18 |
|
|
19 |
=head1 VERSION |
=head1 VERSION |
20 |
|
|
21 |
Version 0.06 |
Version 0.08 |
22 |
|
|
23 |
=cut |
=cut |
24 |
|
|
25 |
our $VERSION = '0.06'; |
our $VERSION = '0.08'; |
26 |
|
|
27 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
28 |
|
|
145 |
$fields->{$f}++; |
$fields->{$f}++; |
146 |
|
|
147 |
if ( ! defined($r->{$f}) ) { |
if ( ! defined($r->{$f}) ) { |
148 |
$errors->{field}->{ $f }->{unexpected} = "this field is not expected"; |
$errors->{ $f }->{unexpected} = "this field is not expected"; |
149 |
next; |
next; |
150 |
} |
} |
151 |
|
|
152 |
|
|
153 |
if (ref($rec->{$f}) ne 'ARRAY') { |
if (ref($rec->{$f}) ne 'ARRAY') { |
154 |
$errors->{field}->{ $f }->{not_repeatable} = "probably bug in parsing input data"; |
$errors->{ $f }->{not_repeatable} = "probably bug in parsing input data"; |
155 |
next; |
next; |
156 |
} |
} |
157 |
|
|
160 |
if (ref($r->{$f}) eq 'ARRAY') { |
if (ref($r->{$f}) eq 'ARRAY') { |
161 |
# are values hashes? (has subfields) |
# are values hashes? (has subfields) |
162 |
if (! defined($v)) { |
if (! defined($v)) { |
163 |
# $errors->{field}->{$f}->{empty} = undef; |
# $errors->{$f}->{empty} = undef; |
164 |
# $errors->{dump} = $rec_dump if ($rec_dump); |
# $errors->{dump} = $rec_dump if ($rec_dump); |
165 |
} elsif (ref($v) ne 'HASH') { |
} elsif (ref($v) ne 'HASH') { |
166 |
$errors->{field}->{$f}->{missing_subfield} = join(",", @{ $r->{$f} }) . " required"; |
$errors->{$f}->{missing_subfield} = join(",", @{ $r->{$f} }) . " required"; |
167 |
next; |
next; |
168 |
} else { |
} else { |
169 |
|
|
185 |
$sf_repeatable->{$sf}++; |
$sf_repeatable->{$sf}++; |
186 |
}; |
}; |
187 |
if (! first { $_ eq $sf } @{ $r->{$f} }) { |
if (! first { $_ eq $sf } @{ $r->{$f} }) { |
188 |
$errors->{field}->{ $f }->{subfield}->{extra}->{$sf}++; |
$errors->{ $f }->{subfield}->{extra}->{$sf}++; |
189 |
} |
} |
190 |
} |
} |
191 |
|
|
193 |
if (my @r_sf = sort keys( %$sf_repeatable )) { |
if (my @r_sf = sort keys( %$sf_repeatable )) { |
194 |
|
|
195 |
foreach my $sf (@r_sf) { |
foreach my $sf (@r_sf) { |
196 |
$errors->{field}->{$f}->{subfield}->{extra_repeatable}->{$sf}++; |
$errors->{$f}->{subfield}->{extra_repeatable}->{$sf}++; |
197 |
$errors->{field}->{$f}->{dump} = |
$errors->{$f}->{dump} = _pack_subfields_hash( $h, 1 ); |
|
join('', _pack_subfields_hash( $h, 1 ) ); |
|
198 |
} |
} |
199 |
|
|
200 |
} |
} |
202 |
if ( defined( $self->{must_exist_sf}->{$f} ) ) { |
if ( defined( $self->{must_exist_sf}->{$f} ) ) { |
203 |
foreach my $sf (sort keys %{ $self->{must_exist_sf}->{$f} }) { |
foreach my $sf (sort keys %{ $self->{must_exist_sf}->{$f} }) { |
204 |
#warn "====> $f $sf must exist\n"; |
#warn "====> $f $sf must exist\n"; |
205 |
$errors->{field}->{$f}->{subfield}->{$sf} = "missing" |
$errors->{$f}->{subfield}->{missing}->{$sf}++ |
206 |
unless defined( $subfields->{$sf} ); |
unless defined( $subfields->{$sf} ); |
207 |
} |
} |
208 |
} |
} |
209 |
|
|
210 |
} |
} |
211 |
} elsif (ref($v) eq 'HASH') { |
} elsif (ref($v) eq 'HASH') { |
212 |
$errors->{field}->{$f}->{unexpected_subfields}++; |
$errors->{$f}->{unexpected_subfields}++; |
213 |
$errors->{field}->{$f}->{dump} = |
$errors->{$f}->{dump} = _pack_subfields_hash( $v, 1 ); |
|
join('', _pack_subfields_hash( $v, 1 ) ); |
|
214 |
} |
} |
215 |
} |
} |
216 |
} |
} |
217 |
|
|
218 |
foreach my $must (sort keys %{ $self->{must_exist} }) { |
foreach my $must (sort keys %{ $self->{must_exist} }) { |
219 |
next if ($fields->{$must}); |
next if ($fields->{$must}); |
220 |
