1 |
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 |
use List::Util qw/first/; |
11 |
use Data::Dump qw/dump/; |
12 |
use WebPAC::Normalize qw/_pack_subfields_hash/; |
13 |
use Storable qw/dclone/; |
14 |
|
15 |
=head1 NAME |
16 |
|
17 |
WebPAC::Validate - provide simple validation for records |
18 |
|
19 |
=head1 VERSION |
20 |
|
21 |
Version 0.06 |
22 |
|
23 |
=cut |
24 |
|
25 |
our $VERSION = '0.06'; |
26 |
|
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 |
# and field e is repeatable |
38 |
200 a b c d e* f g |
39 |
# field 205 can have only subfield a |
40 |
# and must exists |
41 |
205! a |
42 |
# while 210 can have a c or d |
43 |
210 a c d |
44 |
|
45 |
=head1 FUNCTIONS |
46 |
|
47 |
=head2 new |
48 |
|
49 |
Create new validation object |
50 |
|
51 |
my $validate = new WebPAC::Validate( |
52 |
path => 'conf/validate/file', |
53 |
); |
54 |
|
55 |
=cut |
56 |
|
57 |
sub new { |
58 |
my $class = shift; |
59 |
my $self = {@_}; |
60 |
bless($self, $class); |
61 |
|
62 |
my $log = $self->_get_logger(); |
63 |
|
64 |
foreach my $p (qw/path/) { |
65 |
$log->logconfess("need $p") unless ($self->{$p}); |
66 |
} |
67 |
|
68 |
my $v_file = read_file( $self->{path} ) || |
69 |
$log->logdie("can't open validate path $self->{path}: $!"); |
70 |
|
71 |
my $v; |
72 |
my $curr_line = 1; |
73 |
|
74 |
foreach my $l (split(/[\n\r]+/, $v_file)) { |
75 |
$curr_line++; |
76 |
|
77 |
# skip comments and whitespaces |
78 |
next if ($l =~ /^#/ || $l =~ /^\s*$/); |
79 |
|
80 |
$l =~ s/^\s+//; |
81 |
$l =~ s/\s+$//; |
82 |
|
83 |
my @d = split(/\s+/, $l); |
84 |
|
85 |
my $fld = shift @d; |
86 |
|
87 |
if ($fld =~ s/!$//) { |
88 |
$self->{must_exist}->{$fld}++; |
89 |
} |
90 |
|
91 |
$log->logdie("need field name in line $curr_line: $l") unless (defined($fld)); |
92 |
|
93 |
if (@d) { |
94 |
$v->{$fld} = [ map { |
95 |
my $sf = $_; |
96 |
if ( $sf =~ s/!(\*)?$/$1/ ) { |
97 |
$self->{must_exist_sf}->{ $fld }->{ $sf }++; |
98 |
}; |
99 |
$sf; |
100 |
} @d ]; |
101 |
} else { |
102 |
$v->{$fld} = 1; |
103 |
} |
104 |
|
105 |
} |
106 |
|
107 |
$log->debug("current validation rules: ", dump($v)); |
108 |
|
109 |
$self->{rules} = $v; |
110 |
|
111 |
$log->info("validation uses rules from $self->{path}"); |
112 |
|
113 |
$self ? return $self : return undef; |
114 |
} |
115 |
|
116 |
=head2 validate_errors |
117 |
|
118 |
Validate record and return errors |
119 |
|
120 |
my @errors = $validate->validate_errors( $rec ); |
121 |
|
122 |
=cut |
123 |
|
124 |
sub validate_errors { |
125 |
my $self = shift; |
126 |
|
127 |
my $log = $self->_get_logger(); |
128 |
|
129 |
my $rec = shift || $log->logdie("validate_errors need record"); |
130 |
|
131 |
$log->logdie("rec isn't HASH") unless (ref($rec) eq 'HASH'); |
132 |
$log->logdie("can't find validation rules") unless (my $r = $self->{rules}); |
133 |
|
134 |
my $errors; |
135 |
|
136 |
$log->debug("rec = ", sub { dump($rec) }, "keys = ", keys %{ $rec }); |
137 |
|
138 |
my $fields; |
139 |
|
140 |
foreach my $f (keys %{ $rec }) { |
141 |
|
142 |
next if (!defined($f) || $f eq '' || $f eq '000'); |
143 |
|
144 |
$fields->{$f}++; |
