misc/migration_tools/buildEDITORS.pl

   1 #!/usr/bin/perl
   2 # script that rebuild EDITORS
   3
   4 use strict;
   5 #use warnings; FIXME - Bug 2505
   6
   7 # Koha modules used
   8 use MARC::File::USMARC;
   9 use MARC::Record;
  10 use MARC::Batch;
  11 use C4::Context;
  12 use C4::Biblio;
  13 use C4::AuthoritiesMarc;
  14 use Time::HiRes qw(gettimeofday);
  15
  16 use Getopt::Long;
  17 my ( $input_marc_file, $number) = ('',0);
  18 my ($version, $verbose, $test_parameter, $confirm,$delete);
  19 GetOptions(
  20     'h' => \$version,
  21     'd' => \$delete,
  22     't' => \$test_parameter,
  23     'v' => \$verbose,
  24     'c' => \$confirm,
  25 );
  26
  27 if ($version or !$confirm) {
  28         print <<EOF
  29 small script to recreate a authority table into Koha.
  30 This will parse all your biblios to recreate isbn / editor / collections for the unimarc_210c and unimarc_225a plugins.
  31
  32 Remember those plugins will work only if you have an EDITORS authority type, with
  33 \t200a being the first 2 parts of an ISBN
  34 \t200b being the editor name
  35 \t200c (repeatable) being the series title
  36
  37 parameters :
  38 \t-c : confirmation flag. the script will run only with this flag. Otherwise, it will just show this help screen.
  39 \t-d : delete existing EDITORS before rebuilding them
  40 \t-t : test parameters : run the script but don't create really the EDITORS
  41 EOF
  42 ;#'
  43
  44 exit;
  45 }
  46
  47 my $dbh = C4::Context->dbh;
  48 if ($delete) {
  49         print "deleting EDITORS\n";
  50         my $del1 = $dbh->prepare("delete from auth_subfield_table where authid=?");
  51         my $del2 = $dbh->prepare("delete from auth_word where authid=?");
  52         my $sth = $dbh->prepare("select authid from auth_header where authtypecode='EDITORS'");
  53         $sth->execute;
  54         while (my ($authid) = $sth->fetchrow) {
  55                 $del1->execute($authid);
  56                 $del2->execute($authid);
  57         }
  58         $dbh->do("delete from auth_header where authtypecode='EDITORS'");
  59 }
  60
  61 if ($test_parameter) {
  62         print "TESTING MODE ONLY\n    DOING NOTHING\n===============\n";
  63 }
  64 $|=1; # flushes output
  65 my $starttime = gettimeofday;
  66 my $sth = $dbh->prepare("select bibid from marc_biblio");
  67 $sth->execute;
  68 my $i=1;
  69 my %alreadydone;
  70 my $counter;
  71 my %hash;
  72 while (my ($bibid) = $sth->fetchrow) {
  73         my $record = GetMarcBiblio($bibid);
  74         my $isbnField = $record->field('010');
  75         next unless $isbnField;
  76         my $isbn=$isbnField->subfield('a');
  77         my $seg1;
  78         if(substr($isbn, 0, 1) <=7) {
  79                 $seg1 = substr($isbn, 0, 1);
  80         } elsif(substr($isbn, 0, 2) <= 94) {
  81                 $seg1 = substr($isbn, 0, 2);
  82         } elsif(substr($isbn, 0, 3) <= 995) {
  83                 $seg1 = substr($isbn, 0, 3);
  84         } elsif(substr($isbn, 0, 4) <= 9989) {
  85                 $seg1 = substr($isbn, 0, 4);
  86         } else {
  87                 $seg1 = substr($isbn, 0, 5);
  88         }
  89         my $x = substr($isbn, length($seg1));
  90         my $seg2;
  91         if(substr($x, 0, 2) <= 19) {
  92 #               if(sTmp2 < 10) sTmp2 = "0" sTmp2;
  93                 $seg2 = substr($x, 0, 2);
  94         } elsif(substr($x, 0, 3) <= 699) {
  95                 $seg2 = substr($x, 0, 3);
  96         } elsif(substr($x, 0, 4) <= 8399) {
  97                 $seg2 = substr($x, 0, 4);
  98         } elsif(substr($x, 0, 5) <= 89999) {
  99                 $seg2 = substr($x, 0, 5);
 100         } elsif(substr($x, 0, 6) <= 9499999) {
 101                 $seg2 = substr($x, 0, 6);
 102         } else {
 103                 $seg2 = substr($x, 0, 7);
 104         }
 105         $counter++;
 106         print ".";
 107         my $timeneeded = gettimeofday - $starttime;
 108         print "$counter in $timeneeded s\n" unless ($counter % 100);
 109
 110         my $field = $record->field('210');
 111         my $editor;
 112         $editor=$field->subfield('c') if $field;
 113
 114         $field = $record->field('225');
 115         my $collection;
 116         $collection=$field->subfield('a') if $field;
 117
 118 #       print "WARNING : editor empty for ".$record->as_formatted unless $editor and !$verbose;
 119
 120         $hash{$seg1.$seg2}->{editors} = $editor unless ($hash{$seg1.$seg2}->{editors});
 121         $hash{$seg1.$seg2}->{collections}->{$collection}++ if $collection;
 122 }
 123
 124 foreach my $isbnstart (sort keys %hash) {
 125         print "$isbnstart -- ".$hash{$isbnstart}->{editors} if $verbose;
 126         my $collections = $hash{$isbnstart}->{collections};
 127         my $seriestitlelist;
 128         foreach my $collection (sort keys %$collections) {
 129                 print " CC $collection : ".$collections->{$collection} if $verbose;
 130                 $seriestitlelist.=$collection."|";
 131         }
 132         my $authorityRecord = MARC::Record->new();
 133         my $newfield = MARC::Field->new(200,'','','a' => "".$isbnstart,
 134                                                                                                 'b' => "".$hash{$isbnstart}->{editors},
 135                                                                                                 'c' => "".$seriestitlelist);
 136         $authorityRecord->insert_fields_ordered($newfield);
 137         my $authid=AUTHaddauthority($dbh,$authorityRecord,'','EDITORS');
 138
 139 #       print $authorityRecord->as_formatted."\n";
 140         print "\n" if $verbose;
 141 }
 142 exit;
 143
 144 #       my $timeneeded = gettimeofday - $starttime;
 145 #       print "$i in $timeneeded s\n" unless ($i % 50);
 146 #       foreach my $field ($record->field(995)) {
 147 #               $record->delete_field($field);
 148 #       }
 149 #       my $totdone=0;
 150 #       my $authid;
 151 #       foreach my $fieldnumber (('710','711','712')) {
 152 #               foreach my $field ($record->field($fieldnumber)) {
 153 #       #               print "=>".$field->as_formatted."\n";
 154 #                       foreach my $authentry ($field->subfield("a")) {
 155 #                               my $hashentry = $authentry;
 156 #                               # la particularit�de ce script l� c'est que l'entr� dans la table d'autorit�est $a -- $b (et pas $x -- $x -- $x -- $a comme pour les autorit� NC)
 157 #                               # si n�essaire, compl�er avec le $c (n'existe pas dans le fichier que j'ai migr�avec cette moulinette
 158 #                               # supprimer les accents, certaines entr�s sont sans, d'autres avec !
 159 #                               # mysql ne diff�encie pas, mais les hash perl oui !
 160 #                               $hashentry =~ s/���e/g;
 161 #                               $hashentry =~ s/��a/g;
 162 #                               $hashentry =~ s/�i/g;
 163 #                               $hashentry =~ s/�o/g;
 164 #                               $hashentry =~ s/|/u/g;
 165 #                               $hashentry = uc($hashentry);
 166 #                               print "==>$hashentry" if $hashentry =~ /.*ETATS.*/;
 167 #                               $totdone++;
 168 #                               if ($alreadydone{$hashentry}) {
 169 #                                       $authid = $alreadydone{$hashentry};
 170 #                                       print ".";
 171 #                               } else {
 172 #                                       print "*";
 173 #                                       #create authority.
 174 #                                       my $authorityRecord = MARC::Record->new();
 175 #                                       my $newfield = MARC::Field->new(210,'','','a' => "".$authentry,
 176 #                                                                                               'b' => "".$field->subfield('b'),
 177 #                                                                                               'c' => "".$field->subfield('c'),
 178 #                                                                                               );
 179 #                                       $authorityRecord->insert_fields_ordered($newfield);
 180 #                                       $authid=AUTHaddauthority($dbh,$authorityRecord,'','CO');
 181 #                                       $alreadydone{$hashentry} = $authid;
 182 #                                       # OK, on garde la notice d'autorit� on cherche les notices biblio et on les met �jour...
