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