extract year filter into own sub
[koha-bibliografija] / html.pl
diff --git a/html.pl b/html.pl
index 79d56bb..ee49703 100755 (executable)
--- a/html.pl
+++ b/html.pl
@@ -11,6 +11,9 @@ use autodie;
 use locale;
 use Text::Unaccent;
 use Carp qw(confess);
+use utf8;
+use JSON;
+use POSIX qw(strftime);
 
 use lib '/srv/koha_ffzg';
 use C4::Context;
@@ -26,9 +29,23 @@ sub debug {
 
 my $xslfilename = 'compact.xsl';
 
+my $azvo_group_title = {
+'znanstveno nastavni' => qr/(profes|docent|znanstveni savjetnik|znanstveni suradnik)/i,
+'lektori i predavači' => qr/(lektor|predavač)/i,
+'asistenti i novaci' => qr/(asistent|novak)/i,
+};
+
+my $department_groups = {
+'AAB_humanističke'            => qr/(anglistiku|arheologiju|antropologiju|filozofiju|fonetiku|germanistiku|hungarologiju|indologiju|slavenske|filologiju|komparativnu|kroatistiku|lingvistiku|povijest|romanistiku|talijanistiku)/i,
+'AAC_društvene'                       => qr/(informacijske|pedagogiju|psihologiju|sociologiju)/i,
+};
+
 my $auth_header;
 my $auth_department;
+my $auth_group;
 my @authors;
+my $department_in_sum;
+my $department_in_group;
 
 my $skip;
 
@@ -36,7 +53,8 @@ my $sth_auth = $dbh->prepare(q{
 select
        authid,
        ExtractValue(marcxml,'//datafield[@tag="100"]/subfield[@code="a"]') as full_name,
-       ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="a"]') as department
+       ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="a"]') as department,
+       ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="i"]') as academic_title
 from auth_header
 });
 
@@ -49,13 +67,44 @@ while( my $row = $sth_auth->fetchrow_hashref ) {
        $auth_header->{ $row->{authid} } = $row->{full_name};
        $row->{department} =~ s/, Filozofski fakultet u Zagrebu.*$//;
        $row->{department} =~ s/^.+\.\s*//;
-#      warn dump( $row );
+       $row->{department} =~ s/\s+$//s;
+       my $group;
+       foreach my $title ( keys %$azvo_group_title ) {
+               if ( $row->{academic_title} =~ $azvo_group_title->{$title} ) {
+                       $group = $title;
+                       last;
+               }
+       }
+       if ( $group ) {
+               $row->{academic_group} = $group;
+               $auth_group->{ $row->{authid} } = $group;
+               $skip->{group_stat}->{$group}++;
+       } else {
+               push @{ $skip->{no_academic_group} }, $row;
+       }
+
+#      warn "# ", dump( $row );
        push @{ $auth_department->{ $row->{department} } }, $row->{authid};
        push @authors, $row;
+       $department_in_sum->{ $row->{department} }++;
+       foreach my $name ( keys %$department_groups ) {
+               my $regex = $department_groups->{$name};
+               if ( $row->{department} =~ $regex ) {
+                       $department_in_group->{ $row->{department} } = $name;
+                       last;
+               }
+       }
+}
 
+debug 'department_in_group' => $department_in_group;
+
+foreach my $department ( keys %$department_in_sum ) {
+       $department_in_sum->{$department} = 0 unless $department =~ m/(centar|croaticum|katedra|odsjek)/i;
 }
 
 debug 'auth_department' => $auth_department;
+debug 'auth_group' => $auth_group;
+debug 'department_in_sum' => $department_in_sum;
 
 
 my $authors;
@@ -91,6 +140,8 @@ my $parsed = $xslt->parse_stylesheet($style_doc);
 
 my $biblio_html;
 my $biblio_parsed;
+my $biblio_data;
+my $biblio_author_external;
 
 open(my $xml_fh, '>', '/tmp/bibliografija.xml') if $ENV{XML};
 
