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
  15 use lib '/srv/koha_ffzg';
  16 use C4::Context;
  17 use XML::LibXML;
  18 use XML::LibXSLT;
  19
  20 my $dbh = C4::Context->dbh;
  21
  22 sub debug {
  23         my ($title, $data) = @_;
  24         print "# $title ",dump($data), $/;
  25 }
  26
  27 my $xslfilename = 'compact.xsl';
  28
  29 my $auth_header;
  30 my $auth_department;
  31 my @authors;
  32
  33 my $skip;
  34
  35 my $sth_auth = $dbh->prepare(q{
  36 select
  37         authid,
  38         ExtractValue(marcxml,'//datafield[@tag="100"]/subfield[@code="a"]') as full_name,
  39         ExtractValue(marcxml,'//datafield[@tag="680"]/subfield[@code="a"]') as department
  40 from auth_header
  41 });
  42
  43 $sth_auth->execute();
  44 while( my $row = $sth_auth->fetchrow_hashref ) {
  45         if ( $row->{department} !~ m/Filozofski fakultet u Zagrebu/ ) {
  46                 push @{ $skip->{nije_ffzg} }, $row;
  47                 next;
  48         }
  49         $auth_header->{ $row->{authid} } = $row->{full_name};
  50         $row->{department} =~ s/, Filozofski fakultet u Zagrebu.*$//;
  51         $row->{department} =~ s/^.+\.\s*//;
  52 #       warn dump( $row );
  53         push @{ $auth_department->{ $row->{department} } }, $row->{authid};
  54         push @authors, $row;
  55
  56 }
  57
  58 debug 'auth_department' => $auth_department;
  59
  60
  61 my $authors;
  62 my $marcxml;
  63
  64 my $sth_select_authors  = $dbh->prepare(q{
  65 select
  66         biblionumber,
  67         itemtype,
  68         marcxml
  69 from biblioitems
  70 where
  71         agerestriction > 0
  72 });
  73
  74 =for sql
  75 --      ExtractValue(marcxml,'//datafield[@tag="100"]/subfield[@code="9"]') as first_author,
  76 --      ExtractValue(marcxml,'//datafield[@tag="700"]/subfield[@code="9"]') as other_authors,
  77 --      ExtractValue(marcxml,'//datafield[@tag="942"]/subfield[@code="t"]') as category,
  78
  79 --      and SUBSTR(ExtractValue(marcxml,'//controlfield[@tag="008"]'),8,4) between 2008 and 2013
  80 -- order by SUBSTR(ExtractValue(marcxml,'//controlfield[@tag="008"]'),8,4) desc
  81 =cut
  82
  83 my $biblio_year;
  84 my $type_stats;
  85
  86 my $parser = XML::LibXML->new();
  87 $parser->recover_silently(0); # don't die when you find &, >, etc
  88 my $style_doc = $parser->parse_file($xslfilename);
  89 my $xslt = XML::LibXSLT->new();
  90 my $parsed = $xslt->parse_stylesheet($style_doc);
  91
  92 my $biblio_html;
  93
  94 open(my $xml_fh, '>', '/tmp/bibliografija.xml') if $ENV{XML};
  95
  96 sub biblioitem_html {
  97         my $biblionumber = shift;
  98
  99         return $biblio_html->{$biblionumber} if exists $biblio_html->{$biblionumber};
 100
 101         my $xmlrecord = $marcxml->{$biblionumber} || confess "missing $biblionumber marcxml";
 102
 103         print $xml_fh $xmlrecord if $ENV{XML};
 104
 105         my $source = eval { $parser->parse_string($xmlrecord) };
 106         if ( $@ ) {
 107 #               warn "SKIP $biblionumber corrupt XML";
 108                 push @{ $skip->{XML_corrupt} }, $biblionumber;
 109                 return;
 110         }
 111
 112         my $transformed = $parsed->transform($source);
 113         $biblio_html->{$biblionumber} = $parsed->output_string( $transformed );
 114
 115         return ( $biblio_html->{$biblionumber}, $source ) if wantarray;
 116         return $biblio_html->{$biblionumber};
 117 }
 118
 119 $sth_select_authors->execute();
 120 while( my $row = $sth_select_authors->fetchrow_hashref ) {
 121 #       warn dump($row),$/;
 122
 123         my $biblio;
 124
 125         $marcxml->{ $row->{biblionumber} } = $row->{marcxml};
 126
 127         my ( undef, $doc ) = biblioitem_html( $row->{biblionumber} );
 128         if ( ! $doc ) {
 129 #               warn "ERROR can't parse MARCXML ", $row->{biblionumber}, " ", $row->{marcxml}, "\n";
 130                 next;
 131         }
 132
 133         my $root = $doc->documentElement;
 134 =for leader
 135         my @leaders = $root->getElementsByLocalName('leader');
 136         if (@leaders) {
 137                 my $leader = $leaders[0]->textContent;
 138                 warn "leader $leader\n";
 139         }
 140 =cut
