3 # LC_COLLATE=hr_HR.utf8 KOHA_CONF=/etc/koha/sites/ffzg/koha-conf.xml ./html.pl
9 use Data::Dump qw(dump);
16 use POSIX qw(strftime);
19 use lib '/srv/koha_ffzg';
24 my $dbh = C4::Context->dbh;
27 my ($title, $data) = @_;
28 print "# $title ",dump($data), $/ if $ENV{DEBUG};
31 my $xslfilename = 'compact.xsl';
33 my $azvo_group_title = {
34 'znanstveno nastavni' => qr/(profes|docent|znanstveni savjetnik|znanstveni suradnik)/i,
35 'lektori i predavači' => qr/(lektor|predavač)/i,
36 'asistenti i novaci' => qr/(asistent|novak)/i,
39 my $department_groups = {
40 'AAB_humanističke' => qr/(anglistiku|arheologiju|antropologiju|filozofiju|fonetiku|germanistiku|hungarologiju|indologiju|slavenske|filologiju|komparativnu|kroatistiku|lingvistiku|povijest|romanistiku|talijanistiku)/i,
41 'AAC_društvene' => qr/(informacijske|pedagogiju|psihologiju|sociologiju)/i,
48 my $department_in_sum;
49 my $department_in_group;
53 my $sth_auth = $dbh->prepare(q{
56 ExtractValue(marcxml,'//datafield[@tag="100"]/subfield[@code="a"]') as full_name,
57 ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="a"]') as department,
58 ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="i"]') as academic_title
63 while( my $row = $sth_auth->fetchrow_hashref ) {
64 if ( $row->{department} !~ m/Filozofski fakultet u Zagrebu/ ) {
65 push @{ $skip->{nije_ffzg} }, $row;
68 $auth_header->{ $row->{authid} } = $row->{full_name};
69 $row->{department} =~ s/, Filozofski fakultet u Zagrebu.*$//;
70 $row->{department} =~ s/^.+\.\s*//;
71 $row->{department} =~ s/\s+$//s;
73 foreach my $title ( keys %$azvo_group_title ) {
74 if ( $row->{academic_title} =~ $azvo_group_title->{$title} ) {
80 $row->{academic_group} = $group;
81 $auth_group->{ $row->{authid} } = $group;
82 $skip->{group_stat}->{$group}++;
84 push @{ $skip->{no_academic_group} }, $row;
87 # warn "# ", dump( $row );
88 push @{ $auth_department->{ $row->{department} } }, $row->{authid};
90 $department_in_sum->{ $row->{department} }++;
91 foreach my $name ( keys %$department_groups ) {
92 my $regex = $department_groups->{$name};
93 if ( $row->{department} =~ $regex ) {
94 $department_in_group->{ $row->{department} } = $name;
100 debug 'department_in_group' => $department_in_group;
102 foreach my $department ( keys %$department_in_sum ) {
103 $department_in_sum->{$department} = 0 unless $department =~ m/(centar|croaticum|katedra|odsjek)/i;
106 debug 'auth_department' => $auth_department;
107 store $auth_department, '/dev/shm/auth_department.storable';
108 debug 'auth_group' => $auth_group;
109 debug 'department_in_sum' => $department_in_sum;
115 my $sth_select_authors = $dbh->prepare(q{
126 -- ExtractValue(marcxml,'//datafield[@tag="100"]/subfield[@code="9"]') as first_author,
127 -- ExtractValue(marcxml,'//datafield[@tag="700"]/subfield[@code="9"]') as other_authors,
128 -- ExtractValue(marcxml,'//datafield[@tag="942"]/subfield[@code="t"]') as category,
130 -- and SUBSTR(ExtractValue(marcxml,'//controlfield[@tag="008"]'),8,4) between 2008 and 2013
131 -- order by SUBSTR(ExtractValue(marcxml,'//controlfield[@tag="008"]'),8,4) desc
135 my $biblio_full_name;
138 my $parser = XML::LibXML->new();
139 $parser->recover_silently(0); # don't die when you find &, >, etc
140 my $style_doc = $parser->parse_file($xslfilename);
141 my $xslt = XML::LibXSLT->new();
142 my $parsed = $xslt->parse_stylesheet($style_doc);
147 my $biblio_author_external;
149 open(my $xml_fh, '>', '/tmp/bibliografija.xml') if $ENV{XML};
151 sub biblioitem_html {
152 my ($biblionumber, $parse_only) = @_;
154 return $biblio_html->{$biblionumber} if exists $biblio_html->{$biblionumber} && ! $parse_only;
156 my $xmlrecord = $marcxml->{$biblionumber} || confess "missing $biblionumber marcxml";
158 print $xml_fh $xmlrecord if $ENV{XML};
160 my $source = eval { $parser->parse_string($xmlrecord) };
162 # warn "SKIP $biblionumber corrupt XML";
163 push @{ $skip->{XML_corrupt} }, $biblionumber;
168 $biblio_parsed->{$biblionumber} = $source;
172 my $transformed = $parsed->transform($source);
173 $biblio_html->{$biblionumber} = $parsed->output_string( $transformed );
175 delete $biblio_parsed->{$biblionumber};
177 return $biblio_html->{$biblionumber};
180 $sth_select_authors->execute();
181 while( my $row = $sth_select_authors->fetchrow_hashref ) {
182 # warn dump($row),$/;
186 $marcxml->{ $row->{biblionumber} } = $row->{marcxml};
188 my $doc = biblioitem_html( $row->{biblionumber}, 1 );
190 # warn "ERROR can't parse MARCXML ", $row->{biblionumber}, " ", $row->{marcxml}, "\n";
194 my $root = $doc->documentElement;
196 my @leaders = $root->getElementsByLocalName('leader');
198 my $leader = $leaders[0]->textContent;
199 warn "leader $leader\n";
214 foreach my $elt ($root->getChildrenByLocalName('*')) {
215 my $tag = $elt->getAttribute('tag');
217 next unless exists $extract->{ $tag };
219 if ($elt->localname eq 'controlfield') {
220 if ( $tag eq '008' ) {
221 my $content = $elt->textContent;
222 my $year = substr($content, 7, 4 );
223 if ( $year !~ m/^\d+$/ ) {
225 push @{ $skip->{invalid_year} }, $row->{biblionumber};
227 $biblio_year->{ $row->{biblionumber} } = $data->{year} = $year;
228 $data->{'008'} = $content;
231 } elsif ($elt->localname eq 'datafield') {
233 foreach my $sfelt ($elt->getChildrenByLocalName('subfield')) {
234 my $sf = $sfelt->getAttribute('code');
235 next unless $sf =~ m/$extract->{$tag}/;
236 if ( exists $sf_data->{$sf} ) {
237 $sf_data->{$sf} .= " " . $sfelt->textContent();
239 $sf_data->{$sf} = $sfelt->textContent();
242 push @{ $data->{$tag} }, $sf_data if $sf_data;
246 if ( ! defined $data->{year} ) {
247 warn "MISSING year in ", $row->{biblionumber};
248 =for remove-year-limit
249 } elsif ( $data->{year} < 2008 ) {
250 push @{ $skip->{year_lt_2008} }, $row->{biblionumber};
252 } elsif ( $data->{year} > 2013 ) {
253 push @{ $skip->{year_gt_2013} }, $row->{biblionumber};
258 # warn "# ", $row->{biblionumber}, " data ",dump($data);
260 my $category = $data->{942}->[0]->{'t'};
262 # warn "# SKIP ", $row->{biblionumber}, " no category in ", dump($data);
263 push @{ $skip->{no_category} }, $row->{biblionumber};
269 if ( exists $data->{100} ) {
273 if ( ! exists $_->{9} ) {
274 $biblio_author_external->{ $row->{biblionumber} }++;
276 } elsif ( exists $auth_header->{ $_->{9} } ) {
283 foreach my $authid ( @first_author ) {
284 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
286 $biblio_full_name->{ $row->{biblionumber} } = $data->{100}->[0]->{a};
291 $biblio_full_name->{ $row->{biblionumber} } ||= $data->{245}->[0]->{a};
295 if ( exists $data->{700} ) {
298 if ( ! exists $_->{9} ) {
299 $biblio_author_external->{ $row->{biblionumber} }++;
301 } elsif ( exists $auth_header->{ $_->{9} } ) {
308 foreach my $auth ( @other_authors ) {
309 my $authid = $auth->{9} || next;
310 my $type = $auth->{4} || next; #die "no 4 in ",dump($data);
312 $type_stats->{$type}++;
314 if ( $type =~ m/(edt|trl|com|ctb)/ ) {
315 push @{ $authors->{$authid}->{sec}->{ $category } }, $row->{biblionumber};
316 push @{ $authors->{$authid}->{$1}->{ $category } }, $row->{biblionumber};
317 } elsif ( $type =~ m/aut/ ) {
319 $have_edt = grep { exists $_->{4} && $_->{4} =~ m/edt/ } @{ $data->{700} } if ! defined $have_edt;
321 $skip->{ have_700_edt }->{ $row->{biblionumber} }++;
323 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
326 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
329 # warn "# SKIP ", $row->{biblionumber}, ' no 700$4 in ', dump($data);
330 $skip->{ 'no_700$4' }->{ $row->{biblionumber} }++;
336 $biblio_data->{ $row->{biblionumber} } = $data;
340 debug 'authors' => $authors;
341 store $authors, '/dev/shm/authors.storable';
342 debug 'type_stats' => $type_stats;
343 debug 'skip' => $skip;
344 debug 'biblio_year' => $biblio_year;
345 debug 'biblio_full_name' => $biblio_full_name;
346 debug 'biblio_data' => $biblio_data;
347 debug 'biblio_author_external' => $biblio_author_external;
350 my $sth_categories = $dbh->prepare(q{
351 select authorised_value, lib from authorised_values where category = 'BIBCAT'
353 $sth_categories->execute();
354 while( my $row = $sth_categories->fetchrow_hashref ) {
355 $category_label->{ $row->{authorised_value} } = $row->{lib};
358 debug 'category_label' => $category_label;
363 <meta charset="UTF-8">
364 <title>|, join(" ", @_), qq|</title>
365 <link href="style.css" type="text/css" rel="stylesheet" />
366 <script src="//code.jquery.com/jquery-1.11.2.js"></script>
374 qq|<small style="color:gray">Zadnji puta osvježeno: |,
375 strftime("%Y-%m-%d %H:%M:%S\n", localtime()),
376 qq|</body>\n</html>\n|;
379 mkdir 'html' unless -d 'html';
381 open(my $index, '>:encoding(utf-8)', 'html/index.new');
382 print $index html_title('Bibliografija Filozofskog fakulteta');
384 my $first_letter = '';
386 debug 'authors' => \@authors;
389 my ($biblionumber) = @_;
390 return qq|<li class="y|, $biblio_year->{$biblionumber}, qq|">|,
391 qq|<a href="https://koha.ffzg.hr/cgi-bin/koha/opac-detail.pl?biblionumber=$biblionumber">$biblionumber</a>|,
392 biblioitem_html($biblionumber),
393 qq|<a href="https://koha.ffzg.hr:8443/cgi-bin/koha/cataloguing/addbiblio.pl?biblionumber=$biblionumber">edit</a>|,
397 sub unique_biblionumber {
400 $u->{$_}++ foreach @v;
402 $biblio_year->{$b} <=> $biblio_year->{$a} ||
403 $biblio_full_name->{$a} cmp $biblio_full_name->{$b} ||
409 my ( $fh, $authid, $type, $label ) = @_;
411 return unless exists $authors->{$authid}->{$type};
413 print $fh qq|<h2>$label</h2>\n|;
415 foreach my $category ( sort keys %{ $authors->{$authid}->{$type} } ) {
416 my $label = $category_label->{$category} || 'Bez kategorije';
417 print $fh qq|<a name="$type-$category"><h3>$label</h3></a>\n<ol>\n|;
418 foreach my $biblionumber ( unique_biblionumber @{ $authors->{$authid}->{$type}->{$category} } ) {
419 print $fh li_biblio( $biblionumber );
421 print $fh qq|</ol>\n|;
426 sub count_author_years {
429 foreach my $type ( keys %{ $authors->{$authid} } ) {
430 foreach my $category ( keys %{ $authors->{$authid}->{$type} } ) {
431 foreach my $biblionumber ( @{ $authors->{$authid}->{$type}->{$category} } ) {
432 $years->{ $biblio_year->{ $biblionumber } }->{ $type . '-' . $category }++;
439 sub html_year_selection {
443 print $fh qq|<span id="years">Godine:\n|;
444 my $type_cat_count = {};
447 foreach my $authid ( @authids ) {
448 $years = count_author_years( $years, $authid );
451 foreach my $year ( sort { $b <=> $a } keys %$years ) {
452 print $fh qq|<label><input name="year_selection" value="$year" type=checkbox onClick="toggle_year($year, this)" checked="checked">$year</label> \n|;
453 foreach my $type_cat ( keys %{ $years->{$year} } ) {
454 $type_cat_count->{ $type_cat } += $years->{$year}->{$type_cat};
459 <input type=button value="all" onClick="all_years(1)">
460 <input type=button value="none" onClick="all_years(0)">
463 print $fh qq|</span>|;
468 var years = |, encode_json($years), q|;
470 var type_cat_count = |, encode_json($type_cat_count), q|;
472 function year_show(year) {
474 console.debug('show', year);
475 for(var type_cat in years[year]) {
476 if ( ( type_cat_count[ type_cat ] += years[year][type_cat] ) == years[year][type_cat]) {
477 $('a[name="'+type_cat+'"]').show();
478 console.debug(type_cat, 'show');
483 function year_hide(year) {
485 console.debug('hide', year);
486 for(var type_cat in years[year]) {
487 if ( ( type_cat_count[ type_cat ] -= years[year][type_cat] ) == 0 ) {
488 $('a[name="'+type_cat+'"]').hide();
489 console.debug(type_cat, 'hide');
494 function toggle_year(year, el) {
502 function all_years( turn_on ) {
503 $('input[name=year_selection]').each( function(i,el) {
505 if ( ! el.checked ) {
507 year_show( el.value );
512 year_hide( el.value );
518 $(document).ready( function() {
519 console.info('ready');
521 $('input[name=year_selection]').each( function(i, el) {
523 console.debug( 'on load', year, el.checked );
524 if (! el.checked) year_hide(year);
532 foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) {
534 my $first = substr( $row->{full_name}, 0, 1 );
535 if ( $first ne $first_letter ) {
536 print $index qq{</ul>\n} if $first_letter;
537 $first_letter = $first;
538 print $index qq{<h1>$first</h1>\n<ul>\n};
540 print $index qq{<li><a href="}, $row->{authid}, qq{.html">}, $row->{full_name}, "</a></li>\n";
542 my $path = "html/$row->{authid}";
543 open(my $fh, '>:encoding(utf-8)', "$path.new");
544 print $fh html_title($row->{full_name}, "bibliografija");
545 print $fh qq|<h1>$row->{full_name} - bibliografija</h1>\n|;
547 html_year_selection $fh => $row->{authid};
549 author_html( $fh, $row->{authid}, 'aut' => 'Primarno autorstvo' );
550 author_html( $fh, $row->{authid}, 'sec' => 'Uredništva, prijevodi, kritička izdanja' );
554 rename "$path.new", "$path.html";
558 print $index html_end;
560 rename 'html/index.new', 'html/index.html';
562 debug 'auth_header' => $auth_header;
565 my $department_category_author;
566 foreach my $department ( sort keys %$auth_department ) {
567 foreach my $authid ( sort @{ $auth_department->{$department} } ) {
568 my @categories = keys %{ $authors->{$authid}->{aut} };
569 push @categories, keys %{ $authors->{$authid}->{sec} };
570 foreach my $category ( sort @categories ) {
571 push @{ $department_category_author->{$department}->{$category} }, $authid;
572 push @{ $department_category_author->{'AAA_ukupno'}->{$category} }, $authid if $department_in_sum->{$department};
573 if ( my $group = $department_in_group->{ $department } ) {
574 push @{ $department_category_author->{$group}->{$category} }, $authid;
576 $skip->{'department_not_in_group'}->{ $department }++;
582 debug 'department_category_author' => $department_category_author;
585 sub department_html {
586 my ( $fh, $department, $type, $label ) = @_;
588 print $fh qq|<h2>$label</h2>\n|;
590 foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
592 my @authids = @{ $department_category_author->{$department}->{$category} };
593 next unless @authids;
595 my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{$type}->{$category} } } grep { exists $authors->{$_}->{$type}->{$category} } @authids;
597 next unless @biblionumber;
599 my $label = $category_label->{$category} || 'Bez kategorije';
600 print $fh qq|<a name="$type-$category"><h3>$label</h3></a>\n<ol>\n|;
602 print $fh li_biblio( $_ ) foreach @biblionumber;
610 mkdir 'html/departments' unless -d 'html/departments';
612 open(my $dep_fh, '>:encoding(utf-8)', 'html/departments/index.new');
613 print $dep_fh html_title('Odsjeci Filozofskog fakulteta u Zagrebu'), qq|<ul>\n|;
614 foreach my $department ( sort keys %$department_category_author ) {
615 my $dep = $department || 'Nema odsjeka';
616 my $dep_file = unac_string('utf-8',$dep);
617 print $dep_fh qq|<li><a href="$dep_file.html">$dep</a></li>\n|;
618 open(my $fh, '>:encoding(utf-8)', "html/departments/$dep_file.new");
620 print $fh html_title($department . ' bibliografija');
621 print $fh qq|<h1>$department bibliografija</h1>\n|;
624 foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
625 push @authids, @{ $department_category_author->{$department}->{$category} };
627 html_year_selection $fh => @authids;
629 department_html( $fh, $department, 'aut', 'Primarno autorstvo' );
631 department_html( $fh, $department, 'sec', 'Sekundarno autorstvo' );
635 rename "html/departments/$dep_file.new", "html/departments/$dep_file.html";
637 print $dep_fh qq|</ul>\n|, html_end;
639 rename 'html/departments/index.new', 'html/departments/index.html';
641 my $azvo_stat_biblio;
643 foreach my $department ( sort keys %$department_category_author ) {
644 foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
645 foreach my $authid ( @{ $department_category_author->{$department}->{$category} } ) {
646 my $group = $auth_group->{$authid};
648 push @{ $skip->{no_auth_group} }, $authid;
651 foreach my $type ( keys %{ $authors->{$authid} } ) {
652 next unless exists $authors->{$authid}->{$type}->{$category};
653 push @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{$group} }, @{ $authors->{$authid}->{$type}->{$category} };
654 push @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{''} }, @{ $authors->{$authid}->{$type}->{$category} };
657 foreach my $type ( keys %{ $azvo_stat_biblio->{ $department }->{ $category } } ) {
658 foreach my $group ( keys %{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type } } ) {
659 my @biblios = unique_biblionumber @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{ $group } };
660 $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{ $group } = [ @biblios ];
666 debug 'azvo_stat_biblio' => $azvo_stat_biblio;
673 open(my $report, '<:encoding(utf-8)', 'AZVO.txt');
676 if ( /^([^\t]+)\t+(.+)/ ) {
678 push @report_labels, $label;
