18 |
|
|
19 |
=head1 VERSION |
=head1 VERSION |
20 |
|
|
21 |
Version 0.07 |
Version 0.11 |
22 |
|
|
23 |
=cut |
=cut |
24 |
|
|
25 |
our $VERSION = '0.07'; |
our $VERSION = '0.11'; |
26 |
|
|
27 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
28 |
|
|
41 |
205! a |
205! a |
42 |
# while 210 can have a c or d |
# while 210 can have a c or d |
43 |
210 a c d |
210 a c d |
44 |
|
# field which is ignored in validation |
45 |
|
999- |
46 |
|
|
47 |
=head1 FUNCTIONS |
=head1 FUNCTIONS |
48 |
|
|
52 |
|
|
53 |
my $validate = new WebPAC::Validate( |
my $validate = new WebPAC::Validate( |
54 |
path => 'conf/validate/file', |
path => 'conf/validate/file', |
55 |
|
delimiters => [ ' : ', ' / ', ' ; ', ' , ' ], |
56 |
); |
); |
57 |
|
|
58 |
|
Optional parametar C<delimiters> will turn on validating of delimiters. Be |
59 |
|
careful here, those delimiters are just stuck into regex, so they can |
60 |
|
contain L<perlre> regexpes. |
61 |
|
|
62 |
=cut |
=cut |
63 |
|
|
64 |
sub new { |
sub new { |
93 |
|
|
94 |
if ($fld =~ s/!$//) { |
if ($fld =~ s/!$//) { |
95 |
$self->{must_exist}->{$fld}++; |
$self->{must_exist}->{$fld}++; |
96 |
|
} elsif ($fld =~ s/-$//) { |
97 |
|
$self->{dont_validate}->{$fld}++; |
98 |
} |
} |
99 |
|
|
100 |
$log->logdie("need field name in line $curr_line: $l") unless (defined($fld)); |
$log->logdie("need field name in line $curr_line: $l") unless (defined($fld)); |
119 |
|
|
120 |
$log->info("validation uses rules from $self->{path}"); |
$log->info("validation uses rules from $self->{path}"); |
121 |
|
|
122 |
|
if ( $self->{delimiters} ) { |
123 |
|
$self->{delimiters_regex} = '(\^[a-z0-9]|' . join('|', @{ $self->{delimiters} }) . ')'; |
124 |
|
$log->info("validation check delimiters with regex $self->{delimiters_regex}"); |
125 |
|
} |
126 |
|
|
127 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
128 |
} |
} |
129 |
|
|
130 |
=head2 validate_errors |
=head2 validate_rec |
131 |
|
|
132 |
Validate record and return errors |
Validate record and return errors |
133 |
|
|
134 |
my @errors = $validate->validate_errors( $rec, $rec_dump ); |
my @errors = $validate->validate_rec( $rec, $rec_dump ); |
135 |
|
|
136 |
=cut |
=cut |
137 |
|
|
138 |
sub validate_errors { |
sub validate_rec { |
139 |
my $self = shift; |
my $self = shift; |
140 |
|
|
141 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
142 |
|
|
143 |
my $rec = shift || $log->logdie("validate_errors need record"); |
my $rec = shift || $log->logdie("validate_rec need record"); |
144 |
my $rec_dump = shift; |
my $rec_dump = shift; |
145 |
|
|
146 |
$log->logdie("rec isn't HASH") unless (ref($rec) eq 'HASH'); |
$log->logdie("rec isn't HASH") unless (ref($rec) eq 'HASH'); |
156 |
|
|
157 |
next if (!defined($f) || $f eq '' || $f eq '000'); |
next if (!defined($f) || $f eq '' || $f eq '000'); |
158 |
|
|
159 |
|
# first check delimiters |
160 |
|
if ( my $regex = $self->{delimiters_regex} ) { |
161 |
|
|
162 |
|
foreach my $v (@{ $rec->{$f} }) { |
163 |
|
my $l = _pack_subfields_hash( $v, 1 ); |
164 |
|
my $template = ''; |
165 |
|
$l =~ s/$regex/$template.=$1/eg; |
166 |
|
# warn "## template: $template\n"; |
167 |
|
$self->{_delimiters_templates}->{$f}->{$template}++ if ( $template ); |
168 |
|
} |
169 |
|
|
170 |
|
} |
171 |
|
|
172 |
|
next if (defined( $self->{dont_validate}->{$f} )); |
173 |
|
|
174 |
|
# track field usage |
175 |
$fields->{$f}++; |
$fields->{$f}++; |
176 |
|
|
177 |
if ( ! defined($r->{$f}) ) { |
if ( ! defined($r->{$f}) ) { |
178 |
$errors->{field}->{ $f }->{unexpected} = "this field is not expected"; |
$errors->{ $f }->{unexpected} = "this field is not expected"; |
179 |
next; |
next; |
180 |
} |
} |
181 |
|
|
182 |
|
|
183 |
if (ref($rec->{$f}) ne 'ARRAY') { |
if (ref($rec->{$f}) ne 'ARRAY') { |
184 |
$errors->{field}->{ $f }->{not_repeatable} = "probably bug in parsing input data"; |
$errors->{ $f }->{not_repeatable} = "probably bug in parsing input data"; |
185 |
next; |
next; |
186 |
} |
} |
187 |
|
|
190 |
if (ref($r->{$f}) eq 'ARRAY') { |
if (ref($r->{$f}) eq 'ARRAY') { |
191 |
# are values hashes? (has subfields) |
# are values hashes? (has subfields) |
192 |
if (! defined($v)) { |
if (! defined($v)) { |
193 |
# $errors->{field}->{$f}->{empty} = undef; |
# $errors->{$f}->{empty} = undef; |
194 |
# $errors->{dump} = $rec_dump if ($rec_dump); |
# $errors->{dump} = $rec_dump if ($rec_dump); |
195 |
} elsif (ref($v) ne 'HASH') { |
} elsif (ref($v) ne 'HASH') { |
196 |
$errors->{field}->{$f}->{missing_subfield} = join(",", @{ $r->{$f} }) . " required"; |
$errors->{$f}->{missing_subfield} = join(",", @{ $r->{$f} }) . " required"; |
197 |
next; |
next; |
198 |
} else { |
} else { |
199 |
|
|
215 |
$sf_repeatable->{$sf}++; |
$sf_repeatable->{$sf}++; |
216 |
}; |
}; |
217 |
if (! first { $_ eq $sf } @{ $r->{$f} }) { |
if (! first { $_ eq $sf } @{ $r->{$f} }) { |
218 |
$errors->{field}->{ $f }->{subfield}->{extra}->{$sf}++; |
$errors->{ $f }->{subfield}->{extra}->{$sf}++; |
219 |
} |
} |
220 |
} |
} |
221 |
|
|
223 |
if (my @r_sf = sort keys( %$sf_repeatable )) { |
if (my @r_sf = sort keys( %$sf_repeatable )) { |
224 |
|
|
225 |
foreach my $sf (@r_sf) { |
foreach my $sf (@r_sf) { |
226 |
$errors->{field}->{$f}->{subfield}->{extra_repeatable}->{$sf}++; |
$errors->{$f}->{subfield}->{extra_repeatable}->{$sf}++; |
227 |
$errors->{field}->{$f}->{dump} = |
$errors->{$f}->{dump} = _pack_subfields_hash( $h, 1 ); |
|
join('', _pack_subfields_hash( $h, 1 ) ); |
|
228 |
} |
} |
229 |
|
|
230 |
} |
} |
232 |
if ( defined( $self->{must_exist_sf}->{$f} ) ) { |
if ( defined( $self->{must_exist_sf}->{$f} ) ) { |
233 |
foreach my $sf (sort keys %{ $self->{must_exist_sf}->{$f} }) { |
foreach my $sf (sort keys %{ $self->{must_exist_sf}->{$f} }) { |
234 |
#warn "====> $f $sf must exist\n"; |
#warn "====> $f $sf must exist\n"; |
235 |
$errors->{field}->{$f}->{subfield}->{missing}->{$sf}++ |
$errors->{$f}->{subfield}->{missing}->{$sf}++ |
236 |
unless defined( $subfields->{$sf} ); |
unless defined( $subfields->{$sf} ); |
237 |
} |
} |
238 |
} |
} |
239 |
|
|
240 |
} |
} |
241 |
} elsif (ref($v) eq 'HASH') { |
} elsif (ref($v) eq 'HASH') { |
242 |
$errors->{field}->{$f}->{unexpected_subfields}++; |
$errors->{$f}->{unexpected_subfields}++; |
243 |
$errors->{field}->{$f}->{dump} = |
$errors->{$f}->{dump} = _pack_subfields_hash( $v, 1 ); |
|
join('', _pack_subfields_hash( $v, 1 ) ); |
|
244 |
} |
} |
245 |
} |
} |
246 |
} |
} |
247 |
|
|
248 |
|
$log->debug("_delimiters_templates = ", dump( $self->{_delimiters_templates} ) ); |
249 |
|
|
250 |
foreach my $must (sort keys %{ $self->{must_exist} }) { |
foreach my $must (sort keys %{ $self->{must_exist} }) { |
251 |
next if ($fields->{$must}); |
next if ($fields->{$must}); |
252 |
$errors->{field}->{$must}->{missing}++; |
$errors->{$must}->{missing}++; |
253 |
$errors->{dump} = $rec_dump if ($rec_dump); |
$errors->{dump} = $rec_dump if ($rec_dump); |
254 |
} |
} |
255 |
|
|
256 |
if ($errors) { |
if ($errors) { |
257 |
$log->debug("errors: ", sub { dump( $errors ) } ); |
$log->debug("errors: ", $self->report_error( $errors ) ); |
258 |
|
|
259 |
my $mfn = $rec->{'000'}->[0] || $log->logconfess("record ", dump( $rec ), " doesn't have MFN"); |
my $mfn = $rec->{'000'}->[0] || $log->logconfess("record ", dump( $rec ), " doesn't have MFN"); |
260 |
$self->{errors}->{$mfn} = $errors; |
$self->{errors}->{$mfn} = $errors; |
291 |
return $self->{errors}; |
return $self->{errors}; |
292 |
} |
} |
293 |
|
|
294 |
=head2 report |
=head2 report_error |
295 |
|
|
296 |
Produce nice humanly readable report of errors |
Produce nice humanly readable report of single error |
297 |
|
|
298 |
print $validate->report; |
print $validate->report_error( $error_hash ); |
299 |
|
|
300 |
=cut |
=cut |
301 |
|
|
302 |
sub report { |
sub report_error { |
303 |
my $self = shift; |
my $self = shift; |
304 |
|
|
305 |
my $log = $self->_get_logger(); |
my $h = shift || die "no hash?"; |
306 |
|
|
307 |
|
sub _unroll { |
308 |
|
my ($self, $tree, $accumulated) = @_; |
309 |
|
|
310 |
sub unroll { |
my $log = $self->_get_logger(); |
|
my ($tree, $accumulated) = @_; |
|
311 |
|
|
312 |
$log->debug("# ", |
$log->debug("# ", |
313 |
( $tree ? "tree: $tree " : '' ), |
( $tree ? "tree: $tree " : '' ), |
326 |
|
|
327 |
if ($k eq 'dump') { |
if ($k eq 'dump') { |
328 |
$dump = $tree->{dump}; |
$dump = $tree->{dump}; |
329 |
warn "## dump: $dump\n"; |
# warn "## dump: ",dump($dump),"\n"; |
330 |
next; |
next; |
331 |
} |
} |
332 |
|
|
333 |
$log->debug("current: $k"); |
$log->debug("current: $k"); |
334 |
|
|
335 |
my ($new_results, $new_dump) = unroll($tree->{$k}, |
my ($new_results, $new_dump) = $self->_unroll($tree->{$k}, |
336 |
$accumulated ? "$accumulated\t$k" : $k |
$accumulated ? "$accumulated\t$k" : $k |
337 |
); |
); |
338 |
|
|
356 |
} |
} |
357 |
} |
} |
358 |
|
|
359 |
|
|
360 |
|
sub _reformat { |
361 |
|
my $l = shift; |
362 |
|
$l =~ s/\t/ /g; |
363 |
|
$l =~ s/_/ /; |
364 |
|
return $l; |
365 |
|
} |
366 |
|
|
367 |
my $out = ''; |
my $out = ''; |
368 |
|
|
369 |
|
for my $f (sort keys %{ $h }) { |
370 |
|
$out .= "$f: "; |
371 |
|
|
372 |
|
my ($r, $d) = $self->_unroll( $h->{$f} ); |
373 |
|
my $e; |
374 |
|
if (ref($r) eq 'ARRAY') { |
375 |
|
$e .= join(", ", map { _reformat( $_ ) } @$r); |
376 |
|
} else { |
377 |
|
$e .= _reformat( $r ); |
378 |
|
} |
379 |
|
$e .= "\n\t$d" if ($d); |
380 |
|
|
381 |
|
$out .= $e . "\n"; |
382 |
|
} |
383 |
|
return $out; |
384 |
|
} |
385 |
|
|
386 |
|
|
387 |
|
=head2 report |
388 |
|
|
389 |
|
Produce nice humanly readable report of errors |
390 |
|
|
391 |
|
print $validate->report; |
392 |
|
|
393 |
|
=cut |
394 |
|
|
395 |
|
sub report { |
396 |
|
my $self = shift; |
397 |
my $e = $self->{errors} || return; |
my $e = $self->{errors} || return; |
398 |
|
|
399 |
foreach my $mfn (sort keys %$e) { |
my $out; |
400 |
my ($r, $d) = unroll( $e->{$mfn} ); |
foreach my $mfn (sort { $a <=> $b } keys %$e) { |
401 |
$out .= "MFN $mfn\n", dump($r), "\t$d\n\n"; |
$out .= "MFN $mfn\n" . $self->report_error( $e->{$mfn} ) . "\n"; |
402 |
|
} |
403 |
|
|
404 |
|
return $out; |
405 |
|
|
406 |
|
} |
407 |
|
|
408 |
|
=head2 delimiters_templates |
409 |
|
|
410 |
|
|
411 |
|
=cut |
412 |
|
|
413 |
|
sub delimiters_templates { |
414 |
|
my $self = shift; |
415 |
|
|
416 |
|
my $t = $self->{_delimiters_templates}; |
417 |
|
|
418 |
|
my $log = $self->_get_logger; |
419 |
|
|
420 |
|
unless ($t) { |
421 |
|
$log->warn("called without delimiters"); |
422 |
|
return; |
423 |
|
} |
424 |
|
|
425 |
|
my $out; |
426 |
|
|
427 |
|
foreach my $f (sort { $a <=> $b } keys %$t) { |
428 |
|
$out .= "$f\n"; |
429 |
|
foreach my $sft (sort { $a cmp $b } keys %{ $t->{$f} }) { |
430 |
|
$out .= "\t" . $t->{$f}->{$sft} . "\t$sft\n"; |
431 |
|
} |
432 |
} |
} |
433 |
|
|
434 |
return $out; |
return $out; |