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 use POSIX qw(strftime);
  17 use Storable;
  18
  19 use lib '/srv/koha_ffzg';
  20 use C4::Context;
  21 use XML::LibXML;
  22 use XML::LibXSLT;
  23
  24 my $dbh = C4::Context->dbh;
  25
  26 sub debug {
  27         my ($title, $data) = @_;
  28         print "# $title ",dump($data), $/ if $ENV{DEBUG};
  29 }
  30
  31 my $xslfilename = 'compact.xsl';
  32
  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,
  37 };
  38
  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,
  42 };
  43
  44 my $auth_header;
  45 my $auth_department;
  46 my $auth_group;
  47 my @authors;
  48 my $department_in_sum;
  49 my $department_in_group;
  50
  51 my $skip;
  52
  53 my $sth_auth = $dbh->prepare(q{
  54 select
  55         authid,
  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
  59 from auth_header
  60 });
  61
  62 $sth_auth->execute();
  63 while( my $row = $sth_auth->fetchrow_hashref ) {
  64         if ( $row->{department} !~ m/Filozofski fakultet u Zagrebu/ ) {
  65                 push @{ $skip->{nije_ffzg} }, $row;
  66                 next;
  67         }
  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;
  72         my $group;
  73         foreach my $title ( keys %$azvo_group_title ) {
  74                 if ( $row->{academic_title} =~ $azvo_group_title->{$title} ) {
  75                         $group = $title;
  76                         last;
  77                 }
  78         }
  79         if ( $group ) {
  80                 $row->{academic_group} = $group;
  81                 $auth_group->{ $row->{authid} } = $group;
  82                 $skip->{group_stat}->{$group}++;
  83         } else {
  84                 push @{ $skip->{no_academic_group} }, $row;
  85         }
  86
  87 #       warn "# ", dump( $row );
  88         push @{ $auth_department->{ $row->{department} } }, $row->{authid};
  89         push @authors, $row;
  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;
  95                         last;
  96                 }
  97         }
  98 }
  99
 100 debug 'department_in_group' => $department_in_group;
 101
 102 foreach my $department ( keys %$department_in_sum ) {
 103         $department_in_sum->{$department} = 0 unless $department =~ m/(centar|croaticum|katedra|odsjek)/i;
 104 }
 105
 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;
 110
 111
 112 my $authors;
 113 my $marcxml;
 114
 115 my $sth_select_authors  = $dbh->prepare(q{
 116 select
 117         biblionumber,
 118         itemtype,
 119         marcxml
 120 from biblioitems
 121 where
 122         agerestriction > 0
 123 });
 124
 125 =for sql
 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,
 129
 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
 132 =cut
 133
 134 my $biblio_year;
 135 my $biblio_full_name;
 136 my $type_stats;
 137
 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);
 143
 144 my $biblio_html;
 145 my $biblio_parsed;
 146 my $biblio_data;
 147 my $biblio_author_external;
 148
 149 open(my $xml_fh, '>', '/tmp/bibliografija.xml') if $ENV{XML};
 150
 151 sub biblioitem_html {
 152         my ($biblionumber, $parse_only) = @_;
 153
 154         return $biblio_html->{$biblionumber} if exists $biblio_html->{$biblionumber} && ! $parse_only;
 155
 156         my $xmlrecord = $marcxml->{$biblionumber} || confess "missing $biblionumber marcxml";
 157
 158         print $xml_fh $xmlrecord if $ENV{XML};
 159
 160         my $source = eval { $parser->parse_string($xmlrecord) };
 161         if ( $@ ) {
 162 #               warn "SKIP $biblionumber corrupt XML";
 163                 push @{ $skip->{XML_corrupt} }, $biblionumber;
 164                 return;
 165         }
 166
 167         if ( $parse_only ) {
 168                 $biblio_parsed->{$biblionumber} = $source;
 169                 return $source;
 170         }
 171
 172         my $transformed = $parsed->transform($source);
 173         $biblio_html->{$biblionumber} = $parsed->output_string( $transformed );
 174
 175         delete $biblio_parsed->{$biblionumber};
 176
 177         return $biblio_html->{$biblionumber};
 178 }
 179
 180 $sth_select_authors->execute();
 181 while( my $row = $sth_select_authors->fetchrow_hashref ) {
 182 #       warn dump($row),$/;
 183
 184         my $biblio;
 185
 186         $marcxml->{ $row->{biblionumber} } = $row->{marcxml};
 187
 188         my $doc = biblioitem_html( $row->{biblionumber}, 1 );
 189         if ( ! $doc ) {
 190 #               warn "ERROR can't parse MARCXML ", $row->{biblionumber}, " ", $row->{marcxml}, "\n";
