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                                  $biblio_year->{ $row->{biblionumber} } = $data->{year} = substr($elt->textContent, 7, 4 );
 159                         }
 160                         next;
 161         } elsif ($elt->localname eq 'datafield') {
 162                         my $sf_data;
 163             foreach my $sfelt ($elt->getChildrenByLocalName('subfield')) {
 164                 my $sf = $sfelt->getAttribute('code');
 165                                 next unless $sf =~ m/$extract->{$tag}/;
 166                                 if ( exists $sf_data->{$sf} ) {
 167                                         $sf_data->{$sf} .= " " . $sfelt->textContent();
 168                                 } else {
 169                                         $sf_data->{$sf} = $sfelt->textContent();
 170                                 }
 171                         }
 172                         push @{ $data->{$tag} }, $sf_data if $sf_data;
 173         }
 174     }
 175
 176         if ( $data->{year} < 2008 ) {
 177                 push @{ $skip->{year_lt_2008} }, $row->{biblionumber};
 178                 next;
 179         } elsif ( $data->{year} > 2013 ) {
 180                 push @{ $skip->{year_gt_2013} }, $row->{biblionumber};
 181                 next;
 182         }
 183
 184 #       warn "# ", $row->{biblionumber}, " data ",dump($data);
 185
 186         my $category = $data->{942}->[0]->{'t'};
 187         if ( ! $category ) {
 188 #               warn "# SKIP ", $row->{biblionumber}, " no category in ", dump($data);
 189                 push @{ $skip->{no_category} }, $row->{biblionumber};
 190                 next;
 191         }
 192
 193
 194         my $have_100 = 1;
 195
 196         if ( exists $data->{100} ) {
 197                         my @first_author = map { $_->{'9'} } @{ $data->{100} };
 198                         foreach my $authid ( @first_author ) {
 199                                 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 200                         }
 201         } else {
 202                 $have_100 = 0;
 203         }
 204
 205         my $have_edt;
 206
 207         if ( exists $data->{700} ) {
 208                         foreach my $auth ( @{ $data->{700} } ) {
 209                                 my $authid = $auth->{9} || next;
 210                                 my $type   = $auth->{4} || next; #die "no 4 in ",dump($data);
 211
 212                                 $type_stats->{$type}++;
 213
 214                                 if ( $type =~ m/(edt|trl|com|ctb)/ ) {
 215                                         push @{ $authors->{$authid}->{sec}->{ $category } }, $row->{biblionumber};
 216                                         push @{ $authors->{$authid}->{$1}->{ $category } }, $row->{biblionumber};
 217                                 } elsif ( $type =~ m/aut/ ) {
 218                                         if ( ! $have_100 ) {
 219                                                 $have_edt = grep { exists $_->{4} && $_->{4} =~ m/edt/ } @{ $data->{700} } if ! defined $have_edt;
 220                                                 if ( $have_edt ) {
 221                                                         $skip->{ have_700_edt }->{ $row->{biblionumber} }++;
 222                                                 } else {
 223                                                         push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 224                                                 }
 225                                         } else {
 226                                                 push @{ $authors->{$authid}->{aut}->{ $category } }, $row->{biblionumber};
 227                                         }
 228                                 } else {
 229 #                                       warn "# SKIP ", $row->{biblionumber}, ' no 700$4 in ', dump($data);
 230                                         $skip->{ 'no_700$4' }->{ $row->{biblionumber} }++;
 231                                 }
 232                         }
 233         }
 234
 235 }
 236
 237 debug 'authors' => $authors;
 238 debug 'type_stats' => $type_stats;
 239 debug 'skip' => $skip;
 240 debug 'biblio_year' => $biblio_year;
 241
 242 my $category_label;
 243 my $sth_categories = $dbh->prepare(q{
 244 select authorised_value, lib from authorised_values where category = 'BIBCAT'
 245 });
 246 $sth_categories->execute();
 247 while( my $row = $sth_categories->fetchrow_hashref ) {
 248         $category_label->{ $row->{authorised_value} } = $row->{lib};
 249
 250 }
 251 debug 'category_label' => $category_label;
 252
 253 sub html_title {
 254         return qq|<html>
 255 <head>
 256 <meta charset="UTF-8">
 257 <title>|, join(" ", @_), qq|</title>
 258 <link href="style.css" type="text/css" rel="stylesheet" />
 259 </head>
 260 <body>
 261 |;
 262 }
 263
 264 sub html_end {
 265         return qq|</body>\n</html\n|;
 266 }
 267
 268 mkdir 'html' unless -d 'html';
 269
 270 open(my $index, '>:encoding(utf-8)', 'html/index.new');
 271 print $index html_title('Bibliografija Filozofskog fakulteta');
 272
 273 my $first_letter = '';
 274
 275 debug 'authors' => \@authors;
 276
 277 sub li_biblio {
 278         my ($biblionumber) = @_;
 279         return qq|<li>|,
 280                 qq|<a href="https://koha.ffzg.hr/cgi-bin/koha/opac-detail.pl?biblionumber=$biblionumber">$biblionumber</a>|,
 281                 biblioitem_html($biblionumber),
 282                 qq|<a href="https://koha.ffzg.hr:8443/cgi-bin/koha/cataloguing/addbiblio.pl?biblionumber=$biblionumber">edit</a>|,
 283                 qq|</li>\n|;
 284 }
 285
 286 sub author_html {
 287         my ( $fh, $authid, $type, $label ) = @_;
 288
 289         return unless exists $authors->{$authid}->{$type};
 290
 291         print $fh qq|<h2>$label</h2>\n|;
 292
 293         foreach my $category ( sort keys %{ $authors->{$authid}->{$type} } ) {
 294                 my $label = $category_label->{$category} || 'Bez kategorije';
 295                 print $fh qq|<h3>$label</h3>\n<ul>\n|;
 296                 foreach my $biblionumber ( @{ $authors->{$authid}->{$type}->{$category} } ) {
 297                         print $fh li_biblio( $biblionumber );
 298                 }
 299                 print $fh qq|</ul>\n|;
 300         }
 301 }
 302
 303 foreach my $row ( sort { $a->{full_name} cmp $b->{full_name} } @authors ) {
 304
 305         my $first = substr( $row->{full_name}, 0, 1 );
 306         if ( $first ne $first_letter ) {
 307                 print $index qq{</ul>\n} if $first_letter;
 308                 $first_letter = $first;
 309                 print $index qq{<h1>$first</h1>\n<ul>\n};
 310         }
 311         print $index qq{<li><a href="}, $row->{authid}, qq{.html">}, $row->{full_name}, "</a></li>\n";
 312
 313         my $path = "html/$row->{authid}";
 314         open(my $fh, '>:encoding(utf-8)', "$path.new");
 315         print $fh html_title($row->{full_name}, "bibliografija");
 316         print $fh qq|<h1>$row->{full_name} - bibliografija za razdoblje 2008-2013</h1>|;
 317
 318         author_html( $fh, $row->{authid}, 'aut' => 'Primarno autorstvo' );
 319         author_html( $fh, $row->{authid}, 'sec' => 'Sekundarno autorstvo' );
 320
 321         print $fh html_end;
 322         close($fh);
 323         rename "$path.new", "$path.html";
 324
 325 }
 326
 327 print $index html_end;
 328 close($index);
 329 rename 'html/index.new', 'html/index.html';
 330
 331 debug 'auth_header' => $auth_header;
 332
 333
 334 my $department_category_author;
 335 foreach my $department ( sort keys %$auth_department ) {
 336         foreach my $authid ( sort @{ $auth_department->{$department} } ) {
 337                 my   @categories = keys %{ $authors->{$authid}->{aut} };
 338                 push @categories,  keys %{ $authors->{$authid}->{sec} };
 339                 foreach my $category ( sort @categories ) {