@@ -148,9 +199,10 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
 
        my $extract = {
                '008' => undef,
-               '100' => '9',
-               '700' => '(9|4)',
-               '942' => 't'
+               '100' => '(9|a)',
+               '680' => 'i',
+               '700' => '(9|4|a)',
+               '942' => '(t|r|v)'
        };
 
        my $data;
@@ -162,12 +214,14 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
 
         if ($elt->localname eq 'controlfield') {
                        if ( $tag eq '008' ) {
-                               my $year = substr($elt->textContent, 7, 4 );
+                               my $content = $elt->textContent;
+                               my $year = substr($content, 7, 4 );
                                if ( $year !~ m/^\d+$/ ) {
                                        $year = 0;
                                        push @{ $skip->{invalid_year} }, $row->{biblionumber};
                                }
                                $biblio_year->{ $row->{biblionumber} } = $data->{year} = $year;
+                               $data->{'008'} = $content;
                        }
                        next;
         } elsif ($elt->localname eq 'datafield') {
@@ -182,15 +236,19 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
                                }
                        }
                        push @{ $data->{$tag} }, $sf_data if $sf_data;
-        }
-    }
+               }
+       }
 
-       if ( $data->{year} < 2008 ) {
+       if ( ! defined $data->{year} ) {
+               warn "MISSING year in ", $row->{biblionumber};
+=for remove-year-limit
+       } elsif ( $data->{year} < 2008 ) {
                push @{ $skip->{year_lt_2008} }, $row->{biblionumber};
                next;
        } elsif ( $data->{year} > 2013 ) {
                push @{ $skip->{year_gt_2013} }, $row->{biblionumber};
                next;
+=cut
        }
 
 #      warn "# ", $row->{biblionumber}, " data ",dump($data);
@@ -202,11 +260,22 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
                next;
        }
 
-
        my $have_100 = 1;
 
        if ( exists $data->{100} ) {
-                       my @first_author = map { $_->{'9'} } @{ $data->{100} };
+                       my @first_author =
+                               map { $_->{'9'} }
+                               grep {
+                                       if ( ! exists $_->{9} ) {
+                                               $biblio_author_external->{ $row->{biblionumber} }++;
+                                               0;
+                                       } elsif ( exists $auth_header->{ $_->{9} } ) {
+                                               1; # from FFZXG
+                                       } else {
+                                               0;
+                                       }
+                               }
+                               @{ $data->{100} };
                        foreach my $authid ( @first_author ) {
                                push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
                        }
@@ -217,7 +286,19 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
        my $have_edt;
 
        if ( exists $data->{700} ) {
-                       foreach my $auth ( @{ $data->{700} } ) {
+                       my @other_authors =
+                               grep {
+                                       if ( ! exists $_->{9} ) {
+                                               $biblio_author_external->{ $row->{biblionumber} }++;
+                                               0;
+                                       } elsif ( exists $auth_header->{ $_->{9} } ) {
+                                               1; # from FFZXG
+                                       } else {
+                                               0;
+                                       }
+                               }
+                               @{ $data->{700} };
+                       foreach my $auth ( @other_authors ) {
                                my $authid = $auth->{9} || next;
                                my $type   = $auth->{4} || next; #die "no 4 in ",dump($data);
 
@@ -242,14 +323,19 @@ while( my $row = $sth_select_authors->fetchrow_hashref ) {
                                        $skip->{ 'no_700$4' }->{ $row->{biblionumber} }++;
                                }
                        }
+                       delete $data->{700};
        }
 
+       $biblio_data->{ $row->{biblionumber} } = $data;
+
 }
 
 debug 'authors' => $authors;
 debug 'type_stats' => $type_stats;
 debug 'skip' => $skip;
 debug 'biblio_year' => $biblio_year;
+debug 'biblio_data' => $biblio_data;
+debug 'biblio_author_external' => $biblio_author_external;
 
 my $category_label;
 my $sth_categories = $dbh->prepare(q{
@@ -268,13 +354,17 @@ sub html_title {
 <meta charset="UTF-8">
 <title>|, join(" ", @_), qq|</title>
 <link href="style.css" type="text/css" rel="stylesheet" />
+<script src="//code.jquery.com/jquery-1.11.2.js"></script>
 </head>
 <body>
 |;
 }
 
 sub html_end {
-       return qq|</body>\n</html\n|;
+       return
+               qq|<small style="color:gray">Zadnji puta osvježeno: |,
+               strftime("%Y-%m-%d %H:%M:%S\n", localtime()),
+               qq|</body>\n</html>\n|;
 }
 
 mkdir 'html' unless -d 'html';
@@ -288,7 +378,7 @@ debug 'authors' => \@authors;
 
