53 |
my $validate = new WebPAC::Validate( |
my $validate = new WebPAC::Validate( |
54 |
path => 'conf/validate/file', |
path => 'conf/validate/file', |
55 |
delimiters => [ ' : ', ' / ', ' ; ', ' , ' ], |
delimiters => [ ' : ', ' / ', ' ; ', ' , ' ], |
56 |
|
delimiters_path => 'conf/validate/delimiters/file', |
57 |
); |
); |
58 |
|
|
59 |
Optional parametar C<delimiters> will turn on validating of delimiters. Be |
Optional parametar C<delimiters> will turn on validating of delimiters. Be |
69 |
|
|
70 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
71 |
|
|
72 |
foreach my $p (qw/path/) { |
$log->logdie("need path or delimiters_path") unless ( $self->{path} || $self->{delimiters_path} ); |
|
$log->logconfess("need $p") unless ($self->{$p}); |
|
|
} |
|
73 |
|
|
74 |
my $v_file = read_file( $self->{path} ) || |
if ( $self->{path} ) { |
|
$log->logdie("can't open validate path $self->{path}: $!"); |
|
75 |
|
|
76 |
my $v; |
my $v_file = read_file( $self->{path} ) || |
77 |
my $curr_line = 1; |
$log->logdie("can't open validate path $self->{path}: $!"); |
78 |
|
|
79 |
foreach my $l (split(/[\n\r]+/, $v_file)) { |
my $v; |
80 |
$curr_line++; |
my $curr_line = 1; |
81 |
|
|
82 |
# skip comments and whitespaces |
foreach my $l (split(/[\n\r]+/, $v_file)) { |
83 |
next if ($l =~ /^#/ || $l =~ /^\s*$/); |
$curr_line++; |
84 |
|
|
85 |
$l =~ s/^\s+//; |
# skip comments and whitespaces |
86 |
$l =~ s/\s+$//; |
next if ($l =~ /^#/ || $l =~ /^\s*$/); |
87 |
|
|
88 |
my @d = split(/\s+/, $l); |
$l =~ s/^\s+//; |
89 |
|
$l =~ s/\s+$//; |
90 |
|
|
91 |
my $fld = shift @d; |
my @d = split(/\s+/, $l); |
92 |
|
|
93 |
if ($fld =~ s/!$//) { |
my $fld = shift @d; |
|
$self->{must_exist}->{$fld}++; |
|
|
} elsif ($fld =~ s/-$//) { |
|
|
$self->{dont_validate}->{$fld}++; |
|
|
} |
|
94 |
|
|
95 |
$log->logdie("need field name in line $curr_line: $l") unless (defined($fld)); |
if ($fld =~ s/!$//) { |
96 |
|
$self->{must_exist}->{$fld}++; |
97 |
|
} elsif ($fld =~ s/-$//) { |
98 |
|
$self->{dont_validate}->{$fld}++; |
99 |
|
} |
100 |
|
|
101 |
if (@d) { |
$log->logdie("need field name in line $curr_line: $l") unless (defined($fld)); |
|
$v->{$fld} = [ map { |
|
|
my $sf = $_; |
|
|
if ( $sf =~ s/!(\*)?$/$1/ ) { |
|
|
$self->{must_exist_sf}->{ $fld }->{ $sf }++; |
|
|
}; |
|
|
$sf; |
|
|
} @d ]; |
|
|
} else { |
|
|
$v->{$fld} = 1; |
|
|
} |
|
102 |
|
|
103 |
} |
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 |
|
} |
116 |
|
|
117 |
$log->debug("current validation rules: ", dump($v)); |
$log->debug("current validation rules: ", dump($v)); |
118 |
|
|
119 |
$self->{rules} = $v; |
$self->{rules} = $v; |
120 |
|
|
121 |
$log->info("validation uses rules from $self->{path}"); |
$log->info("validation uses rules from $self->{path}"); |
122 |
|
} |
123 |
|
|
124 |
if ( $self->{delimiters} ) { |
if ( $self->{delimiters} ) { |
125 |
$self->{delimiters_regex} = '(\^[a-z0-9]|' . join('|', @{ $self->{delimiters} }) . ')'; |
$self->{delimiters_regex} = '(\^[a-z0-9]|' . join('|', @{ $self->{delimiters} }) . ')'; |
126 |
$log->info("validation check delimiters with regex $self->{delimiters_regex}"); |
$log->info("validation check delimiters with regex $self->{delimiters_regex}"); |
127 |
} |
} |
128 |
|
|
129 |
|
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 |
|
#warn "_validate_delimiters_templates = ",dump( $self->{_validate_delimiters_templates} ); |
144 |
|
} else { |
145 |
|
$log->warn("delimiters path $path doesn't exist, it will be created after this run"); |
146 |
|
} |
147 |
|
} |
148 |
|
|
149 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
150 |
} |
} |
151 |
|
|
166 |
my $rec_dump = shift; |
my $rec_dump = shift; |
167 |
|
|
168 |
$log->logdie("rec isn't HASH") unless (ref($rec) eq 'HASH'); |
$log->logdie("rec isn't HASH") unless (ref($rec) eq 'HASH'); |
169 |
$log->logdie("can't find validation rules") unless (my $r = $self->{rules}); |
# $log->logdie("can't find validation rules") unless (my $r = $self->{rules}); |
170 |
|
my $r = $self->{rules}; |
171 |
|
|
172 |
my $errors; |
my $errors; |
173 |
|
|
183 |
if ( my $regex = $self->{delimiters_regex} ) { |
if ( my $regex = $self->{delimiters_regex} ) { |
184 |
|
|
185 |
foreach my $v (@{ $rec->{$f} }) { |
foreach my $v (@{ $rec->{$f} }) { |
186 |
my $l = _pack_subfields_hash( $v, 1 ); |
my $l = _pack_subfields_hash( $v, 1 ); |
187 |
my $template = ''; |
my $subfield_dump = $l; |
188 |
$l =~ s/$regex/$template.=$1/eg; |
my $template = ''; |
189 |
# warn "## template: $template\n"; |
$l =~ s/$regex/$template.=$1/eg; |
190 |
$self->{_delimiters_templates}->{$f}->{$template}++ if ( $template ); |
#warn "## template: $template\n"; |
191 |
|
|
192 |
|
if ( $template ) { |
193 |
|
$self->{_delimiters_templates}->{$f}->{$template}++; |
194 |
|
|
195 |
|
if ( my $v = $self->{_validate_delimiters_templates} ) { |
196 |
|
if ( ! defined( $v->{$f}->{$template} ) ) { |
197 |
|
$errors->{$f}->{invalid_delimiters_combination} = $template; |
198 |
|
$errors->{$f}->{dump} = $subfield_dump; |
199 |
|
#} else { |
200 |
|
# warn "## $f $template ok\n"; |
201 |
|
} |
202 |
|
} |
203 |
|
} |
204 |
} |
} |
|
|
|
205 |
} |
} |
206 |
|
|
207 |
|
next unless ( $r ); # skip validation of no rules are specified |
208 |
|
|
209 |
next if (defined( $self->{dont_validate}->{$f} )); |
next if (defined( $self->{dont_validate}->{$f} )); |
210 |
|
|
211 |
# track field usage |
# track field usage |
363 |
|
|
364 |
if ($k eq 'dump') { |
if ($k eq 'dump') { |
365 |
$dump = $tree->{dump}; |
$dump = $tree->{dump}; |
366 |
# warn "## dump: ",dump($dump),"\n"; |
#warn "## dump ",dump($dump),"\n"; |
367 |
next; |
next; |
368 |
} |
} |
369 |
|
|
397 |
sub _reformat { |
sub _reformat { |
398 |
my $l = shift; |
my $l = shift; |
399 |
$l =~ s/\t/ /g; |
$l =~ s/\t/ /g; |
400 |
$l =~ s/_/ /; |
$l =~ s/_/ /g; |
401 |
return $l; |
return $l; |
402 |
} |
} |
403 |
|
|
484 |
my $count = $t->{$f}->{$template}; |
my $count = $t->{$f}->{$template}; |
485 |
$out .= |
$out .= |
486 |
( $count ? "" : "# " ) . |
( $count ? "" : "# " ) . |
487 |
( $args->{report} ? "" : "$f\t" ) . |
( $args->{report} ? "" : "$f" ) . |
488 |
"\t$count\t$template\n"; |
"\t$count\t$template\n"; |
489 |
} |
} |
490 |
} |
} |
492 |
return $out; |
return $out; |
493 |
} |
} |
494 |
|
|
495 |
|
=head2 save_delimiters_templates |
496 |
|
|
497 |
|
=cut |
498 |
|
|
499 |
|
sub save_delimiters_templates { |
500 |
|
my $self = shift; |
501 |
|
|
502 |
|
my $path = $self->{delimiters_path}; |
503 |
|
|
504 |
|
return unless ( $path ); |
505 |
|
|
506 |
|
my $log = $self->_get_logger; |
507 |
|
|
508 |
|
$path .= '.new' if ( -e $path ); |
509 |
|
|
510 |
|
open(my $d, '>', $path) || $log->fatal("can't open $path: $!"); |
511 |
|
print $d $self->delimiters_templates; |
512 |
|
close($d); |
513 |
|
|
514 |
|
$log->info("new delimiters templates saved to $path"); |
515 |
|
} |
516 |
|
|
517 |
=head1 AUTHOR |
=head1 AUTHOR |
518 |
|
|
519 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |