--- trunk/lib/WebPAC/Validate.pm 2006/09/13 17:44:57 677 +++ trunk/lib/WebPAC/Validate.pm 2007/05/27 11:14:40 851 @@ -18,11 +18,11 @@ =head1 VERSION -Version 0.09 +Version 0.12 =cut -our $VERSION = '0.09'; +our $VERSION = '0.12'; =head1 SYNOPSIS @@ -52,8 +52,14 @@ my $validate = new WebPAC::Validate( path => 'conf/validate/file', + delimiters => [ ' : ', ' / ', ' ; ', ' , ' ], + delimiters_path => 'conf/validate/delimiters/file', ); +Optional parametar C will turn on validating of delimiters. Be +careful here, those delimiters are just stuck into regex, so they can +contain L regexpes. + =cut sub new { @@ -63,78 +69,105 @@ my $log = $self->_get_logger(); - foreach my $p (qw/path/) { - $log->logconfess("need $p") unless ($self->{$p}); - } + $log->logdie("need path or delimiters_path") unless ( $self->{path} || $self->{delimiters_path} ); - my $v_file = read_file( $self->{path} ) || - $log->logdie("can't open validate path $self->{path}: $!"); + if ( $self->{path} ) { - my $v; - my $curr_line = 1; + my $v_file = read_file( $self->{path} ) || + $log->logdie("can't open validate path $self->{path}: $!"); - foreach my $l (split(/[\n\r]+/, $v_file)) { - $curr_line++; + my $v; + my $curr_line = 1; - # skip comments and whitespaces - next if ($l =~ /^#/ || $l =~ /^\s*$/); + foreach my $l (split(/[\n\r]+/, $v_file)) { + $curr_line++; - $l =~ s/^\s+//; - $l =~ s/\s+$//; + # skip comments and whitespaces + next if ($l =~ /^#/ || $l =~ /^\s*$/); - my @d = split(/\s+/, $l); + $l =~ s/^\s+//; + $l =~ s/\s+$//; - my $fld = shift @d; + my @d = split(/\s+/, $l); - if ($fld =~ s/!$//) { - $self->{must_exist}->{$fld}++; - } elsif ($fld =~ s/-$//) { - $self->{dont_validate}->{$fld}++; - } + my $fld = shift @d; - $log->logdie("need field name in line $curr_line: $l") unless (defined($fld)); + if ($fld =~ s/!$//) { + $self->{must_exist}->{$fld}++; + } elsif ($fld =~ s/-$//) { + $self->{dont_validate}->{$fld}++; + } + + $log->logdie("need field name in line $curr_line: $l") unless (defined($fld)); + + if (@d) { + $v->{$fld} = [ map { + my $sf = $_; + if ( $sf =~ s/!(\*)?$/$1/ ) { + $self->{must_exist_sf}->{ $fld }->{ $sf }++; + }; + $sf; + } @d ]; + } else { + $v->{$fld} = 1; + } - if (@d) { - $v->{$fld} = [ map { - my $sf = $_; - if ( $sf =~ s/!(\*)?$/$1/ ) { - $self->{must_exist_sf}->{ $fld }->{ $sf }++; - }; - $sf; - } @d ]; - } else { - $v->{$fld} = 1; } - } + $log->debug("current validation rules: ", dump($v)); - $log->debug("current validation rules: ", dump($v)); + $self->{rules} = $v; - $self->{rules} = $v; + $log->info("validation uses rules from $self->{path}"); + } - $log->info("validation uses rules from $self->{path}"); + if ( $self->{delimiters} ) { + $self->{delimiters_regex} = '(\^[a-z0-9]|' . join('|', @{ $self->{delimiters} }) . ')'; + $log->info("validation check delimiters with regex $self->{delimiters_regex}"); + } + + if ( my $path = $self->{delimiters_path} ) { + if ( -e $path ) { + $log->info("using delimiter validation rules from $path"); + open(my $d, $path) || $log->fatal("can't open $path: $!"); + while(<$d>) { + chomp($d); + if (/^\s*(#*)\s*(\d+)\t+(\d+)\t+(.*)$/) { + my ($comment,$field,$count,$template) = ($1,$2,$3,$4); + $self->{_validate_delimiters_templates}->{$field}->{$template} = $count unless ($comment); + } else { + warn "## ignored $d\n"; + } + } + close($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"); + } + } $self ? return $self : return undef; } -=head2 validate_errors +=head2 validate_rec Validate record and return errors - my @errors = $validate->validate_errors( $rec, $rec_dump ); + my @errors = $validate->validate_rec( $rec, $rec_dump ); =cut -sub validate_errors { +sub validate_rec { my $self = shift; my $log = $self->_get_logger(); - my $rec = shift || $log->logdie("validate_errors need record"); + my $rec = shift || $log->logdie("validate_rec 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}); +# $log->logdie("can't find validation rules") unless (my $r = $self->{rules}); + my $r = $self->{rules}; my $errors; @@ -146,6 +179,33 @@ next if (!defined($f) || $f eq '' || $f eq '000'); + # first check delimiters + if ( my $regex = $self->{delimiters_regex} ) { + + foreach my $v (@{ $rec->{$f} }) { + my $l = _pack_subfields_hash( $v, 1 ); + my $subfield_dump = $l; + my $template = ''; + $l =~ s/$regex/$template.