 183 #                                       if ($fieldnumber eq '710') {
 184 #                                               $sthBIBLIOS710->execute($authentry);
 185 #                                               while (my ($bibid,$tag,$tagorder,$subfieldorder) = $sthBIBLIOS710->fetchrow) {
 186 #                                                       my $inbiblio = GetMarcBiblio($bibid);
 187 #                                                       my $isOK = 0;
 188 #                                                       foreach my $in7xx ($inbiblio->field($fieldnumber)) {
 189 #                                                               # !!!!! ici, il faut reconstruire l'entr� de la table de hachage comme ci dessus
 190 #                                                               # sinon,
 191 #                                                               my $inEntry = $in7xx->subfield('a');
 192 #                                                               $inEntry =~ s/���e/g;
 193 #                                                               $inEntry =~ s/��a/g;
 194 #                                                               $inEntry =~ s/�i/g;
 195 #                                                               $inEntry =~ s/�o/g;
 196 #                                                               $inEntry =~ s/|/u/g;
 197 #                                                               $inEntry = uc($inEntry);
 198 #                                                               $isOK=1 if $inEntry eq $hashentry;
 199 #                                                       }
 200 #                                                       C4::Biblio::MARCaddsubfield($dbh,$bibid,$tag,'',$tagorder,9,$subfieldorder,$authid) if $isOK;
 201 #                                               }
 202 #                                       }
 203 #                                       if ($fieldnumber eq '711') {
 204 #                                               $sthBIBLIOS711->execute($authentry);
 205 #                                               while (my ($bibid,$tag,$tagorder,$subfieldorder) = $sthBIBLIOS711->fetchrow) {
 206 #                                                       my $inbiblio = GetMarcBiblio($bibid);
 207 #                                                       my $isOK = 0;
 208 #                                                       foreach my $in7xx ($inbiblio->field($fieldnumber)) {
 209 #                                                               # !!!!! ici, il faut reconstruire l'entr� de la table de hachage comme ci dessus
 210 #                                                               # sinon,
 211 #                                                               my $inEntry = $in7xx->subfield('a');
 212 #                                                               $inEntry =~ s/���e/g;
 213 #                                                               $inEntry =~ s/��a/g;
 214 #                                                               $inEntry =~ s/�i/g;
 215 #                                                               $inEntry =~ s/�o/g;
 216 #                                                               $inEntry =~ s/|/u/g;
 217 #                                                               $inEntry = uc($inEntry);
 218 #                                                               $isOK=1 if $inEntry eq $hashentry;
 219 #                                                       }
 220 #                                                       C4::Biblio::MARCaddsubfield($dbh,$bibid,$tag,'',$tagorder,9,$subfieldorder,$authid) if $isOK;
 221 #                                               }
 222 #                                       }
 223 #                                       if ($fieldnumber eq '712') {
 224 #                                               $sthBIBLIOS712->execute($authentry);
 225 #                                               while (my ($bibid,$tag,$tagorder,$subfieldorder) = $sthBIBLIOS712->fetchrow) {
 226 #                                                       my $inbiblio = GetMarcBiblio($bibid);
 227 #                                                       my $isOK = 0;
 228 #                                                       foreach my $in7xx ($inbiblio->field($fieldnumber)) {
 229 #                                                               # !!!!! ici, il faut reconstruire l'entr� de la table de hachage comme ci dessus
 230 #                                                               # sinon,
 231 #                                                               my $inEntry = $in7xx->subfield('a');
 232 #                                                               $inEntry =~ s/���e/g;
 233 #                                                               $inEntry =~ s/��a/g;
 234 #                                                               $inEntry =~ s/�i/g;
 235 #                                                               $inEntry =~ s/�o/g;
 236 #                                                               $inEntry =~ s/|/u/g;
 237 #                                                               $inEntry = uc($inEntry);
 238 #                                                               $isOK=1 if $inEntry eq $hashentry;
 239 #                                                       }
 240 #                                                       C4::Biblio::MARCaddsubfield($dbh,$bibid,$tag,'',$tagorder,9,$subfieldorder,$authid) if $isOK;
 241 #                                               }
 242 #                                       }
 243 #                               }
 244 #                       }
 245 #               }
 246 #       }
 247 #       $i++;
 248 # }
 249 # my $timeneeded = gettimeofday - $starttime;
 250 # print "$i entries done in $timeneeded seconds (".($i/$timeneeded)." per second)\n";