misc/migration_tools/buildEDITORS.pl

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