145 |
|
146 |
if ( ! defined($r->{$f}) ) { |
147 |
$errors->{field}->{ $f }->{extra} = "not expected"; |
148 |
next; |
149 |
} |
150 |
|
151 |
|
152 |
if (ref($rec->{$f}) ne 'ARRAY') { |
153 |
$errors->{field}->{ $f }->{not_repeatable} = "probably bug in parsing input data"; |
154 |
next; |
155 |
} |
156 |
|
157 |
foreach my $v (@{ $rec->{$f} }) { |
158 |
# can we have subfields? |
159 |
if (ref($r->{$f}) eq 'ARRAY') { |
160 |
# are values hashes? (has subfields) |
161 |
if (! defined($v)) { |
162 |
$errors->{field}->{$f}->{empty} = undef; |
163 |
$errors->{dump}->{record}++; |
164 |
} elsif (ref($v) ne 'HASH') { |
165 |
$errors->{field}->{$f}->{missing_subfield} = "value without subfields: $v"; |
166 |
next; |
167 |
} else { |
168 |
|
169 |
my $h = dclone( $v ); |
170 |
|
171 |
my $sf_repeatable; |
172 |
|
173 |
delete($v->{subfields}) if (defined($v->{subfields})); |
174 |
|
175 |
my $subfields; |
176 |
|
177 |
foreach my $sf (keys %{ $v }) { |
178 |
|
179 |
$subfields->{ $sf }++; |
180 |
|
181 |
# is non-repeatable but with multiple values? |
182 |
if ( ! first { $_ eq $sf.'*' } @{$r->{$f}} ) { |
183 |
if ( ref($v->{$sf}) eq 'ARRAY' ) { |
184 |
$sf_repeatable->{$sf}++; |
185 |
}; |
186 |
if (! first { $_ eq $sf } @{ $r->{$f} }) { |
187 |
$errors->{field}->{ $f }->{subfield}->{$sf} = "unknown"; |
188 |
} |
189 |
} |
190 |
|
191 |
} |
192 |
if (my @r_sf = sort keys( %$sf_repeatable )) { |
193 |
|
194 |
foreach my $sf (@r_sf) { |
195 |
$errors->{field}->{$f}->{subfield}->{$sf} = "repeatable"; |
196 |
$errors->{dump}->{field}->{$f} = |
197 |
join('', _pack_subfields_hash( $h, 1 ) ); |
198 |
} |
199 |
|
200 |
} |
201 |
|
202 |
if ( defined( $self->{must_exist_sf}->{$f} ) ) { |
203 |
foreach my $sf (sort keys %{ $self->{must_exist_sf}->{$f} }) { |
204 |
#warn "====> $f $sf must exist\n"; |
205 |
$errors->{field}->{$f}->{subfield}->{$sf} = "missing" |
206 |
unless defined( $subfields->{$sf} ); |
207 |
} |
208 |
} |
209 |
|
210 |
} |
211 |
} elsif (ref($v) eq 'HASH') { |
212 |
$errors->{field}->{$f}->{unexpected_subfields}++; |
213 |
$errors->{dump}->{field}->{$f} = |
214 |
join('', _pack_subfields_hash( $v, 1 ) ); |
215 |
} |
216 |
} |
217 |
} |
218 |
|
219 |
foreach my $must (sort keys %{ $self->{must_exist} }) { |
220 |
next if ($fields->{$must}); |
221 |
$errors->{field}->{$must}->{missing}++; |
222 |
$errors->{dump}->{record}++; |
223 |
} |
224 |
|
225 |
if ($errors) { |
226 |
$log->debug("errors: ", sub { dump( $errors ) } ); |
227 |
|
228 |
my $mfn = $rec->{'000'}->[0]; |
229 |
$self->{errors}->{$mfn} = $errors; |
230 |
} |
231 |
|
232 |
#$log->logcluck("return from this function is ARRAY") unless wantarray; |
233 |
|
234 |
return $errors; |
235 |
} |
236 |
|
237 |
=head1 AUTHOR |
238 |
|
239 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
240 |
|
241 |
=head1 COPYRIGHT & LICENSE |
242 |
|
243 |
Copyright 2006 Dobrica Pavlinusic, All Rights Reserved. |
244 |
|
245 |
This program is free software; you can redistribute it and/or modify it |
246 |
under the same terms as Perl itself. |
247 |
|
248 |
=cut |
249 |
|
250 |
1; # End of WebPAC::Validate |