ($self, $tree, $accumulated) = @_;
366	dpavlin	664
367	dpavlin	666	my $log = $self->_get_logger();
368
369	dpavlin	665	$log->debug("# ",
370			( $tree ? "tree: $tree " : '' ),
371			( $accumulated ? "accumulated: $accumulated " : '' ),
372			);
373	dpavlin	664
374	dpavlin	665	my $results;
375	dpavlin	664
376	dpavlin	665	if (ref($tree) ne 'HASH') {
377			return ("$accumulated\t($tree)", undef);
378	dpavlin	664	}
379
380	dpavlin	665	my $dump;
381	dpavlin	664
382	dpavlin	665	foreach my $k (sort keys %{ $tree }) {
383
384	dpavlin	664	if ($k eq 'dump') {
385	dpavlin	665	$dump = $tree->{dump};
386	dpavlin	835	#warn "## dump ",dump($dump),"\n";
387	dpavlin	664	next;
388			}
389
390	dpavlin	665	$log->debug("current: $k");
391	dpavlin	664
392	dpavlin	671	my ($new_results, $new_dump) = $self->_unroll($tree->{$k},
393	dpavlin	665	$accumulated ? "$accumulated\t$k" : $k
394			);
395
396			$log->debug(
397			( $new_results ? "new_results: " . dump($new_results) ." " : '' ),
398			);
399
400			push @$results, $new_results if ($new_results);
401			$dump = $new_dump if ($new_dump);
402
403	dpavlin	664	}
404	dpavlin	665
405			$log->debug(
406			( $results ? "results: " . dump($results) ." " : '' ),
407			);
408
409			if ($#$results == 0) {
410			return ($results->[0], $dump);
411			} else {
412			return ($results, $dump);
413			}
414	dpavlin	664	}
415
416	dpavlin	666
417	dpavlin	671	sub _reformat {
418	dpavlin	666	my $l = shift;
419			$l =~ s/\t/ /g;
420	dpavlin	835	$l =~ s/_/ /g;
421	dpavlin	666	return $l;
422			}
423
424	dpavlin	671	my $out = '';
425	dpavlin	666
426	dpavlin	671	for my $f (sort keys %{ $h }) {
427			$out .= "$f: ";
428
429			my ($r, $d) = $self->_unroll( $h->{$f} );
430			my $e;
431			if (ref($r) eq 'ARRAY') {
432			$e .= join(", ", map { _reformat( $_ ) } @$r);
433			} else {
434			$e .= _reformat( $r );
435	dpavlin	666	}
436	dpavlin	671	$e .= "\n\t$d" if ($d);
437
438			$out .= $e . "\n";
439	dpavlin	664	}
440	dpavlin	671	return $out;
441			}
442	dpavlin	664
443	dpavlin	671
444			=head2 report
445
446			Produce nice humanly readable report of errors
447
448			print $validate->report;
449
450			=cut
451
452			sub report {
453			my $self = shift;
454			my $e = $self->{errors} \|\| return;
455
456			my $out;
457			foreach my $mfn (sort { $a <=> $b } keys %$e) {
458			$out .= "MFN $mfn\n" . $self->report_error( $e->{$mfn} ) . "\n";
459			}
460
461	dpavlin	664	return $out;
462	dpavlin	671
463	dpavlin	664	}
464
465	dpavlin	832	=head2 delimiters_templates
466
467	dpavlin	834	Generate report of delimiter tamplates
468	dpavlin	832
469	dpavlin	834	my $report = $validate->delimiter_teplates(
470			report => 1,
471	dpavlin	849	accumulated => 1,
472	dpavlin	834	);
473
474			Options:
475
476			=over 4
477
478			=item report
479
480			Generate humanly readable report with single fields
481
482	dpavlin	849	=item accumulated
483
484			Use accumulated data from all inputs
485
486	dpavlin	834	=back
487
488	dpavlin	832	=cut
489
490			sub delimiters_templates {
491			my $self = shift;
492
493	dpavlin	834	my $args = {@_};
494
495	dpavlin	849	my $t;
496			if ( $args->{accumulated} ) {
497			$t = $self->{_accumulated_delimiters_templates};
498			} else {
499			$t = $self->{_delimiters_templates};
500			}
501	dpavlin	832
502			my $log = $self->_get_logger;
503
504			unless ($t) {
505	dpavlin	834	$log->error("called without delimiters");
506	dpavlin	832	return;
507			}
508
509			my $out;
510
511			foreach my $f (sort { $a <=> $b } keys %$t) {
512	dpavlin	834	$out .= "$f\n" if ( $args->{report} );
513			foreach my $template (sort { $a cmp $b } keys %{ $t->{$f} }) {
514			my $count = $t->{$f}->{$template};
515			$out .=
516			( $count ? "" : "# " ) .
517	dpavlin	836	( $args->{report} ? "" : "$f" ) .
518	dpavlin	834	"\t$count\t$template\n";
519	dpavlin	832	}
520			}
521
522			return $out;
523			}
524
525	dpavlin	837	=head2 save_delimiters_templates
526
527	dpavlin	849	Save accumulated delimiter templates
528
529			$validator->save_delimiters_template( '/path/to/validate/delimiters' );
530
531	dpavlin	837	=cut
532
533			sub save_delimiters_templates {
534			my $self = shift;
535
536			my $path = $self->{delimiters_path};
537
538			return unless ( $path );
539
540			my $log = $self->_get_logger;
541
542	dpavlin	849	if ( ! $self->{_accumulated_delimiters_templates} ) {
543			$log->error('no _accumulated_delimiters_templates found, reset');
544			$self->reset;
545			}
546
547			if ( ! $self->{_delimiters_templates} ) {
548			$log->error('found _delimiters_templates, calling reset');
549			$self->reset;
550			}
551
552	dpavlin	840	$path .= '.new' if ( -e $path );
553
554	dpavlin	837	open(my $d, '>', $path) \|\| $log->fatal("can't open $path: $!");
555	dpavlin	849	print $d $self->delimiters_templates( accumulated => 1 );
556	dpavlin	837	close($d);
557
558			$log->info("new delimiters templates saved to $path");
559			}
560
561	dpavlin	514	=head1 AUTHOR
562
563			Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
564
565			=head1 COPYRIGHT & LICENSE
566
567			Copyright 2006 Dobrica Pavlinusic, All Rights Reserved.
568
569			This program is free software; you can redistribute it and/or modify it
570			under the same terms as Perl itself.
571
572			=cut
573
574			1; # End of WebPAC::Validate