 340                         push @{ $department_category_author->{$department}->{$category} }, $authid;
 341                 }
 342         }
 343 }
 344
 345 debug 'department_category_author' => $department_category_author;
 346
 347 mkdir 'html/departments' unless -d 'html/departments';
 348
 349 sub unique_biblionumber {
 350         my @v = @_;
 351         my $u;
 352         $u->{$_}++ foreach @v;
 353         return sort { $biblio_year->{$b} <=> $biblio_year->{$a} || $a <=> $b } keys %$u;
 354 }
 355
 356 open(my $dep_fh, '>:encoding(utf-8)', 'html/departments/index.new');
 357 print $dep_fh html_title('Odsijeci Filozofskog fakulteta u Zagrebu'), qq|<ul>\n|;
 358 foreach my $department ( sort keys %$department_category_author ) {
 359         my $dep = $department || 'Nema odsjeka';
 360         my $dep_file = unac_string('utf-8',$dep);
 361         print $dep_fh qq|<li><a href="$dep_file.html">$dep</a></li>\n|;
 362         open(my $fh, '>:encoding(utf-8)', "html/departments/$dep_file.new");
 363
 364         print $fh html_title($department . ' bibliografija');
 365         print $fh qq|<h1>$department bibliografija</h1>\n|;
 366
 367         print $fh qq|<h2>Primarno autorstvo</h2>\n|;
 368
 369         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 370
 371                 my @authids = @{ $department_category_author->{$department}->{$category} };
 372                 next unless @authids;
 373
 374                 my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{aut}->{$category} } } grep { exists $authors->{$_}->{aut}->{$category} } @authids;
 375                 my $unique;
 376                 $unique->{$_}++ foreach @biblionumber;
 377
 378                 next unless @biblionumber;
 379
 380                 my $label = $category_label->{$category} || 'Bez kategorije';
 381                 print $fh qq|<h3>$label</h3>\n<ul>\n|;
 382
 383                 print $fh li_biblio( $_ ) foreach @biblionumber;
 384
 385                 print $fh qq|</ul>|;
 386         }
 387
 388
 389         print $fh qq|<h2>Sekundarno autorstvo</h2>\n|;
 390
 391         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 392
 393                 my @authids = @{ $department_category_author->{$department}->{$category} };
 394                 next unless @authids;
 395
 396                 my @biblionumber = unique_biblionumber map { @{ $authors->{$_}->{sec}->{$category} } } grep { exists $authors->{$_}->{sec}->{$category} } @authids;
 397
 398                 next unless @biblionumber;
 399
 400                 my $label = $category_label->{$category} || 'Bez kategorije';
 401                 print $fh qq|<h3>$label</h3>\n<ul>\n|;
 402
 403                 print $fh li_biblio( $_ ) foreach @biblionumber;
 404
 405                 print $fh qq|</ul>|;
 406         }
 407
 408
 409         print $fh html_end;
 410         close($fh);
 411         rename "html/departments/$dep_file.new", "html/departments/$dep_file.html";
 412 }
 413 print $dep_fh qq|</ul>\n|, html_end;
 414 close($dep_fh);
 415 rename 'html/departments/index.new', 'html/departments/index.html';
 416
 417 my $azvo_stat;
 418
 419 foreach my $department ( sort keys %$department_category_author ) {
 420         foreach my $category ( sort keys %{ $department_category_author->{$department} } ) {
 421                 foreach my $authid ( @{ $department_category_author->{$department}->{$category} } ) {
 422                         foreach my $type ( keys %{ $authors->{$authid} } ) {
 423                                 next unless exists $authors->{$authid}->{$type}->{$category};
 424                                 $azvo_stat->{ $department }->{ $category }->{ $type } += $#{ $authors->{$authid}->{$type}->{$category} } + 1;
 425                         }
 426                 }
 427         }
 428 }
 429
 430 debug 'azvo_stat' => $azvo_stat;
 431
 432 =for later
 433 open(my $fh, '>', 'html/azvo.new');
 434
 435
 436
 437 close($fh);
 438 rename 'html/azvo.new', 'html/azvo.html';
 439 =cut
 440