 191                 next;
 192         }
 193
 194         my $root = $doc->documentElement;
 195 =for leader
 196         my @leaders = $root->getElementsByLocalName('leader');
 197         if (@leaders) {
 198                 my $leader = $leaders[0]->textContent;
 199                 warn "leader $leader\n";
 200         }
 201 =cut
 202
 203         my $extract = {
 204                 '008' => undef,
 205                 '100' => '(9|a)',
 206                 '245' => 'a',
 207                 '680' => 'i',
 208                 '700' => '(9|4|a)',
 209                 '942' => '(t|r|v)'
 210         };
 211
 212         my $data;
 213
 214         foreach my $elt ($root->getChildrenByLocalName('*')) {
 215                 my $tag = $elt->getAttribute('tag');
 216                 next if ! $tag;
 217                 next unless exists $extract->{ $tag };
 218
 219         if ($elt->localname eq 'controlfield') {
 220                         if ( $tag eq '008' ) {
 221                                 my $content = $elt->textContent;
 222                                 my $year = substr($content, 7, 4 );
 223                                 if ( $year !~ m/^\d+$/ ) {
 224                                         $year = 0;
 225                                         push @{ $skip->{invalid_year} }, $row->{biblionumber};
 226                                 }
 227                                 $biblio_year->{ $row->{biblionumber} } = $data->{year} = $year;
 228                                 $data->{'008'} = $content;
 229                         }
 230                         next;
 231         } elsif ($elt->localname eq 'datafield') {
 232                         my $sf_data;
 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();
 238                                 } else {
 239                                         $sf_data->{$sf} = $sfelt->textContent();
 240                                 }
 241                         }
 242                         push @{ $data->{$tag} }, $sf_data if $sf_data;
 243                 }
 244         }
 245
 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};
 251                 next;
 252         } elsif ( $data->{year} > 2013 ) {
 253                 push @{ $skip->{year_gt_2013} }, $row->{biblionumber};
 254                 next;
 255 =cut
 256         }
 257
 258 #       warn "# ", $row->{biblionumber}, " data ",dump($data);
 259
 260         my $category = $data->{942}->[0]->{'t'};
 261         if ( ! $category ) {
 262 #               warn "# SKIP ", $row->{biblionumber}, " no category in ", dump($data);
 263                 push @{ $skip->{no_category} }, $row->{biblionumber};
 264                 next;
 265         }
 266
 267         my $have_100 = 1;
 268
 269         if ( exists $data->{100} ) {
 270                         my @first_author =
 271                                 map { $_->{'9'} }
 272                                 grep {
 273                                         if ( ! exists $_->{9} ) {
 274                                                 $biblio_author_external->{ $row->{biblionumber} }++;
 275                                                 0;
 276                                         } elsif ( exists $auth_header->{ $_->{9} } ) {
 277                                                 1; # from FFZXG
 278                                         } else {
 279                                                 0;
 280                                         }
 281                                 }
 282                                 @{ $data->{100} };
 283                         foreach my $authid ( @first_author ) {
 284                                 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 285                         }
 286                         $biblio_full_name->{ $row->{biblionumber} } = $data->{100}->[0]->{a};
 287         } else {
 288                 $have_100 = 0;
 289         }
 290
 291         $biblio_full_name->{ $row->{biblionumber} } ||= $data->{245}->[0]->{a};
 292
 293         my $have_edt;
 294
 295         if ( exists $data->{700} ) {
 296                         my @other_authors =
 297                                 grep {
 298                                         if ( ! exists $_->{9} ) {
 299                                                 $biblio_author_external->{ $row->{biblionumber} }++;
 300                                                 0;
 301                                         } elsif ( exists $auth_header->{ $_->{9} } ) {
