tools/export.pl

   1 #!/usr/bin/perl
   2
   3 #
   4 # This file is part of Koha.
   5 #
   6 # Koha is free software; you can redistribute it and/or modify it under the
   7 # terms of the GNU General Public License as published by the Free Software
   8 # Foundation; either version 2 of the License, or (at your option) any later
   9 # version.
  10 #
  11 # Koha is distributed in the hope that it will be useful, but WITHOUT ANY
  12 # WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  13 # A PARTICULAR PURPOSE.  See the GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License along with
  16 # Koha; if not, write to the Free Software Foundation, Inc., 59 Temple Place,
  17 # Suite 330, Boston, MA  02111-1307 USA
  18
  19 use Modern::Perl;
  20 use MARC::File::XML;
  21 use List::MoreUtils qw(uniq);
  22 use Getopt::Long;
  23 use CGI;
  24 use C4::Auth;
  25 use C4::AuthoritiesMarc;    # GetAuthority
  26 use C4::Biblio;             # GetMarcBiblio
  27 use C4::Branch;             # GetBranches
  28 use C4::Csv;
  29 use C4::Koha;               # GetItemTypes
  30 use C4::Output;
  31 use C4::Record;
  32
  33 my $query = new CGI;
  34
  35 my $clean;
  36 my $output_format;
  37 my $dont_export_items;
  38 my $deleted_barcodes;
  39 my $timestamp;
  40 my $record_type;
  41 my $help;
  42 my $op       = $query->param("op")       || '';
  43 my $filename = $query->param("filename") || 'koha.mrc';
  44 my $dbh      = C4::Context->dbh;
  45 my $marcflavour = C4::Context->preference("marcflavour");
  46 my $format = $query->param("format") || 'iso2709';
  47
  48 # Checks if the script is called from commandline
  49 my $commandline = not defined $ENV{GATEWAY_INTERFACE};
  50
  51 if ( $commandline ) {
  52
  53     # Getting parameters
  54     $op = 'export';
  55     GetOptions(
  56         'format=s'          => \$output_format,
  57         'date=s'            => \$timestamp,
  58         'dont_export_items' => \$dont_export_items,
  59         'deleted_barcodes'  => \$deleted_barcodes,
  60         'clean'             => \$clean,
  61         'filename=s'        => \$filename,
  62         'record-type=s'     => \$record_type,
  63         'help|?'            => \$help
  64     );
  65
  66     if ($help) {
  67         print <<_USAGE_;
  68 export.pl [--format=format] [--date=date] [--record-type=TYPE] [--dont_export_items] [--deleted_barcodes] [--clean] --filename=outputfile
  69
  70
  71  --format=FORMAT        FORMAT is either 'xml' or 'marc' (default)
  72
  73  --date=DATE            DATE should be entered as the 'dateformat' syspref is
  74                         set (dd/mm/yyyy for metric, yyyy-mm-dd for iso,
  75                         mm/dd/yyyy for us) records exported are the ones that
  76                         have been modified since DATE
  77
  78  --record-type=TYPE     TYPE is 'bibs' or 'auths'
  79
  80  --deleted_barcodes     If used, a list of barcodes of items deleted since DATE
  81                         is produced (or from all deleted items if no date is
  82                         specified). Used only if TYPE is 'bibs'
  83
  84  --clean                removes NSE/NSB
  85 _USAGE_
  86         exit;
  87     }
  88
  89     # Default parameters values :
  90     $output_format     ||= 'marc';
  91     $timestamp         ||= '';
  92     $dont_export_items ||= 0;
  93     $deleted_barcodes  ||= 0;
  94     $clean             ||= 0;
  95     $record_type       ||= "bibs";
  96
  97     # Redirect stdout
  98     open STDOUT, '>', $filename if $filename;
  99
 100 }
 101 else {
 102
 103     $op       = $query->param("op")       || '';
 104     $filename = $query->param("filename") || 'koha.mrc';
 105     $filename =~ s/(\r|\n)//;
 106
 107 }
 108
 109 my ( $template, $loggedinuser, $cookie, $flags ) = get_template_and_user(
 110     {
 111         template_name   => "tools/export.tmpl",
 112         query           => $query,
 113         type            => "intranet",
 114         authnotrequired => $commandline,
 115         flagsrequired   => { tools => 'export_catalog' },
 116         debug           => 1,
 117     }
 118 );
 119
 120 my $limit_ind_branch =
 121   (      C4::Context->preference('IndependentBranches')
 122       && C4::Context->userenv
 123       && !C4::Context->IsSuperLibrarian()
 124       && C4::Context->userenv->{branch} ) ? 1 : 0;
 125
 126 my $branch = $query->param("branch") || '';
 127 if (   C4::Context->preference("IndependentBranches")
 128     && C4::Context->userenv
 129     && !C4::Context->IsSuperLibrarian() )
 130 {
 131     $branch = C4::Context->userenv->{'branch'};
 132 }
 133
 134 my $backupdir = C4::Context->config('backupdir');
 135
 136 if ( $op eq "export" ) {
 137     if ( $format eq "iso2709" or $format eq "xml" ) {
 138         my $charset  = 'utf-8';
 139         my $mimetype = 'application/octet-stream';
 140         binmode STDOUT, ':encoding(UTF-8)';
 141         if ( $filename =~ m/\.gz$/ ) {
 142             $mimetype = 'application/x-gzip';
 143             $charset  = '';
 144             binmode STDOUT;
 145         }
 146         elsif ( $filename =~ m/\.bz2$/ ) {
 147             $mimetype = 'application/x-bzip2';
 148             binmode STDOUT;
 149             $charset = '';
 150         }
 151         print $query->header(
 152             -type       => $mimetype,
 153             -charset    => $charset,
 154             -attachment => $filename
 155         ) unless ($commandline);
 156
 157         $record_type = $query->param("record_type") unless ($commandline);
 158         $output_format = $query->param("output_format") || 'marc'
 159           unless ($commandline);
 160         my $export_remove_fields = $query->param("export_remove_fields");
 161         my @biblionumbers      = $query->param("biblionumbers");
 162         my @itemnumbers        = $query->param("itemnumbers");
 163         my @sql_params;
 164         my $sql_query;
 165         my @recordids;
 166
 167         my $StartingBiblionumber = $query->param("StartingBiblionumber");
 168         my $EndingBiblionumber   = $query->param("EndingBiblionumber");
 169         my $itemtype             = $query->param("itemtype");
 170         my $start_callnumber     = $query->param("start_callnumber");
 171         my $end_callnumber       = $query->param("end_callnumber");
 172         $timestamp = ($timestamp) ? C4::Dates->new($timestamp) : ''
 173           if ($commandline);
 174         my $start_accession =
 175           ( $query->param("start_accession") )
 176           ? C4::Dates->new( $query->param("start_accession") )
 177           : '';
 178         my $end_accession =
 179           ( $query->param("end_accession") )
 180           ? C4::Dates->new( $query->param("end_accession") )
 181           : '';
 182         $dont_export_items = $query->param("dont_export_item")
 183           unless ($commandline);
 184
 185         my $strip_nonlocal_items = $query->param("strip_nonlocal_items");
 186
 187         my $biblioitemstable =
 188           ( $commandline and $deleted_barcodes )
 189           ? 'deletedbiblioitems'
 190           : 'biblioitems';
 191         my $itemstable =
 192           ( $commandline and $deleted_barcodes )
 193           ? 'deleteditems'
 194           : 'items';
 195
 196         my $starting_authid = $query->param('starting_authid');
 197         my $ending_authid   = $query->param('ending_authid');
 198         my $authtype        = $query->param('authtype');
 199
 200         if ( $record_type eq 'bibs' and not @biblionumbers ) {
 201             if ($timestamp) {
 202
 203             # Specific query when timestamp is used
 204             # Actually it's used only with CLI and so all previous filters
 205             # are not used.
 206             # If one day timestamp is used via the web interface, this part will
 207             # certainly have to be rewrited
 208                 my ( $query, $params ) = construct_query(
 209                     {
 210                         recordtype       => $record_type,
 211                         timestamp        => $timestamp,
 212                         biblioitemstable => $biblioitemstable,
 213                     }
 214                 );
 215                 $sql_query  = $query;
 216                 @sql_params = @$params;
 217
 218             }
 219             else {
 220                 my ( $query, $params ) = construct_query(
 221                     {
 222                         recordtype           => $record_type,
 223                         biblioitemstable     => $biblioitemstable,
 224                         itemstable           => $itemstable,
 225                         StartingBiblionumber => $StartingBiblionumber,
 226                         EndingBiblionumber   => $EndingBiblionumber,
 227                         branch               => $branch,
 228                         start_callnumber     => $start_callnumber,
 229                         end_callnumber       => $end_callnumber,
 230                         start_accession      => $start_accession,
 231                         end_accession        => $end_accession,
 232                         itemtype             => $itemtype,
 233                     }
 234                 );
 235                 $sql_query  = $query;
 236                 @sql_params = @$params;
 237             }
 238         }
 239         elsif ( $record_type eq 'auths' ) {
 240             my ( $query, $params ) = construct_query(
 241                 {
 242                     recordtype      => $record_type,
 243                     starting_authid => $starting_authid,
 244                     ending_authid   => $ending_authid,
 245                     authtype        => $authtype,
 246                 }
 247             );
 248             $sql_query  = $query;
 249             @sql_params = @$params;
 250
 251         }
 252         elsif ( $record_type eq 'db' ) {
 253             my $successful_export;
 254             if ( $flags->{superlibrarian}
 255                 && C4::Context->config('backup_db_via_tools') )
 256             {
 257                 $successful_export = download_backup(
 258                     {
 259                         directory => "$backupdir",
 260                         extension => 'sql',
 261                         filename  => "$filename"
 262                     }
 263                 );
 264             }
 265             unless ($successful_export) {
 266                 my $remotehost = $query->remote_host();
 267                 $remotehost =~ s/(\n|\r)//;
 268                 warn
 269 "A suspicious attempt was made to download the db at '$filename' by someone at "
 270                   . $remotehost . "\n";
 271             }
 272             exit;
 273         }
 274         elsif ( $record_type eq 'conf' ) {
 275             my $successful_export;
 276             if ( $flags->{superlibrarian}
 277                 && C4::Context->config('backup_conf_via_tools') )
 278             {
 279                 $successful_export = download_backup(
 280                     {
 281                         directory => "$backupdir",
 282                         extension => 'tar',
 283                         filename  => "$filename"
 284                     }
 285                 );
 286             }
 287             unless ($successful_export) {
 288                 my $remotehost = $query->remote_host();
 289                 $remotehost =~ s/(\n|\r)//;
 290                 warn
 291 "A suspicious attempt was made to download the configuration at '$filename' by someone at "
 292                   . $remotehost . "\n";
 293             }
 294             exit;
 295         }
 296         elsif (@biblionumbers) {
 297             push @recordids, (@biblionumbers);
 298         }
 299         else {
 300
 301             # Someone is trying to mess us up
 302             exit;
 303         }
 304
 305         unless (@biblionumbers) {
 306             my $sth = $dbh->prepare($sql_query);
 307             $sth->execute(@sql_params);
 308             push @recordids, map {
 309                 map { $$_[0] } $_
 310             } @{ $sth->fetchall_arrayref };
 311         }
 312
 313         my $xml_header_written = 0;
 314         for my $recordid ( uniq @recordids ) {
 315             if ($deleted_barcodes) {
 316                 my $q = "
 317                     SELECT DISTINCT barcode
 318                     FROM deleteditems
 319                     WHERE deleteditems.biblionumber = ?
 320                 ";
 321                 my $sth = $dbh->prepare($q);
 322                 $sth->execute($recordid);
 323                 while ( my $row = $sth->fetchrow_array ) {
 324                     print "$row\n";
 325                 }
 326             }
 327             else {
 328                 my $record;
 329                 if ( $record_type eq 'bibs' ) {
 330                     $record = eval { GetMarcBiblio($recordid); };
 331
 332                     next if $@;
 333                     next if not defined $record;
 334                     C4::Biblio::EmbedItemsInMarcBiblio( $record, $recordid,
 335                         \@itemnumbers )
 336                       unless $dont_export_items;
 337                     if (   $strip_nonlocal_items
 338                         || $limit_ind_branch
 339                         || $dont_export_items )
 340                     {
 341                         my ( $homebranchfield, $homebranchsubfield ) =
 342                           GetMarcFromKohaField( 'items.homebranch', '' );
 343                         for my $itemfield ( $record->field($homebranchfield) ) {
 344
 345 # if stripping nonlocal items, use loggedinuser's branch if they didn't select one
 346                             $branch = C4::Context->userenv->{'branch'}
 347                               unless $branch;
 348                             $record->delete_field($itemfield)
 349                               if ( $dont_export_items
 350                                 || $itemfield->subfield($homebranchsubfield) ne
 351                                 $branch );
 352                         }
 353                     }
 354                 }
 355                 elsif ( $record_type eq 'auths' ) {
 356                     $record = C4::AuthoritiesMarc::GetAuthority($recordid);
 357                     next if not defined $record;
 358                 }
 359
 360                 if ($export_remove_fields) {
 361                     for my $f ( split / /, $export_remove_fields ) {
 362                         if ( $f =~ m/^(\d{3})(.)?$/ ) {
 363                             my ( $field, $subfield ) = ( $1, $2 );
 364
 365                             # skip if this record doesn't have this field
 366                             if ( defined $record->field($field) ) {
 367                                 if ( defined $subfield ) {
 368                                     my @tags = $record->field($field);
 369                                     foreach my $t (@tags) {
 370                                         $t->delete_subfields($subfield);
 371                                     }
 372                                 }
 373                                 else {
 374                                     $record->delete_fields($field);
 375                                 }
 376                             }
 377                         }
 378                     }
 379                 }
 380                 RemoveAllNsb($record) if ($clean);
 381                 if ( $output_format eq "xml" ) {
 382                     unless ($xml_header_written) {
 383                         MARC::File::XML->default_record_format(
 384                             (
 385                                      $marcflavour eq 'UNIMARC'
 386                                   && $record_type eq 'auths'
 387                             ) ? 'UNIMARCAUTH' : $marcflavour
 388                         );
 389                         print MARC::File::XML::header();
 390                         print "\n";
 391                         $xml_header_written = 1;
 392                     }
 393                     print MARC::File::XML::record($record);
 394                     print "\n";
 395                 }
 396                 else {
 397                     print $record->as_usmarc();
 398                 }
 399             }
 400         }
 401         if ($xml_header_written) {
 402             print MARC::File::XML::footer();
 403             print "\n";
 404         }
 405
 406         exit;
 407     }
 408     elsif ( $format eq "csv" ) {
 409         my @biblionumbers = uniq $query->param("biblionumbers");
 410         my @itemnumbers   = $query->param("itemnumbers");
 411         my $output =
 412           marc2csv( \@biblionumbers,
 413             GetCsvProfileId( C4::Context->preference('ExportWithCsvProfile') ),
 414             \@itemnumbers, );
 415         print $query->header(
 416             -type                        => 'application/octet-stream',
 417             -'Content-Transfer-Encoding' => 'binary',
 418             -attachment                  => "export.csv"
 419         );
 420         print $output;
 421         exit;
 422     }
 423 }    # if export
 424
 425 else {
 426
 427     my $itemtypes = GetItemTypes;
 428     my @itemtypesloop;
 429     foreach my $thisitemtype ( sort keys %$itemtypes ) {
 430         my %row = (
 431             value       => $thisitemtype,
 432             description => $itemtypes->{$thisitemtype}->{'description'},
 433         );
 434         push @itemtypesloop, \%row;
 435     }
 436     my $branches = GetBranches($limit_ind_branch);
 437     my @branchloop;
 438     for my $thisbranch (
 439         sort { $branches->{$a}->{branchname} cmp $branches->{$b}->{branchname} }
 440         keys %{$branches}
 441       )
 442     {
 443         push @branchloop,
 444           {
 445             value      => $thisbranch,
 446             selected   => $thisbranch eq $branch,
 447             branchname => $branches->{$thisbranch}->{'branchname'},
 448           };
 449     }
 450
 451     my $authtypes = getauthtypes;
 452     my @authtypesloop;
 453     foreach my $thisauthtype ( sort keys %$authtypes ) {
 454         next unless $thisauthtype;
 455         my %row = (
 456             value       => $thisauthtype,
 457             description => $authtypes->{$thisauthtype}->{'authtypetext'},
 458         );
 459         push @authtypesloop, \%row;
 460     }
 461
 462     if (   $flags->{superlibrarian}
 463         && C4::Context->config('backup_db_via_tools')
 464         && $backupdir
 465         && -d $backupdir )
 466     {
 467         $template->{VARS}->{'allow_db_export'} = 1;
 468         $template->{VARS}->{'dbfiles'}         = getbackupfilelist(
 469             { directory => "$backupdir", extension => 'sql' } );
 470     }
 471
 472     if (   $flags->{superlibrarian}
 473         && C4::Context->config('backup_conf_via_tools')
 474         && $backupdir
 475         && -d $backupdir )
 476     {
 477         $template->{VARS}->{'allow_conf_export'} = 1;
 478         $template->{VARS}->{'conffiles'}         = getbackupfilelist(
 479             { directory => "$backupdir", extension => 'tar' } );
 480     }
 481
 482     $template->param(
 483         branchloop               => \@branchloop,
 484         itemtypeloop             => \@itemtypesloop,
 485         authtypeloop             => \@authtypesloop,
 486         export_remove_fields     => C4::Context->preference("ExportRemoveFields"),
 487     );
 488
 489     output_html_with_http_headers $query, $cookie, $template->output;
 490 }
 491
 492 sub construct_query {
 493     my ($params) = @_;
 494
 495     my ( $sql_query, @sql_params );
 496
 497     if ( $params->{recordtype} eq "bibs" ) {
 498         if ( $params->{timestamp} ) {
 499             my $biblioitemstable = $params->{biblioitemstable};
 500             $sql_query = " (
 501                 SELECT biblionumber
 502                 FROM $biblioitemstable
 503                   LEFT JOIN items USING(biblionumber)
 504                 WHERE $biblioitemstable.timestamp >= ?
 505                   OR items.timestamp >= ?
 506             ) UNION (
 507                 SELECT biblionumber
 508                 FROM $biblioitemstable
 509                   LEFT JOIN deleteditems USING(biblionumber)
 510                 WHERE $biblioitemstable.timestamp >= ?
 511                   OR deleteditems.timestamp >= ?
 512             ) ";
 513             my $ts = $timestamp->output('iso');
 514             @sql_params = ( $ts, $ts, $ts, $ts );
 515         }
 516         else {
 517             my $biblioitemstable     = $params->{biblioitemstable};
 518             my $itemstable           = $params->{itemstable};
 519             my $StartingBiblionumber = $params->{StartingBiblionumber};
 520             my $EndingBiblionumber   = $params->{EndingBiblionumber};
 521             my $branch               = $params->{branch};
 522             my $start_callnumber     = $params->{start_callnumber};
 523             my $end_callnumber       = $params->{end_callnumber};
 524             my $start_accession      = $params->{start_accession};
 525             my $end_accession        = $params->{end_accession};
 526             my $itemtype             = $params->{itemtype};
 527             my $items_filter =
 528                  $branch
 529               || $start_callnumber
