r1232@llin: dpavlin | 2007-05-24 14:44:53 +0200

 added storing and retriving of delimiters templates to file and some basic tests

git-svn-id: svn+ssh://mjesec/home/dpavlin/svn/webpac2/trunk@837 07558da8-63fa-0310-ba24-9fe276d99e06

diff --git a/lib/WebPAC/Validate.pm b/lib/WebPAC/Validate.pm
index e816db0..8dbc24f 100644 (file)
--- a/lib/WebPAC/Validate.pm
+++ b/lib/WebPAC/Validate.pm
@@ -53,6 +53,7 @@ Create new validation object
   my $validate = new WebPAC::Validate(
        path => 'conf/validate/file',
        delimiters => [ ' : ', ' / ', ' ; ', ' , ' ],
+       delimiters_path => 'conf/validate/delimiters/file',
   );
 
 Optional parametar C<delimiters> will turn on validating of delimiters. Be
@@ -66,64 +67,85 @@ sub new {
        my $self = {@_};
        bless($self, $class);
 
+warn dump( @_ );
+
        my $log = $self->_get_logger();
 
-       foreach my $p (qw/path/) {
-               $log->logconfess("need $p") unless ($self->{$p});
-       }
+       if ( $self->{path} ) {
 
-       my $v_file = read_file( $self->{path} ) ||
-               $log->logdie("can't open validate path $self->{path}: $!");
+               my $v_file = read_file( $self->{path} ) ||
+                       $log->logdie("can't open validate path $self->{path}: $!");
 
-       my $v;
-       my $curr_line = 1;
+               my $v;
+               my $curr_line = 1;
 
-       foreach my $l (split(/[\n\r]+/, $v_file)) {
-               $curr_line++;
+               foreach my $l (split(/[\n\r]+/, $v_file)) {
+                       $curr_line++;
 
-               # skip comments and whitespaces
-               next if ($l =~ /^#/ || $l =~ /^\s*$/);
+                       # skip comments and whitespaces
+                       next if ($l =~ /^#/ || $l =~ /^\s*$/);
 
-               $l =~ s/^\s+//;
-               $l =~ s/\s+$//;
+                       $l =~ s/^\s+//;
+                       $l =~ s/\s+$//;
 
-               my @d = split(/\s+/, $l);
+                       my @d = split(/\s+/, $l);
 
-               my $fld = shift @d;
+                       my $fld = shift @d;
 
-               if ($fld =~ s/!$//) {
-                       $self->{must_exist}->{$fld}++;
-               } elsif ($fld =~ s/-$//) {
-                       $self->{dont_validate}->{$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));
+                       $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;
 }
 
@@ -144,7 +166,8 @@ sub validate_rec {
        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;
 
@@ -181,6 +204,8 @@ sub validate_rec {
                        }
                }
 
+               next unless ( $r );     # skip validation of no rules are specified
+
                next if (defined( $self->{dont_validate}->{$f} ));
 
                # track field usage
@@ -467,6 +492,26 @@ sub delimiters_templates {
        return $out;
 }
 
+=head2 save_delimiters_templates
+
+=cut
+
+sub save_delimiters_templates {
+       my $self = shift;
+
+       my $path = $self->{delimiters_path};
+
+       return unless ( $path );
+
+       my $log = $self->_get_logger;
+
+       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<< <dpavlin@rot13.org> >>

diff --git a/run.pl b/run.pl
index 21fb841..a98aadc 100755 (executable)
--- a/run.pl
+++ b/run.pl
@@ -68,6 +68,11 @@ and subfield usage for each input
 
 turn on extra validation of imput records, see L<WebPAC::Validation>
 
+=item --validate-delimiters path/to/validate_delimiters_file
+
+this option is used with C<--validate> to turn on extra validation of
+delimiters. If file is non existant, it will be created on first run.
+
 =item --marc-generate
 
 Generate MARC file. This will automatically be on if file contains C<marc*> directives.
@@ -108,6 +113,7 @@ my $debug = 0;
 my $only_filter;
 my $stats = 0;
 my $validate_path;
+my $validate_delimiters_path;
 my $marc_generate = 1;
 my $marc_lint = 1;
 my $marc_dump = 0;
@@ -128,6 +134,7 @@ GetOptions(
        "debug+" => \$debug,
        "stats" => \$stats,
        "validate=s" => \$validate_path,
+       "validate-delimiters=s" => \$validate_delimiters_path,
        "marc-generate!" => \$marc_generate,
        "marc-lint!" => \$marc_lint,
        "marc-dump!" => \$marc_dump,
@@ -163,13 +170,12 @@ if ($merge) {
        $log->info("created merge batch file $estcmd_path");
 }
 
-
 my $validate;
 $validate = new WebPAC::Validate(
        path => $validate_path,
        delimiters => $config->webpac('delimiters'),
-) if ($validate_path);
-
+       delimiters_path => $validate_delimiters_path,
+) if ($validate_path || $validate_delimiters_path);
 
 my $use_indexer = $config->use_indexer;
 $stats ||= $validate;
@@ -603,6 +609,9 @@ if ($parallel) {
        $log->info("all parallel processes finished");
 }
 
+# save new delimiters if needed
+$validate->save_delimiters_templates;
+
 #
 # handle links or merge after indexing
 #

diff --git a/t/1-validate-delimiters.t b/t/1-validate-delimiters.t
index a01ed96..2478ba1 100755 (executable)
--- a/t/1-validate-delimiters.t
+++ b/t/1-validate-delimiters.t
@@ -1,12 +1,13 @@
 #!/usr/bin/perl -w
 
 use strict;
-use Test::More tests => 7;
+use Test::More tests => 10;
 use Test::Exception;
 use blib;
 
 use Data::Dump qw/dump/;
 use Cwd qw/abs_path/;
+use File::Temp qw/tempfile/;
 
 BEGIN {
 use_ok( 'WebPAC::Validate' );
@@ -17,9 +18,12 @@ my $debug = shift @ARGV;
 ok(my $abs_path = abs_path($0), "abs_path");
 $abs_path =~ s#/[^/]*$#/#;
 
+my ( $fh, $path ) = tempfile();
+
 ok(my $v = new WebPAC::Validate(
        path => "$abs_path/data/validate_test",
        delimiters => [ ' : ', ' ; ', ' / ', ' \. ', ' = ' ],
+       delimiters_path => $path,
        debug => $debug,
 ), "new");
 
@@ -87,3 +91,13 @@ ok (my $e = $v->report, 'report');
 
 diag $e;
 diag dump( $v->{errors} );
+
+ok( $v->save_delimiters_templates, 'save_delimiters_templates' );
+
+ok(my $v2 = new WebPAC::Validate(
+       delimiters => [ ' : ', ' ; ', ' / ', ' \. ', ' = ' ],
+       delimiters_path => $path,
+       debug => $debug,
+), "new");
+
+is_deeply( $v->{_delimiters_templates}, $v2->{_validate_delimiters_templates}, 'save/load ok');