 302                                                 1; # from FFZXG
 303                                         } else {
 304                                                 0;
 305                                         }
 306                                 }
 307                                 @{ $data->{700} };
 308                         foreach my $auth ( @other_authors ) {
 309                                 my $authid = $auth->{9} || next;
 310                                 my $type   = $auth->{4} || next; #die "no 4 in ",dump($data);
 311
 312                                 $type_stats->{$type}++;
 313
 314                                 my @types = split(/[\s\/]+/, $type);
 315
 316                                 foreach my $type ( @types ) {
 317                                         my $type = substr($type,0,3);
 318                                         $type_stats->{_count_each_type}->{$type}++;
 319
 320                                         if ( $type =~ m/(edt|trl|com|ctb)/ ) {
 321                                                 push @{ $authors->{$authid}->{__sec}->{ $category } }, $row->{biblionumber};
 322                                                 push @{ $authors->{$authid}->{$type}->{ $category } }, $row->{biblionumber};
 323                                                 $type =~ s/(com|ctb)/_ostalo/;
 324                                                 push @{ $authors->{$authid}->{$type}->{ $category } }, $row->{biblionumber};
 325
 326                                         } elsif ( $type =~ m/aut/ ) {
 327                                                 if ( ! $have_100 ) {
 328                                                         $have_edt = grep { exists $_->{4} && $_->{4} =~ m/edt/ } @{ $data->{700} } if ! defined $have_edt;
 329                                                         if ( $have_edt ) {
 330                                                                 $skip->{ have_700_edt }->{ $row->{biblionumber} }++;
 331                                                         } else {
 332                                                                 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 333                                                         }
 334                                                 } else {
 335                                                         push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 336                                                 }
 337                                         } else {
 338 #                                               warn "# SKIP ", $row->{biblionumber}, ' no 700$4 in ', dump($data);
 339                                                 $skip->{ 'no_700$4' }->{ $row->{biblionumber} }++;
 340                                         }
 341                                 }
 342                         }
 343                         delete $data->{700};
 344         }
 345
 346         $biblio_data->{ $row->{biblionumber} } = $data;
 347
 348 }
 349
 350 debug 'authors' => $authors;
 351 store $authors, '/dev/shm/authors.storable';
 352 debug 'type_stats' => $type_stats;
 353 debug 'skip' => $skip;
 354 debug 'biblio_year' => $biblio_year;
 355 debug 'biblio_full_name' => $biblio_full_name;
 356 debug 'biblio_data' => $biblio_data;
 357 debug 'biblio_author_external' => $biblio_author_external;
 358
 359 my $category_label;
 360 my $sth_categories = $dbh->prepare(q{
 361 select authorised_value, lib from authorised_values where category = 'BIBCAT'
 362 });
 363 $sth_categories->execute();
 364 while( my $row = $sth_categories->fetchrow_hashref ) {
 365         $category_label->{ $row->{authorised_value} } = $row->{lib};
 366
 367 }
 368 debug 'category_label' => $category_label;
 369
 370 sub html_title {
 371         return qq|<html>
 372 <head>
 373 <meta charset="UTF-8">
 374 <title>|, join(" ", @_), qq|</title>
 375 <link href="style.css" type="text/css" rel="stylesheet" />
 376 <script src="//code.jquery.com/jquery-1.11.2.js"></script>
 377 <script src="filters.js"></script>
 378 </head>
 379 <body>
 380 |;
 381 }
 382
 383 sub html_end {
 384         return
 385                 qq|<small style="color:gray">Zadnji puta osvježeno: |,
 386                 strftime("%Y-%m-%d %H:%M:%S\n", localtime()),
 387                 qq|</body>\n</html>\n|;
 388 }
 389
 390 mkdir 'html' unless -d 'html';
 391
 392 open(my $index, '>:encoding(utf-8)', 'html/index.new');
 393 print $index html_title('Bibliografija Filozofskog fakulteta');
 394
 395 my $first_letter = '';
 396
