git.rot13.org Git - koha-eprints/blob - tsv2eprints.pl

   1 #!/usr/bin/perl -I /usr/share/eprints3/perl_lib
   2 use warnings;
   3 use strict;
   4 use autodie;
   5 use utf8;
   6
   7 # export single record to get structure:
   8 # sudo -u eprints /usr/share/eprints3/bin/export snz archive XMLFiles 20 > /tmp/20.xml
   9
  10 # Import procedure:
  11
  12 # 3. import with:
  13 # sudo -u eprints /usr/share/eprints3/bin/import --verbose --migration --enable-file-imports --update --enable-import-fields grf eprint XML /tmp/xml
  14 #
  15 # 4. re-run view generation
  16 # sudo -u eprints /usr/share/eprints3/bin/generate_views grf --verbose
  17
  18 use Encode;
  19 use Data::Dump qw(dump);
  20 use Storable;
  21
  22 =for eprints-api
  23
  24 use EPrints;
  25 my $institution = 'Grafički fakultet';
  26
  27 my $ep = EPrints->new();
  28 my $repo = $ep->repository( 'grf' );
  29 $repo->{config}->{enable_file_imports} = 1;
  30 $repo->{config}->{enable_web_imports} = 1;
  31
  32 my $dataset = $repo->dataset( 'eprint' );
  33 my $list = $dataset->search;
  34 my $count = $list->count;
  35 warn "# found [$count] eprints\n";
  36
  37 my $eprint = $dataset->dataobj( 21 );
  38 warn dump( $eprint->get_value('institution'), $institution );
  39
  40 warn dump( $eprint );
  41
  42 if ( $eprint->get_value( 'institution' ) ne $institution ) {
  43         $eprint->set_value( 'institution' => $institution );
  44         $eprint->save_revision();
  45         $eprint->commit();
  46 }
  47
  48 $repo->terminate();
  49
  50 =cut
  51
  52 our $eprintid = 1;
  53
  54 my $files;
  55
  56 my $mkp_path = "/mnt/share/MKP/ELEKTRONIČKI DOKUMENTI/EL.DOKUMENTI PO BIBLIOBROJU/do18052015/";
  57 my $koha_path = "/tmp/koha_ffzg";
  58
  59 open(my $fh, '-|:encoding(UTF-8)', 'find "' . $mkp_path . '" -iname "*.pdf"');
  60 while(my $full_path = <$fh>) {
  61         chomp $full_path;
  62
  63         my $file = $1 if $full_path =~ m{/([^/]+)\.pdf}i;
  64
  65         my $file_id;
  66         if ( $file =~ m/^(\d+)/ ) {
  67                 $file_id = $1;
  68         } else {
  69                 $file_id = $file;
  70         }
  71
  72         warn "# $file_id\t$full_path\n";
  73         $files->{ $file_id } = $full_path;
  74
  75 }
  76
  77 warn "# got ", scalar keys %$files, " files\n";
  78
  79 store $files, "$koha_path.biblionumber.file";
  80
  81 my $stat;
  82 my $this_id = '';
  83 my $item;
  84
  85 binmode STDOUT, ":utf8";
  86
  87 sub dump_item {
  88         my $item = shift || return;
  89
  90         my $f200 = $item->{200}->[0] || die "no 200 in ",dump($item);
  91
  92         if ( $f200 =~ s/\s*;\s*([^;]+?)$//i ) {
  93                 $item->{mentor} = $1;
  94                 $item->{mentor} =~ s/^\s*voditelji?\s*(?:rada)\s*//i;
  95         } else {
  96                 warn "MISSING ; voditelj [$f200]\n";
  97         }
  98
  99         if ( $f200 =~ s{\s*/\s*([^/]+?)$}{} ) {
 100                 $item->{autor} = $1;
 101         } else {
 102                 warn "MISSING / autor [$f200]\n";
 103         }
 104
 105         if ( $f200 =~ s{\s*:\s*([^:]+?)$}{} ) {
 106                 $item->{tip} = lc($1);
 107         } else {
 108                 warn "MISSING : tip [$f200]\n";
 109         }
 110
 111         $item->{title} = $f200;
 112
 113         if ( exists $item->{991} ) {
 114                 my $file_id = $item->{991}->[0];
 115                 if ( exists $files->{ $file_id } ) {
 116                         $item->{full_path} = delete $files->{ $file_id };
 117                 } elsif ( $file_id =~ s/(\w)0*(\d)/$1$2/ ) {
 118                         if ( exists $files->{ $file_id } ) {
 119                                 $item->{full_path} = delete $files->{ $file_id };
 120                         }
 121                 }
 122         }
 123
 124         if ( ! exists $item->{full_path} ) {
 125                 my $file_id = ucfirst( $item->{300}->[0] . ' ' . $item->{700}->[0] );
 126                 $file_id =~ s/[\.\,]//g;
 127                 if ( exists $files->{ $file_id } ) {
 128                         $item->{full_path} = delete $files->{ $file_id };
 129                 }
 130         }
 131
 132         warn "MISSING file for $eprintid\n" unless exists $item->{full_path};
 133
 134         warn "# item ",dump($item);
 135
 136         my $eprint = {
 137                 eprintid => $eprintid++
 138         };
 139
 140         $eprint->{filename} = $1 if $item->{full_path} =~ m{/([^/]+)$};
 141         $eprint->{full_path} = $item->{full_path};
 142
 143         $eprint->{date} = $1 if $item->{210}->[0] =~ m/\$d(\d+)/;
 144         $eprint->{pages} = $1 if $item->{215}->[0] =~ m/^(\d+)/;
 145
 146         ( $eprint->{creators_family}, $eprint->{creators_given} ) = split(/,\s*/, $item->{700}->[0] );
 147
 148         $eprint->{title} = $item->{title};
 149
 150         $eprint->{keywords} = join(", ", @{ $item->{610} }) if exists $item->{610};
 151
 152         if ( exists $item->{700}->[1] ) {
 153                 ( $eprint->{thesis_mentor_family}, $eprint->{thesis_mentor_given} ) = split(/,\s*/, $item->{700}->[1] );
 154         } elsif ( $item->{mentor} ) {
 155                 ( $eprint->{thesis_mentor_given}, $eprint->{thesis_mentor_family} ) = split(/\s+/, $item->{mentor} );
 156         }
 157
 158         $eprint->{thesis_mentor_family} =~ s/(\S+)\s*-\s*(\S+)/$1-$2/; # fix spaces between dash in double surname
 159
 160         $eprint->{thesis_callnumber} = $item->{990}->[0];
 161         $eprint->{thesis_invnumber} =  $item->{991}->[0]; # FIXME?
 162
 163         # fallback za radove bez datuma na godinu
 164         if ( ! $eprint->{thesis_date} && $item->{990}->[0] =~ m{/(\d\d\d\d)/} ) {
 165                 $eprint->{thesis_date} = $1;
 166         }
 167
 168         if ( ! $eprint->{date} ) {
 169                 $eprint->{date} = $eprint->{thesis_date};
 170         }
 171
 172         warn "# eprint ",dump($eprint);
 173
 174         print qq|
 175
 176   <eprint>
 177     <eprintid>$eprint->{eprintid}</eprintid>
 178
 179         |;
 180
 181         if ( $eprint->{full_path} ) {
 182                 print qq|
 183
 184     <documents>
 185       <document>
 186
 187         <files>
 188           <file>
 189             <datasetid>document</datasetid>
 190             <filename>$eprint->{filename}</filename>
 191             <mime_type>application/pdf</mime_type>
 192             <url>file://$eprint->{full_path}</url>
 193           </file>
 194         </files>
 195         <mime_type>application/pdf</mime_type>
 196         <format>application/pdf</format>
 197         <language>hr</language>
 198         <security>validuser</security>
 199         <main>$eprint->{filename}</main>
 200       </document>
 201     </documents>
 202
 203                 |;
 204         }
 205         print qq|
 206
 207     <eprint_status>archive</eprint_status>
 208     <type>$eprint->{type}</type>
 209     <metadata_visibility>show</metadata_visibility>
 210     <creators>
 211       <item>
 212         <name>
 213           <family>$eprint->{creators_family}</family>
 214           <given>$eprint->{creators_given}</given>
 215         </name>
 216       </item>
 217     </creators>
 218     <title>$eprint->{title}</title>
 219     <ispublished>unpub</ispublished>
 220     <subjects>
 221       <item>2.06</item>
 222     </subjects>
 223     <full_text_status>restricted</full_text_status>
 224     <keywords>$eprint->{keywords}</keywords>
 225     <date>$eprint->{date}</date>
 226     <date_type>completed</date_type>
 227     <pages>$eprint->{pages}</pages>
 228     <institution>Grafički fakultet</institution>
 229 <!--
 230     <department>strojevi</department>
 231 -->
 232     <thesis_date>$eprint->{thesis_date}</thesis_date>
 233     <thesis_callnumber>$eprint->{thesis_callnumber}</thesis_callnumber>
 234     <thesis_invnumber>$eprint->{thesis_invnumber}</thesis_invnumber>
 235     <thesis_mentor>
 236       <name>
 237         <family>$eprint->{thesis_mentor_family}</family>
 238         <given>$eprint->{thesis_mentor_given}</given>
 239       </name>
 240     </thesis_mentor>
 241   </eprint>
 242
 243         |;
 244
 245 }
 246
 247 print qq{<?xml version="1.0" encoding="utf-8" ?>
 248 <eprints>
 249 };
 250
 251
 252 open(my $tsv_fh,   '<:encoding(UTF-8)', "$koha_path.tsv");
 253 open(my $tsv_marc, '<:encoding(UTF-8)', "$koha_path.marc");
 254
 255 while(<$tsv_fh>) {
 256         my $line = $_;
 257         $line =~ s/[\n\r]+$//;
 258
 259         my ($offset, $biblionumber, $title) = split(/\t/,$_,3);
 260
 261         warn "# $offset $biblionumber $title\n";
 262
 263         exit if $ENV{LAST} && $eprintid >= $ENV{LAST};
 264
 265         my $item;
 266
 267         if ( $item->{full_path} = $files->{$biblionumber} ) {
 268                 $stat->{file}++;
 269         } else {
 270                 $stat->{missing}++;
 271         }
 272
 273 }
 274
 275 print qq{
 276 </eprints>
 277 };
 278
 279
 280 warn "# files left ", dump($files);
 281
 282 warn "# stat ", dump($stat);