 sub li_biblio {
        my ($biblionumber) = @_;
-       return qq|<li>|,
+       return qq|<li class="y|, $biblio_year->{$biblionumber}, qq|">|,
                qq|<a href="https://koha.ffzg.hr/cgi-bin/koha/opac-detail.pl?biblionumber=$biblionumber">$biblionumber</a>|,
                biblioitem_html($biblionumber),
                qq|<a href="https://koha.ffzg.hr:8443/cgi-bin/koha/cataloguing/addbiblio.pl?biblionumber=$biblionumber">edit</a>|,
@@ -304,14 +394,122 @@ sub author_html {
 
        foreach my $category ( sort keys %{ $authors->{$authid}->{$type} } ) {
                my $label = $category_label->{$category} || 'Bez kategorije';
-               print $fh qq|<h3>$label</h3>\n<ul>\n|;
-               foreach my $biblionumber ( @{ $authors->{$authid}->{$type}->{$category} } ) {
+               print $fh qq|<a name="$type-$category"><h3>$label</h3></a>\n<ul>\n|;
+               foreach my $biblionumber ( sort {
+                               $biblio_year->{$b} <=> $biblio_year->{$a} || $a <=> $b
+                       } @{ $authors->{$authid}->{$type}->{$category} } ) {
                        print $fh li_biblio( $biblionumber );
                }
                print $fh qq|</ul>\n|;
        }
 }
 
+sub count_author_years {
+       my $years = shift;
+       my ($authid) = @_;
+       foreach my $type ( keys %{ $authors->{$authid} } ) {
+               foreach my $category ( keys %{ $authors->{$authid}->{$type} } ) {
+                       foreach my $biblionumber ( @{ $authors->{$authid}->{$type}->{$category} } ) {
+                               $years->{ $biblio_year->{ $biblionumber } }->{ $type . '-' . $category }++;
+                       }
+               }
+       }
+       return $years;
+}
+
+sub html_year_selection {
+       my $fh = shift;
+       my @authids = @_;
+
+       print $fh qq|<span id="years">Godine:\n|;
+       my $type_cat_count = {};
+       my $years;
+
+       foreach my $authid ( @authids ) {
+               $years = count_author_years( $years, $authid );
+       }
+
+       foreach my $year ( sort { $b <=> $a } keys %$years ) {
+               print $fh qq|<label><input name="year_selection" value="$year" type=checkbox onClick="toggle_year($year, this)" checked="checked">$year</label>&nbsp;\n|;
+               foreach my $type_cat ( keys %{ $years->{$year} } ) {
+                       $type_cat_count->{ $type_cat } += $years->{$year}->{$type_cat};
+               }
+       }
+
+       print $fh qq|
+<input type=button value="all" onClick="all_years(1)">
+<input type=button value="none" onClick="all_years(0)">
+       |;
+
+       print $fh qq|</span>|;
+
+       print $fh q|
+<script>
+
+var years = |, encode_json($years), q|;
+
+var type_cat_count = |, encode_json($type_cat_count), q|;
+
+function year_show(year) {
+       $('.y'+year).show();
+       console.debug('show', year);
+       for(var type_cat in years[year]) {
+               if ( ( type_cat_count[ type_cat ] += years[year][type_cat] ) == years[year][type_cat]) {
+                       $('a[name="'+type_cat+'"]').show();
+                       console.debug(type_cat, 'show');
+               }
+       }
+}
+
+function year_hide(year) {
+       $('.y'+year).hide();
+       console.debug('hide', year);
+       for(var type_cat in years[year]) {
+               if ( ( type_cat_count[ type_cat ] -= years[year][type_cat] ) == 0 ) {
+                       $('a[name="'+type_cat+'"]').hide();
+                       console.debug(type_cat, 'hide');
+               }
+       }
+}
+
+function toggle_year(year, el) {
+       if ( el.checked ) {
+               year_show(year);
+       } else {
+               year_hide(year);
+       }
+}
+
+function all_years( turn_on ) {
+       $('input[name=year_selection]').each( function(i,el) {
+               if ( turn_on ) {
+                       if ( ! el.checked ) {
+                               el.checked = true;
+                               year_show( el.value );
+                       }
+               } else {
+                       if ( el.checked ) {
+                               el.checked = false;
+                               year_hide( el.value );
+                       }
+               }
+       } );
+}
+
+$(document).ready( function() {
+       console.info('ready');
+
+       $('input[name=year_selection]').each( function(i, el) {
+               var year = el.value;
+               console.debug( 'on load', year, el.checked );
+               if (! el.checked) year_hide(year);
+       });
+
+});
+</script>
+       |;
+}
+
 foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) {
 