 141
 142         my $extract = {
 143                 '008' => undef,
 144                 '100' => '9',
 145                 '700' => '(9|4)',
 146                 '942' => 't'
 147         };
 148
 149         my $data;
 150
 151         foreach my $elt ($root->getChildrenByLocalName('*')) {
 152                 my $tag = $elt->getAttribute('tag');
 153                 next if ! $tag;
 154                 next unless exists $extract->{ $tag };
 155
 156         if ($elt->localname eq 'controlfield') {
 157                         if ( $tag eq '008' ) {
 158                                 my $year = substr($elt->textContent, 7, 4 );
 159                                 if ( $year !~ m/^\d+$/ ) {
 160                                         $year = 0;
 161                                         push @{ $skip->{invalid_year} }, $row->{biblionumber};
 162                                 }
 163                                 $biblio_year->{ $row->{biblionumber} } = $data->{year} = $year;
 164                         }
 165                         next;
 166         } elsif ($elt->localname eq 'datafield') {
 167                         my $sf_data;
 168             foreach my $sfelt ($elt->getChildrenByLocalName('subfield')) {
 169                 my $sf = $sfelt->getAttribute('code');
 170                                 next unless $sf =~ m/$extract->{$tag}/;
 171                                 if ( exists $sf_data->{$sf} ) {
 172                                         $sf_data->{$sf} .= " " . $sfelt->textContent();
 173                                 } else {
 174                                         $sf_data->{$sf} = $sfelt->textContent();
 175                                 }
 176                         }
 177                         push @{ $data->{$tag} }, $sf_data if $sf_data;
 178         }
 179     }
 180
 181         if ( $data->{year} < 2008 ) {
 182                 push @{ $skip->{year_lt_2008} }, $row->{biblionumber};
 183                 next;
 184         } elsif ( $data->{year} > 2013 ) {
 185                 push @{ $skip->{year_gt_2013} }, $row->{biblionumber};
 186                 next;
 187         }
 188
 189 #       warn "# ", $row->{biblionumber}, " data ",dump($data);
 190
 191         my $category = $data->{942}->[0]->{'t'};
 192         if ( ! $category ) {
 193 #               warn "# SKIP ", $row->{biblionumber}, " no category in ", dump($data);
 194                 push @{ $skip->{no_category} }, $row->{biblionumber};
 195                 next;
 196         }
 197
 198
 199         my $have_100 = 1;
 200
 201         if ( exists $data->{100} ) {
 202                         my @first_author = map { $_->{'9'} } @{ $data->{100} };
 203                         foreach my $authid ( @first_author ) {
 204                                 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 205                         }
 206         } else {
 207                 $have_100 = 0;
 208         }
 209
 210         my $have_edt;
 211
 212         if ( exists $data->{700} ) {
 213                         foreach my $auth ( @{ $data->{700} } ) {
 214                                 my $authid = $auth->{9} || next;
 215                                 my $type   = $auth->{4} || next; #die "no 4 in ",dump($data);
 216
 217                                 $type_stats->{$type}++;
 218
 219                                 if ( $type =~ m/(edt|trl|com|ctb)/ ) {
 220                                         push @{ $authors->{$authid}->{sec}->{ $category } }, $row->{biblionumber};
 221                                         push @{ $authors->{$authid}->{$1}->{ $category } }, $row->{biblionumber};
 222                                 } elsif ( $type =~ m/aut/ ) {
 223                                         if ( ! $have_100 ) {
 224                                                 $have_edt = grep { exists $_->{4} && $_->{4} =~ m/edt/ } @{ $data->{700} } if ! defined $have_edt;
 225                                                 if ( $have_edt ) {
 226                                                         $skip->{ have_700_edt }->{ $row->{biblionumber} }++;
 227                                                 } else {
