[18457] | 1 | ###########################################################################
|
---|
| 2 | #
|
---|
[22413] | 3 | # inexport.pm -- useful class to support import.pl and export.pl
|
---|
[18457] | 4 | # A component of the Greenstone digital library software
|
---|
| 5 | # from the New Zealand Digital Library Project at the
|
---|
| 6 | # University of Waikato, New Zealand.
|
---|
| 7 | #
|
---|
| 8 | # Copyright (C) 1999 New Zealand Digital Library Project
|
---|
| 9 | #
|
---|
| 10 | # This program is free software; you can redistribute it and/or modify
|
---|
| 11 | # it under the terms of the GNU General Public License as published by
|
---|
| 12 | # the Free Software Foundation; either version 2 of the License, or
|
---|
| 13 | # (at your option) any later version.
|
---|
| 14 | #
|
---|
| 15 | # This program is distributed in the hope that it will be useful,
|
---|
| 16 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 17 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 18 | # GNU General Public License for more details.
|
---|
| 19 | #
|
---|
| 20 | # You should have received a copy of the GNU General Public License
|
---|
| 21 | # along with this program; if not, write to the Free Software
|
---|
| 22 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
| 23 | #
|
---|
| 24 | ###########################################################################
|
---|
| 25 |
|
---|
| 26 | package inexport;
|
---|
| 27 |
|
---|
| 28 | use strict;
|
---|
| 29 |
|
---|
[22413] | 30 | no strict 'refs'; # allow filehandles to be variables and vice versa
|
---|
| 31 | no strict 'subs'; # allow barewords (eg STDERR) as function arguments
|
---|
[19789] | 32 |
|
---|
[22413] | 33 | use arcinfo;
|
---|
| 34 | use colcfg;
|
---|
[21553] | 35 | use dbutil;
|
---|
[22464] | 36 | use doc;
|
---|
[22413] | 37 | use plugin;
|
---|
| 38 | use plugout;
|
---|
| 39 | use manifest;
|
---|
| 40 | use inexport;
|
---|
[18457] | 41 | use util;
|
---|
[22413] | 42 | use scriptutil;
|
---|
| 43 | use FileHandle;
|
---|
| 44 | use gsprintf 'gsprintf';
|
---|
| 45 | use printusage;
|
---|
| 46 | use parse2;
|
---|
[18457] | 47 |
|
---|
[22413] | 48 | use File::Basename;
|
---|
[21563] | 49 |
|
---|
[22413] | 50 | sub new
|
---|
| 51 | {
|
---|
| 52 | my $class = shift (@_);
|
---|
[22421] | 53 | my ($mode,$argv,$options,$opt_listall_options) = @_;
|
---|
[22413] | 54 |
|
---|
[22421] | 55 | my $self = { 'xml' => 0, 'mode' => $mode };
|
---|
[22413] | 56 |
|
---|
| 57 | # general options available to all plugins
|
---|
| 58 | my $arguments = $options->{'args'};
|
---|
| 59 | my $intArgLeftinAfterParsing = parse2::parse($argv,$arguments,$self,"allow_extra_options");
|
---|
| 60 | # Parse returns -1 if something has gone wrong
|
---|
| 61 | if ($intArgLeftinAfterParsing == -1)
|
---|
| 62 | {
|
---|
| 63 | &PrintUsage::print_txt_usage($options, "{import.params}");
|
---|
| 64 | die "\n";
|
---|
| 65 | }
|
---|
| 66 |
|
---|
| 67 | my $language = $self->{'language'};
|
---|
| 68 | # If $language has been specified, load the appropriate resource bundle
|
---|
| 69 | # (Otherwise, the default resource bundle will be loaded automatically)
|
---|
| 70 | if ($language && $language =~ /\S/) {
|
---|
| 71 | &gsprintf::load_language_specific_resource_bundle($language);
|
---|
| 72 | }
|
---|
| 73 |
|
---|
[22421] | 74 | if ($self->{'listall'}) {
|
---|
| 75 | if ($self->{'xml'}) {
|
---|
| 76 | &PrintUsage::print_xml_usage($opt_listall_options);
|
---|
| 77 | }
|
---|
| 78 | else
|
---|
| 79 | {
|
---|
| 80 | &PrintUsage::print_txt_usage($opt_listall_options,"{export.params}");
|
---|
| 81 | }
|
---|
| 82 | die "\n";
|
---|
| 83 | }
|
---|
| 84 |
|
---|
| 85 |
|
---|
[22413] | 86 | if ($self->{'xml'}) {
|
---|
| 87 | &PrintUsage::print_xml_usage($options);
|
---|
| 88 | print "\n";
|
---|
[22460] | 89 | return bless $self, $class;
|
---|
[22413] | 90 | }
|
---|
| 91 |
|
---|
| 92 | if ($self->{'gli'}) { # the gli wants strings to be in UTF-8
|
---|
| 93 | &gsprintf::output_strings_in_UTF8;
|
---|
| 94 | }
|
---|
| 95 |
|
---|
| 96 | # now check that we had exactly one leftover arg, which should be
|
---|
| 97 | # the collection name. We don't want to do this earlier, cos
|
---|
| 98 | # -xml arg doesn't need a collection name
|
---|
| 99 | # Or if the user specified -h, then we output the usage also
|
---|
[22460] | 100 |
|
---|
[22413] | 101 | if ($intArgLeftinAfterParsing != 1 || (@$argv && $argv->[0] =~ /^\-+h/))
|
---|
| 102 | {
|
---|
| 103 | &PrintUsage::print_txt_usage($options, "{import.params}");
|
---|
| 104 | die "\n";
|
---|
| 105 | }
|
---|
| 106 |
|
---|
| 107 | $self->{'close_out'} = 0;
|
---|
| 108 | my $out = $self->{'out'};
|
---|
| 109 | if ($out !~ /^(STDERR|STDOUT)$/i) {
|
---|
| 110 | open (OUT, ">$out") ||
|
---|
| 111 | (&gsprintf(STDERR, "{common.cannot_open_output_file}: $!\n", $out) && die);
|
---|
[23042] | 112 | $out = 'inexport::OUT';
|
---|
[22413] | 113 | $self->{'close_out'} = 1;
|
---|
| 114 | }
|
---|
| 115 | $out->autoflush(1);
|
---|
| 116 | $self->{'out'} = $out;
|
---|
| 117 |
|
---|
| 118 | # @ARGV should be only one item, the name of the collection
|
---|
| 119 | $self->{'collection'} = shift @$argv;
|
---|
| 120 |
|
---|
[27302] | 121 | # Unless otherwise stated all manifests are considered version 1---where
|
---|
| 122 | # they act more like an advanced process expression---as compared to newer
|
---|
| 123 | # manifest files that act as an explicit (and exhaustive) list of files to
|
---|
| 124 | # process [jmt12]
|
---|
| 125 | $self->{'manifest_version'} = 1;
|
---|
[22445] | 126 |
|
---|
[22413] | 127 | return bless $self, $class;
|
---|
| 128 | }
|
---|
| 129 |
|
---|
[23402] | 130 | # Simplified version of the contstructor for use with CGI scripts
|
---|
| 131 | sub newCGI
|
---|
| 132 | {
|
---|
| 133 | my $class = shift (@_);
|
---|
[23767] | 134 | my ($mode,$collect,$gsdl_cgi,$opt_site) = @_;
|
---|
[23402] | 135 |
|
---|
| 136 | my $self = { 'xml' => 0, 'mode' => $mode };
|
---|
| 137 |
|
---|
| 138 | $self->{'out'} = STDERR;
|
---|
[23767] | 139 |
|
---|
| 140 | if (defined $gsdl_cgi) {
|
---|
| 141 | $self->{'site'} = $opt_site;
|
---|
| 142 | my $collect_dir = $gsdl_cgi->get_collection_dir($opt_site);
|
---|
| 143 | $self->{'collectdir'} = $collect_dir;
|
---|
| 144 | }
|
---|
| 145 | else {
|
---|
| 146 | $self->{'site'} = "";
|
---|
[27302] | 147 | $self->{'collectdir'} = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"collect");
|
---|
[23767] | 148 | }
|
---|
[23402] | 149 | $self->{'faillog'} = "";
|
---|
| 150 |
|
---|
| 151 | $self->{'collection'} = $collect;
|
---|
| 152 |
|
---|
| 153 | return bless $self, $class;
|
---|
| 154 | }
|
---|
[22413] | 155 | sub get_collection
|
---|
| 156 | {
|
---|
| 157 | my $self = shift @_;
|
---|
| 158 |
|
---|
| 159 | return $self->{'collection'};
|
---|
| 160 | }
|
---|
| 161 |
|
---|
| 162 |
|
---|
| 163 | sub read_collection_cfg
|
---|
| 164 | {
|
---|
| 165 | my $self = shift @_;
|
---|
| 166 | my ($collection,$options) = @_;
|
---|
| 167 |
|
---|
| 168 | my $collectdir = $self->{'collectdir'};
|
---|
| 169 | my $site = $self->{'site'};
|
---|
| 170 | my $out = $self->{'out'};
|
---|
[23767] | 171 |
|
---|
[22413] | 172 | if (($collection = &colcfg::use_collection($site, $collection, $collectdir)) eq "") {
|
---|
| 173 | &PrintUsage::print_txt_usage($options, "{import.params}");
|
---|
| 174 | die "\n";
|
---|
| 175 | }
|
---|
| 176 |
|
---|
[26567] | 177 | # set gs_version 2/3
|
---|
[25957] | 178 | $self->{'gs_version'} = "2";
|
---|
| 179 | if ((defined $site) && ($site ne "")) {
|
---|
| 180 | # gs3
|
---|
| 181 | $self->{'gs_version'} = "3";
|
---|
| 182 | }
|
---|
[27302] | 183 |
|
---|
| 184 | # add collection's perllib dir into include path in
|
---|
[22413] | 185 | # case we have collection specific modules
|
---|
[27302] | 186 | &util::augmentINC(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, 'perllib'));
|
---|
[22413] | 187 |
|
---|
| 188 | # check that we can open the faillog
|
---|
| 189 | my $faillog = $self->{'faillog'};
|
---|
| 190 | if ($faillog eq "") {
|
---|
[27302] | 191 | $faillog = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "etc", "fail.log");
|
---|
[22413] | 192 | }
|
---|
| 193 | open (FAILLOG, ">$faillog") ||
|
---|
| 194 | (&gsprintf(STDERR, "{import.cannot_open_fail_log}\n", $faillog) && die);
|
---|
| 195 |
|
---|
| 196 |
|
---|
| 197 | my $faillogname = $faillog;
|
---|
| 198 | $faillog = 'inexport::FAILLOG';
|
---|
| 199 | $faillog->autoflush(1);
|
---|
| 200 | $self->{'faillog'} = $faillog;
|
---|
| 201 | $self->{'faillogname'} = $faillogname;
|
---|
[27302] | 202 | $self->{'close_faillog'} = 1;
|
---|
[22413] | 203 |
|
---|
| 204 | # Read in the collection configuration file.
|
---|
[26567] | 205 | my $gs_mode = "gs".$self->{'gs_version'}; #gs2 or gs3
|
---|
| 206 | my $config_filename = &colcfg::get_collect_cfg_name($out, $gs_mode);
|
---|
[22421] | 207 | my $collectcfg = &colcfg::read_collection_cfg ($config_filename, $gs_mode);
|
---|
[22413] | 208 |
|
---|
[22421] | 209 | return ($config_filename,$collectcfg);
|
---|
[22413] | 210 | }
|
---|
| 211 |
|
---|
| 212 | sub set_collection_options
|
---|
| 213 | {
|
---|
| 214 | my $self = shift @_;
|
---|
[22421] | 215 | my ($collectcfg) = @_;
|
---|
[22413] | 216 |
|
---|
[22421] | 217 | my $inexport_mode = $self->{'mode'};
|
---|
| 218 |
|
---|
[22413] | 219 | my $verbosity = $self->{'verbosity'};
|
---|
| 220 | my $debug = $self->{'debug'};
|
---|
| 221 | my $importdir = $self->{'importdir'};
|
---|
[22466] | 222 | my $archivedir = $self->{'archivedir'} || $self->{'exportdir'} || "";
|
---|
[22413] | 223 | my $out = $self->{'out'};
|
---|
| 224 |
|
---|
| 225 | # If the infodbtype value wasn't defined in the collect.cfg file, use the default
|
---|
| 226 | if (!defined($collectcfg->{'infodbtype'}))
|
---|
| 227 | {
|
---|
| 228 | $collectcfg->{'infodbtype'} = &dbutil::get_default_infodb_type();
|
---|
| 229 | }
|
---|
[23170] | 230 | if ($collectcfg->{'infodbtype'} eq "gdbm-txtgz") {
|
---|
| 231 | # we can't use the text version for archives dbs.
|
---|
| 232 | $collectcfg->{'infodbtype'} = "gdbm";
|
---|
| 233 | }
|
---|
[26450] | 234 |
|
---|
[22413] | 235 | if (defined $collectcfg->{'importdir'} && $importdir eq "") {
|
---|
| 236 | $importdir = $collectcfg->{'importdir'};
|
---|
| 237 | }
|
---|
| 238 | if (defined $collectcfg->{'archivedir'} && $archivedir eq "") {
|
---|
| 239 | $archivedir = $collectcfg->{'archivedir'};
|
---|
| 240 | }
|
---|
| 241 | # fill in the default import and archives directories if none
|
---|
| 242 | # were supplied, turn all \ into / and remove trailing /
|
---|
[27394] | 243 | if ($importdir eq "")
|
---|
| 244 | {
|
---|
| 245 | $importdir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "import");
|
---|
[22413] | 246 | }
|
---|
[27394] | 247 | else
|
---|
| 248 | {
|
---|
| 249 | # Don't do this - it kills protocol prefixes
|
---|
| 250 | #$importdir =~ s/[\\\/]+/\//g;
|
---|
| 251 | #$importdir =~ s/\/$//;
|
---|
| 252 | # Do this instead
|
---|
| 253 | &FileUtils::sanitizePath($importdir);
|
---|
| 254 | }
|
---|
| 255 | if (!&FileUtils::directoryExists($importdir))
|
---|
| 256 | {
|
---|
| 257 | &gsprintf($out, "{import.no_import_dir}\n\n", $importdir);
|
---|
| 258 | die "\n";
|
---|
| 259 | }
|
---|
[22413] | 260 | $self->{'importdir'} = $importdir;
|
---|
| 261 |
|
---|
| 262 | if ($archivedir eq "") {
|
---|
| 263 | if ($inexport_mode eq "import") {
|
---|
[27302] | 264 | $archivedir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "archives");
|
---|
[22413] | 265 | }
|
---|
| 266 | elsif ($inexport_mode eq "export") {
|
---|
[27302] | 267 | $archivedir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "export");
|
---|
[22413] | 268 | }
|
---|
| 269 | else {
|
---|
| 270 | print STDERR "Warning: Unrecognized import/export mode '$inexport_mode'\n";
|
---|
| 271 | print STDERR " Defaulting to 'archives' for file output\n";
|
---|
[27302] | 272 | $archivedir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "archives");
|
---|
[22413] | 273 | }
|
---|
| 274 | }
|
---|
| 275 |
|
---|
[27483] | 276 | $archivedir = &FileUtils::sanitizePath($archivedir);
|
---|
[27482] | 277 | #$archivedir =~ s/[\\\/]+/\//g;
|
---|
| 278 | #$archivedir =~ s/\/$//;
|
---|
[22413] | 279 | $self->{'archivedir'} = $archivedir;
|
---|
| 280 |
|
---|
| 281 | if ($verbosity !~ /\d+/) {
|
---|
| 282 | if (defined $collectcfg->{'verbosity'} && $collectcfg->{'verbosity'} =~ /\d+/) {
|
---|
| 283 | $verbosity = $collectcfg->{'verbosity'};
|
---|
| 284 | } else {
|
---|
| 285 | $verbosity = 2; # the default
|
---|
| 286 | }
|
---|
| 287 | }
|
---|
[22421] | 288 | $self->{'verbosity'} = $verbosity;
|
---|
| 289 |
|
---|
[22413] | 290 | if (defined $collectcfg->{'manifest'} && $self->{'manifest'} eq "") {
|
---|
| 291 | $self->{'manifest'} = $collectcfg->{'manifest'};
|
---|
| 292 | }
|
---|
| 293 |
|
---|
| 294 | if (defined $collectcfg->{'gzip'} && !$self->{'gzip'}) {
|
---|
| 295 | if ($collectcfg->{'gzip'} =~ /^true$/i) {
|
---|
| 296 | $self->{'gzip'} = 1;
|
---|
| 297 | }
|
---|
| 298 | }
|
---|
| 299 |
|
---|
| 300 | if ($self->{'maxdocs'} !~ /\-?\d+/) {
|
---|
| 301 | if (defined $collectcfg->{'maxdocs'} && $collectcfg->{'maxdocs'} =~ /\-?\d+/) {
|
---|
| 302 | $self->{'maxdocs'} = $collectcfg->{'maxdocs'};
|
---|
| 303 | } else {
|
---|
| 304 | $self->{'maxdocs'} = -1; # the default
|
---|
| 305 | }
|
---|
| 306 | }
|
---|
[22421] | 307 |
|
---|
| 308 | if ((defined $self->{'groupsize'}) && ($self->{'groupsize'} == 1)) {
|
---|
[22413] | 309 | if (defined $collectcfg->{'groupsize'} && $collectcfg->{'groupsize'} =~ /\d+/) {
|
---|
| 310 | $self->{'groupsize'} = $collectcfg->{'groupsize'};
|
---|
| 311 | }
|
---|
| 312 | }
|
---|
| 313 |
|
---|
| 314 | if (!defined $self->{'OIDtype'}
|
---|
[27352] | 315 | || ($self->{'OIDtype'} !~ /^(hash|hash_on_full_filename|incremental|assigned|filename|dirname|full_filename)$/ )) {
|
---|
[26536] | 316 | # OIDtype was either not defined on the command-line, or if it was not one of the recognized values
|
---|
[22413] | 317 | if (defined $collectcfg->{'OIDtype'}
|
---|
[27352] | 318 | && $collectcfg->{'OIDtype'} =~ /^(hash|hash_on_full_filename|incremental|assigned|filename|dirname|full_filename)$/) {
|
---|
[22413] | 319 | $self->{'OIDtype'} = $collectcfg->{'OIDtype'};
|
---|
| 320 | } else {
|
---|
| 321 | $self->{'OIDtype'} = "hash"; # the default
|
---|
| 322 | }
|
---|
| 323 | }
|
---|
| 324 |
|
---|
| 325 | if ((!defined $self->{'OIDmetadata'}) || ($self->{'OIDmetadata'} eq "")) {
|
---|
| 326 | if (defined $collectcfg->{'OIDmetadata'}) {
|
---|
| 327 | $self->{'OIDmetadata'} = $collectcfg->{'OIDmetadata'};
|
---|
| 328 | } else {
|
---|
| 329 | $self->{'OIDmetadata'} = "dc.Identifier"; # the default
|
---|
| 330 | }
|
---|
| 331 | }
|
---|
| 332 |
|
---|
| 333 | my $sortmeta = $self->{'sortmeta'};
|
---|
| 334 | if (defined $collectcfg->{'sortmeta'} && (!defined $sortmeta || $sortmeta eq "")) {
|
---|
| 335 | $sortmeta = $collectcfg->{'sortmeta'};
|
---|
| 336 | }
|
---|
| 337 | # sortmeta cannot be used with group size
|
---|
| 338 | $sortmeta = undef unless defined $sortmeta && $sortmeta =~ /\S/;
|
---|
| 339 | if (defined $sortmeta && $self->{'groupsize'} > 1) {
|
---|
| 340 | &gsprintf($out, "{import.cannot_sort}\n\n");
|
---|
| 341 | $sortmeta = undef;
|
---|
| 342 | }
|
---|
| 343 | $self->{'sortmeta'} = $sortmeta;
|
---|
| 344 |
|
---|
| 345 | if (defined $collectcfg->{'removeprefix'} && $self->{'removeprefix'} eq "") {
|
---|
| 346 | $self->{'removeprefix'} = $collectcfg->{'removeprefix'};
|
---|
| 347 | }
|
---|
| 348 |
|
---|
| 349 | if (defined $collectcfg->{'removesuffix'} && $self->{'removesuffix'} eq "") {
|
---|
| 350 | $self->{'removesuffix'} = $collectcfg->{'removesuffix'};
|
---|
| 351 | }
|
---|
| 352 | if (defined $collectcfg->{'debug'} && $collectcfg->{'debug'} =~ /^true$/i) {
|
---|
| 353 | $self->{'debug'} = 1;
|
---|
| 354 | }
|
---|
| 355 | if (defined $collectcfg->{'gli'} && $collectcfg->{'gli'} =~ /^true$/i) {
|
---|
| 356 | $self->{'gli'} = 1;
|
---|
| 357 | }
|
---|
| 358 | $self->{'gli'} = 0 unless defined $self->{'gli'};
|
---|
| 359 |
|
---|
| 360 | # check keepold and removeold
|
---|
[22421] | 361 | my $checkdir = ($inexport_mode eq "import") ? "archives" : "export";
|
---|
| 362 |
|
---|
[22413] | 363 | my ($removeold, $keepold, $incremental, $incremental_mode)
|
---|
| 364 | = &scriptutil::check_removeold_and_keepold($self->{'removeold'}, $self->{'keepold'},
|
---|
[22421] | 365 | $self->{'incremental'}, $checkdir,
|
---|
[22413] | 366 | $collectcfg);
|
---|
| 367 |
|
---|
| 368 | $self->{'removeold'} = $removeold;
|
---|
| 369 | $self->{'keepold'} = $keepold;
|
---|
| 370 | $self->{'incremental'} = $incremental;
|
---|
| 371 | $self->{'incremental_mode'} = $incremental_mode;
|
---|
[27302] | 372 |
|
---|
| 373 | # Since this wasted my morning, let's at least warn a user that manifest
|
---|
| 374 | # files now *only* work if keepold is set [jmt12]
|
---|
| 375 | if ($self->{'manifest'} && !$self->{'keepold'})
|
---|
| 376 | {
|
---|
| 377 | print STDERR "Warning: -manifest flag should not be specified without also setting -keepold or -incremental\n";
|
---|
| 378 | }
|
---|
[22413] | 379 | }
|
---|
| 380 |
|
---|
| 381 | sub process_files
|
---|
| 382 | {
|
---|
| 383 | my $self = shift @_;
|
---|
[22421] | 384 | my ($config_filename,$collectcfg) = @_;
|
---|
[22413] | 385 |
|
---|
[22421] | 386 | my $inexport_mode = $self->{'mode'};
|
---|
| 387 |
|
---|
[22413] | 388 | my $verbosity = $self->{'verbosity'};
|
---|
| 389 | my $debug = $self->{'debug'};
|
---|
| 390 |
|
---|
| 391 | my $importdir = $self->{'importdir'};
|
---|
[22460] | 392 | my $archivedir = $self->{'archivedir'} || $self->{'exportdir'};
|
---|
[22413] | 393 |
|
---|
| 394 | my $incremental = $self->{'incremental'};
|
---|
| 395 | my $incremental_mode = $self->{'incremental_mode'};
|
---|
| 396 |
|
---|
[25957] | 397 | my $gs_version = $self->{'gs_version'};
|
---|
| 398 |
|
---|
[22413] | 399 | my $removeold = $self->{'removeold'};
|
---|
| 400 | my $keepold = $self->{'keepold'};
|
---|
| 401 |
|
---|
| 402 | my $saveas = $self->{'saveas'};
|
---|
| 403 | my $OIDtype = $self->{'OIDtype'};
|
---|
| 404 | my $OIDmetadata = $self->{'OIDmetadata'};
|
---|
| 405 |
|
---|
| 406 | my $out = $self->{'out'};
|
---|
| 407 | my $faillog = $self->{'faillog'};
|
---|
| 408 |
|
---|
| 409 | my $maxdocs = $self->{'maxdocs'};
|
---|
| 410 | my $gzip = $self->{'gzip'};
|
---|
| 411 | my $groupsize = $self->{'groupsize'};
|
---|
| 412 | my $sortmeta = $self->{'sortmeta'};
|
---|
| 413 |
|
---|
| 414 | my $removeprefix = $self->{'removeprefix'};
|
---|
| 415 | my $removesuffix = $self->{'removesuffix'};
|
---|
| 416 |
|
---|
[22421] | 417 | my $gli = $self->{'gli'};
|
---|
[22413] | 418 |
|
---|
[22421] | 419 | # related to export
|
---|
| 420 | my $xsltfile = $self->{'xsltfile'};
|
---|
| 421 | my $group_marc = $self->{'group_marc'};
|
---|
| 422 | my $mapping_file = $self->{'mapping_file'};
|
---|
| 423 | my $xslt_mets = $self->{'xslt_mets'};
|
---|
| 424 | my $xslt_txt = $self->{'xslt_txt'};
|
---|
| 425 | my $fedora_namespace = $self->{'fedora_namespace'};
|
---|
[23825] | 426 | my $metadata_prefix = $self->{'metadata_prefix'};
|
---|
[22421] | 427 |
|
---|
| 428 | if ($inexport_mode eq "import") {
|
---|
| 429 | print STDERR "<Import>\n" if $gli;
|
---|
| 430 | }
|
---|
| 431 | else {
|
---|
| 432 | print STDERR "<export>\n" if $gli;
|
---|
| 433 | }
|
---|
[24344] | 434 |
|
---|
[22413] | 435 | my $manifest_lookup = new manifest($collectcfg->{'infodbtype'},$archivedir);
|
---|
[23053] | 436 | if ($self->{'manifest'} ne "") {
|
---|
[22413] | 437 | my $manifest_filename = $self->{'manifest'};
|
---|
| 438 |
|
---|
[27302] | 439 | if (!&FileUtils::isFilenameAbsolute($manifest_filename)) {
|
---|
| 440 | $manifest_filename = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, $manifest_filename);
|
---|
[22413] | 441 | }
|
---|
| 442 |
|
---|
[27482] | 443 | $self->{'manifest'} = &FileUtils::sanitizePath($self->{'manifest'});
|
---|
| 444 | #$self->{'manifest'} =~ s/[\\\/]+/\//g;
|
---|
| 445 | #$self->{'manifest'} =~ s/\/$//;
|
---|
[22413] | 446 |
|
---|
| 447 | $manifest_lookup->parse($manifest_filename);
|
---|
[27302] | 448 |
|
---|
| 449 | # manifests may now include a version number [jmt12]
|
---|
| 450 | $self->{'manifest_version'} = $manifest_lookup->get_version();
|
---|
[22413] | 451 | }
|
---|
| 452 |
|
---|
| 453 | my $manifest = $self->{'manifest'};
|
---|
| 454 |
|
---|
| 455 | # load all the plugins
|
---|
| 456 | my $plugins = [];
|
---|
| 457 | if (defined $collectcfg->{'plugin'}) {
|
---|
| 458 | $plugins = $collectcfg->{'plugin'};
|
---|
| 459 | }
|
---|
| 460 |
|
---|
[23053] | 461 | my $plugin_incr_mode = $incremental_mode;
|
---|
| 462 | if ($manifest ne "") {
|
---|
| 463 | # if we have a manifest file, then we pretend we are fully incremental for plugins
|
---|
| 464 | $plugin_incr_mode = "all";
|
---|
| 465 | }
|
---|
[22413] | 466 | #some global options for the plugins
|
---|
| 467 | my @global_opts = ();
|
---|
| 468 |
|
---|
[25957] | 469 | my $pluginfo = &plugin::load_plugins ($plugins, $verbosity, $out, $faillog, \@global_opts, $plugin_incr_mode, $gs_version);
|
---|
[22413] | 470 | if (scalar(@$pluginfo) == 0) {
|
---|
| 471 | &gsprintf($out, "{import.no_plugins_loaded}\n");
|
---|
| 472 | die "\n";
|
---|
| 473 | }
|
---|
| 474 |
|
---|
[24344] | 475 | # remove the old contents of the archives directory (and tmp
|
---|
| 476 | # directory) if needed
|
---|
| 477 |
|
---|
[22413] | 478 | if ($removeold) {
|
---|
[27302] | 479 | if (&FileUtils::directoryExists($archivedir)) {
|
---|
[22413] | 480 | &gsprintf($out, "{import.removing_archives}\n");
|
---|
[27302] | 481 | &FileUtils::removeFilesRecursive($archivedir);
|
---|
[22413] | 482 | }
|
---|
[27302] | 483 | my $tmpdir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "tmp");
|
---|
[22413] | 484 | $tmpdir =~ s/[\\\/]+/\//g;
|
---|
| 485 | $tmpdir =~ s/\/$//;
|
---|
[27302] | 486 | if (&FileUtils::directoryExists($tmpdir)) {
|
---|
[22413] | 487 | &gsprintf($out, "{import.removing_tmpdir}\n");
|
---|
[27334] | 488 | &FileUtils::removeFilesRecursive($tmpdir);
|
---|
[22413] | 489 | }
|
---|
| 490 | }
|
---|
| 491 |
|
---|
| 492 | # create the archives dir if needed
|
---|
[27302] | 493 | &FileUtils::makeAllDirectories($archivedir);
|
---|
[22413] | 494 |
|
---|
| 495 | # read the archive information file
|
---|
| 496 |
|
---|
| 497 | # BACKWARDS COMPATIBILITY: Just in case there are old .ldb/.bdb files (won't do anything for other infodbtypes)
|
---|
[27302] | 498 | &util::rename_ldb_or_bdb_file(&FileUtils::filenameConcatenate($archivedir, "archiveinf-doc"));
|
---|
| 499 | &util::rename_ldb_or_bdb_file(&FileUtils::filenameConcatenate($archivedir, "archiveinf-src"));
|
---|
[22413] | 500 |
|
---|
[27302] | 501 | # When we make these initial calls to determine the archive information doc
|
---|
| 502 | # and src databases we pass through a '1' to indicate this is the first
|
---|
| 503 | # time we are referring to these databases. When using dynamic dbutils
|
---|
| 504 | # (available in extensions) this indicates to some database types (for
|
---|
| 505 | # example, persistent servers) that this is a good time to perform any
|
---|
| 506 | # one time initialization. The argument has no effect on vanilla dbutils
|
---|
| 507 | # [jmt12]
|
---|
| 508 | my $perform_firsttime_init = 1;
|
---|
| 509 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($collectcfg->{'infodbtype'}, "archiveinf-doc", $archivedir, $perform_firsttime_init);
|
---|
| 510 | my $arcinfo_src_filename = &dbutil::get_infodb_file_path($collectcfg->{'infodbtype'}, "archiveinf-src", $archivedir, $perform_firsttime_init);
|
---|
| 511 |
|
---|
[22413] | 512 | my $archive_info = new arcinfo ($collectcfg->{'infodbtype'});
|
---|
| 513 | $archive_info->load_info ($arcinfo_doc_filename);
|
---|
| 514 |
|
---|
| 515 | if ($manifest eq "") {
|
---|
| 516 | # Load in list of files in import folder from last import (if present)
|
---|
| 517 | $archive_info->load_prev_import_filelist ($arcinfo_src_filename);
|
---|
| 518 | }
|
---|
| 519 |
|
---|
| 520 | ####Use Plugout####
|
---|
[22421] | 521 | my $plugout;
|
---|
| 522 |
|
---|
| 523 | if ($inexport_mode eq "import") {
|
---|
| 524 | if (defined $collectcfg->{'plugout'}) {
|
---|
| 525 | # If a plugout was specified in the collect.cfg file, assume it is sensible
|
---|
| 526 | # We can't check the name because it could be anything, if it is a custom plugout
|
---|
| 527 | $plugout = $collectcfg->{'plugout'};
|
---|
| 528 | }
|
---|
| 529 | else{
|
---|
| 530 | if ($saveas !~ /^(GreenstoneXML|GreenstoneMETS)$/) {
|
---|
| 531 | push @$plugout,"GreenstoneXMLPlugout";
|
---|
| 532 | }
|
---|
| 533 | else{
|
---|
| 534 | push @$plugout,$saveas."Plugout";
|
---|
| 535 | }
|
---|
| 536 | }
|
---|
[22413] | 537 | }
|
---|
[22421] | 538 | else {
|
---|
| 539 | if (defined $collectcfg->{'plugout'} && $collectcfg->{'plugout'} =~ /^(.*METS|DSpace|MARCXML)Plugout/) {
|
---|
| 540 | $plugout = $collectcfg->{'plugout'};
|
---|
[22413] | 541 | }
|
---|
| 542 | else{
|
---|
[22421] | 543 | if ($saveas !~ /^(GreenstoneMETS|FedoraMETS|DSpace|MARCXML)$/) {
|
---|
| 544 | push @$plugout,"GreenstoneMETSPlugout";
|
---|
| 545 | }
|
---|
| 546 | else{
|
---|
| 547 | push @$plugout,$saveas."Plugout";
|
---|
| 548 | }
|
---|
[22413] | 549 | }
|
---|
| 550 | }
|
---|
[22421] | 551 |
|
---|
| 552 | my $plugout_name = $plugout->[0];
|
---|
[22413] | 553 |
|
---|
[22421] | 554 | push @$plugout,("-output_info",$archive_info) if (defined $archive_info);
|
---|
| 555 | push @$plugout,("-verbosity",$verbosity) if (defined $verbosity);
|
---|
| 556 | push @$plugout,("-debug") if ($debug);
|
---|
| 557 | push @$plugout,("-group_size",$groupsize) if (defined $groupsize);
|
---|
| 558 | push @$plugout,("-gzip_output") if ($gzip);
|
---|
| 559 | push @$plugout,("-output_handle",$out) if (defined $out);
|
---|
| 560 |
|
---|
| 561 | push @$plugout,("-xslt_file",$xsltfile) if (defined $xsltfile && $xsltfile ne "");
|
---|
| 562 |
|
---|
| 563 | if ($plugout_name =~ m/^MARCXMLPlugout$/) {
|
---|
| 564 | push @$plugout,("-group") if ($group_marc);
|
---|
| 565 | push @$plugout,("-mapping_file",$mapping_file) if (defined $mapping_file && $mapping_file ne "");
|
---|
| 566 | }
|
---|
| 567 | if ($plugout_name =~ m/^.*METSPlugout$/) {
|
---|
| 568 | push @$plugout,("-xslt_mets",$xslt_mets) if (defined $xslt_mets && $xslt_mets ne "");
|
---|
| 569 | push @$plugout,("-xslt_txt",$xslt_txt) if (defined $xslt_txt && $xslt_txt ne "");
|
---|
| 570 | }
|
---|
| 571 |
|
---|
| 572 | if ($plugout_name eq "FedoraMETSPlugout") {
|
---|
| 573 | push @$plugout,("-fedora_namespace",$fedora_namespace) if (defined $fedora_namespace && $fedora_namespace ne "");
|
---|
| 574 | }
|
---|
[23825] | 575 |
|
---|
| 576 | if ($plugout_name eq "DSpacePlugout") {
|
---|
| 577 | push @$plugout,("-metadata_prefix",$metadata_prefix) if (defined $metadata_prefix && $metadata_prefix ne "");
|
---|
| 578 | }
|
---|
[22421] | 579 |
|
---|
[27302] | 580 | my $processor = &plugout::load_plugout($plugout);
|
---|
[22413] | 581 | $processor->setoutputdir ($archivedir);
|
---|
| 582 | $processor->set_sortmeta ($sortmeta, $removeprefix, $removesuffix) if defined $sortmeta;
|
---|
[26536] | 583 |
|
---|
[22413] | 584 | $processor->set_OIDtype ($OIDtype, $OIDmetadata);
|
---|
| 585 |
|
---|
| 586 | &plugin::begin($pluginfo, $importdir, $processor, $maxdocs, $gli);
|
---|
| 587 |
|
---|
| 588 | if ($removeold) {
|
---|
[24344] | 589 | # occasionally, plugins may want to do something on remove
|
---|
| 590 | # old, eg pharos image indexing
|
---|
[22413] | 591 | &plugin::remove_all($pluginfo, $importdir, $processor, $maxdocs, $gli);
|
---|
| 592 | }
|
---|
| 593 |
|
---|
[23053] | 594 | # process the import directory
|
---|
| 595 | my $block_hash = {};
|
---|
| 596 | $block_hash->{'new_files'} = {};
|
---|
| 597 | $block_hash->{'reindex_files'} = {};
|
---|
[27302] | 598 | # all of these are set somewhere else, so it's more readable to define them
|
---|
| 599 | # here [jmt12]
|
---|
| 600 | $block_hash->{'all_files'} = {};
|
---|
| 601 | $block_hash->{'deleted_files'} = {};
|
---|
| 602 | $block_hash->{'file_blocks'} = {};
|
---|
| 603 | $block_hash->{'metadata_files'} = {};
|
---|
| 604 | $block_hash->{'shared_fileroot'} = '';
|
---|
| 605 | # a new flag so we can tell we had a manifest way down in the plugins
|
---|
| 606 | # [jmt12]
|
---|
| 607 | $block_hash->{'manifest'} = 'false';
|
---|
[23053] | 608 | my $metadata = {};
|
---|
| 609 |
|
---|
[23943] | 610 | # global blocking pass may set up some metadata
|
---|
[27302] | 611 | # - when we have a newer manifest file we don't do this -unless- the
|
---|
| 612 | # collection configuration indicates this collection contains complex
|
---|
| 613 | # (inherited) metadata [jmt12]
|
---|
| 614 | if ($manifest eq '' || (defined $collectcfg->{'complexmeta'} && $collectcfg->{'complexmeta'} eq 'true'))
|
---|
| 615 | {
|
---|
| 616 | &plugin::file_block_read($pluginfo, $importdir, "", $block_hash, $metadata, $gli);
|
---|
| 617 | }
|
---|
| 618 | else
|
---|
| 619 | {
|
---|
| 620 | print STDERR "Skipping global file scan due to manifest and complexmeta configuration\n";
|
---|
| 621 | }
|
---|
| 622 |
|
---|
[23053] | 623 | if ($manifest ne "") {
|
---|
[27302] | 624 |
|
---|
| 625 | # mark that we are using a manifest - information that might be needed
|
---|
| 626 | # down in plugins (for instance DirectoryPlugin)
|
---|
| 627 | $block_hash->{'manifest'} = $self->{'manifest_version'};
|
---|
| 628 |
|
---|
[23053] | 629 | #
|
---|
| 630 | # 1. Process delete files first
|
---|
| 631 | #
|
---|
| 632 | my @deleted_files = keys %{$manifest_lookup->{'delete'}};
|
---|
| 633 | my @full_deleted_files = ();
|
---|
| 634 |
|
---|
| 635 | # ensure all filenames are absolute
|
---|
| 636 | foreach my $df (@deleted_files) {
|
---|
| 637 | my $full_df =
|
---|
[27302] | 638 | (&FileUtils::isFilenameAbsolute($df))
|
---|
[23053] | 639 | ? $df
|
---|
[27302] | 640 | : &FileUtils::filenameConcatenate($importdir,$df);
|
---|
[23053] | 641 |
|
---|
| 642 | if (-d $full_df) {
|
---|
| 643 | &add_dir_contents_to_list($full_df, \@full_deleted_files);
|
---|
| 644 | } else {
|
---|
| 645 | push(@full_deleted_files,$full_df);
|
---|
| 646 | }
|
---|
| 647 | }
|
---|
| 648 |
|
---|
| 649 | &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@full_deleted_files);
|
---|
| 650 | mark_docs_for_deletion($archive_info,{},
|
---|
| 651 | \@full_deleted_files,
|
---|
| 652 | $archivedir, $verbosity, "delete");
|
---|
| 653 |
|
---|
| 654 |
|
---|
| 655 | #
|
---|
| 656 | # 2. Now files for reindexing
|
---|
| 657 | #
|
---|
| 658 |
|
---|
| 659 | my @reindex_files = keys %{$manifest_lookup->{'reindex'}};
|
---|
| 660 | my @full_reindex_files = ();
|
---|
| 661 | # ensure all filenames are absolute
|
---|
| 662 | foreach my $rf (@reindex_files) {
|
---|
| 663 | my $full_rf =
|
---|
[27302] | 664 | (&FileUtils::isFilenameAbsolute($rf))
|
---|
[23053] | 665 | ? $rf
|
---|
[27302] | 666 | : &FileUtils::filenameConcatenate($importdir,$rf);
|
---|
[23053] | 667 |
|
---|
| 668 | if (-d $full_rf) {
|
---|
| 669 | &add_dir_contents_to_list($full_rf, \@full_reindex_files);
|
---|
| 670 | } else {
|
---|
| 671 | push(@full_reindex_files,$full_rf);
|
---|
| 672 | }
|
---|
| 673 | }
|
---|
| 674 |
|
---|
| 675 | &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@full_reindex_files);
|
---|
| 676 | mark_docs_for_deletion($archive_info,{},\@full_reindex_files, $archivedir,$verbosity, "reindex");
|
---|
| 677 |
|
---|
| 678 | # And now to ensure the new version of the file processed by
|
---|
| 679 | # appropriate plugin, we need to add it to block_hash reindex list
|
---|
| 680 | foreach my $full_rf (@full_reindex_files) {
|
---|
| 681 | $block_hash->{'reindex_files'}->{$full_rf} = 1;
|
---|
| 682 | }
|
---|
| 683 |
|
---|
| 684 |
|
---|
| 685 | #
|
---|
| 686 | # 3. Now finally any new files - add to block_hash new_files list
|
---|
| 687 | #
|
---|
| 688 |
|
---|
| 689 | my @new_files = keys %{$manifest_lookup->{'index'}};
|
---|
| 690 | my @full_new_files = ();
|
---|
| 691 |
|
---|
| 692 | foreach my $nf (@new_files) {
|
---|
| 693 | # ensure filename is absolute
|
---|
| 694 | my $full_nf =
|
---|
[27302] | 695 | (&FileUtils::isFilenameAbsolute($nf))
|
---|
[23053] | 696 | ? $nf
|
---|
[27302] | 697 | : &FileUtils::filenameConcatenate($importdir,$nf);
|
---|
[23053] | 698 |
|
---|
| 699 | if (-d $full_nf) {
|
---|
| 700 | &add_dir_contents_to_list($full_nf, \@full_new_files);
|
---|
| 701 | } else {
|
---|
| 702 | push(@full_new_files,$full_nf);
|
---|
| 703 | }
|
---|
| 704 | }
|
---|
| 705 |
|
---|
[23132] | 706 | my $arcinfo_src_filename = &dbutil::get_infodb_file_path($collectcfg->{'infodbtype'}, "archiveinf-src", $archivedir);
|
---|
[27302] | 707 | # need to check this file exists before trying to read it - in the past
|
---|
| 708 | # it wasn't possible to have a manifest unless keepold was also set so
|
---|
| 709 | # you were pretty much guarenteed arcinfo existed
|
---|
| 710 | # [jmt12]
|
---|
| 711 | # @todo &FileUtils::fileExists($arcinfo_src_filename) [jmt12]
|
---|
| 712 | if (-e $arcinfo_src_filename)
|
---|
| 713 | {
|
---|
[23132] | 714 | my $arcinfodb_map = {};
|
---|
| 715 | &dbutil::read_infodb_file($collectcfg->{'infodbtype'}, $arcinfo_src_filename, $arcinfodb_map);
|
---|
[23053] | 716 | foreach my $f (@full_new_files) {
|
---|
[23132] | 717 | # check that we haven't seen it already
|
---|
| 718 | if (defined $arcinfodb_map->{$f}) {
|
---|
| 719 | # TODO make better warning
|
---|
| 720 | print STDERR "Warning: $f already in src archive, \n";
|
---|
| 721 | } else {
|
---|
| 722 | $block_hash->{'new_files'}->{$f} = 1;
|
---|
| 723 | }
|
---|
[23053] | 724 | }
|
---|
[23132] | 725 |
|
---|
| 726 | undef $arcinfodb_map;
|
---|
[27302] | 727 | }
|
---|
| 728 | # no existing files - so we can just add all the files [jmt12]
|
---|
| 729 | else
|
---|
| 730 | {
|
---|
| 731 | foreach my $f (@full_new_files)
|
---|
| 732 | {
|
---|
| 733 | $block_hash->{'new_files'}->{$f} = 1;
|
---|
| 734 | }
|
---|
| 735 | }
|
---|
| 736 |
|
---|
| 737 | # If we are not using complex inherited metadata (and thus have skipped
|
---|
| 738 | # the global file scan) we need to at least check for a matching
|
---|
| 739 | # metadata.xml for the files being indexed/reindexed
|
---|
| 740 | # - unless we are using the newer version of Manifests, which are treated
|
---|
| 741 | # verbatim, and should have a metadata element for metadata files (so
|
---|
| 742 | # we can explicitly process metadata files other than metadata.xml)
|
---|
| 743 | # [jmt12]
|
---|
| 744 | if ($self->{'manifest_version'} < 1 && (!defined $collectcfg->{'complexmeta'} || $collectcfg->{'complexmeta'} ne 'true'))
|
---|
| 745 | {
|
---|
| 746 | my @all_files_to_import = (keys %{$block_hash->{'reindex_files'}}, keys %{$block_hash->{'new_files'}});
|
---|
| 747 | foreach my $file_to_import (@all_files_to_import)
|
---|
| 748 | {
|
---|
| 749 | my $metadata_xml_path = $file_to_import;
|
---|
| 750 | $metadata_xml_path =~ s/[^\\\/]*$/metadata.xml/;
|
---|
| 751 | if (&FileUtils::fileExists($metadata_xml_path))
|
---|
| 752 | {
|
---|
| 753 | &plugin::file_block_read($pluginfo, '', $metadata_xml_path, $block_hash, $metadata, $gli);
|
---|
| 754 | }
|
---|
| 755 | }
|
---|
| 756 | }
|
---|
| 757 |
|
---|
| 758 | # new version manifest files explicitly list metadata files to be
|
---|
| 759 | # processed (ignoring complexmeta if set)
|
---|
| 760 | # [jmt12]
|
---|
| 761 | if ($self->{'manifest_version'} > 1)
|
---|
| 762 | {
|
---|
| 763 | # Process metadata files
|
---|
| 764 | foreach my $file_to_import (keys %{$block_hash->{'reindex_files'}}, keys %{$block_hash->{'new_files'}})
|
---|
| 765 | {
|
---|
| 766 | $self->perform_process_files($manifest, $pluginfo, '', $file_to_import, $block_hash, $metadata, $processor, $maxdocs);
|
---|
| 767 | }
|
---|
| 768 | }
|
---|
[23053] | 769 | }
|
---|
| 770 | else {
|
---|
| 771 | # if incremental, we read through the import folder to see whats changed.
|
---|
| 772 |
|
---|
[22413] | 773 | if ($incremental || $incremental_mode eq "onlyadd") {
|
---|
| 774 | prime_doc_oid_count($archivedir);
|
---|
| 775 |
|
---|
| 776 | # Can now work out which files were new, already existed, and have
|
---|
| 777 | # been deleted
|
---|
| 778 |
|
---|
| 779 | new_vs_old_import_diff($archive_info,$block_hash,$importdir,
|
---|
| 780 | $archivedir,$verbosity,$incremental_mode);
|
---|
| 781 |
|
---|
| 782 | my @new_files = sort keys %{$block_hash->{'new_files'}};
|
---|
| 783 | if (scalar(@new_files>0)) {
|
---|
| 784 | print STDERR "New files and modified metadata files since last import:\n ";
|
---|
| 785 | print STDERR join("\n ",@new_files), "\n";
|
---|
| 786 | }
|
---|
| 787 |
|
---|
| 788 | if ($incremental) {
|
---|
| 789 | # only look for deletions if we are truely incremental
|
---|
| 790 | my @deleted_files = sort keys %{$block_hash->{'deleted_files'}};
|
---|
| 791 | # Filter out any in gsdl/tmp area
|
---|
| 792 | my @filtered_deleted_files = ();
|
---|
[27302] | 793 | my $gsdl_tmp_area = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "tmp");
|
---|
| 794 | my $collect_tmp_area = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "tmp");
|
---|
[22413] | 795 | $gsdl_tmp_area = &util::filename_to_regex($gsdl_tmp_area);
|
---|
| 796 | $collect_tmp_area = &util::filename_to_regex($collect_tmp_area);
|
---|
| 797 |
|
---|
| 798 | foreach my $df (@deleted_files) {
|
---|
| 799 | next if ($df =~ m/^$gsdl_tmp_area/);
|
---|
| 800 | next if ($df =~ m/^$collect_tmp_area/);
|
---|
| 801 |
|
---|
| 802 | push(@filtered_deleted_files,$df);
|
---|
| 803 | }
|
---|
| 804 |
|
---|
| 805 |
|
---|
| 806 | @deleted_files = @filtered_deleted_files;
|
---|
| 807 |
|
---|
| 808 | if (scalar(@deleted_files)>0) {
|
---|
| 809 | print STDERR "Files deleted since last import:\n ";
|
---|
| 810 | print STDERR join("\n ",@deleted_files), "\n";
|
---|
| 811 |
|
---|
| 812 |
|
---|
| 813 | &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@deleted_files);
|
---|
| 814 |
|
---|
| 815 | mark_docs_for_deletion($archive_info,$block_hash,\@deleted_files, $archivedir,$verbosity, "delete");
|
---|
| 816 | }
|
---|
| 817 |
|
---|
| 818 | my @reindex_files = sort keys %{$block_hash->{'reindex_files'}};
|
---|
| 819 |
|
---|
| 820 | if (scalar(@reindex_files)>0) {
|
---|
| 821 | print STDERR "Files to reindex since last import:\n ";
|
---|
| 822 | print STDERR join("\n ",@reindex_files), "\n";
|
---|
| 823 | &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@reindex_files);
|
---|
| 824 | mark_docs_for_deletion($archive_info,$block_hash,\@reindex_files, $archivedir,$verbosity, "reindex");
|
---|
| 825 | }
|
---|
| 826 |
|
---|
[23053] | 827 | }
|
---|
[22413] | 828 | }
|
---|
| 829 | }
|
---|
[23939] | 830 |
|
---|
| 831 | # Check for existence of the file that's to contain earliestDateStamp in archivesdir
|
---|
| 832 | # Do nothing if the file already exists (file exists on incremental build).
|
---|
| 833 | # If the file doesn't exist, as happens on full build, create it and write out the current datestamp into it
|
---|
| 834 | # In buildcol, read the file's contents and set the earliestdateStamp in GS2's build.cfg / GS3's buildconfig.xml
|
---|
| 835 | # In doc.pm have set_oaiLastModified similar to set_lastmodified, and create the doc fields
|
---|
| 836 | # oailastmodified and oailastmodifieddate
|
---|
[27302] | 837 | my $earliestDatestampFile = &FileUtils::filenameConcatenate($archivedir, "earliestDatestamp");
|
---|
[23943] | 838 | if (!-f $earliestDatestampFile && -d $archivedir) {
|
---|
[23939] | 839 | my $current_time_in_seconds = time; # in seconds
|
---|
[23946] | 840 |
|
---|
| 841 | if(open(FOUT, ">$earliestDatestampFile")) {
|
---|
| 842 | # || (&gsprintf(STDERR, "{common.cannot_open}: $!\n", $earliestDatestampFile) && die);
|
---|
| 843 | print FOUT $current_time_in_seconds;
|
---|
| 844 | close(FOUT);
|
---|
| 845 | }
|
---|
| 846 | else {
|
---|
| 847 | &gsprintf(STDERR, "{import.cannot_write_earliestdatestamp}\n", $earliestDatestampFile);
|
---|
| 848 | }
|
---|
| 849 |
|
---|
[23939] | 850 | }
|
---|
| 851 |
|
---|
[23053] | 852 |
|
---|
[27302] | 853 | $self->perform_process_files($manifest, $pluginfo, $importdir, '', $block_hash, $metadata, $processor, $maxdocs);
|
---|
| 854 |
|
---|
[22421] | 855 | if ($saveas eq "FedoraMETS") {
|
---|
| 856 | # create collection "doc obj" for Fedora that contains
|
---|
| 857 | # collection-level metadata
|
---|
| 858 |
|
---|
| 859 | my $doc_obj = new doc($config_filename,"nonindexed_doc","none");
|
---|
| 860 | $doc_obj->set_OID("collection");
|
---|
| 861 |
|
---|
| 862 | my $col_name = undef;
|
---|
| 863 | my $col_meta = $collectcfg->{'collectionmeta'};
|
---|
| 864 |
|
---|
| 865 | if (defined $col_meta) {
|
---|
| 866 | store_collectionmeta($col_meta,"collectionname",$doc_obj); # in GS3 this is a collection's name
|
---|
| 867 | store_collectionmeta($col_meta,"collectionextra",$doc_obj); # in GS3 this is a collection's description
|
---|
| 868 | }
|
---|
| 869 | $processor->process($doc_obj);
|
---|
| 870 | }
|
---|
| 871 |
|
---|
[22413] | 872 | &plugin::end($pluginfo, $processor);
|
---|
| 873 |
|
---|
| 874 | &plugin::deinit($pluginfo, $processor);
|
---|
| 875 |
|
---|
| 876 | # Store the value of OIDCount (used in doc.pm) so it can be
|
---|
| 877 | # restored correctly to this value on an incremental build
|
---|
[27302] | 878 | # - this OIDcount file should only be generated for numerical oids [jmt12]
|
---|
| 879 | if ($self->{'OIDtype'} eq 'incremental')
|
---|
| 880 | {
|
---|
| 881 | store_doc_oid_count($archivedir);
|
---|
| 882 | }
|
---|
[22413] | 883 |
|
---|
| 884 | # write out the archive information file
|
---|
[22464] | 885 | $processor->close_file_output() if (defined $groupsize) && ($groupsize > 1);
|
---|
[22413] | 886 | $processor->close_group_output() if $processor->is_group();
|
---|
| 887 |
|
---|
| 888 | # for backwards compatability with archvies.inf file
|
---|
| 889 | if ($arcinfo_doc_filename =~ m/(contents)|(\.inf)$/) {
|
---|
| 890 | $archive_info->save_info($arcinfo_doc_filename);
|
---|
| 891 | }
|
---|
| 892 | else {
|
---|
| 893 | $archive_info->save_revinfo_db($arcinfo_src_filename);
|
---|
| 894 | }
|
---|
| 895 |
|
---|
| 896 | return $pluginfo;
|
---|
| 897 | }
|
---|
| 898 |
|
---|
[27302] | 899 | # @function perform_process_files()
|
---|
| 900 | # while process_files() above prepares the system to import files this is the
|
---|
| 901 | # function that actually initiates the plugin pipeline to process the files.
|
---|
| 902 | # This function the therefore be overridden in subclasses of inexport.pm should
|
---|
| 903 | # they wish to do different or further processing
|
---|
| 904 | # @author jmt12
|
---|
| 905 | sub perform_process_files
|
---|
| 906 | {
|
---|
| 907 | my $self = shift(@_);
|
---|
| 908 | my ($manifest, $pluginfo, $importdir, $file_to_import, $block_hash, $metadata, $processor, $maxdocs) = @_;
|
---|
| 909 | my $gli = $self->{'gli'};
|
---|
| 910 | # specific file to process - via manifest version 2+
|
---|
| 911 | if ($file_to_import ne '')
|
---|
| 912 | {
|
---|
| 913 | &plugin::read ($pluginfo, '', $file_to_import, $block_hash, $metadata, $processor, $maxdocs, 0, $gli);
|
---|
| 914 | }
|
---|
| 915 | # global file scan - if we are using a new version manifest, files would have
|
---|
| 916 | # been read above. Older manifests use extra settings in the $block_hash to
|
---|
| 917 | # control what is imported, while non-manifest imports use a regular
|
---|
| 918 | # $block_hash (so obeying process_exp and block_exp) [jmt12]
|
---|
| 919 | elsif ($manifest eq '' || $self->{'manifest_version'} < 1)
|
---|
| 920 | {
|
---|
| 921 | &plugin::read ($pluginfo, $importdir, '', $block_hash, $metadata, $processor, $maxdocs, 0, $gli);
|
---|
| 922 | }
|
---|
| 923 | else
|
---|
| 924 | {
|
---|
| 925 | print STDERR "Skipping perform_process_files() due to manifest presence and version\n";
|
---|
| 926 | }
|
---|
| 927 | }
|
---|
| 928 | # perform_process_files()
|
---|
[22413] | 929 |
|
---|
[27302] | 930 | # @function generate_statistics()
|
---|
[22413] | 931 | sub generate_statistics
|
---|
| 932 | {
|
---|
[27302] | 933 | my $self = shift @_;
|
---|
| 934 | my ($pluginfo) = @_;
|
---|
[22413] | 935 |
|
---|
[27302] | 936 | my $inexport_mode = $self->{'mode'};
|
---|
| 937 | my $out = $self->{'out'};
|
---|
| 938 | my $faillogname = $self->{'faillogname'};
|
---|
| 939 | my $gli = $self->{'gli'};
|
---|
[22421] | 940 |
|
---|
[27302] | 941 | &gsprintf($out, "\n");
|
---|
| 942 | &gsprintf($out, "*********************************************\n");
|
---|
| 943 | &gsprintf($out, "{$inexport_mode.complete}\n");
|
---|
| 944 | &gsprintf($out, "*********************************************\n");
|
---|
[22413] | 945 |
|
---|
[27302] | 946 | &plugin::write_stats($pluginfo, 'STDERR', $faillogname, $gli);
|
---|
| 947 | }
|
---|
| 948 | # generate_statistics()
|
---|
[22413] | 949 |
|
---|
| 950 |
|
---|
[27302] | 951 | # @function deinit()
|
---|
| 952 | # Close down any file handles that we opened (and hence are responsible for
|
---|
| 953 | # closing
|
---|
| 954 | sub deinit
|
---|
| 955 | {
|
---|
| 956 | my $self = shift(@_);
|
---|
| 957 | close OUT if $self->{'close_out'};
|
---|
| 958 | close FAILLOG if $self->{'close_faillog'};
|
---|
[22413] | 959 | }
|
---|
[27302] | 960 | # deinit()
|
---|
[22413] | 961 |
|
---|
| 962 |
|
---|
[22464] | 963 | sub store_collectionmeta
|
---|
| 964 | {
|
---|
| 965 | my ($collectionmeta,$field,$doc_obj) = @_;
|
---|
| 966 |
|
---|
| 967 | my $section = $doc_obj->get_top_section();
|
---|
| 968 |
|
---|
| 969 | my $field_hash = $collectionmeta->{$field};
|
---|
| 970 |
|
---|
| 971 | foreach my $k (keys %$field_hash)
|
---|
| 972 | {
|
---|
| 973 | my $val = $field_hash->{$k};
|
---|
| 974 |
|
---|
| 975 | ### print STDERR "*** $k = $field_hash->{$k}\n";
|
---|
| 976 |
|
---|
| 977 | my $md_label = "ex.$field";
|
---|
| 978 |
|
---|
| 979 |
|
---|
| 980 | if ($k =~ m/^\[l=(.*?)\]$/)
|
---|
| 981 | {
|
---|
| 982 |
|
---|
| 983 | my $md_suffix = $1;
|
---|
| 984 | $md_label .= "^$md_suffix";
|
---|
| 985 | }
|
---|
| 986 |
|
---|
| 987 |
|
---|
| 988 | $doc_obj->add_utf8_metadata($section,$md_label, $val);
|
---|
| 989 |
|
---|
| 990 | # see collConfigxml.pm: GS2's "collectionextra" is called "description" in GS3,
|
---|
| 991 | # while "collectionname" in GS2 is called "name" in GS3.
