generate biblio_in_database from 942$r

[koha-bibliografija] / html.pl

diff --git a/html.pl b/html.pl
index a638c23..ce3f30f 100755 (executable)
--- a/html.pl
+++ b/html.pl
@@ -21,7 +21,7 @@ my $dbh = C4::Context->dbh;
 
 sub debug {
        my ($title, $data) = @_;
-       print "# $title ",dump($data), $/;
+       print "# $title ",dump($data), $/ if $ENV{DEBUG};
 }
 
 my $xslfilename = 'compact.xsl';
@@ -42,11 +42,15 @@ from auth_header
 
 $sth_auth->execute();
 while( my $row = $sth_auth->fetchrow_hashref ) {
+       if ( $row->{department} !~ m/Filozofski fakultet u Zagrebu/ ) {
+               push @{ $skip->{nije_ffzg} }, $row;
+               next;
+       }
        $auth_header->{ $row->{authid} } = $row->{full_name};
-       $row->{department} =~ s/, Filozofski fakultet u Zagrebu\s*// || next;
+       $row->{department} =~ s/, Filozofski fakultet u Zagrebu.*$//;
        $row->{department} =~ s/^.+\.\s*//;
-       push @{ $auth_department->{ $row->{department} } }, $row->{authid};
 #      warn dump( $row );
+       push @{ $auth_department->{ $row->{department} } }, $row->{authid};
        push @authors, $row;
 
 }
@@ -86,13 +90,15 @@ my $xslt = XML::LibXSLT->new();
 my $parsed = $xslt->parse_stylesheet($style_doc);
 
 my $biblio_html;
+my $biblio_parsed;
+my $biblio_in_database;
 
 open(my $xml_fh, '>', '/tmp/bibliografija.xml') if $ENV{XML};
 
 sub biblioitem_html {
-       my $biblionumber = shift;
+       my ($biblionumber, $parse_only) = @_;
 
-       return $biblio_html->{$biblionumber} if exists $biblio_html->{$biblionumber};
+       return $biblio_html->{$biblionumber} if exists $biblio_html->{$biblionumber} && ! $parse_only;
 
        my $xmlrecord = $marcxml->{$biblionumber} || confess "missing $biblionumber marcxml";
 
@@ -105,10 +111,16 @@ sub biblioitem_html {
                return;
        }
 
+       if ( $parse_only ) {
+               $biblio_parsed->{$biblionumber} = $source;
+               return $source;
+       }
+
        my $transformed = $parsed->transform($source);
        $biblio_html->{$biblionumber} = $parsed->output_string( $transformed );
 
-       return ( $biblio_html->{$biblionumber}, $source ) if wantarray;
+       delete $biblio_parsed->{$biblionumber};
+
        return $biblio_html->{$biblionumber};
 }
 
@@ -120,9 +132,9 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
 
        $marcxml->{ $row->{biblionumber} } = $row->{marcxml};
 
-       my ( undef, $doc ) = biblioitem_html( $row->{biblionumber} );
+       my $doc = biblioitem_html( $row->{biblionumber}, 1 );
        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;
        }
 
@@ -139,7 +151,7 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
                '008' => undef,
                '100' => '9',
                '700' => '(9|4)',
-               '942' => 't'
+               '942' => '(t|r)'
        };
 
        my $data;
@@ -151,7 +163,12 @@ 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') {
@@ -169,6 +186,14 @@ 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'};
@@ -178,6 +203,9 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
                next;
        }
 
+       if ( my $in_database = $data->{942}->[0]->{'r'} ) {
+               $biblio_in_database->{ $row->{biblionumber} } = $in_database;
+       }
 
        my $have_100 = 1;
 
@@ -225,6 +253,8 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
 debug 'authors' => $authors;
 debug 'type_stats' => $type_stats;
 debug 'skip' => $skip;
+debug 'biblio_year' => $biblio_year;
+debug 'biblio_in_database' => $biblio_in_database;
 
 my $category_label;
 my $sth_categories = $dbh->prepare(q{
@@ -303,7 +333,7 @@ foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) {
        print $fh qq|<h1>$row->{full_name} - bibliografija za razdoblje 2008-2013</h1>|;
 
        author_html( $fh, $row->{authid}, 'aut' => 'Primarno autorstvo' );
-       author_html( $fh, $row->{authid}, 'sec' => 'Sekundarno autorstvo' );
+       author_html( $fh, $row->{authid}, 'sec' => 'Uredništva, prijevodi, krička izdanja' );
 
        print $fh html_end;
        close($fh);
@@ -333,6 +363,13 @@ 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 ) {
@@ -351,7 +388,9 @@ 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;
 
@@ -371,7 +410,7 @@ 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;
 
@@ -392,20 +431,27 @@ print $dep_fh qq|</ul>\n|, html_end;
 close($dep_fh);
 rename 'html/departments/index.new', 'html/departments/index.html';
 
-my $azvo_stat;
+my $azvo_stat_authors;
+my $azvo_stat_biblio;
 
 foreach my $department ( sort keys %$department_category_author ) {
        foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
                foreach my $authid ( @{ $department_category_author->{$department}->{$category} } ) {
                        foreach my $type ( keys %{ $authors->{$authid} } ) {
                                next unless exists $authors->{$authid}->{$type}->{$category};
-                               $azvo_stat->{ $department }->{ $category }->{ $type } += $#{ $authors->{$authid}->{$type}->{$category} } + 1;
+                               $azvo_stat_authors->{ $department }->{ $category }->{ $type } += $#{ $authors->{$authid}->{$type}->{$category} } + 1;
+                               push @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type } },  @{ $authors->{$authid}->{$type}->{$category} };
                        }
                }
+               foreach my $type ( keys %{ $azvo_stat_biblio->{ $department }->{ $category } } ) {
+                               my @biblios = unique_biblionumber @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type } };
+                               $azvo_stat_biblio->{ $department }->{ $category }->{ $type } = $#biblios + 1;
+               }
        }
 }
 
-debug 'azvo_stat' => $azvo_stat;
+debug 'azvo_stat_authors' => $azvo_stat_authors;
+debug 'azvo_stat_biblio' => $azvo_stat_biblio;
 
 =for later
 open(my $fh, '>', 'html/azvo.new');