--- trunk/lib/WebPAC/Validate.pm 2007/05/24 12:44:43 836 +++ trunk/lib/WebPAC/Validate.pm 2007/05/27 10:50:37 849 @@ -18,11 +18,11 @@ =head1 VERSION -Version 0.11 +Version 0.12 =cut -our $VERSION = '0.11'; +our $VERSION = '0.12'; =head1 SYNOPSIS @@ -53,6 +53,7 @@ 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 @@ -68,62 +69,83 @@ 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}++; + } - if (@d) { - $v->{$fld} = [ map { - my $sf = $_; - if ( $sf =~ s/!(\*)?$/$1/ ) { - $self->{must_exist_sf}->{ $fld }->{ $sf }++; - }; - $sf; - } @d ]; - } else { - $v->{$fld} = 1; - } + $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; + } - $log->debug("current validation rules: ", dump($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; } @@ -144,7 +166,8 @@ 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; @@ -170,17 +193,19 @@ $self->{_delimiters_templates}->{$f}->{$template}++; if ( my $v = $self->{_validate_delimiters_templates} ) { - if ( ! defined( $v->{$template} ) ) { - $errors->{$f}->{invalid_delimiters_combination} = $template; + if ( ! defined( $v->{$f}->{$template} ) ) { + $errors->{$f}->{potentially_invalid_combination} = $template; $errors->{$f}->{dump} = $subfield_dump; - } else { - warn "## $f $template ok\n"; + #} 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 @@ -277,17 +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 @@ -423,6 +468,7 @@ my $report = $validate->delimiter_teplates( report => 1, + accumulated => 1, ); Options: @@ -433,6 +479,10 @@ Generate humanly readable report with single fields +=item accumulated + +Use accumulated data from all inputs + =back =cut @@ -442,7 +492,12 @@ my $args = {@_}; - my $t = $self->{_delimiters_templates}; + my $t; + if ( $args->{accumulated} ) { + $t = $self->{_accumulated_delimiters_templates}; + } else { + $t = $self->{_delimiters_templates}; + } my $log = $self->_get_logger; @@ -467,6 +522,42 @@ 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 $self->delimiters_templates( accumulated => 1 ); + close($d); + + $log->info("new delimiters templates saved to $path"); +} + =head1 AUTHOR Dobrica Pavlinusic, C<< >>