 530               || $end_callnumber
 531               || $start_accession
 532               || $end_accession
 533               || ( $itemtype && C4::Context->preference('item-level_itypes') );
 534             $sql_query = $items_filter
 535               ? "SELECT DISTINCT $biblioitemstable.biblionumber
 536                 FROM $biblioitemstable JOIN $itemstable
 537                 USING (biblionumber) WHERE 1"
 538               : "SELECT $biblioitemstable.biblionumber FROM $biblioitemstable WHERE biblionumber >0 ";
 539
 540             if ($StartingBiblionumber) {
 541                 $sql_query .= " AND $biblioitemstable.biblionumber >= ? ";
 542                 push @sql_params, $StartingBiblionumber;
 543             }
 544
 545             if ($EndingBiblionumber) {
 546                 $sql_query .= " AND $biblioitemstable.biblionumber <= ? ";
 547                 push @sql_params, $EndingBiblionumber;
 548             }
 549
 550             if ($branch) {
 551                 $sql_query .= " AND homebranch = ? ";
 552                 push @sql_params, $branch;
 553             }
 554
 555             if ($start_callnumber) {
 556                 $sql_query .= " AND itemcallnumber >= ? ";
 557                 push @sql_params, $start_callnumber;
 558             }
 559
 560             if ($end_callnumber) {
 561                 $sql_query .= " AND itemcallnumber <= ? ";
 562                 push @sql_params, $end_callnumber;
 563             }
 564             if ($start_accession) {
 565                 $sql_query .= " AND dateaccessioned >= ? ";
 566                 push @sql_params, $start_accession->output('iso');
 567             }
 568
 569             if ($end_accession) {
 570                 $sql_query .= " AND dateaccessioned <= ? ";
 571                 push @sql_params, $end_accession->output('iso');
 572             }
 573
 574             if ($itemtype) {
 575                 $sql_query .=
 576                   ( C4::Context->preference('item-level_itypes') )
 577                   ? " AND items.itype = ? "
 578                   : " AND biblioitems.itemtype = ?";
 579                 push @sql_params, $itemtype;
 580             }
 581         }
 582     }
 583     elsif ( $params->{recordtype} eq "auths" ) {
 584         if ( $params->{timestamp} ) {
 585
 586             #TODO
 587         }
 588         else {
 589             my $starting_authid = $params->{starting_authid};
 590             my $ending_authid   = $params->{ending_authid};
 591             my $authtype        = $params->{authtype};
 592             $sql_query =
 593               "SELECT DISTINCT auth_header.authid FROM auth_header WHERE 1";
 594
 595             if ($starting_authid) {
 596                 $sql_query .= " AND auth_header.authid >= ? ";
 597                 push @sql_params, $starting_authid;
 598             }
 599
 600             if ($ending_authid) {
 601                 $sql_query .= " AND auth_header.authid <= ? ";
 602                 push @sql_params, $ending_authid;
 603             }
 604
 605             if ($authtype) {
 606                 $sql_query .= " AND auth_header.authtypecode = ? ";
 607                 push @sql_params, $authtype;
 608             }
 609         }
 610     }
 611     return ( $sql_query, \@sql_params );
 612 }
 613
 614 sub getbackupfilelist {
 615     my $args      = shift;
 616     my $directory = $args->{directory};
 617     my $extension = $args->{extension};
 618     my @files;
 619
 620     if ( opendir( my $dir, $directory ) ) {
 621         while ( my $file = readdir($dir) ) {
 622             next unless ( $file =~ m/\.$extension(\.(gz|bz2|xz))?/ );
 623             push @files, $file
 624               if ( -f "$directory/$file" && -r "$directory/$file" );
 625         }
 626         closedir($dir);
 627     }
 628     return \@files;
 629 }
 630
 631 sub download_backup {
 632     my $args      = shift;
 633     my $directory = $args->{directory};
 634     my $extension = $args->{extension};
 635     my $filename  = $args->{filename};
 636
 637     return unless ( $directory && -d $directory );
 638     return unless ( $filename =~ m/\.$extension(\.(gz|bz2|xz))?$/ );
 639     return if ( $filename =~ m#/# );
 640     $filename = "$directory/$filename";
 641     return unless ( -f $filename && -r $filename );
 642     return unless ( open( my $dump, '<', $filename ) );
 643     binmode $dump;
 644
 645     while ( read( $dump, my $data, 64 * 1024 ) ) {
 646         print $data;
 647     }
 648     close($dump);
 649     return 1;
 650 }