git.rot13.org Git - koha-bibliografija/blob - html.pl

   1 #!/usr/bin/perl
   2
   3 # LC_COLLATE=hr_HR.utf8 KOHA_CONF=/etc/koha/sites/ffzg/koha-conf.xml ./html.pl
   4
   5 use warnings;
   6 use strict;
   7
   8 use DBI;
   9 use Data::Dump qw(dump);
  10 use autodie;
  11 use locale;
  12 use Text::Unaccent;
  13 use Carp qw(confess);
  14 use utf8;
  15 use JSON;
  16
  17 use lib '/srv/koha_ffzg';
  18 use C4::Context;
  19 use XML::LibXML;
  20 use XML::LibXSLT;
  21
  22 my $dbh = C4::Context->dbh;
  23
  24 sub debug {
  25         my ($title, $data) = @_;
  26         print "# $title ",dump($data), $/ if $ENV{DEBUG};
  27 }
  28
  29 my $xslfilename = 'compact.xsl';
  30
  31 my $azvo_group_title = {
  32 'znanstveno nastavni' => qr/(profes|docent|znanstveni savjetnik|znanstveni suradnik)/i,
  33 'lektori i predavači' => qr/(lektor|predavač)/i,
  34 'asistenti i novaci' => qr/(asistent|novak)/i,
  35 };
  36
  37 my $department_groups = {
  38 'AAB_humanističke'             => qr/(anglistiku|arheologiju|antropologiju|filozofiju|fonetiku|germanistiku|hungarologiju|indologiju|slavenske|filologiju|komparativnu|kroatistiku|lingvistiku|povijest|romanistiku|talijanistiku)/i,
  39 'AAC_društvene'                        => qr/(informacijske|pedagogiju|psihologiju|sociologiju)/i,
  40 };
  41
  42 my $auth_header;
  43 my $auth_department;
  44 my $auth_group;
  45 my @authors;
  46 my $department_in_sum;
  47 my $department_in_group;
  48
  49 my $skip;
  50
  51 my $sth_auth = $dbh->prepare(q{
  52 select
  53         authid,
  54         ExtractValue(marcxml,'//datafield[@tag="100"]/subfield[@code="a"]') as full_name,
  55         ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="a"]') as department,
  56         ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="i"]') as academic_title
  57 from auth_header
  58 });
  59
  60 $sth_auth->execute();
  61 while( my $row = $sth_auth->fetchrow_hashref ) {
  62         if ( $row->{department} !~ m/Filozofski fakultet u Zagrebu/ ) {
  63                 push @{ $skip->{nije_ffzg} }, $row;
  64                 next;
  65         }
  66         $auth_header->{ $row->{authid} } = $row->{full_name};
  67         $row->{department} =~ s/, Filozofski fakultet u Zagrebu.*$//;
  68         $row->{department} =~ s/^.+\.\s*//;
  69         $row->{department} =~ s/\s+$//s;
  70         my $group;
  71         foreach my $title ( keys %$azvo_group_title ) {
  72                 if ( $row->{academic_title} =~ $azvo_group_title->{$title} ) {
  73                         $group = $title;
  74                         last;
  75                 }
  76         }
  77         if ( $group ) {
  78                 $row->{academic_group} = $group;
  79                 $auth_group->{ $row->{authid} } = $group;
  80                 $skip->{group_stat}->{$group}++;
  81         } else {
  82                 push @{ $skip->{no_academic_group} }, $row;
  83         }
  84
  85 #       warn "# ", dump( $row );
  86         push @{ $auth_department->{ $row->{department} } }, $row->{authid};
  87         push @authors, $row;
  88         $department_in_sum->{ $row->{department} }++;
  89         foreach my $name ( keys %$department_groups ) {
  90                 my $regex = $department_groups->{$name};
  91                 if ( $row->{department} =~ $regex ) {
  92                         $department_in_group->{ $row->{department} } = $name;
  93                         last;
  94                 }
  95         }
  96 }
  97
  98 debug 'department_in_group' => $department_in_group;
  99
 100 foreach my $department ( keys %$department_in_sum ) {
 101         $department_in_sum->{$department} = 0 unless $department =~ m/(centar|croaticum|katedra|odsjek)/i;
 102 }
 103
 104 debug 'auth_department' => $auth_department;
 105 debug 'auth_group' => $auth_group;
 106 debug 'department_in_sum' => $department_in_sum;
 107
 108
 109 my $authors;
 110 my $marcxml;
 111
 112 my $sth_select_authors  = $dbh->prepare(q{
 113 select
 114         biblionumber,
 115         itemtype,
 116         marcxml
 117 from biblioitems
 118 where
 119         agerestriction > 0
 120 });
 121
 122 =for sql
 123 --      ExtractValue(marcxml,'//datafield[@tag="100"]/subfield[@code="9"]') as first_author,
 124 --      ExtractValue(marcxml,'//datafield[@tag="700"]/subfield[@code="9"]') as other_authors,
 125 --      ExtractValue(marcxml,'//datafield[@tag="942"]/subfield[@code="t"]') as category,
 126
 127 --      and SUBSTR(ExtractValue(marcxml,'//controlfield[@tag="008"]'),8,4) between 2008 and 2013
 128 -- order by SUBSTR(ExtractValue(marcxml,'//controlfield[@tag="008"]'),8,4) desc
 129 =cut
 130
 131 my $biblio_year;
 132 my $type_stats;
 133
 134 my $parser = XML::LibXML->new();
 135 $parser->recover_silently(0); # don't die when you find &, >, etc
 136 my $style_doc = $parser->parse_file($xslfilename);
 137 my $xslt = XML::LibXSLT->new();
 138 my $parsed = $xslt->parse_stylesheet($style_doc);
 139
 140 my $biblio_html;
 141 my $biblio_parsed;
 142 my $biblio_data;
 143 my $biblio_author_external;
 144
 145 open(my $xml_fh, '>', '/tmp/bibliografija.xml') if $ENV{XML};
 146
 147 sub biblioitem_html {
 148         my ($biblionumber, $parse_only) = @_;
 149
 150         return $biblio_html->{$biblionumber} if exists $biblio_html->{$biblionumber} && ! $parse_only;
 151
 152         my $xmlrecord = $marcxml->{$biblionumber} || confess "missing $biblionumber marcxml";
 153
 154         print $xml_fh $xmlrecord if $ENV{XML};
 155
 156         my $source = eval { $parser->parse_string($xmlrecord) };
 157         if ( $@ ) {
 158 #               warn "SKIP $biblionumber corrupt XML";
 159                 push @{ $skip->{XML_corrupt} }, $biblionumber;
 160                 return;
 161         }
 162
 163         if ( $parse_only ) {
 164                 $biblio_parsed->{$biblionumber} = $source;
 165                 return $source;
 166         }
 167
 168         my $transformed = $parsed->transform($source);
 169         $biblio_html->{$biblionumber} = $parsed->output_string( $transformed );
 170
 171         delete $biblio_parsed->{$biblionumber};
 172
 173         return $biblio_html->{$biblionumber};
 174 }
 175
 176 $sth_select_authors->execute();
 177 while( my $row = $sth_select_authors->fetchrow_hashref ) {
 178 #       warn dump($row),$/;
 179
 180         my $biblio;
 181
 182         $marcxml->{ $row->{biblionumber} } = $row->{marcxml};
 183
 184         my $doc = biblioitem_html( $row->{biblionumber}, 1 );
 185         if ( ! $doc ) {
 186 #               warn "ERROR can't parse MARCXML ", $row->{biblionumber}, " ", $row->{marcxml}, "\n";
 187                 next;
 188         }
 189
 190         my $root = $doc->documentElement;
 191 =for leader
 192         my @leaders = $root->getElementsByLocalName('leader');
 193         if (@leaders) {
 194                 my $leader = $leaders[0]->textContent;
 195                 warn "leader $leader\n";
 196         }
 197 =cut
 198
 199         my $extract = {
 200                 '008' => undef,
 201                 '100' => '(9|a)',
 202                 '680' => 'i',
 203                 '700' => '(9|4|a)',
 204                 '942' => '(t|r|v)'
 205         };
 206
 207         my $data;
 208
 209         foreach my $elt ($root->getChildrenByLocalName('*')) {
 210                 my $tag = $elt->getAttribute('tag');
 211                 next if ! $tag;
 212                 next unless exists $extract->{ $tag };
 213
 214         if ($elt->localname eq 'controlfield') {
 215                         if ( $tag eq '008' ) {
 216                                 my $content = $elt->textContent;
 217                                 my $year = substr($content, 7, 4 );
 218                                 if ( $year !~ m/^\d+$/ ) {
 219                                         $year = 0;
 220                                         push @{ $skip->{invalid_year} }, $row->{biblionumber};
 221                                 }
 222                                 $biblio_year->{ $row->{biblionumber} } = $data->{year} = $year;