$errors->{field}->{$must}->{missing}++; |
$errors->{$must}->{missing}++; |
221 |
$errors->{dump} = $rec_dump if ($rec_dump); |
$errors->{dump} = $rec_dump if ($rec_dump); |
222 |
} |
} |
223 |
|
|
224 |
if ($errors) { |
if ($errors) { |
225 |
$log->debug("errors: ", sub { dump( $errors ) } ); |
$log->debug("errors: ", $self->report_error( $errors ) ); |
226 |
|
|
227 |
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"); |
228 |
$self->{errors}->{$mfn} = $errors; |
$self->{errors}->{$mfn} = $errors; |
259 |
return $self->{errors}; |
return $self->{errors}; |
260 |
} |
} |
261 |
|
|
262 |
|
=head2 report_error |
263 |
|
|
264 |
|
Produce nice humanly readable report of single error |
265 |
|
|
266 |
|
print $validate->report_error( $error_hash ); |
267 |
|
|
268 |
|
=cut |
269 |
|
|
270 |
|
sub report_error { |
271 |
|
my $self = shift; |
272 |
|
|
273 |
|
my $h = shift || die "no hash?"; |
274 |
|
|
275 |
|
sub _unroll { |
276 |
|
my ($self, $tree, $accumulated) = @_; |
277 |
|
|
278 |
|
my $log = $self->_get_logger(); |
279 |
|
|
280 |
|
$log->debug("# ", |
281 |
|
( $tree ? "tree: $tree " : '' ), |
282 |
|
( $accumulated ? "accumulated: $accumulated " : '' ), |
283 |
|
); |
284 |
|
|
285 |
|
my $results; |
286 |
|
|
287 |
|
if (ref($tree) ne 'HASH') { |
288 |
|
return ("$accumulated\t($tree)", undef); |
289 |
|
} |
290 |
|
|
291 |
|
my $dump; |
292 |
|
|
293 |
|
foreach my $k (sort keys %{ $tree }) { |
294 |
|
|
295 |
|
if ($k eq 'dump') { |
296 |
|
$dump = $tree->{dump}; |
297 |
|
# warn "## dump: ",dump($dump),"\n"; |
298 |
|
next; |
299 |
|
} |
300 |
|
|
301 |
|
$log->debug("current: $k"); |
302 |
|
|
303 |
|
my ($new_results, $new_dump) = $self->_unroll($tree->{$k}, |
304 |
|
$accumulated ? "$accumulated\t$k" : $k |
305 |
|
); |
306 |
|
|
307 |
|
$log->debug( |
308 |
|
( $new_results ? "new_results: " . dump($new_results) ." " : '' ), |
309 |
|
); |
310 |
|
|
311 |
|
push @$results, $new_results if ($new_results); |
312 |
|
$dump = $new_dump if ($new_dump); |
313 |
|
|
314 |
|
} |
315 |
|
|
316 |
|
$log->debug( |
317 |
|
( $results ? "results: " . dump($results) ." " : '' ), |
318 |
|
); |
319 |
|
|
320 |
|
if ($#$results == 0) { |
321 |
|
return ($results->[0], $dump); |
322 |
|
} else { |
323 |
|
return ($results, $dump); |
324 |
|
} |
325 |
|
} |
326 |
|
|
327 |
|
|
328 |
|
sub _reformat { |
329 |
|
my $l = shift; |
330 |
|
$l =~ s/\t/ /g; |
331 |
|
$l =~ s/_/ /; |
332 |
|
return $l; |
333 |
|
} |
334 |
|
|
335 |
|
my $out = ''; |
336 |
|
|
337 |
|
for my $f (sort keys %{ $h }) { |
338 |
|
$out .= "$f: "; |
339 |
|
|
340 |
|
my ($r, $d) = $self->_unroll( $h->{$f} ); |
341 |
|
my $e; |
342 |
|
if (ref($r) eq 'ARRAY') { |
343 |
|
$e .= join(", ", map { _reformat( $_ ) } @$r); |
344 |
|
} else { |
345 |
|
$e .= _reformat( $r ); |
346 |
|
} |
347 |
|
$e .= "\n\t$d" if ($d); |
348 |
|
|
349 |
|
$out .= $e . "\n"; |
350 |
|
} |
351 |
|
return $out; |
352 |
|
} |
353 |
|
|
354 |
|
|
355 |
|
=head2 report |
356 |
|
|
357 |
|
Produce nice humanly readable report of errors |
358 |
|
|
359 |
|
print $validate->report; |
360 |
|
|
361 |
|
=cut |
362 |
|
|
363 |
|
sub report { |
364 |
|
my $self = shift; |
365 |
|
my $e = $self->{errors} || return; |
366 |
|
|
367 |
|
my $out; |
368 |
|
foreach my $mfn (sort { $a <=> $b } keys %$e) { |
369 |
|
$out .= "MFN $mfn\n" . $self->report_error( $e->{$mfn} ) . "\n"; |
370 |
|
} |
371 |
|
|
372 |
|
return $out; |
373 |
|
|
374 |
|
} |
375 |
|
|
376 |
=head1 AUTHOR |
=head1 AUTHOR |
377 |
|
|
378 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |