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 { |
72 |
|
|
73 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
74 |
|
|
75 |
$log->logdie("need path or delimiters_path") unless ( $self->{path} || $self->{delimiters_path} ); |
$self->read_validate_file( $self->{path} ) if ( $self->{path} ); |
76 |
|
|
77 |
if ( $self->{path} ) { |
if ( $self->{delimiters} ) { |
78 |
|
$self->{delimiters_regex} = '(\^[a-z0-9]|' . join('|', @{ $self->{delimiters} }) . ')'; |
79 |
|
$log->info("validation check delimiters with regex $self->{delimiters_regex}"); |
80 |
|
} |
81 |
|
|
82 |
my $v_file = read_file( $self->{path} ) || |
$self->read_validate_delimiters_file( $self->{delimiters_path} ) if ( $self->{delimiters_path} ); |
|
$log->logdie("can't open validate path $self->{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: $!"); |
|
$sf; |
|
|
} @d ]; |
|
|
} else { |
|
|
$v->{$fld} = 1; |
|
|
} |
|
107 |
|
|
108 |
|
my $v; |
109 |
|
delete( $self->{must_exist} ); |
110 |
|
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 |
324 |
} |
} |
325 |
} |
} |
326 |
|
|
327 |
$log->debug("_delimiters_templates = ", dump( $self->{_delimiters_templates} ) ); |
$log->debug("_delimiters_templates = ", sub { dump( $self->{_delimiters_templates} ) } ); |
328 |
|
|
329 |
foreach my $must (sort keys %{ $self->{must_exist} }) { |
foreach my $must (sort keys %{ $self->{must_exist} }) { |
330 |
next if ($fields->{$must}); |
next if ($fields->{$must}); |
335 |
if ($errors) { |
if ($errors) { |
336 |
$log->debug("errors: ", $self->report_error( $errors ) ); |
$log->debug("errors: ", $self->report_error( $errors ) ); |
337 |
|
|
338 |
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"); |
339 |
$self->{errors}->{$mfn} = $errors; |
$self->{errors}->{$mfn} = $errors; |
340 |
} |
} |
341 |
|
|
373 |
$self->{_delimiters_templates}->{$f}->{$t}; |
$self->{_delimiters_templates}->{$f}->{$t}; |
374 |
} |
} |
375 |
} |
} |
376 |
$log->debug("_accumulated_delimiters_templates = ", dump( $self->{_accumulated_delimiter_templates} )); |
$log->debug("_accumulated_delimiters_templates = ", sub { dump( $self->{_accumulated_delimiters_templates} ) } ); |
377 |
delete ($self->{_delimiters_templates}); |
delete ($self->{_delimiters_templates}); |
378 |
} |
} |
379 |
|
|
435 |
$accumulated ? "$accumulated\t$k" : $k |
$accumulated ? "$accumulated\t$k" : $k |
436 |
); |
); |
437 |
|
|
438 |
$log->debug( |
$log->debug( "new_results: ", sub { dump($new_results) } ) if ( $new_results ); |
|
( $new_results ? "new_results: " . dump($new_results) ." " : '' ), |
|
|
); |
|
439 |
|
|
440 |
push @$results, $new_results if ($new_results); |
push @$results, $new_results if ($new_results); |
441 |
$dump = $new_dump if ($new_dump); |
$dump = $new_dump if ($new_dump); |
442 |
|
|
443 |
} |
} |
444 |
|
|
445 |
$log->debug( |
$log->debug( "results: ", sub { dump($results) } ) if ( $results ); |
|
( $results ? "results: " . dump($results) ." " : '' ), |
|
|
); |
|
446 |
|
|
447 |
if ($#$results == 0) { |
if ($#$results == 0) { |
448 |
return ($results->[0], $dump); |
return ($results->[0], $dump); |
506 |
|
|
507 |
my $report = $validate->delimiter_teplates( |
my $report = $validate->delimiter_teplates( |
508 |
report => 1, |
report => 1, |
509 |
accumulated => 1, |
current_input => 1, |
510 |
); |
); |
511 |
|
|
512 |
Options: |
Options: |
517 |
|
|
518 |
Generate humanly readable report with single fields |
Generate humanly readable report with single fields |
519 |
|
|
520 |
=item accumulated |
=item current_input |
521 |
|
|
522 |
Use accumulated data from all inputs |
Report just current_input and not accumulated data |
523 |
|
|
524 |
=back |
=back |
525 |
|
|
530 |
|
|
531 |
my $args = {@_}; |
my $args = {@_}; |
532 |
|
|
533 |
my $t; |
my $t = $self->{_accumulated_delimiters_templates}; |
534 |
if ( $args->{accumulated} ) { |
$t = $self->{_delimiters_templates} if ( $args->{current_input} ); |
|
$t = $self->{_accumulated_delimiters_templates}; |
|
|
} else { |
|
|
$t = $self->{_delimiters_templates}; |
|
|
} |
|
535 |
|
|
536 |
my $log = $self->_get_logger; |
my $log = $self->_get_logger; |
537 |
|
|
567 |
sub save_delimiters_templates { |
sub save_delimiters_templates { |
568 |
my $self = shift; |
my $self = shift; |
569 |
|
|
570 |
my $path = $self->{delimiters_path}; |
my $path = shift; |
571 |
|
$path ||= $self->{delimiters_path}; |
|
return unless ( $path ); |
|
572 |
|
|
573 |
my $log = $self->_get_logger; |
my $log = $self->_get_logger; |
574 |
|
|
575 |
|
$log->logdie("need path") unless ( $path ); |
576 |
|
|
577 |
|
|
578 |
if ( ! $self->{_accumulated_delimiters_templates} ) { |
if ( ! $self->{_accumulated_delimiters_templates} ) { |
579 |
$log->error('no _accumulated_delimiters_templates found, reset'); |
$log->error('no _accumulated_delimiters_templates found, reset'); |
580 |
$self->reset; |
$self->reset; |
581 |
} |
} |
582 |
|
|
583 |
if ( ! $self->{_delimiters_templates} ) { |
if ( $self->{_delimiters_templates} ) { |
584 |
$log->error('found _delimiters_templates, calling reset'); |
$log->error('found _delimiters_templates, calling reset'); |
585 |
$self->reset; |
$self->reset; |
586 |
} |
} |
588 |
$path .= '.new' if ( -e $path ); |
$path .= '.new' if ( -e $path ); |
589 |
|
|
590 |
open(my $d, '>', $path) || $log->fatal("can't open $path: $!"); |
open(my $d, '>', $path) || $log->fatal("can't open $path: $!"); |
591 |
print $d $self->delimiters_templates( accumulated => 1 ); |
print $d $self->delimiters_templates; |
592 |
close($d); |
close($d); |
593 |
|
|
594 |
$log->info("new delimiters templates saved to $path"); |
$log->info("new delimiters templates saved to $path"); |