check year validity and report invalid_year in $skip

[koha-bibliografija] / html.pl

diff --git a/html.pl b/html.pl
index 0c55a7a..84d54e2 100755 (executable)
--- a/html.pl
+++ b/html.pl
@@ -126,7 +126,7 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
 
        my ( undef, $doc ) = biblioitem_html( $row->{biblionumber} );
        if ( ! $doc ) {
-               warn "ERROR can't parse MARCXML ", $row->{biblionumber}, " ", $row->{marcxml}, "\n";
+#              warn "ERROR can't parse MARCXML ", $row->{biblionumber}, " ", $row->{marcxml}, "\n";
                next;
        }
 
@@ -155,7 +155,12 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
 
         if ($elt->localname eq 'controlfield') {
                        if ( $tag eq '008' ) {
-                                $biblio_year->{ $row->{biblionumber} } = $elt->textContent;
+                               my $year = substr($elt->textContent, 7, 4 );
+                               if ( $year !~ m/^\d+$/ ) {
+                                       $year = 0;
+                                       push @{ $skip->{invalid_year} }, $row->{biblionumber};
+                               }
+                               $biblio_year->{ $row->{biblionumber} } = $data->{year} = $year;
                        }
                        next;
         } elsif ($elt->localname eq 'datafield') {
@@ -173,6 +178,14 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
         }
     }
 
+       if ( $data->{year} < 2008 ) {
+               push @{ $skip->{year_lt_2008} }, $row->{biblionumber};
+               next;
+       } elsif ( $data->{year} > 2013 ) {
+               push @{ $skip->{year_gt_2013} }, $row->{biblionumber};
+               next;
+       }
+
 #      warn "# ", $row->{biblionumber}, " data ",dump($data);
 
        my $category = $data->{942}->[0]->{'t'};
@@ -229,6 +242,7 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
 debug 'authors' => $authors;
 debug 'type_stats' => $type_stats;
 debug 'skip' => $skip;
+debug 'biblio_year' => $biblio_year;
 
 my $category_label;
 my $sth_categories = $dbh->prepare(q{
@@ -337,6 +351,13 @@ debug 'department_category_author' => $department_category_author;
 
 mkdir 'html/departments' unless -d 'html/departments';
 
+sub unique_biblionumber {
+       my @v = @_;
+       my $u;
+       $u->{$_}++ foreach @v;
+       return sort { $biblio_year->{$b} <=> $biblio_year->{$a} || $a <=> $b } keys %$u;
+}
+
 open(my $dep_fh, '>:encoding(utf-8)', 'html/departments/index.new');
 print $dep_fh html_title('Odsijeci Filozofskog fakulteta u Zagrebu'), qq|<ul>\n|;
 foreach my $department ( sort keys %$department_category_author ) {
@@ -355,7 +376,9 @@ foreach my $department ( sort keys %$department_category_author ) {
                my @authids = @{ $department_category_author->{$department}->{$category} };
                next unless @authids;
 
-               my @biblionumber = map { @{ $authors->{$_}->{aut}->{$category} } } grep { exists $authors->{$_}->{aut}->{$category} } @authids;
+               my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{aut}->{$category} } } grep { exists $authors->{$_}->{aut}->{$category} } @authids;
+               my $unique;
+               $unique->{$_}++ foreach @biblionumber;
 
                next unless @biblionumber;
 
@@ -375,7 +398,7 @@ foreach my $department ( sort keys %$department_category_author ) {
                my @authids = @{ $department_category_author->{$department}->{$category} };
                next unless @authids;
 
-               my @biblionumber = map { @{ $authors->{$_}->{sec}->{$category} } } grep { exists $authors->{$_}->{sec}->{$category} } @authids;
+               my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{sec}->{$category} } } grep { exists $authors->{$_}->{sec}->{$category} } @authids;
 
                next unless @biblionumber;