 228                                                         push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 229                                                 }
 230                                         } else {
 231                                                 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 232                                         }
 233                                 } else {
 234 #                                       warn "# SKIP ", $row->{biblionumber}, ' no 700$4 in ', dump($data);
 235                                         $skip->{ 'no_700$4' }->{ $row->{biblionumber} }++;
 236                                 }
 237                         }
 238         }
 239
 240 }
 241
 242 debug 'authors' => $authors;
 243 debug 'type_stats' => $type_stats;
 244 debug 'skip' => $skip;
 245 debug 'biblio_year' => $biblio_year;
 246
 247 my $category_label;
 248 my $sth_categories = $dbh->prepare(q{
 249 select authorised_value, lib from authorised_values where category = 'BIBCAT'
 250 });
 251 $sth_categories->execute();
 252 while( my $row = $sth_categories->fetchrow_hashref ) {
 253         $category_label->{ $row->{authorised_value} } = $row->{lib};
 254
 255 }
 256 debug 'category_label' => $category_label;
 257
 258 sub html_title {
 259         return qq|<html>
 260 <head>
 261 <meta charset="UTF-8">
 262 <title>|, join(" ", @_), qq|</title>
 263 <link href="style.css" type="text/css" rel="stylesheet" />
 264 </head>
 265 <body>
 266 |;
 267 }
 268
 269 sub html_end {
 270         return qq|</body>\n</html\n|;
 271 }
 272
 273 mkdir 'html' unless -d 'html';
 274
 275 open(my $index, '>:encoding(utf-8)', 'html/index.new');
 276 print $index html_title('Bibliografija Filozofskog fakulteta');
 277
 278 my $first_letter = '';
 279
 280 debug 'authors' => \@authors;
 281
 282 sub li_biblio {
 283         my ($biblionumber) = @_;
 284         return qq|<li>|,
 285                 qq|<a href="https://koha.ffzg.hr/cgi-bin/koha/opac-detail.pl?biblionumber=$biblionumber">$biblionumber</a>|,
 286                 biblioitem_html($biblionumber),
 287                 qq|<a href="https://koha.ffzg.hr:8443/cgi-bin/koha/cataloguing/addbiblio.pl?biblionumber=$biblionumber">edit</a>|,
 288                 qq|</li>\n|;
 289 }
 290
 291 sub author_html {
 292         my ( $fh, $authid, $type, $label ) = @_;
 293
 294         return unless exists $authors->{$authid}->{$type};
 295
 296         print $fh qq|<h2>$label</h2>\n|;
 297
 298         foreach my $category ( sort keys %{ $authors->{$authid}->{$type} } ) {
 299                 my $label = $category_label->{$category} || 'Bez kategorije';
 300                 print $fh qq|<h3>$label</h3>\n<ul>\n|;
 301                 foreach my $biblionumber ( @{ $authors->{$authid}->{$type}->{$category} } ) {
 302                         print $fh li_biblio( $biblionumber );
 303                 }
 304                 print $fh qq|</ul>\n|;
 305         }
 306 }
 307
 308 foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) {
 309
 310         my $first = substr( $row->{full_name}, 0, 1 );
 311         if ( $first ne $first_letter ) {
 312                 print $index qq{</ul>\n} if $first_letter;
 313                 $first_letter = $first;
 314                 print $index qq{<h1>$first</h1>\n<ul>\n};
 315         }
 316         print $index qq{<li><a href="}, $row->{authid}, qq{.html">}, $row->{full_name}, "</a></li>\n";
 317
 318         my $path = "html/$row->{authid}";
 319         open(my $fh, '>:encoding(utf-8)', "$path.new");
 320         print $fh html_title($row->{full_name}, "bibliografija");
 321         print $fh qq|<h1>$row->{full_name} - bibliografija za razdoblje 2008-2013</h1>|;
 322
 323         author_html( $fh, $row->{authid}, 'aut' => 'Primarno autorstvo' );
 324         author_html( $fh, $row->{authid}, 'sec' => 'Sekundarno autorstvo' );
 325
 326         print $fh html_end;
 327         close($fh);
 328         rename "$path.new", "$path.html";
 329
 330 }
 331
 332 print $index html_end;
 333 close($index);
 334 rename 'html/index.new', 'html/index.html';
 335
 336 debug 'auth_header' => $auth_header;
 337
 338
 339 my $department_category_author;
 340 foreach my $department ( sort keys %$auth_department ) {