        my $first = substr( $row->{full_name}, 0, 1 );
@@ -325,10 +523,12 @@ foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) {
        my $path = "html/$row->{authid}";
        open(my $fh, '>:encoding(utf-8)', "$path.new");
        print $fh html_title($row->{full_name}, "bibliografija");
-       print $fh qq|<h1>$row->{full_name} - bibliografija za razdoblje 2008-2013</h1>|;
+       print $fh qq|<h1>$row->{full_name} - bibliografija</h1>\n|;
+
+       html_year_selection $fh => $row->{authid};
 
        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);
@@ -350,6 +550,12 @@ foreach my $department ( sort keys %$auth_department ) {
                push @categories,  keys %{ $authors->{$authid}->{sec} };
                foreach my $category ( sort @categories ) {
                        push @{ $department_category_author->{$department}->{$category} }, $authid;
+                       push @{ $department_category_author->{'AAA_ukupno'}->{$category} }, $authid if $department_in_sum->{$department};
+                       if ( my $group = $department_in_group->{ $department } ) {
+                               push @{ $department_category_author->{$group}->{$category} }, $authid;
+                       } else {
+                               $skip->{'department_not_in_group'}->{ $department }++;
+                       }
                }
        }
 }
@@ -426,27 +632,169 @@ 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_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} } ) {
+                       my $group = $auth_group->{$authid};
+                       if ( ! $group ) {
+                               push @{ $skip->{no_auth_group} }, $authid;
+                               next;
+                       }
                        foreach my $type ( keys %{ $authors->{$authid} } ) {
                                next unless exists $authors->{$authid}->{$type}->{$category};
-                               $azvo_stat->{ $department }->{ $category }->{ $type } += $#{ $authors->{$authid}->{$type}->{$category} } + 1;
+                               push @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{$group} },  @{ $authors->{$authid}->{$type}->{$category} };
+                               push @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{''} },  @{ $authors->{$authid}->{$type}->{$category} };
+                       }
+               }
+               foreach my $type ( keys %{ $azvo_stat_biblio->{ $department }->{ $category } } ) {
+                       foreach my $group ( keys %{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type } } ) {
+                               my @biblios = unique_biblionumber @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{ $group } };
+                               $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{ $group } = [ @biblios ];
                        }
                }
        }
 }
 
-debug 'azvo_stat' => $azvo_stat;
+debug 'azvo_stat_biblio' => $azvo_stat_biblio;
+
+my @report_lines;
+my @report_labels;
+
+my $label;
+my $sub_labels;
+open(my $report, '<:encoding(utf-8)', 'AZVO.txt');
+while( <$report> ) {
+       chomp;
+       if ( /^([^\t]+)\t+(.+)/ ) {
+               $label = $1;
+               push @report_labels, $label;
+               my $type = [ map { m/\s+/ ? [ split(/\s+/,$_) ] : [ $_, 'aut' ] } split (/\s*\+\s*/, $2) ];
+               push @report_lines, [ $label, @$type ];
+       } elsif ( /^\t+([^\t]+):\t+(\d+)(\w*)\t*(.*)$/ ) {
+               push @{ $sub_labels->{$label} }, [ $1, $2, $3, $4 ];
+               my $sub_label = $1;
+               pop (@report_labels) if ( $report_labels[ $#report_labels ] =~ m/^$label$/ ); # remove partial name
+               push @report_labels, $label . $sub_label;
+       } else {
+               die "ERROR: [$_]\n";
+       }
+}
+
+debug 'report_lines', \@report_lines;
+debug 'sub_labels', $sub_labels;
+debug 'report_labels', \@report_labels;
+
+my @departments = ( sort { lc($a) cmp lc($b) } keys %$azvo_stat_biblio );
 
-=for later
-open(my $fh, '>', 'html/azvo.new');
+debug 'departments' => \@departments;
 
+my $department2col;
+$department2col->{ $departments[$_] } = $_ foreach ( 0 .. $#departments );
+my $label2row;
+$label2row->{ $report_labels[$_] } = $_ foreach ( 0 .. $#report_labels );
 
+my $table;
 
+sub table_count {
+       my $label = shift @_;
+       my $department = shift @_;
+       my $group = shift @_;
+       my @biblionumbers = @_;
+       my $unique;
+       $unique->{$_}++ foreach @biblionumbers;
+       my @bibs = keys %$unique;
+       $table->{ffzg}->{$group}->[ $label2row->{ $label } ]->[ $department2col->{$department} ] = scalar @bibs;
+       $table->{external}->{$group}->[ $label2row->{ $label } ]->[ $department2col->{$department} ] = scalar grep { $biblio_author_external->{$_} } @bibs;
+}
+
+foreach my $group ( '', keys %$azvo_group_title ) {
+
+foreach my $department ( @departments ) {
+       foreach my $line ( @report_lines ) {
+               my $label = $line->[0];
+               my @biblionumbers;
+               foreach ( 1 .. $#$line ) {
+                       my ( $category, $type ) = @{ $line->[ $_ ] };
+                       my $b = $azvo_stat_biblio->{ $department }->{$category}->{$type}->{$group};
+                       push @biblionumbers, @$b if $b;
+               }
+               if ( $sub_labels->{$label} ) {
+                       my $sub_stats;
+                       foreach my $biblionumber ( @biblionumbers ) {
+                               my $data = $biblio_data->{$biblionumber} || die "can't find biblionumber $biblionumber";
+                               foreach my $sub_label ( @{ $sub_labels->{$label} } ) {
+                                       my ( $sub_label, $field, $sf, $regex ) = @$sub_label;
+                                       if ( ! $regex ) {
+                                               push @{ $sub_stats->{ $sub_label } }, $biblionumber;
+                                               last;
+                                       }
+                                       if ( $field < 100 ) {
+                                               if ( $data->{$field} =~ m/$regex/ ) {
+                                                       push @{ $sub_stats->{ $sub_label } }, $biblionumber;
+                                                       last;
+                                               }
+                                       } else {
+                                               if ( exists $data->{$field}->[0]->{$sf} && $data->{$field}->[0]->{$sf} =~ m/$regex/ ) {
+                                                       push @{ $sub_stats->{ $sub_label } }, $biblionumber;
+                                                       last;
+                                               }
+                                       }
+                               }
+                       }
+                       foreach my $sub_label ( keys %$sub_stats ) {
+                               my $full_label = $label . $sub_label;
+                               table_count $full_label, $department, $group, @{ $sub_stats->{$sub_label} };
+                       }
+               } else {
+                       table_count $label, $department, $group, @biblionumbers;
+               }
+       }
+}
+
+} # group
+
+debug 'table', $table;
+
+open(my $fh, '>:encoding(utf-8)', 'html/azvo.new');
+open(my $fh2, '>:encoding(utf-8)', 'html/azvo2.new');
+
+sub print_fh {
+       print $fh @_;
+       print $fh2 @_;
+}
+
+print $fh html_title('AZVO tablica - FFZG');
+print $fh2 html_title('AZVO tablica - kolaboracija sa FFZG');
+
+foreach my $group ( keys %{ $table->{ffzg} } ) {
+
+               print_fh "<h1>$group</h1>" if $group;
+
+               print_fh "<table border=1>\n";
+               print_fh "<tr><th></th>";
+               print_fh "<th>$_</th>" foreach @departments;
+               print_fh "</tr>\n";
+
+               foreach my $row ( 0 .. $#{ $table->{ffzg}->{$group} } ) {
+                       print_fh "<tr><th>", $report_labels[$row], "</th>\n";
+                       foreach ( 0 .. $#departments ) {
+                               print_fh "<td>";
+                               print $fh $table->{ffzg}->{$group}->[ $row ]->[ $_ ] || '';
+                               print $fh2 $table->{external}->{$group}->[ $row ]->[ $_ ] || '';
+                               print_fh "</td>\n"
+                       }
+                       print_fh "</tr>\n";
+               }
+
+               print_fh "</table>\n";
+
+} # group
+
+print_fh html_end;
 close($fh);
+close($fh2);
 rename 'html/azvo.new', 'html/azvo.html';
-=cut
+rename 'html/azvo2.new', 'html/azvo2.html';