use base 'WebPAC::Common';
use File::Slurp;
use List::Util qw/first/;
-use Data::Dumper;
+use Data::Dump qw/dump/;
+use WebPAC::Normalize qw/_pack_subfields_hash/;
+use Storable qw/dclone/;
=head1 NAME
=head1 VERSION
-Version 0.01
+Version 0.07
=cut
-our $VERSION = '0.01';
+our $VERSION = '0.07';
=head1 SYNOPSIS
# same with 101
101
# field 200 have valid subfields a-g
- 200 a b c d e f g
+ # and field e is repeatable
+ 200 a b c d e* f g
# field 205 can have only subfield a
- 205 a
+ # and must exists
+ 205! a
# while 210 can have a c or d
210 a c d
Create new validation object
my $validate = new WebPAC::Validate(
- path => '/path/to/input/validate_file',
+ path => 'conf/validate/file',
);
=cut
my $fld = shift @d;
+ if ($fld =~ s/!$//) {
+ $self->{must_exist}->{$fld}++;
+ }
+
$log->logdie("need field name in line $curr_line: $l") unless (defined($fld));
if (@d) {
- $v->{$fld} = \@d;
+ $v->{$fld} = [ map {
+ my $sf = $_;
+ if ( $sf =~ s/!(\*)?$/$1/ ) {
+ $self->{must_exist_sf}->{ $fld }->{ $sf }++;
+ };
+ $sf;
+ } @d ];
} else {
$v->{$fld} = 1;
}
}
- $log->debug("current validation rules: ", Dumper($v));
+ $log->debug("current validation rules: ", dump($v));
$self->{rules} = $v;
Validate record and return errors
- my @errors = $validate->validate_errors( $rec );
+ my @errors = $validate->validate_errors( $rec, $rec_dump );
=cut
my $log = $self->_get_logger();
my $rec = shift || $log->logdie("validate_errors need record");
+ my $rec_dump = shift;
$log->logdie("rec isn't HASH") unless (ref($rec) eq 'HASH');
$log->logdie("can't find validation rules") unless (my $r = $self->{rules});
- my @errors;
+ my $errors;
+
+ $log->debug("rec = ", sub { dump($rec) }, "keys = ", keys %{ $rec });
- $log->debug("rec = ", sub { Dumper($rec) }, "keys = ", keys %{ $rec });
+ my $fields;
foreach my $f (keys %{ $rec }) {
next if (!defined($f) || $f eq '' || $f eq '000');
- if (! defined($r->{$f})) {
- push @errors, "field '$f' shouldn't exists";
+ $fields->{$f}++;
+
+ if ( ! defined($r->{$f}) ) {
+ $errors->{field}->{ $f }->{unexpected} = "this field is not expected";
next;
}
+
if (ref($rec->{$f}) ne 'ARRAY') {
- push @errors, "field '$f' isn't repetable, probably bug in parsing input data";
+ $errors->{field}->{ $f }->{not_repeatable} = "probably bug in parsing input data";
next;
}
# can we have subfields?
if (ref($r->{$f}) eq 'ARRAY') {
# are values hashes? (has subfields)
- if (ref($v) ne 'HASH') {
- push @errors, "$f has value without subfields: $v";
+ if (! defined($v)) {
+# $errors->{field}->{$f}->{empty} = undef;
+# $errors->{dump} = $rec_dump if ($rec_dump);
+ } elsif (ref($v) ne 'HASH') {
+ $errors->{field}->{$f}->{missing_subfield} = join(",", @{ $r->{$f} }) . " required";
next;
} else {
+
+ my $h = dclone( $v );
+
+ my $sf_repeatable;
+
+ delete($v->{subfields}) if (defined($v->{subfields}));
+
+ my $subfields;
+
foreach my $sf (keys %{ $v }) {
- # permited subfield?
- if (! first { $_ eq $sf } @{ $r->{$f} }) {
- push @errors, "$f has unknown subfield: $sf" if ($sf ne 'subfields');
+
+ $subfields->{ $sf }++;
+
+ # is non-repeatable but with multiple values?