 397 debug 'authors' => \@authors;
 398
 399 sub li_biblio {
 400         my ($biblionumber) = @_;
 401         return qq|<li class="y|, $biblio_year->{$biblionumber}, qq|">|,
 402                 qq|<a href="https://koha.ffzg.hr/cgi-bin/koha/opac-detail.pl?biblionumber=$biblionumber">$biblionumber</a>|,
 403                 biblioitem_html($biblionumber),
 404                 qq|<a href="https://koha.ffzg.hr:8443/cgi-bin/koha/cataloguing/addbiblio.pl?biblionumber=$biblionumber">edit</a>|,
 405                 qq|</li>\n|;
 406 }
 407
 408 sub unique {
 409         my $unique;
 410         $unique->{$_}++ foreach @_;
 411         return keys %$unique;
 412 }
 413
 414 sub unique_biblionumber {
 415         my @v = unique @_;
 416         return sort {
 417                 $biblio_year->{$b} <=> $biblio_year->{$a} ||
 418                 $biblio_full_name->{$a} cmp $biblio_full_name->{$b} ||
 419                 $a <=> $b
 420         } @v;
 421 }
 422
 423 sub author_html {
 424         my ( $fh, $authid, $type, $label ) = @_;
 425
 426         return unless exists $authors->{$authid}->{$type};
 427
 428         print $fh qq|<a name="$type"><h2>$label</h2></a>\n|;
 429
 430         foreach my $category ( sort keys %{ $authors->{$authid}->{$type} } ) {
 431                 my $label = $category_label->{$category} || 'Bez kategorije';
 432                 print $fh qq|<a name="$type-$category"><h3>$label</h3></a>\n<ol>\n|;
 433                 foreach my $biblionumber ( unique_biblionumber @{ $authors->{$authid}->{$type}->{$category} } ) {
 434                         print $fh li_biblio( $biblionumber );
 435                 }
 436                 print $fh qq|</ol>\n|;
 437         }
 438 }
 439
 440 my @toc_type_label = (
 441 'aut' => 'Primarno autorstvo',
 442 'edt' => 'Uredništva',
 443 'trl' => 'Prijevodi',
 444 '_ostalo' => 'Ostalo',
 445 );
 446
 447
 448 sub count_author_years {
 449         my $years = shift;
 450         my ($authid) = @_;
 451         foreach my $type ( keys %{ $authors->{$authid} } ) {
 452                 next if $type =~ m/^_/;
 453                 foreach my $category ( keys %{ $authors->{$authid}->{$type} } ) {
 454                         foreach my $biblionumber ( unique_biblionumber @{ $authors->{$authid}->{$type}->{$category} } ) {
 455                                 $years->{ $biblio_year->{ $biblionumber } }->{ $type . '-' . $category }++;
 456                         }
 457                 }
 458         }
 459         return $years;
 460 }
 461
 462 sub html_year_selection {
 463         my $fh = shift;
 464         my @authids = unique @_;
 465
 466         debug 'html_year_selection authids=', [ @authids ];
 467
 468         print $fh qq|<span id="years">Godine:\n|;
 469         my $type_cat_count = {};
 470         my $years;
 471
 472         foreach my $authid ( @authids ) {
 473                 $years = count_author_years( $years, $authid );
 474         }
 475
 476         debug 'years' => $years;
 477
 478         foreach my $year ( sort { $b <=> $a } keys %$years ) {
 479                 print $fh qq|<label><input name="year_selection" value="$year" type=checkbox onClick="toggle_year($year, this)" checked="checked">$year</label>&nbsp;\n|;
 480                 foreach my $type_cat ( keys %{ $years->{$year} } ) {
 481                         $type_cat_count->{ $type_cat } += $years->{$year}->{$type_cat};
 482                         my ($type,$cat) = split(/-/, $type_cat);
 483                         $type_cat_count->{_toc}->{$type}->{$cat}++;
 484                         $type_cat_count->{_toc_count}->{$type} += $years->{$year}->{$type_cat};
 485                 }
 486         }
 487
 488         print $fh qq|
 489 <input type=button value="all" onClick="all_years(1)">
 490 <input type=button value="none" onClick="all_years(0)">
 491         |;
 492
 493         print $fh qq|</span>|;
 494
 495         print $fh q|
 496 <script>
 497
 498 var years = |, encode_json($years), q|;
 499
 500 var type_cat_count = |, encode_json($type_cat_count), q|;
 501
 502 </script>
 503
 504         |;
 505
 506         debug 'type_cat_count' => $type_cat_count;
 507
 508         # TOC
 509         print $fh qq|<ul id="toc">\n|;
 510         my $i = 0;
 511         while ( $i < $#toc_type_label ) {
 512                 my $type  = $toc_type_label[$i++] || die "type";
 513                 my $label = $toc_type_label[$i++] || die "label";
 514                 next unless exists $type_cat_count->{_toc}->{$type};