 223                                 $data->{'008'} = $content;
 224                         }
 225                         next;
 226         } elsif ($elt->localname eq 'datafield') {
 227                         my $sf_data;
 228             foreach my $sfelt ($elt->getChildrenByLocalName('subfield')) {
 229                 my $sf = $sfelt->getAttribute('code');
 230                                 next unless $sf =~ m/$extract->{$tag}/;
 231                                 if ( exists $sf_data->{$sf} ) {
 232                                         $sf_data->{$sf} .= " " . $sfelt->textContent();
 233                                 } else {
 234                                         $sf_data->{$sf} = $sfelt->textContent();
 235                                 }
 236                         }
 237                         push @{ $data->{$tag} }, $sf_data if $sf_data;
 238                 }
 239         }
 240
 241         if ( ! defined $data->{year} ) {
 242                 warn "MISSING year in ", $row->{biblionumber};
 243 =for remove-year-limit
 244         } elsif ( $data->{year} < 2008 ) {
 245                 push @{ $skip->{year_lt_2008} }, $row->{biblionumber};
 246                 next;
 247         } elsif ( $data->{year} > 2013 ) {
 248                 push @{ $skip->{year_gt_2013} }, $row->{biblionumber};
 249                 next;
 250 =cut
 251         }
 252
 253 #       warn "# ", $row->{biblionumber}, " data ",dump($data);
 254
 255         my $category = $data->{942}->[0]->{'t'};
 256         if ( ! $category ) {
 257 #               warn "# SKIP ", $row->{biblionumber}, " no category in ", dump($data);
 258                 push @{ $skip->{no_category} }, $row->{biblionumber};
 259                 next;
 260         }
 261
 262         my $have_100 = 1;
 263
 264         if ( exists $data->{100} ) {
 265                         my @first_author =
 266                                 map { $_->{'9'} }
 267                                 grep {
 268                                         if ( ! exists $_->{9} ) {
 269                                                 $biblio_author_external->{ $row->{biblionumber} }++;
 270                                                 0;
 271                                         } elsif ( exists $auth_header->{ $_->{9} } ) {
 272                                                 1; # from FFZXG
 273                                         } else {
 274                                                 0;
 275                                         }
 276                                 }
 277                                 @{ $data->{100} };
 278                         foreach my $authid ( @first_author ) {
 279                                 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 280                         }
 281         } else {
 282                 $have_100 = 0;
 283         }
 284
 285         my $have_edt;
 286
 287         if ( exists $data->{700} ) {
 288                         my @other_authors =
 289                                 grep {
 290                                         if ( ! exists $_->{9} ) {
 291                                                 $biblio_author_external->{ $row->{biblionumber} }++;
 292                                                 0;
 293                                         } elsif ( exists $auth_header->{ $_->{9} } ) {
 294                                                 1; # from FFZXG
 295                                         } else {
 296                                                 0;
 297                                         }
 298                                 }
 299                                 @{ $data->{700} };
 300                         foreach my $auth ( @other_authors ) {
 301                                 my $authid = $auth->{9} || next;
 302                                 my $type   = $auth->{4} || next; #die "no 4 in ",dump($data);
 303
 304                                 $type_stats->{$type}++;
 305
 306                                 if ( $type =~ m/(edt|trl|com|ctb)/ ) {