+ if ( ! first { $_ eq $sf.'*' } @{$r->{$f}} ) {
+ if ( ref($v->{$sf}) eq 'ARRAY' ) {
+ $sf_repeatable->{$sf}++;
+ };
+ if (! first { $_ eq $sf } @{ $r->{$f} }) {
+ $errors->{field}->{ $f }->{subfield}->{extra}->{$sf}++;
+ }
+ }
+
+ }
+ if (my @r_sf = sort keys( %$sf_repeatable )) {
+
+ foreach my $sf (@r_sf) {
+ $errors->{field}->{$f}->{subfield}->{extra_repeatable}->{$sf}++;
+ $errors->{field}->{$f}->{dump} =
+ join('', _pack_subfields_hash( $h, 1 ) );
+ }
+
+ }
+
+ if ( defined( $self->{must_exist_sf}->{$f} ) ) {
+ foreach my $sf (sort keys %{ $self->{must_exist_sf}->{$f} }) {
+#warn "====> $f $sf must exist\n";
+ $errors->{field}->{$f}->{subfield}->{missing}->{$sf}++
+ unless defined( $subfields->{$sf} );
}
}
+
}
} elsif (ref($v) eq 'HASH') {
- push @errors, "$f has subfields which is not valid";
+ $errors->{field}->{$f}->{unexpected_subfields}++;
+ $errors->{field}->{$f}->{dump} =
+ join('', _pack_subfields_hash( $v, 1 ) );
}
}
}
+ foreach my $must (sort keys %{ $self->{must_exist} }) {
+ next if ($fields->{$must});
+ $errors->{field}->{$must}->{missing}++;
+ $errors->{dump} = $rec_dump if ($rec_dump);
+ }
+
+ if ($errors) {
+ $log->debug("errors: ", sub { dump( $errors ) } );
+
+ my $mfn = $rec->{'000'}->[0] || $log->logconfess("record ", dump( $rec ), " doesn't have MFN");
+ $self->{errors}->{$mfn} = $errors;
+ }
+
#$log->logcluck("return from this function is ARRAY") unless wantarray;
- $log->debug("errors: ", join(", ", @errors)) if (@errors);
+ return $errors;
+}
+
+=head2 reset_errors
+
+Clean all accumulated errors for this input
+
+ $validate->reset_errors;
+
+=cut
+
+sub reset_errors {
+ my $self = shift;
+ delete ($self->{errors});
+}
+
+=head2 all_errors
+
+Return hash with all errors
+
+ print dump( $validate->all_errors );
+
+=cut
+
+sub all_errors {
+ my $self = shift;
+ return $self->{errors};
+}
+
+=head2 report
+
+Produce nice humanly readable report of errors
+
+ print $validate->report;
+
+=cut
+
+sub report {
+ my $self = shift;
+
+ my $log = $self->_get_logger();
+
+ sub unroll {
+ my ($tree, $accumulated) = @_;
+
+ $log->debug("# ",
+ ( $tree ? "tree: $tree " : '' ),
+ ( $accumulated ? "accumulated: $accumulated " : '' ),
+ );
+
+ my $results;
+
+ if (ref($tree) ne 'HASH') {
+ return ("$accumulated\t($tree)", undef);
+ }
+
+ my $dump;
+
+ foreach my $k (sort keys %{ $tree }) {
+
+ if ($k eq 'dump') {
+ $dump = $tree->{dump};
+ warn "## dump: $dump\n";
+ next;
+ }
+
+ $log->debug("current: $k");
+
+ my ($new_results, $new_dump) = unroll($tree->{$k},
+ $accumulated ? "$accumulated\t$k" : $k
+ );
+
+ $log->debug(
+ ( $new_results ? "new_results: " . dump($new_results) ." " : '' ),
+ );
+
+ push @$results, $new_results if ($new_results);
+ $dump = $new_dump if ($new_dump);
+
+ }
+
+ $log->debug(
+ ( $results ? "results: " . dump($results) ." " : '' ),
+ );
+
+ if ($#$results == 0) {
+ return ($results->[0], $dump);
+ } else {
+ return ($results, $dump);
+ }
+ }
+
+ my $out = '';
+ my $e = $self->{errors} || return;
+
+ foreach my $mfn (sort keys %$e) {
+ my ($r, $d) = unroll( $e->{$mfn} );
+ $out .= "MFN $mfn\n", dump($r), "\t$d\n\n";
+ }
- return @errors;
+ return $out;
}
=head1 AUTHOR