 515                 print $fh qq| <li class="toc" id="toc-$type"><a href="#$type">$label</a> <tt id="toc-count-$type">$type_cat_count->{_toc_count}->{$type}</tt></li>\n <ul>\n|;
 516                 foreach my $category ( sort keys %{ $type_cat_count->{_toc}->{$type} } ) {
 517                         my $label = $category_label->{$category} || 'Bez kategorije';
 518                         my $count = $type_cat_count->{ $type . '-' . $category };
 519                         my $cat_html = $category;
 520                         $cat_html =~ s/\./-/g;
 521                         print $fh qq|  <li class="toc" id="toc-$category"><a href="#$type-$category">$label</a> <tt id="toc-count-$type-$cat_html">$count</tt></li>\n|;
 522                 }
 523                 print $fh qq| </ul>\n|;
 524         }
 525         print $fh qq|</ul>\n|;
 526
 527 }
 528
 529
 530 foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) {
 531
 532         my $first = substr( $row->{full_name}, 0, 1 );
 533         if ( $first ne $first_letter ) {
 534                 print $index qq{</ul>\n} if $first_letter;
 535                 $first_letter = $first;
 536                 print $index qq{<h1>$first</h1>\n<ul>\n};
 537         }
 538         print $index qq{<li><a href="}, $row->{authid}, qq{.html">}, $row->{full_name}, "</a></li>\n";
 539
 540         my $path = "html/$row->{authid}";
 541         open(my $fh, '>:encoding(utf-8)', "$path.new");
 542         print $fh html_title($row->{full_name}, "bibliografija");
 543         print $fh qq|<h1>$row->{full_name} - bibliografija</h1>\n|;
 544
 545         html_year_selection $fh => $row->{authid};
 546
 547         my $i = 0;
 548         while ( $i < $#toc_type_label ) {
 549                 my $type  = $toc_type_label[$i++] || die "type";
 550                 my $label = $toc_type_label[$i++] || die "label";
 551                 author_html( $fh, $row->{authid}, $type => $label );
 552         }
 553
 554         print $fh html_end;
 555         close($fh);
 556         rename "$path.new", "$path.html";
 557
 558 }
 559
 560 print $index html_end;
 561 close($index);
 562 rename 'html/index.new', 'html/index.html';
 563
 564 debug 'auth_header' => $auth_header;
 565
 566
 567 my $department_category_author;
 568 foreach my $department ( sort keys %$auth_department ) {
 569         foreach my $authid ( sort @{ $auth_department->{$department} } ) {
 570                 my   @categories = keys %{ $authors->{$authid}->{aut} };
 571                 push @categories,  keys %{ $authors->{$authid}->{__sec} };
 572                 foreach my $category ( sort @categories ) {
 573                         push @{ $department_category_author->{$department}->{$category} }, $authid;
 574                         push @{ $department_category_author->{'AAA_ukupno'}->{$category} }, $authid if $department_in_sum->{$department};
 575                         if ( my $group = $department_in_group->{ $department } ) {
 576                                 push @{ $department_category_author->{$group}->{$category} }, $authid;
 577                         } else {
 578                                 $skip->{'department_not_in_group'}->{ $department }++;
 579                         }
 580                 }
 581         }
 582 }
 583
 584 debug 'department_category_author' => $department_category_author;
 585
 586
 587 sub department_html {
 588         my ( $fh, $department, $type, $label ) = @_;
 589
 590         print $fh qq|<a name="$type"><h2>$label</h2></a>\n|;
 591
 592         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 593
 594                 my @authids = @{ $department_category_author->{$department}->{$category} };
 595                 next unless @authids;
 596
 597                 my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{$type}->{$category} } } grep { exists $authors->{$_}->{$type}->{$category} } @authids;
 598
 599                 next unless @biblionumber;
 600
 601                 my $label = $category_label->{$category} || 'Bez kategorije';
 602                 print $fh qq|<a name="$type-$category"><h3>$label</h3></a>\n<ol>\n|;
 603
 604                 print $fh li_biblio( $_ ) foreach @biblionumber;
 605
 606                 print $fh qq|</ol>|;
 607         }
 608
 609 }
 610
 611
 612 mkdir 'html/departments' unless -d 'html/departments';
 613
 614 open(my $dep_fh, '>:encoding(utf-8)', 'html/departments/index.new');
 615 print $dep_fh html_title('Odsjeci Filozofskog fakulteta u Zagrebu'), qq|<ul>\n|;
 616 foreach my $department ( sort keys %$department_category_author ) {
 617         my $dep = $department || 'Nema odsjeka';
 618         my $dep_file = unac_string('utf-8',$dep);
 619         print $dep_fh qq|<li><a href="$dep_file.html">$dep</a></li>\n|;
 620         open(my $fh, '>:encoding(utf-8)', "html/departments/$dep_file.new");
 621
 622         print $fh html_title($department . ' bibliografija');
 623         print $fh qq|<h1>$department bibliografija</h1>\n|;
 624
 625         my @authids;
 626         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 627                 push @authids, @{ $department_category_author->{$department}->{$category} };
 628         }
 629         html_year_selection $fh => @authids;
 630
 631         my $i = 0;
 632         while ( $i < $#toc_type_label ) {
 633                 my $type  = $toc_type_label[$i++] || die "type";
 634                 my $label = $toc_type_label[$i++] || die "label";
 635                 department_html( $fh, $department, $type, $label );
 636         }
 637
 638         print $fh html_end;
 639         close($fh);
 640         rename "html/departments/$dep_file.new", "html/departments/$dep_file.html";
 641 }
 642 print $dep_fh qq|</ul>\n|, html_end;
 643 close($dep_fh);
 644 rename 'html/departments/index.new', 'html/departments/index.html';
 645
 646 my $azvo_stat_biblio;
 647
 648 foreach my $department ( sort keys %$department_category_author ) {
 649         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 650                 foreach my $authid ( @{ $department_category_author->{$department}->{$category} } ) {
 651                         my $group = $auth_group->{$authid};
 652                         if ( ! $group ) {
 653                                 push @{ $skip->{no_auth_group} }, $authid;
 654                                 next;
 655                         }
 656                         foreach my $type ( keys %{ $authors->{$authid} } ) {
 657                                 next unless exists $authors->{$authid}->{$type}->{$category};
 658                                 push @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{$group} },  @{ $authors->{$authid}->{$type}->{$category} };
 659                                 push @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{''} },  @{ $authors->{$authid}->{$type}->{$category} };
 660                         }
 661                 }
 662                 foreach my $type ( keys %{ $azvo_stat_biblio->{ $department }->{ $category } } ) {
 663                         foreach my $group ( keys %{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type } } ) {
 664                                 my @biblios = unique_biblionumber @{ $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{ $group } };
 665                                 $azvo_stat_biblio->{ $department }->{ $category }->{ $type }->{ $group } = [ @biblios ];
 666                         }
 667                 }
 668         }
 669 }
 670
 671 debug 'azvo_stat_biblio' => $azvo_stat_biblio;
 672
 673 my @report_lines;
 674 my @report_labels;
 675
 676 my $label;
 677 my $sub_labels;
 678 open(my $report, '<:encoding(utf-8)', 'AZVO.txt');
 679 while( <$report> ) {
 680         chomp;
 681         if ( /^([^\t]+)\t+(.+)/ ) {
 682                 $label = $1;
 683                 push @report_labels, $label;
 684                 my $type = [ map { m/\s+/ ? [ split(/\s+/,$_) ] : [ $_, 'aut' ] } split (/\s*\+\s*/, $2) ];
 685                 push @report_lines, [ $label, @$type ];
 686         } elsif ( /^\t+([^\t]+):\t+(\d+)(\w*)\t*(.*)$/ ) {
 687                 push @{ $sub_labels->{$label} }, [ $1, $2, $3, $4 ];
 688                 my $sub_label = $1;
 689                 pop (@report_labels) if ( $report_labels[ $#report_labels ] =~ m/^$label$/ ); # remove partial name
 690                 push @report_labels, $label . $sub_label;
 691         } else {
 692                 die "ERROR: [$_]\n";
 693         }
 694 }
 695
 696 debug 'report_lines', \@report_lines;
 697 debug 'sub_labels', $sub_labels;
 698 debug 'report_labels', \@report_labels;
 699
 700 my @departments = ( sort { lc($a) cmp lc($b) } keys %$azvo_stat_biblio );
 701
 702 debug 'departments' => \@departments;
 703
 704 my $department2col;
 705 $department2col->{ $departments[$_] } = $_ foreach ( 0 .. $#departments );
 706 my $label2row;
 707 $label2row->{ $report_labels[$_] } = $_ foreach ( 0 .. $#report_labels );
 708
 709 my $table;
 710
 711 sub table_count {
 712         my $label = shift @_;
 713         my $department = shift @_;
 714         my $group = shift @_;
 715         my @biblionumbers = unique @_;
 716         $table->{ffzg}->{$group}->[ $label2row->{ $label } ]->[ $department2col->{$department} ] = scalar @biblionumbers;
 717         $table->{external}->{$group}->[ $label2row->{ $label } ]->[ $department2col->{$department} ] = scalar grep { $biblio_author_external->{$_} } @biblionumbers;
 718 }
 719
 720 foreach my $group ( '', keys %$azvo_group_title ) {
 721
 722 foreach my $department ( @departments ) {
 723         foreach my $line ( @report_lines ) {
 724                 my $label = $line->[0];
 725                 my @biblionumbers;
 726                 foreach ( 1 .. $#$line ) {
 727                         my ( $category, $type ) = @{ $line->[ $_ ] };
 728                         my $b = $azvo_stat_biblio->{ $department }->{$category}->{$type}->{$group};
 729                         push @biblionumbers, @$b if $b;
 730                 }
 731                 if ( $sub_labels->{$label} ) {
 732                         my $sub_stats;
 733                         foreach my $biblionumber ( @biblionumbers ) {
 734                                 my $data = $biblio_data->{$biblionumber} || die "can't find biblionumber $biblionumber";
 735                                 foreach my $sub_label ( @{ $sub_labels->{$label} } ) {
 736                                         my ( $sub_label, $field, $sf, $regex ) = @$sub_label;
 737                                         if ( ! $regex ) {
 738                                                 push @{ $sub_stats->{ $sub_label } }, $biblionumber;
 739                                                 last;
 740                                         }
 741                                         if ( $field < 100 ) {
 742                                                 if ( $data->{$field} =~ m/$regex/ ) {
 743                                                         push @{ $sub_stats->{ $sub_label } }, $biblionumber;
 744                                                         last;
 745                                                 }
 746                                         } else {
 747                                                 if ( exists $data->{$field}->[0]->{$sf} && $data->{$field}->[0]->{$sf} =~ m/$regex/ ) {
 748                                                         push @{ $sub_stats->{ $sub_label } }, $biblionumber;
 749                                                         last;
 750                                                 }
 751                                         }
 752                                 }
 753                         }
 754                         foreach my $sub_label ( keys %$sub_stats ) {
 755                                 my $full_label = $label . $sub_label;
 756                                 table_count $full_label, $department, $group, @{ $sub_stats->{$sub_label} };
 757                         }
 758                 } else {
 759                         table_count $label, $department, $group, @biblionumbers;
 760                 }
 761         }
 762 }
 763
 764 } # group
 765
 766 debug 'table', $table;
 767
 768 open(my $fh, '>:encoding(utf-8)', 'html/azvo.new');
 769 open(my $fh2, '>:encoding(utf-8)', 'html/azvo2.new');
 770
 771 sub print_fh {
 772         print $fh @_;
 773         print $fh2 @_;
 774 }
 775
 776 print $fh html_title('AZVO tablica - FFZG');
 777 print $fh2 html_title('AZVO tablica - kolaboracija sa FFZG');
 778
 779 foreach my $group ( keys %{ $table->{ffzg} } ) {
 780
 781                 print_fh "<h1>$group</h1>" if $group;
 782
 783                 print_fh "<table border=1>\n";
 784                 print_fh "<tr><th></th>";
 785                 print_fh "<th>$_</th>" foreach @departments;
 786                 print_fh "</tr>\n";
 787
 788                 foreach my $row ( 0 .. $#{ $table->{ffzg}->{$group} } ) {
 789                         print_fh "<tr><th>", $report_labels[$row], "</th>\n";
 790                         foreach ( 0 .. $#departments ) {
 791                                 print_fh "<td>";
 792                                 print $fh $table->{ffzg}->{$group}->[ $row ]->[ $_ ] || '';
 793                                 print $fh2 $table->{external}->{$group}->[ $row ]->[ $_ ] || '';
 794                                 print_fh "</td>\n"
 795                         }
 796                         print_fh "</tr>\n";
 797                 }
 798
 799                 print_fh "</table>\n";
 800
 801 } # group
 802
 803 print_fh html_end;
 804 close($fh);
 805 close($fh2);
 806 rename 'html/azvo.new', 'html/azvo.html';
 807 rename 'html/azvo2.new', 'html/azvo2.html';
 808