679 my $type = [ map { m/\s+/ ? [ split(/\s+/,$_) ] : [ $_, 'aut' ] } split (/\s*\+\s*/, $2) ];
680 push @report_lines, [ $label, @$type ];
681 } elsif ( /^\t+([^\t]+):\t+(\d+)(\w*)\t*(.*)$/ ) {
682 push @{ $sub_labels->{$label} }, [ $1, $2, $3, $4 ];
684 pop (@report_labels) if ( $report_labels[ $#report_labels ] =~ m/^$label$/ ); # remove partial name
685 push @report_labels, $label . $sub_label;
691 debug 'report_lines', \@report_lines;
692 debug 'sub_labels', $sub_labels;
693 debug 'report_labels', \@report_labels;
695 my @departments = ( sort { lc($a) cmp lc($b) } keys %$azvo_stat_biblio );
697 debug 'departments' => \@departments;
700 $department2col->{ $departments[$_] } = $_ foreach ( 0 .. $#departments );
702 $label2row->{ $report_labels[$_] } = $_ foreach ( 0 .. $#report_labels );
707 my $label = shift @_;
708 my $department = shift @_;
709 my $group = shift @_;
710 my @biblionumbers = @_;
712 $unique->{$_}++ foreach @biblionumbers;
713 my @bibs = keys %$unique;
714 $table->{ffzg}->{$group}->[ $label2row->{ $label } ]->[ $department2col->{$department} ] = scalar @bibs;
715 $table->{external}->{$group}->[ $label2row->{ $label } ]->[ $department2col->{$department} ] = scalar grep { $biblio_author_external->{$_} } @bibs;
718 foreach my $group ( '', keys %$azvo_group_title ) {
720 foreach my $department ( @departments ) {
721 foreach my $line ( @report_lines ) {
722 my $label = $line->[0];
724 foreach ( 1 .. $#$line ) {
725 my ( $category, $type ) = @{ $line->[ $_ ] };
726 my $b = $azvo_stat_biblio->{ $department }->{$category}->{$type}->{$group};
727 push @biblionumbers, @$b if $b;
729 if ( $sub_labels->{$label} ) {
731 foreach my $biblionumber ( @biblionumbers ) {
732 my $data = $biblio_data->{$biblionumber} || die "can't find biblionumber $biblionumber";
733 foreach my $sub_label ( @{ $sub_labels->{$label} } ) {
734 my ( $sub_label, $field, $sf, $regex ) = @$sub_label;
736 push @{ $sub_stats->{ $sub_label } }, $biblionumber;
739 if ( $field < 100 ) {
740 if ( $data->{$field} =~ m/$regex/ ) {
741 push @{ $sub_stats->{ $sub_label } }, $biblionumber;
745 if ( exists $data->{$field}->[0]->{$sf} && $data->{$field}->[0]->{$sf} =~ m/$regex/ ) {
746 push @{ $sub_stats->{ $sub_label } }, $biblionumber;
752 foreach my $sub_label ( keys %$sub_stats ) {
753 my $full_label = $label . $sub_label;
754 table_count $full_label, $department, $group, @{ $sub_stats->{$sub_label} };
757 table_count $label, $department, $group, @biblionumbers;
764 debug 'table', $table;
766 open(my $fh, '>:encoding(utf-8)', 'html/azvo.new');
767 open(my $fh2, '>:encoding(utf-8)', 'html/azvo2.new');
774 print $fh html_title('AZVO tablica - FFZG');
775 print $fh2 html_title('AZVO tablica - kolaboracija sa FFZG');
777 foreach my $group ( keys %{ $table->{ffzg} } ) {
779 print_fh "<h1>$group</h1>" if $group;
781 print_fh "<table border=1>\n";
782 print_fh "<tr><th></th>";
783 print_fh "<th>$_</th>" foreach @departments;
786 foreach my $row ( 0 .. $#{ $table->{ffzg}->{$group} } ) {
787 print_fh "<tr><th>", $report_labels[$row], "</th>\n";
788 foreach ( 0 .. $#departments ) {
790 print $fh $table->{ffzg}->{$group}->[ $row ]->[ $_ ] || '';
791 print $fh2 $table->{external}->{$group}->[ $row ]->[ $_ ] || '';
797 print_fh "</table>\n";
804 rename 'html/azvo.new', 'html/azvo.html';
805 rename 'html/azvo2.new', 'html/azvo2.html';