 307                                         push @{ $authors->{$authid}->{sec}->{ $category } }, $row->{biblionumber};
 308                                         push @{ $authors->{$authid}->{$1}->{ $category } }, $row->{biblionumber};
 309                                 } elsif ( $type =~ m/aut/ ) {
 310                                         if ( ! $have_100 ) {
 311                                                 $have_edt = grep { exists $_->{4} && $_->{4} =~ m/edt/ } @{ $data->{700} } if ! defined $have_edt;
 312                                                 if ( $have_edt ) {
 313                                                         $skip->{ have_700_edt }->{ $row->{biblionumber} }++;
 314                                                 } else {
 315                                                         push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 316                                                 }
 317                                         } else {
 318                                                 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 319                                         }
 320                                 } else {
 321 #                                       warn "# SKIP ", $row->{biblionumber}, ' no 700$4 in ', dump($data);
 322                                         $skip->{ 'no_700$4' }->{ $row->{biblionumber} }++;
 323                                 }
 324                         }
 325                         delete $data->{700};
 326         }
 327
 328         $biblio_data->{ $row->{biblionumber} } = $data;
 329
 330 }
 331
 332 debug 'authors' => $authors;
 333 debug 'type_stats' => $type_stats;
 334 debug 'skip' => $skip;
 335 debug 'biblio_year' => $biblio_year;
 336 debug 'biblio_data' => $biblio_data;
 337 debug 'biblio_author_external' => $biblio_author_external;
 338
 339 my $category_label;
 340 my $sth_categories = $dbh->prepare(q{
 341 select authorised_value, lib from authorised_values where category = 'BIBCAT'
 342 });
 343 $sth_categories->execute();
 344 while( my $row = $sth_categories->fetchrow_hashref ) {
 345         $category_label->{ $row->{authorised_value} } = $row->{lib};
 346
 347 }
 348 debug 'category_label' => $category_label;
 349
 350 sub html_title {
 351         return qq|<html>
 352 <head>
 353 <meta charset="UTF-8">
 354 <title>|, join(" ", @_), qq|</title>
 355 <link href="style.css" type="text/css" rel="stylesheet" />
 356 <script src="//code.jquery.com/jquery-1.11.2.js"></script>
 357 </head>
 358 <body>
 359 |;
 360 }
 361
 362 sub html_end {
 363         return qq|</body>\n</html>\n|;
 364 }
 365
 366 mkdir 'html' unless -d 'html';
 367
 368 open(my $index, '>:encoding(utf-8)', 'html/index.new');
 369 print $index html_title('Bibliografija Filozofskog fakulteta');
 370
 371 my $first_letter = '';
 372
 373 debug 'authors' => \@authors;
 374
 375 sub li_biblio {
 376         my ($biblionumber) = @_;
 377         return qq|<li class="y|, $biblio_year->{$biblionumber}, qq|">|,
 378                 qq|<a href="https://koha.ffzg.hr/cgi-bin/koha/opac-detail.pl?biblionumber=$biblionumber">$biblionumber</a>|,
 379                 biblioitem_html($biblionumber),
 380                 qq|<a href="https://koha.ffzg.hr:8443/cgi-bin/koha/cataloguing/addbiblio.pl?biblionumber=$biblionumber">edit</a>|,
 381                 qq|</li>\n|;
 382 }
 383
 384 sub author_html {
 385         my ( $fh, $authid, $type, $label ) = @_;
 386
 387         return unless exists $authors->{$authid}->{$type};
 388
 389         print $fh qq|<h2>$label</h2>\n|;
 390
 391         foreach my $category ( sort keys %{ $authors->{$authid}->{$type} } ) {
 392                 my $label = $category_label->{$category} || 'Bez kategorije';
 393                 print $fh qq|<a name="$type-$category"><h3>$label</h3></a>\n<ul>\n|;
 394                 foreach my $biblionumber ( sort {
 395                                 $biblio_year->{$b} <=> $biblio_year->{$a} || $a <=> $b
 396                         } @{ $authors->{$authid}->{$type}->{$category} } ) {
 397                         print $fh li_biblio( $biblionumber );
 398                 }
 399                 print $fh qq|</ul>\n|;
 400         }
 401 }
 402
 403 sub count_author_years {
 404         my ($authid) = @_;
 405         my $years;
 406         foreach my $type ( keys %{ $authors->{$authid} } ) {
 407                 foreach my $category ( keys %{ $authors->{$authid}->{$type} } ) {
 408                         foreach my $biblionumber ( @{ $authors->{$authid}->{$type}->{$category} } ) {
 409                                 $years->{ $biblio_year->{ $biblionumber } }->{ $type . '-' . $category }++;
 410                         }
 411                 }
 412         }
 413         return $years;
 414 }
 415
 416 foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) {
 417
 418         my $first = substr( $row->{full_name}, 0, 1 );
 419         if ( $first ne $first_letter ) {
 420                 print $index qq{</ul>\n} if $first_letter;
 421                 $first_letter = $first;
 422                 print $index qq{<h1>$first</h1>\n<ul>\n};
 423         }
 424         print $index qq{<li><a href="}, $row->{authid}, qq{.html">}, $row->{full_name}, "</a></li>\n";
 425
 426         my $path = "html/$row->{authid}";
 427         open(my $fh, '>:encoding(utf-8)', "$path.new");
 428         print $fh html_title($row->{full_name}, "bibliografija");
 429         print $fh qq|<h1>$row->{full_name} - bibliografija</h1>|;
 430
 431         my $years = count_author_years( $row->{authid} );
 432         print $fh qq|<span id="years">Godine:|;
 433         my $type_cat_count = {};
 434         foreach my $year ( sort { $b <=> $a } keys %$years ) {
 435                 print $fh qq|<label><input name="year_selection" value="$year" type=checkbox onClick="toggle_year($year, this)" checked>$year</label>&nbsp;\n|;
 436                 foreach my $type_cat ( keys %{ $years->{$year} } ) {
 437                         $type_cat_count->{ $type_cat } += $years->{$year}->{$type_cat};
 438                 }
 439         }
 440         print $fh qq|</span>|;
 441
 442         print $fh q|
 443 <script>
 444
 445 var years = |, encode_json($years), q|;
 446
 447 var type_cat_count = |, encode_json($type_cat_count), q|;
 448
 449 function year_show(year) {
 450         $('.y'+year).show();
 451         console.debug('show', year);
 452         for(var type_cat in years[year]) {
 453                 if ( ( type_cat_count[ type_cat ] += years[year][type_cat] ) == years[year][type_cat]) {
 454                         $('a[name="'+type_cat+'"]').show();
 455                         console.debug(type_cat, 'show');
 456                 }
 457         }
 458 }
 459
 460 function year_hide(year) {
 461         $('.y'+year).hide();
 462         console.debug('hide', year);
 463         for(var type_cat in years[year]) {
 464                 if ( ( type_cat_count[ type_cat ] -= years[year][type_cat] ) == 0 ) {
 465                         $('a[name="'+type_cat+'"]').hide();
 466                         console.debug(type_cat, 'hide');
 467                 }
 468         }
 469 }
 470
 471 function toggle_year(year, el) {
 472         if ( el.checked ) {
 473                 year_show(year);
 474         } else {
 475                 year_hide(year);
 476         }
 477 }
 478
 479 $(document).ready( function() {
 480         console.info('ready');
 481
 482         $('input[name=year_selection]').each( function(i, el) {
 483                 var year = el.value;
 484                 console.debug( 'on load', year, el.checked );
 485                 if (! el.checked) year_hide(year);
 486         });
 487
 488 });
 489 </script>
 490         |;
 491
 492         author_html( $fh, $row->{authid}, 'aut' => 'Primarno autorstvo' );
 493         author_html( $fh, $row->{authid}, 'sec' => 'Uredništva, prijevodi, krička izdanja' );
 494
 495         print $fh html_end;
 496         close($fh);
 497         rename "$path.new", "$path.html";
 498
 499 }
 500
 501 print $index html_end;
 502 close($index);
 503 rename 'html/index.new', 'html/index.html';
 504
 505 debug 'auth_header' => $auth_header;
 506
 507
 508 my $department_category_author;
 509 foreach my $department ( sort keys %$auth_department ) {
 510         foreach my $authid ( sort @{ $auth_department->{$department} } ) {
 511                 my   @categories = keys %{ $authors->{$authid}->{aut} };
 512                 push @categories,  keys %{ $authors->{$authid}->{sec} };
 513                 foreach my $category ( sort @categories ) {
 514                         push @{ $department_category_author->{$department}->{$category} }, $authid;
 515                         push @{ $department_category_author->{'AAA_ukupno'}->{$category} }, $authid if $department_in_sum->{$department};
 516                         if ( my $group = $department_in_group->{ $department } ) {
 517                                 push @{ $department_category_author->{$group}->{$category} }, $authid;
 518                         } else {
 519                                 $skip->{'department_not_in_group'}->{ $department }++;
 520                         }
 521                 }
 522         }
 523 }
 524
 525 debug 'department_category_author' => $department_category_author;
 526
 527 mkdir 'html/departments' unless -d 'html/departments';
 528
 529 sub unique_biblionumber {
 530         my @v = @_;
 531         my $u;
 532         $u->{$_}++ foreach @v;
 533         return sort { $biblio_year->{$b} <=> $biblio_year->{$a} || $a <=> $b } keys %$u;
 534 }
 535
 536 open(my $dep_fh, '>:encoding(utf-8)', 'html/departments/index.new');
 537 print $dep_fh html_title('Odsijeci Filozofskog fakulteta u Zagrebu'), qq|<ul>\n|;
 538 foreach my $department ( sort keys %$department_category_author ) {
 539         my $dep = $department || 'Nema odsjeka';
 540         my $dep_file = unac_string('utf-8',$dep);
 541         print $dep_fh qq|<li><a href="$dep_file.html">$dep</a></li>\n|;
 542         open(my $fh, '>:encoding(utf-8)', "html/departments/$dep_file.new");
 543
 544         print $fh html_title($department . ' bibliografija');
 545         print $fh qq|<h1>$department bibliografija</h1>\n|;
 546
 547         print $fh qq|<h2>Primarno autorstvo</h2>\n|;
 548
 549         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 550
 551                 my @authids = @{ $department_category_author->{$department}->{$category} };
 552                 next unless @authids;
 553
 554                 my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{aut}->{$category} } } grep { exists $authors->{$_}->{aut}->{$category} } @authids;
 555                 my $unique;
 556                 $unique->{$_}++ foreach @biblionumber;
 557
 558                 next unless @biblionumber;
 559
 560                 my $label = $category_label->{$category} || 'Bez kategorije';
 561                 print $fh qq|<h3>$label</h3>\n<ul>\n|;
 562
 563                 print $fh li_biblio( $_ ) foreach @biblionumber;
 564
 565                 print $fh qq|</ul>|;
 566         }
 567
 568
 569         print $fh qq|<h2>Sekundarno autorstvo</h2>\n|;
 570
 571         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 572
 573                 my @authids = @{ $department_category_author->{$department}->{$category} };
 574                 next unless @authids;
 575
 576                 my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{sec}->{$category} } } grep { exists $authors->{$_}->{sec}->{$category} } @authids;
 577
 578                 next unless @biblionumber;
 579
 580                 my $label = $category_label->{$category} || 'Bez kategorije';
 581                 print $fh qq|<h3>$label</h3>\n<ul>\n|;
 582
 583                 print $fh li_biblio( $_ ) foreach @biblionumber;
 584
 585                 print $fh qq|</ul>|;
 586         }
 587
 588
 589         print $fh html_end;
 590         close($fh);
 591         rename "html/departments/$dep_file.new", "html/departments/$dep_file.html";
 592 }
 593 print $dep_fh qq|</ul>\n|, html_end;
 594 close($dep_fh);
 595 rename 'html/departments/index.new', 'html/departments/index.html';
 596
 597 my $azvo_stat_biblio;
 598
 599 foreach my $department ( sort keys %$department_category_author ) {
 600         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 601                 foreach my $authid ( @{ $department_category_author->{$department}->{$category} } ) {
 602                         my $group = $auth_group->{$authid};
 603                         if ( ! $group ) {
 604                                 push @{ $skip->{no_auth_group} }, $authid;
 605                                 next;
 606                         }
 607                         foreach my $type ( keys %{ $authors->{$authid} } ) {
 608                                 next unless exists $authors->{$authid}->{$type}->{$category};
 609                                 push @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{$group} },  @{ $authors->{$authid}->{$type}->{$category} };
 610                                 push @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{''} },  @{ $authors->{$authid}->{$type}->{$category} };
 611                         }
 612                 }
 613                 foreach my $type ( keys %{ $azvo_stat_biblio->{ $department }->{ $category } } ) {
 614                         foreach my $group ( keys %{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type } } ) {
 615                                 my @biblios = unique_biblionumber @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{ $group } };
 616                                 $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{ $group } = [ @biblios ];
 617                         }
 618                 }
 619         }
 620 }
 621
 622 debug 'azvo_stat_biblio' => $azvo_stat_biblio;
 623
 624 my @report_lines;
 625 my @report_labels;
 626
 627 my $label;
 628 my $sub_labels;
 629 open(my $report, '<:encoding(utf-8)', 'AZVO.txt');
 630 while( <$report> ) {
 631         chomp;
 632         if ( /^([^\t]+)\t+(.+)/ ) {
 633                 $label = $1;
 634                 push @report_labels, $label;
 635                 my $type = [ map { m/\s+/ ? [ split(/\s+/,$_) ] : [ $_, 'aut' ] } split (/\s*\+\s*/, $2) ];
 636                 push @report_lines, [ $label, @$type ];
 637         } elsif ( /^\t+([^\t]+):\t+(\d+)(\w*)\t*(.*)$/ ) {
 638                 push @{ $sub_labels->{$label} }, [ $1, $2, $3, $4 ];
 639                 my $sub_label = $1;
 640                 pop (@report_labels) if ( $report_labels[ $#report_labels ] =~ m/^$label$/ ); # remove partial name
 641                 push @report_labels, $label . $sub_label;
 642         } else {
 643                 die "ERROR: [$_]\n";
 644         }
 645 }
 646
 647 debug 'report_lines', \@report_lines;
 648 debug 'sub_labels', $sub_labels;
 649 debug 'report_labels', \@report_labels;
 650
 651 my @departments = ( sort { lc($a) cmp lc($b) } keys %$azvo_stat_biblio );
 652
 653 debug 'departments' => \@departments;
 654
 655 my $department2col;
 656 $department2col->{ $departments[$_] } = $_ foreach ( 0 .. $#departments );
 657 my $label2row;
 658 $label2row->{ $report_labels[$_] } = $_ foreach ( 0 .. $#report_labels );
 659
 660 my $table;
 661
 662 sub table_count {
 663         my $label = shift @_;
 664         my $department = shift @_;
 665         my $group = shift @_;
 666         my @biblionumbers = @_;
 667         my $unique;
 668         $unique->{$_}++ foreach @biblionumbers;
 669         my @bibs = keys %$unique;
 670         $table->{ffzg}->{$group}->[ $label2row->{ $label } ]->[ $department2col->{$department} ] = scalar @bibs;
 671         $table->{external}->{$group}->[ $label2row->{ $label } ]->[ $department2col->{$department} ] = scalar grep { $biblio_author_external->{$_} } @bibs;
 672 }
 673
 674 foreach my $group ( '', keys %$azvo_group_title ) {
 675
 676 foreach my $department ( @departments ) {
 677         foreach my $line ( @report_lines ) {
 678                 my $label = $line->[0];
 679                 my @biblionumbers;
 680                 foreach ( 1 .. $#$line ) {
 681                         my ( $category, $type ) = @{ $line->[ $_ ] };
 682                         my $b = $azvo_stat_biblio->{ $department }->{$category}->{$type}->{$group};
 683                         push @biblionumbers, @$b if $b;
 684                 }
 685                 if ( $sub_labels->{$label} ) {
 686                         my $sub_stats;
 687                         foreach my $biblionumber ( @biblionumbers ) {
 688                                 my $data = $biblio_data->{$biblionumber} || die "can't find biblionumber $biblionumber";
 689                                 foreach my $sub_label ( @{ $sub_labels->{$label} } ) {
 690                                         my ( $sub_label, $field, $sf, $regex ) = @$sub_label;
 691                                         if ( ! $regex ) {
 692                                                 push @{ $sub_stats->{ $sub_label } }, $biblionumber;
 693                                                 last;
 694                                         }
 695                                         if ( $field < 100 ) {
 696                                                 if ( $data->{$field} =~ m/$regex/ ) {
 697                                                         push @{ $sub_stats->{ $sub_label } }, $biblionumber;
 698                                                         last;
 699                                                 }
 700                                         } else {
 701                                                 if ( exists $data->{$field}->[0]->{$sf} && $data->{$field}->[0]->{$sf} =~ m/$regex/ ) {
 702                                                         push @{ $sub_stats->{ $sub_label } }, $biblionumber;
 703                                                         last;
 704                                                 }
 705                                         }
 706                                 }
 707                         }
 708                         foreach my $sub_label ( keys %$sub_stats ) {
 709                                 my $full_label = $label . $sub_label;
 710                                 table_count $full_label, $department, $group, @{ $sub_stats->{$sub_label} };
 711                         }
 712                 } else {
 713                         table_count $label, $department, $group, @biblionumbers;
 714                 }
 715         }
 716 }
 717
 718 } # group
 719
 720 debug 'table', $table;
 721
 722 open(my $fh, '>:encoding(utf-8)', 'html/azvo.new');
 723 open(my $fh2, '>:encoding(utf-8)', 'html/azvo2.new');
 724
 725 sub print_fh {
 726         print $fh @_;
 727         print $fh2 @_;
 728 }
 729
 730 print $fh html_title('AZVO tablica - FFZG');
 731 print $fh2 html_title('AZVO tablica - kolaboracija sa FFZG');
 732
 733 foreach my $group ( keys %{ $table->{ffzg} } ) {
 734
 735                 print_fh "<h1>$group</h1>" if $group;
 736
 737                 print_fh "<table border=1>\n";
 738                 print_fh "<tr><th></th>";
 739                 print_fh "<th>$_</th>" foreach @departments;
 740                 print_fh "</tr>\n";
 741
 742                 foreach my $row ( 0 .. $#{ $table->{ffzg}->{$group} } ) {
 743                         print_fh "<tr><th>", $report_labels[$row], "</th>\n";
 744                         foreach ( 0 .. $#departments ) {
 745                                 print_fh "<td>";
 746                                 print $fh $table->{ffzg}->{$group}->[ $row ]->[ $_ ] || '';
 747                                 print $fh2 $table->{external}->{$group}->[ $row ]->[ $_ ] || '';
 748                                 print_fh "</td>\n"
 749                         }
 750                         print_fh "</tr>\n";
 751                 }
 752
 753                 print_fh "</table>\n";
 754
 755 } # group
 756
 757 print_fh html_end;
 758 close($fh);
 759 close($fh2);
 760 rename 'html/azvo.new', 'html/azvo.html';
 761 rename 'html/azvo2.new', 'html/azvo2.html';
 762