18 |
|
|
19 |
=head1 VERSION |
=head1 VERSION |
20 |
|
|
21 |
Version 0.11 |
Version 0.12 |
22 |
|
|
23 |
=cut |
=cut |
24 |
|
|
25 |
our $VERSION = '0.11'; |
our $VERSION = '0.12'; |
26 |
|
|
27 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
28 |
|
|
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}->{potentially_invalid_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 |
282 |
} |
} |
283 |
} |
} |
284 |
|
|
285 |
$log->debug("_delimiters_templates = ", dump( $self->{_delimiters_templates} ) ); |
$log->debug("_delimiters_templates = ", sub { dump( $self->{_delimiters_templates} ) } ); |
286 |
|
|
287 |
foreach my $must (sort keys %{ $self->{must_exist} }) { |
foreach my $must (sort keys %{ $self->{must_exist} }) { |
288 |
next if ($fields->{$must}); |
next if ($fields->{$must}); |
293 |
if ($errors) { |
if ($errors) { |
294 |
$log->debug("errors: ", $self->report_error( $errors ) ); |
$log->debug("errors: ", $self->report_error( $errors ) ); |
295 |
|
|
296 |
my $mfn = $rec->{'000'}->[0] || $log->logconfess("record ", dump( $rec ), " doesn't have MFN"); |
my $mfn = $rec->{'000'}->[0] || $log->logconfess("record ", sub { dump( $rec ) }, " doesn't have MFN"); |
297 |
$self->{errors}->{$mfn} = $errors; |
$self->{errors}->{$mfn} = $errors; |
298 |
} |
} |
299 |
|
|
302 |
return $errors; |
return $errors; |
303 |
} |
} |
304 |
|
|
305 |
=head2 reset_errors |
=head2 reset |
306 |
|
|
307 |
Clean all accumulated errors for this input |
Clean all accumulated errors for this input and remember delimiter templates |
308 |
|
for L<save_delimiters_templates> |
309 |
|
|
310 |
$validate->reset_errors; |
$validate->reset; |
311 |
|
|
312 |
|
This function B<must> be called after each input to provide accurate statistics. |
313 |
|
|
314 |
=cut |
=cut |
315 |
|
|
316 |
sub reset_errors { |
sub reset { |
317 |
my $self = shift; |
my $self = shift; |
318 |
|
|
319 |
|
my $log = $self->_get_logger; |
320 |
|
|
321 |
delete ($self->{errors}); |
delete ($self->{errors}); |
322 |
|
|
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 = ", sub { dump( $self->{_accumulated_delimiter_templates} ) } ); |
335 |
|
delete ($self->{_delimiters_templates}); |
336 |
} |
} |
337 |
|
|
338 |
=head2 all_errors |
=head2 all_errors |
383 |
|
|
384 |
if ($k eq 'dump') { |
if ($k eq 'dump') { |
385 |
$dump = $tree->{dump}; |
$dump = $tree->{dump}; |
386 |
# warn "## dump: ",dump($dump),"\n"; |
#warn "## dump ",dump($dump),"\n"; |
387 |
next; |
next; |
388 |
} |
} |
389 |
|
|
393 |
$accumulated ? "$accumulated\t$k" : $k |
$accumulated ? "$accumulated\t$k" : $k |
394 |
); |
); |
395 |
|
|
396 |
$log->debug( |
$log->debug( "new_results: ", sub { dump($new_results) } ) if ( $new_results ); |
|
( $new_results ? "new_results: " . dump($new_results) ." " : '' ), |
|
|
); |
|
397 |
|
|
398 |
push @$results, $new_results if ($new_results); |
push @$results, $new_results if ($new_results); |
399 |
$dump = $new_dump if ($new_dump); |
$dump = $new_dump if ($new_dump); |
400 |
|
|
401 |
} |
} |
402 |
|
|
403 |
$log->debug( |
$log->debug( "results: ", sub { dump($results) } ) if ( $results ); |
|
( $results ? "results: " . dump($results) ." " : '' ), |
|
|
); |
|
404 |
|
|
405 |
if ($#$results == 0) { |
if ($#$results == 0) { |
406 |
return ($results->[0], $dump); |
return ($results->[0], $dump); |
413 |
sub _reformat { |
sub _reformat { |
414 |
my $l = shift; |
my $l = shift; |
415 |
$l =~ s/\t/ /g; |
$l =~ s/\t/ /g; |
416 |
$l =~ s/_/ /; |
$l =~ s/_/ /g; |
417 |
return $l; |
return $l; |
418 |
} |
} |
419 |
|
|
460 |
|
|
461 |
=head2 delimiters_templates |
=head2 delimiters_templates |
462 |
|
|
463 |
|
Generate report of delimiter tamplates |
464 |
|
|
465 |
|
my $report = $validate->delimiter_teplates( |
466 |
|
report => 1, |
467 |
|
current_input => 1, |
468 |
|
); |
469 |
|
|
470 |
|
Options: |
471 |
|
|
472 |
|
=over 4 |
473 |
|
|
474 |
|
=item report |
475 |
|
|
476 |
|
Generate humanly readable report with single fields |
477 |
|
|
478 |
|
=item current_input |
479 |
|
|
480 |
|
Report just current_input and not accumulated data |
481 |
|
|
482 |
|
=back |
483 |
|
|
484 |
=cut |
=cut |
485 |
|
|
486 |
sub delimiters_templates { |
sub delimiters_templates { |
487 |
my $self = shift; |
my $self = shift; |
488 |
|
|
489 |
my $t = $self->{_delimiters_templates}; |
my $args = {@_}; |
490 |
|
|
491 |
|
my $t = $self->{_accumulated_delimiters_templates}; |
492 |
|
$t = $self->{_delimiters_templates} if ( $args->{current_input} ); |
493 |
|
|
494 |
my $log = $self->_get_logger; |
my $log = $self->_get_logger; |
495 |
|
|
496 |
unless ($t) { |
unless ($t) { |
497 |
$log->warn("called without delimiters"); |
$log->error("called without delimiters"); |
498 |
return; |
return; |
499 |
} |
} |
500 |
|
|
501 |
my $out; |
my $out; |
502 |
|
|
503 |
foreach my $f (sort { $a <=> $b } keys %$t) { |
foreach my $f (sort { $a <=> $b } keys %$t) { |
504 |
$out .= "$f\n"; |
$out .= "$f\n" if ( $args->{report} ); |
505 |
foreach my $sft (sort { $a cmp $b } keys %{ $t->{$f} }) { |
foreach my $template (sort { $a cmp $b } keys %{ $t->{$f} }) { |
506 |
$out .= "\t" . $t->{$f}->{$sft} . "\t$sft\n"; |
my $count = $t->{$f}->{$template}; |
507 |
|
$out .= |
508 |
|
( $count ? "" : "# " ) . |
509 |
|
( $args->{report} ? "" : "$f" ) . |
510 |
|
"\t$count\t$template\n"; |
511 |
} |
} |
512 |
} |
} |
513 |
|
|
514 |
return $out; |
return $out; |
515 |
} |
} |
516 |
|
|
517 |
|
=head2 save_delimiters_templates |
518 |
|
|
519 |
|
Save accumulated delimiter templates |
520 |
|
|
521 |
|
$validator->save_delimiters_template( '/path/to/validate/delimiters' ); |
522 |
|
|
523 |
|
=cut |
524 |
|
|
525 |
|
sub save_delimiters_templates { |
526 |
|
my $self = shift; |
527 |
|
|
528 |
|
my $path = $self->{delimiters_path}; |
529 |
|
|
530 |
|
return unless ( $path ); |
531 |
|
|
532 |
|
my $log = $self->_get_logger; |
533 |
|
|
534 |
|
if ( ! $self->{_accumulated_delimiters_templates} ) { |
535 |
|
$log->error('no _accumulated_delimiters_templates found, reset'); |
536 |
|
$self->reset; |
537 |
|
} |
538 |
|
|
539 |
|
if ( ! $self->{_delimiters_templates} ) { |
540 |
|
$log->error('found _delimiters_templates, calling reset'); |
541 |
|
$self->reset; |
542 |
|
} |
543 |
|
|
544 |
|
$path .= '.new' if ( -e $path ); |
545 |
|
|
546 |
|
open(my $d, '>', $path) || $log->fatal("can't open $path: $!"); |
547 |
|
print $d $self->delimiters_templates; |
548 |
|
close($d); |
549 |
|
|
550 |
|
$log->info("new delimiters templates saved to $path"); |
551 |
|
} |
552 |
|
|
553 |
=head1 AUTHOR |
=head1 AUTHOR |
554 |
|
|
555 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |