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 |
|
|
60 |
careful here, those delimiters are just stuck into regex, so they can |
careful here, those delimiters are just stuck into regex, so they can |
61 |
contain L<perlre> regexpes. |
contain L<perlre> regexpes. |
62 |
|
|
63 |
|
C<path> and C<delimiters_path> can be specified by L<read_validate_file> and |
64 |
|
L<read_validate_delimiters> calls. |
65 |
|
|
66 |
=cut |
=cut |
67 |
|
|
68 |
sub new { |
sub new { |
70 |
my $self = {@_}; |
my $self = {@_}; |
71 |
bless($self, $class); |
bless($self, $class); |
72 |
|
|
|
warn dump( @_ ); |
|
|
|
|
73 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
74 |
|
|
75 |
if ( $self->{path} ) { |
$self->read_validate_file( $self->{path} ) if ( $self->{path} ); |
76 |
|
|
77 |
my $v_file = read_file( $self->{path} ) || |
if ( $self->{delimiters} ) { |
78 |
$log->logdie("can't open validate path $self->{path}: $!"); |
$self->{delimiters_regex} = '(\^[a-z0-9]|' . join('|', @{ $self->{delimiters} }) . ')'; |
79 |
|
$log->info("validation check delimiters with regex $self->{delimiters_regex}"); |
80 |
|
} |
81 |
|
|
82 |
|
$self->read_validate_delimiters_file( $self->{delimiters_path} ) if ( $self->{delimiters_path} ); |
83 |
|
|
84 |
my $v; |
return $self; |
85 |
my $curr_line = 1; |
} |
86 |
|
|
|
foreach my $l (split(/[\n\r]+/, $v_file)) { |
|
|
$curr_line++; |
|
87 |
|
|
88 |
# skip comments and whitespaces |
=head2 read_validate_file |
|
next if ($l =~ /^#/ || $l =~ /^\s*$/); |
|
89 |
|
|
90 |
$l =~ s/^\s+//; |
Specify validate rules file |
|
$l =~ s/\s+$//; |
|
91 |
|
|
92 |
my @d = split(/\s+/, $l); |
$validate->read_validate_file( 'conf/validate/file' ); |
93 |
|
|
94 |
my $fld = shift @d; |
Returns number of lines in file |
95 |
|
|
96 |
if ($fld =~ s/!$//) { |
=cut |
|
$self->{must_exist}->{$fld}++; |
|
|
} elsif ($fld =~ s/-$//) { |
|
|
$self->{dont_validate}->{$fld}++; |
|
|
} |
|
97 |
|
|
98 |
$log->logdie("need field name in line $curr_line: $l") unless (defined($fld)); |
sub read_validate_file { |
99 |
|
my $self = shift; |
100 |
|
|
101 |
if (@d) { |
my $path = shift || die "no path?"; |
102 |
$v->{$fld} = [ map { |
|
103 |
my $sf = $_; |
my $log = $self->_get_logger(); |
104 |
if ( $sf =~ s/!(\*)?$/$1/ ) { |
|
105 |
$self->{must_exist_sf}->{ $fld }->{ $sf }++; |
my $v_file = read_file( $path ) || |
106 |
}; |
$log->logdie("can't open validate path $path: $!"); |
107 |
$sf; |
|
108 |
} @d ]; |
my $v; |
109 |
} else { |
delete( $self->{must_exist} ); |
110 |
$v->{$fld} = 1; |
delete( $self->{must_exist_sf} ); |
111 |
} |
delete( $self->{dont_validate} ); |
112 |
|
my $curr_line = 1; |
113 |
|
|
114 |
|
foreach my $l (split(/[\n\r]+/, $v_file)) { |
115 |
|
$curr_line++; |
116 |
|
|
117 |
|
# skip comments and whitespaces |
118 |
|
next if ($l =~ /^#/ || $l =~ /^\s*$/); |
119 |
|
|
120 |
|
$l =~ s/^\s+//; |
121 |
|
$l =~ s/\s+$//; |
122 |
|
|
123 |
|
my @d = split(/\s+/, $l); |
124 |
|
|
125 |
|
my $fld = shift @d; |
126 |
|
|
127 |
|
if ($fld =~ s/!$//) { |
128 |
|
$self->{must_exist}->{$fld}++; |
129 |
|
} elsif ($fld =~ s/-$//) { |
130 |
|
$self->{dont_validate}->{$fld}++; |
131 |
} |
} |
132 |
|
|
133 |
$log->debug("current validation rules: ", dump($v)); |
$log->logdie("need field name in line $curr_line: $l") unless (defined($fld)); |
134 |
|
|
135 |
$self->{rules} = $v; |
if (@d) { |
136 |
|
$v->{$fld} = [ map { |
137 |
|
my $sf = $_; |
138 |
|
if ( $sf =~ s/!(\*)?$/$1/ ) { |
139 |
|
$self->{must_exist_sf}->{ $fld }->{ $sf }++; |
140 |
|
}; |
141 |
|
$sf; |
142 |
|
} @d ]; |
143 |
|
} else { |
144 |
|
$v->{$fld} = 1; |
145 |
|
} |
146 |
|
|
|
$log->info("validation uses rules from $self->{path}"); |
|
147 |
} |
} |
148 |
|
|
149 |
if ( $self->{delimiters} ) { |
$log->debug("current validation rules: ", dump($v)); |
|
$self->{delimiters_regex} = '(\^[a-z0-9]|' . join('|', @{ $self->{delimiters} }) . ')'; |
|
|
$log->info("validation check delimiters with regex $self->{delimiters_regex}"); |
|
|
} |
|
150 |
|
|
151 |
if ( my $path = $self->{delimiters_path} ) { |
$self->{rules} = $v; |
152 |
if ( -e $path ) { |
|
153 |
$log->info("using delimiter validation rules from $path"); |
$log->info("validation uses rules from $path"); |
154 |
open(my $d, $path) || $log->fatal("can't open $path: $!"); |
|
155 |
while(<$d>) { |
return $curr_line; |
156 |
chomp($d); |
} |
157 |
if (/^\s*(#*)\s*(\d+)\t+(\d+)\t+(.*)$/) { |
|
158 |
my ($comment,$field,$count,$template) = ($1,$2,$3,$4); |
=head2 read_validate_delimiters_file |
159 |
$self->{_validate_delimiters_templates}->{$field}->{$template} = $count unless ($comment); |
|
160 |
} else { |
$validate->read_validate_delimiters_file( 'conf/validate/delimiters/file' ); |
161 |
warn "## ignored $d\n"; |
|
162 |
} |
=cut |
163 |
|
|
164 |
|
sub read_validate_delimiters_file { |
165 |
|
my $self = shift; |
166 |
|
|
167 |
|
my $path = shift || die "no path?"; |
168 |
|
|
169 |
|
my $log = $self->_get_logger(); |
170 |
|
|
171 |
|
delete( $self->{_validate_delimiters_templates} ); |
172 |
|
delete( $self->{_delimiters_templates} ); |
173 |
|
|
174 |
|
if ( -e $path ) { |
175 |
|
$log->info("using delimiter validation rules from $path"); |
176 |
|
open(my $d, $path) || $log->fatal("can't open $path: $!"); |
177 |
|
while(<$d>) { |
178 |
|
chomp($d); |
179 |
|
if (/^\s*(#*)\s*(\d+)\t+(\d+)\t+(.*)$/) { |
180 |
|
my ($comment,$field,$count,$template) = ($1,$2,$3,$4); |
181 |
|
$self->{_validate_delimiters_templates}->{$field}->{$template} = $count unless ($comment); |
182 |
|
} else { |
183 |
|
warn "## ignored $d\n"; |
184 |
} |
} |
|
close($d); |
|
|
warn "_validate_delimiters_templates = ",dump( $self->{_validate_delimiters_templates} ); |
|
|
} else { |
|
|
$log->warn("delimiters path $path doesn't exist, it will be created after this run"); |
|
185 |
} |
} |
186 |
|
close($d); |
187 |
|
#warn "_validate_delimiters_templates = ",dump( $self->{_validate_delimiters_templates} ); |
188 |
|
} else { |
189 |
|
$log->warn("delimiters path $path doesn't exist, it will be created after this run"); |
190 |
} |
} |
191 |
|
$self->{delimiters_path} = $path; |
|
$self ? return $self : return undef; |
|
192 |
} |
} |
193 |
|
|
194 |
=head2 validate_rec |
=head2 validate_rec |
236 |
|
|
237 |
if ( my $v = $self->{_validate_delimiters_templates} ) { |
if ( my $v = $self->{_validate_delimiters_templates} ) { |
238 |
if ( ! defined( $v->{$f}->{$template} ) ) { |
if ( ! defined( $v->{$f}->{$template} ) ) { |
239 |
$errors->{$f}->{invalid_delimiters_combination} = $template; |
$errors->{$f}->{potentially_invalid_combination} = $template; |
240 |
$errors->{$f}->{dump} = $subfield_dump; |
$errors->{$f}->{dump} = $subfield_dump; |
241 |
#} else { |
#} else { |
242 |
# warn "## $f $template ok\n"; |
# warn "## $f $template ok\n"; |
273 |
# $errors->{dump} = $rec_dump if ($rec_dump); |
# $errors->{dump} = $rec_dump if ($rec_dump); |
274 |
} elsif (ref($v) ne 'HASH') { |
} elsif (ref($v) ne 'HASH') { |
275 |
$errors->{$f}->{missing_subfield} = join(",", @{ $r->{$f} }) . " required"; |
$errors->{$f}->{missing_subfield} = join(",", @{ $r->{$f} }) . " required"; |
276 |
|
$errors->{$f}->{dump} = $v; |
277 |
next; |
next; |
278 |
} else { |
} else { |
279 |
|
|
325 |
} |
} |
326 |
} |
} |
327 |
|
|
328 |
$log->debug("_delimiters_templates = ", dump( $self->{_delimiters_templates} ) ); |
$log->debug("_delimiters_templates = ", sub { dump( $self->{_delimiters_templates} ) } ); |
329 |
|
|
330 |
foreach my $must (sort keys %{ $self->{must_exist} }) { |
foreach my $must (sort keys %{ $self->{must_exist} }) { |
331 |
next if ($fields->{$must}); |
next if ($fields->{$must}); |
336 |
if ($errors) { |
if ($errors) { |
337 |
$log->debug("errors: ", $self->report_error( $errors ) ); |
$log->debug("errors: ", $self->report_error( $errors ) ); |
338 |
|
|
339 |
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"); |
340 |
$self->{errors}->{$mfn} = $errors; |
$self->{errors}->{$mfn} = $errors; |
341 |
} |
} |
342 |
|
|
345 |
return $errors; |
return $errors; |
346 |
} |
} |
347 |
|
|
348 |
=head2 reset_errors |
=head2 reset |
349 |
|
|
350 |
|
Clean all accumulated errors for this input and remember delimiter templates |
351 |
|
for L<save_delimiters_templates> |
352 |
|
|
353 |
Clean all accumulated errors for this input |
$validate->reset; |
354 |
|
|
355 |
$validate->reset_errors; |
This function B<must> be called after each input to provide accurate statistics. |
356 |
|
|
357 |
=cut |
=cut |
358 |
|
|
359 |
sub reset_errors { |
sub reset { |
360 |
my $self = shift; |
my $self = shift; |
361 |
|
|
362 |
|
my $log = $self->_get_logger; |
363 |
|
|
364 |
delete ($self->{errors}); |
delete ($self->{errors}); |
365 |
|
|
366 |
|
if ( ! $self->{_delimiters_templates} ) { |
367 |
|
$log->debug("called without _delimiters_templates?"); |
368 |
|
return; |
369 |
|
} |
370 |
|
|
371 |
|
foreach my $f ( keys %{ $self->{_delimiters_templates} } ) { |
372 |
|
foreach my $t ( keys %{ $self->{_delimiters_templates}->{$f} } ) { |
373 |
|
$self->{_accumulated_delimiters_templates}->{$f}->{$t} += |
374 |
|
$self->{_delimiters_templates}->{$f}->{$t}; |
375 |
|
} |
376 |
|
} |
377 |
|
$log->debug("_accumulated_delimiters_templates = ", sub { dump( $self->{_accumulated_delimiters_templates} ) } ); |
378 |
|
delete ($self->{_delimiters_templates}); |
379 |
} |
} |
380 |
|
|
381 |
=head2 all_errors |
=head2 all_errors |
436 |
$accumulated ? "$accumulated\t$k" : $k |
$accumulated ? "$accumulated\t$k" : $k |
437 |
); |
); |
438 |
|
|
439 |
$log->debug( |
$log->debug( "new_results: ", sub { dump($new_results) } ) if ( $new_results ); |
|
( $new_results ? "new_results: " . dump($new_results) ." " : '' ), |
|
|
); |
|
440 |
|
|
441 |
push @$results, $new_results if ($new_results); |
push @$results, $new_results if ($new_results); |
442 |
$dump = $new_dump if ($new_dump); |
$dump = $new_dump if ($new_dump); |
443 |
|
|
444 |
} |
} |
445 |
|
|
446 |
$log->debug( |
$log->debug( "results: ", sub { dump($results) } ) if ( $results ); |
|
( $results ? "results: " . dump($results) ." " : '' ), |
|
|
); |
|
447 |
|
|
448 |
if ($#$results == 0) { |
if ($#$results == 0) { |
449 |
return ($results->[0], $dump); |
return ($results->[0], $dump); |
507 |
|
|
508 |
my $report = $validate->delimiter_teplates( |
my $report = $validate->delimiter_teplates( |
509 |
report => 1, |
report => 1, |
510 |
|
current_input => 1, |
511 |
); |
); |
512 |
|
|
513 |
Options: |
Options: |
518 |
|
|
519 |
Generate humanly readable report with single fields |
Generate humanly readable report with single fields |
520 |
|
|
521 |
|
=item current_input |
522 |
|
|
523 |
|
Report just current_input and not accumulated data |
524 |
|
|
525 |
=back |
=back |
526 |
|
|
527 |
=cut |
=cut |
531 |
|
|
532 |
my $args = {@_}; |
my $args = {@_}; |
533 |
|
|
534 |
my $t = $self->{_delimiters_templates}; |
my $t = $self->{_accumulated_delimiters_templates}; |
535 |
|
$t = $self->{_delimiters_templates} if ( $args->{current_input} ); |
536 |
|
|
537 |
my $log = $self->_get_logger; |
my $log = $self->_get_logger; |
538 |
|
|
559 |
|
|
560 |
=head2 save_delimiters_templates |
=head2 save_delimiters_templates |
561 |
|
|
562 |
|
Save accumulated delimiter templates |
563 |
|
|
564 |
|
$validator->save_delimiters_template( '/path/to/validate/delimiters' ); |
565 |
|
|
566 |
=cut |
=cut |
567 |
|
|
568 |
sub save_delimiters_templates { |
sub save_delimiters_templates { |
569 |
my $self = shift; |
my $self = shift; |
570 |
|
|
571 |
my $path = $self->{delimiters_path}; |
my $path = shift; |
572 |
|
$path ||= $self->{delimiters_path}; |
|
return unless ( $path ); |
|
573 |
|
|
574 |
my $log = $self->_get_logger; |
my $log = $self->_get_logger; |
575 |
|
|
576 |
|
$log->logdie("need path") unless ( $path ); |
577 |
|
|
578 |
|
|
579 |
|
if ( ! $self->{_accumulated_delimiters_templates} ) { |
580 |
|
$log->error('no _accumulated_delimiters_templates found, reset'); |
581 |
|
$self->reset; |
582 |
|
} |
583 |
|
|
584 |
|
if ( $self->{_delimiters_templates} ) { |
585 |
|
$log->error('found _delimiters_templates, calling reset'); |
586 |
|
$self->reset; |
587 |
|
} |
588 |
|
|
589 |
|
$path .= '.new' if ( -e $path ); |
590 |
|
|
591 |
open(my $d, '>', $path) || $log->fatal("can't open $path: $!"); |
open(my $d, '>', $path) || $log->fatal("can't open $path: $!"); |
592 |
print $d $self->delimiters_templates; |
print $d $self->delimiters_templates; |
593 |
close($d); |
close($d); |
594 |
|
|
595 |
$log->info("new delimiters templates saved to $path"); |
$log->info("new delimiters templates saved to $path"); |
596 |
|
|
597 |
|
return 1; |
598 |
} |
} |
599 |
|
|
600 |
=head1 AUTHOR |
=head1 AUTHOR |