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::Dumper;
use WebPAC::Normalize qw/_pack_subfields_hash/;
use Storable qw/dclone/;

=head1 NAME

WebPAC::Validate - provide simple validation for records

=head1 VERSION

Version 0.02

=cut

our $VERSION = '0.02';

=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
  200 a b c d e f g
  # field 205 can have only subfield a
  205 a
  # while 210 can have a c or d
  210 a c d

=head1 FUNCTIONS

=head2 new

Create new validation object

  my $validate = new WebPAC::Validate(
        path => 'conf/validate/file',
  );

=cut

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

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

        foreach my $p (qw/path/) {
                $log->logconfess("need $p") unless ($self->{$p});
        }

        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;

                $log->logdie("need field name in line $curr_line: $l") unless (defined($fld));

                if (@d) {
                        $v->{$fld} = \@d;
                } else {
                        $v->{$fld} = 1;
                }

        }

        $log->debug("current validation rules: ", Dumper($v));

        $self->{rules} = $v;

        $log->info("validation uses rules from $self->{path}");

        $self ? return $self : return undef;
}

=head2 validate_errors

Validate record and return errors

  my @errors = $validate->validate_errors( $rec );

=cut

sub validate_errors {
        my $self = shift;

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

        my $rec = shift || $log->logdie("validate_errors need record");

        $log->logdie("rec isn't HASH") unless (ref($rec) eq 'HASH');
        $log->logdie("can't find validation rules") unless (my $r = $self->{rules});

        my @errors;

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

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

                next if (!defined($f) || $f eq '' || $f eq '000');

                if (! defined($r->{$f})) {
                        push @errors, "field '$f' shouldn't exists";
                        next;
                }

                if (ref($rec->{$f}) ne 'ARRAY') {
                        push @errors, "field '$f' isn't repetable, 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 (ref($v) ne 'HASH') {
                                        push @errors, "$f has value without subfields: $v";
                                        next;
                                } else {

                                        my $h = dclone( $v );

                                        delete($v->{subfields}) if (defined($v->{subfields}));

                                        foreach my $sf (keys %{ $v }) {

                                                # permited subfield?
                                                if (! first { $_ eq $sf } @{ $r->{$f} }) {
                                                        push @errors, "$f has unknown subfield: $sf";
                                                }

                                                # is repeatable?
                                                if ( ref($v->{$sf}) eq 'ARRAY' ) {
                                                        push @errors, "$f subfield $sf is repeatable: " .
                                                                join('', _pack_subfields_hash( dclone($h), 1) );
                                                        ### FIXME
                                                }
                                        }
                                }
                        } elsif (ref($v) eq 'HASH') {
                                push @errors, "$f has subfields which is not valid";
                        }
                }
        }

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

        $log->debug("errors: ", join(", ", @errors)) if (@errors);

        return @errors;
}

=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	514	use Data::Dumper;
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	643	Version 0.02
22	dpavlin	514
23			=cut
24
25	dpavlin	643	our $VERSION = '0.02';
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			200 a b c d e f g
38			# field 205 can have only subfield a
39			205 a
40			# while 210 can have a c or d
41			210 a c d
42
43			=head1 FUNCTIONS
44
45			=head2 new
46
47			Create new validation object
48
49			my $validate = new WebPAC::Validate(
50	dpavlin	643	path => 'conf/validate/file',
51	dpavlin	514	);
52
53			=cut
54
55			sub new {
56			my $class = shift;
57			my $self = {@_};
58			bless($self, $class);
59
60			my $log = $self->_get_logger();
61
62			foreach my $p (qw/path/) {
63			$log->logconfess("need $p") unless ($self->{$p});
64			}
65
66			my $v_file = read_file( $self->{path} ) \|\|
67			$log->logdie("can't open validate path $self->{path}: $!");
68
69			my $v;
70			my $curr_line = 1;
71
72			foreach my $l (split(/[\n\r]+/, $v_file)) {
73			$curr_line++;
74
75	dpavlin	516	# skip comments and whitespaces
76			next if ($l =~ /^#/ \|\| $l =~ /^\s*$/);
77
78	dpavlin	514	$l =~ s/^\s+//;
79			$l =~ s/\s+$//;
80
81			my @d = split(/\s+/, $l);
82
83	dpavlin	516	my $fld = shift @d;
84	dpavlin	514
85	dpavlin	516	$log->logdie("need field name in line $curr_line: $l") unless (defined($fld));
86
87	dpavlin	514	if (@d) {
88	dpavlin	515	$v->{$fld} = \@d;
89	dpavlin	514	} else {
90	dpavlin	515	$v->{$fld} = 1;
91	dpavlin	514	}
92
93			}
94
95			$log->debug("current validation rules: ", Dumper($v));
96
97			$self->{rules} = $v;
98
99	dpavlin	516	$log->info("validation uses rules from $self->{path}");
100
101	dpavlin	514	$self ? return $self : return undef;
102			}
103
104	dpavlin	515	=head2 validate_errors
105
106			Validate record and return errors
107
108			my @errors = $validate->validate_errors( $rec );
109
110			=cut
111
112			sub validate_errors {
113			my $self = shift;
114
115			my $log = $self->_get_logger();
116
117			my $rec = shift \|\| $log->logdie("validate_errors need record");
118
119			$log->logdie("rec isn't HASH") unless (ref($rec) eq 'HASH');
120			$log->logdie("can't find validation rules") unless (my $r = $self->{rules});
121
122			my @errors;
123
124			$log->debug("rec = ", sub { Dumper($rec) }, "keys = ", keys %{ $rec });
125
126			foreach my $f (keys %{ $rec }) {
127
128	dpavlin	516	next if (!defined($f) \|\| $f eq '' \|\| $f eq '000');
129	dpavlin	515
130			if (! defined($r->{$f})) {
131			push @errors, "field '$f' shouldn't exists";
132			next;
133			}
134
135			if (ref($rec->{$f}) ne 'ARRAY') {
136			push @errors, "field '$f' isn't repetable, probably bug in parsing input data";
137			next;
138			}
139
140			foreach my $v (@{ $rec->{$f} }) {
141			# can we have subfields?
142			if (ref($r->{$f}) eq 'ARRAY') {
143			# are values hashes? (has subfields)
144			if (ref($v) ne 'HASH') {
145			push @errors, "$f has value without subfields: $v";
146			next;
147			} else {
148	dpavlin	643
149			my $h = dclone( $v );
150
151			delete($v->{subfields}) if (defined($v->{subfields}));
152
153	dpavlin	515	foreach my $sf (keys %{ $v }) {
154	dpavlin	643
155	dpavlin	515	# permited subfield?
156			if (! first { $_ eq $sf } @{ $r->{$f} }) {
157	dpavlin	643	push @errors, "$f has unknown subfield: $sf";
158	dpavlin	515	}
159	dpavlin	643
160			# is repeatable?
161			if ( ref($v->{$sf}) eq 'ARRAY' ) {
162			push @errors, "$f subfield $sf is repeatable: " .
163			join('', _pack_subfields_hash( dclone($h), 1) );
164			### FIXME
165			}
166	dpavlin	515	}
167			}
168			} elsif (ref($v) eq 'HASH') {
169			push @errors, "$f has subfields which is not valid";
170			}
171			}
172			}
173
174			#$log->logcluck("return from this function is ARRAY") unless wantarray;
175
176			$log->debug("errors: ", join(", ", @errors)) if (@errors);
177
178			return @errors;
179			}
180
181	dpavlin	514	=head1 AUTHOR
182
183			Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
184
185			=head1 COPYRIGHT & LICENSE
186
187			Copyright 2006 Dobrica Pavlinusic, All Rights Reserved.
188
189			This program is free software; you can redistribute it and/or modify it
190			under the same terms as Perl itself.
191
192			=cut
193
194			1; # End of WebPAC::Validate