=$1/eg; + #warn "## template: $template\n"; + + if ( $template ) { + $self->{_delimiters_templates}->{$f}->{$template}++; + + if ( my $v = $self->{_validate_delimiters_templates} ) { + if ( ! defined( $v->{$f}->{$template} ) ) { + $errors->{$f}->{potentially_invalid_combination} = $template; + $errors->{$f}->{dump} = $subfield_dump; + #} else { + # warn "## $f $template ok\n"; + } + } + } + } + } + + next unless ( $r ); # skip validation of no rules are specified + next if (defined( $self->{dont_validate}->{$f} )); # track field usage @@ -222,6 +282,8 @@ } } + $log->debug("_delimiters_templates = ", dump( $self->{_delimiters_templates} ) ); + foreach my $must (sort keys %{ $self->{must_exist} }) { next if ($fields->{$must}); $errors->{$must}->{missing}++; @@ -240,17 +302,37 @@ return $errors; } -=head2 reset_errors +=head2 reset + +Clean all accumulated errors for this input and remember delimiter templates +for L -Clean all accumulated errors for this input + $validate->reset; - $validate->reset_errors; +This function B be called after each input to provide accurate statistics. =cut -sub reset_errors { +sub reset { my $self = shift; + + my $log = $self->_get_logger; + delete ($self->{errors}); + + if ( ! $self->{_delimiters_templates} ) { + $log->debug("called without _delimiters_templates?"); + return; + } + + foreach my $f ( keys %{ $self->{_delimiters_templates} } ) { + foreach my $t ( keys %{ $self->{_delimiters_templates}->{$f} } ) { + $self->{_accumulated_delimiters_templates}->{$f}->{$t} += + $self->{_delimiters_templates}->{$f}->{$t}; + } + } + $log->debug("_accumulated_delimiters_templates = ", dump( $self->{_accumulated_delimiter_templates} )); + delete ($self->{_delimiters_templates}); } =head2 all_errors @@ -301,7 +383,7 @@ if ($k eq 'dump') { $dump = $tree->{dump}; -# warn "## dump: ",dump($dump),"\n"; + #warn "## dump ",dump($dump),"\n"; next; } @@ -335,7 +417,7 @@ sub _reformat { my $l = shift; $l =~ s/\t/ /g; - $l =~ s/_/ /; + $l =~ s/_/ /g; return $l; } @@ -380,6 +462,98 @@ } +=head2 delimiters_templates + +Generate report of delimiter tamplates + + my $report = $validate->delimiter_teplates( + report => 1, + current_input => 1, + ); + +Options: + +=over 4 + +=item report + +Generate humanly readable report with single fields + +=item current_input + +Report just current_input and not accumulated data + +=back + +=cut + +sub delimiters_templates { + my $self = shift; + + my $args = {@_}; + + my $t = $self->{_accumulated_delimiters_templates}; + $t = $self->{_delimiters_templates} if ( $args->{current_input} ); + + my $log = $self->_get_logger; + + unless ($t) { + $log->error("called without delimiters"); + return; + } + + my $out; + + foreach my $f (sort { $a <=> $b } keys %$t) { + $out .= "$f\n" if ( $args->{report} ); + foreach my $template (sort { $a cmp $b } keys %{ $t->{$f} }) { + my $count = $t->{$f}->{$template}; + $out .= + ( $count ? "" : "# " ) . + ( $args->{report} ? "" : "$f" ) . + "\t$count\t$template\n"; + } + } + + return $out; +} + +=head2 save_delimiters_templates + +Save accumulated delimiter templates + + $validator->save_delimiters_template( '/path/to/validate/delimiters' ); + +=cut + +sub save_delimiters_templates { + my $self = shift; + + my $path = $self->{delimiters_path}; + + return unless ( $path ); + + my $log = $self->_get_logger; + + if ( ! $self->{_accumulated_delimiters_templates} ) { + $log->error('no _accumulated_delimiters_templates found, reset'); + $self->reset; + } + + if ( ! $self->{_delimiters_templates} ) { + $log->error('found _delimiters_templates, calling reset'); + $self->reset; + } + + $path .= '.new' if ( -e $path ); + + open(my $d, '>', $path) || $log->fatal("can't open $path: $!"); + print $d $self->delimiters_templates; + close($d); + + $log->info("new delimiters templates saved to $path"); +} + =head1 AUTHOR Dobrica Pavlinusic, C<< >>