 341         foreach my $authid ( sort @{ $auth_department->{$department} } ) {
 342                 my   @categories = keys %{ $authors->{$authid}->{aut} };
 343                 push @categories,  keys %{ $authors->{$authid}->{sec} };
 344                 foreach my $category ( sort @categories ) {
 345                         push @{ $department_category_author->{$department}->{$category} }, $authid;
 346                 }
 347         }
 348 }
 349
 350 debug 'department_category_author' => $department_category_author;
 351
 352 mkdir 'html/departments' unless -d 'html/departments';
 353
 354 sub unique_biblionumber {
 355         my @v = @_;
 356         my $u;
 357         $u->{$_}++ foreach @v;
 358         return sort { $biblio_year->{$b} <=> $biblio_year->{$a} || $a <=> $b } keys %$u;
 359 }
 360
 361 open(my $dep_fh, '>:encoding(utf-8)', 'html/departments/index.new');
 362 print $dep_fh html_title('Odsijeci Filozofskog fakulteta u Zagrebu'), qq|<ul>\n|;
 363 foreach my $department ( sort keys %$department_category_author ) {
 364         my $dep = $department || 'Nema odsjeka';
 365         my $dep_file = unac_string('utf-8',$dep);
 366         print $dep_fh qq|<li><a href="$dep_file.html">$dep</a></li>\n|;
 367         open(my $fh, '>:encoding(utf-8)', "html/departments/$dep_file.new");
 368
 369         print $fh html_title($department . ' bibliografija');
 370         print $fh qq|<h1>$department bibliografija</h1>\n|;
 371
 372         print $fh qq|<h2>Primarno autorstvo</h2>\n|;
 373
 374         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 375
 376                 my @authids = @{ $department_category_author->{$department}->{$category} };
 377                 next unless @authids;
 378
 379                 my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{aut}->{$category} } } grep { exists $authors->{$_}->{aut}->{$category} } @authids;
 380                 my $unique;
 381                 $unique->{$_}++ foreach @biblionumber;
 382
 383                 next unless @biblionumber;
 384
 385                 my $label = $category_label->{$category} || 'Bez kategorije';
 386                 print $fh qq|<h3>$label</h3>\n<ul>\n|;
 387
 388                 print $fh li_biblio( $_ ) foreach @biblionumber;
 389
 390                 print $fh qq|</ul>|;
 391         }
 392
 393
 394         print $fh qq|<h2>Sekundarno autorstvo</h2>\n|;
 395
 396         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 397
 398                 my @authids = @{ $department_category_author->{$department}->{$category} };
 399                 next unless @authids;
 400
 401                 my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{sec}->{$category} } } grep { exists $authors->{$_}->{sec}->{$category} } @authids;
 402
 403                 next unless @biblionumber;
 404
 405                 my $label = $category_label->{$category} || 'Bez kategorije';
 406                 print $fh qq|<h3>$label</h3>\n<ul>\n|;
 407
 408                 print $fh li_biblio( $_ ) foreach @biblionumber;
 409
 410                 print $fh qq|</ul>|;
 411         }
 412
 413
 414         print $fh html_end;
 415         close($fh);
 416         rename "html/departments/$dep_file.new", "html/departments/$dep_file.html";
 417 }
 418 print $dep_fh qq|</ul>\n|, html_end;
 419 close($dep_fh);
 420 rename 'html/departments/index.new', 'html/departments/index.html';
 421
 422 my $azvo_stat;
 423
 424 foreach my $department ( sort keys %$department_category_author ) {
 425         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 426                 foreach my $authid ( @{ $department_category_author->{$department}->{$category} } ) {
 427                         foreach my $type ( keys %{ $authors->{$authid} } ) {
 428                                 next unless exists $authors->{$authid}->{$type}->{$category};
 429                                 $azvo_stat->{ $department }->{ $category }->{ $type } += $#{ $authors->{$authid}->{$type}->{$category} } + 1;
 430                         }
 431                 }
 432         }
 433 }
 434
 435 debug 'azvo_stat' => $azvo_stat;
 436
 437 =for later
 438 open(my $fh, '>', 'html/azvo.new');
 439
 440
 441
 442 close($fh);
 443 rename 'html/azvo.new', 'html/azvo.html';
 444 =cut
 445