|
---|
| 992 | # Variable $nameMap variable in collConfigxml.pm maps between GS2 and GS3
|
---|
| 993 | if (($md_label eq "ex.collectionname^en") || ($md_label eq "ex.collectionname"))
|
---|
| 994 | {
|
---|
| 995 | $doc_obj->add_utf8_metadata($section,"dc.Title", $val);
|
---|
| 996 | }
|
---|
| 997 |
|
---|
| 998 | }
|
---|
| 999 | }
|
---|
[22413] | 1000 |
|
---|
| 1001 |
|
---|
[21306] | 1002 | sub oid_count_file {
|
---|
| 1003 | my ($archivedir) = @_;
|
---|
[27302] | 1004 | return &FileUtils::filenameConcatenate($archivedir, "OIDcount");
|
---|
[21306] | 1005 | }
|
---|
| 1006 |
|
---|
| 1007 |
|
---|
[18528] | 1008 | sub prime_doc_oid_count
|
---|
| 1009 | {
|
---|
| 1010 | my ($archivedir) = @_;
|
---|
[21306] | 1011 | my $oid_count_filename = &oid_count_file($archivedir);
|
---|
[18528] | 1012 |
|
---|
| 1013 | if (-e $oid_count_filename) {
|
---|
| 1014 | if (open(OIDIN,"<$oid_count_filename")) {
|
---|
| 1015 | my $OIDcount = <OIDIN>;
|
---|
| 1016 | chomp $OIDcount;
|
---|
| 1017 | close(OIDIN);
|
---|
| 1018 |
|
---|
| 1019 | $doc::OIDcount = $OIDcount;
|
---|
| 1020 | }
|
---|
[23946] | 1021 | else {
|
---|
| 1022 | &gsprintf(STDERR, "{import.cannot_read_OIDcount}\n", $oid_count_filename);
|
---|
[18528] | 1023 | }
|
---|
| 1024 | }
|
---|
| 1025 |
|
---|
| 1026 | }
|
---|
| 1027 |
|
---|
| 1028 | sub store_doc_oid_count
|
---|
| 1029 | {
|
---|
| 1030 | # Use the file "OIDcount" in the archives directory to record
|
---|
| 1031 | # what value doc.pm got up to
|
---|
| 1032 |
|
---|
| 1033 | my ($archivedir) = @_;
|
---|
[21306] | 1034 | my $oid_count_filename = &oid_count_file($archivedir);
|
---|
[18528] | 1035 |
|
---|
[27302] | 1036 | # @todo $oidout = &FileUtils::openFileDescriptor($oid_count_filename, 'w') [jmt12]
|
---|
[18528] | 1037 | if (open(OIDOUT,">$oid_count_filename")) {
|
---|
| 1038 | print OIDOUT $doc::OIDcount, "\n";
|
---|
| 1039 |
|
---|
| 1040 | close(OIDOUT);
|
---|
| 1041 | }
|
---|
| 1042 | else {
|
---|
[23946] | 1043 | &gsprintf(STDERR, "{import.cannot_write_OIDcount}\n", $oid_count_filename);
|
---|
[18528] | 1044 | }
|
---|
| 1045 | }
|
---|
| 1046 |
|
---|
| 1047 |
|
---|
| 1048 |
|
---|
[18457] | 1049 | sub new_vs_old_import_diff
|
---|
| 1050 | {
|
---|
[20578] | 1051 | my ($archive_info,$block_hash,$importdir,$archivedir,$verbosity,$incremental_mode) = @_;
|
---|
[18457] | 1052 |
|
---|
[21620] | 1053 | # Get the infodbtype value for this collection from the arcinfo object
|
---|
| 1054 | my $infodbtype = $archive_info->{'infodbtype'};
|
---|
| 1055 |
|
---|
[20776] | 1056 | # in this method, we want to know if metadata files are modified or not.
|
---|
[21620] | 1057 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archivedir);
|
---|
[20776] | 1058 |
|
---|
| 1059 | my $archiveinf_timestamp = -M $arcinfo_doc_filename;
|
---|
| 1060 |
|
---|
[18457] | 1061 | # First convert all files to absolute form
|
---|
| 1062 | # This is to support the situation where the import folder is not
|
---|
| 1063 | # the default
|
---|
| 1064 |
|
---|
| 1065 | my $prev_all_files = $archive_info->{'prev_import_filelist'};
|
---|
| 1066 | my $full_prev_all_files = {};
|
---|
| 1067 |
|
---|
| 1068 | foreach my $prev_file (keys %$prev_all_files) {
|
---|
| 1069 |
|
---|
[27302] | 1070 | if (!&FileUtils::isFilenameAbsolute($prev_file)) {
|
---|
| 1071 | my $full_prev_file = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},$prev_file);
|
---|
[18457] | 1072 | $full_prev_all_files->{$full_prev_file} = $prev_file;
|
---|
| 1073 | }
|
---|
| 1074 | else {
|
---|
| 1075 | $full_prev_all_files->{$prev_file} = $prev_file;
|
---|
| 1076 | }
|
---|
| 1077 | }
|
---|
| 1078 |
|
---|
[18469] | 1079 |
|
---|
[18457] | 1080 | # Figure out which are the new files, existing files and so
|
---|
| 1081 | # by implication the files from the previous import that are not
|
---|
| 1082 | # there any more => mark them for deletion
|
---|
| 1083 | foreach my $curr_file (keys %{$block_hash->{'all_files'}}) {
|
---|
| 1084 |
|
---|
| 1085 | my $full_curr_file = $curr_file;
|
---|
| 1086 |
|
---|
| 1087 | # entry in 'all_files' is moved to either 'existing_files',
|
---|
[20776] | 1088 | # 'deleted_files', 'new_files', or 'new_or_modified_metadata_files'
|
---|
[18457] | 1089 |
|
---|
[27302] | 1090 | if (!&FileUtils::isFilenameAbsolute($curr_file)) {
|
---|
[18457] | 1091 | # add in import dir to make absolute
|
---|
[27302] | 1092 | $full_curr_file = &FileUtils::filenameConcatenate($importdir,$curr_file);
|
---|
[18457] | 1093 | }
|
---|
| 1094 |
|
---|
[19498] | 1095 | # figure out if new file or not
|
---|
[18457] | 1096 | if (defined $full_prev_all_files->{$full_curr_file}) {
|
---|
[20776] | 1097 | # delete it so that only files that need deleting are left
|
---|
| 1098 | delete $full_prev_all_files->{$full_curr_file};
|
---|
| 1099 |
|
---|
| 1100 | # had it before. is it a metadata file?
|
---|
| 1101 | if ($block_hash->{'metadata_files'}->{$full_curr_file}) {
|
---|
[20578] | 1102 |
|
---|
[20776] | 1103 | # is it modified??
|
---|
| 1104 | if (-M $full_curr_file < $archiveinf_timestamp) {
|
---|
[23485] | 1105 | print STDERR "*** Detected a *modified metadata* file: $full_curr_file\n" if $verbosity >= 2;
|
---|
[20776] | 1106 | # its newer than last build
|
---|
| 1107 | $block_hash->{'new_or_modified_metadata_files'}->{$full_curr_file} = 1;
|
---|
| 1108 | }
|
---|
[20578] | 1109 | }
|
---|
| 1110 | else {
|
---|
[20776] | 1111 | if ($incremental_mode eq "all") {
|
---|
| 1112 |
|
---|
| 1113 | # had it before
|
---|
| 1114 | $block_hash->{'existing_files'}->{$full_curr_file} = 1;
|
---|
| 1115 |
|
---|
| 1116 | }
|
---|
| 1117 | else {
|
---|
| 1118 | # Warning in "onlyadd" mode, but had it before!
|
---|
| 1119 | print STDERR "Warning: File $full_curr_file previously imported.\n";
|
---|
| 1120 | print STDERR " Treating as new file\n";
|
---|
| 1121 |
|
---|
| 1122 | $block_hash->{'new_files'}->{$full_curr_file} = 1;
|
---|
| 1123 |
|
---|
| 1124 | }
|
---|
[20578] | 1125 | }
|
---|
| 1126 | }
|
---|
| 1127 | else {
|
---|
| 1128 | if ($block_hash->{'metadata_files'}->{$full_curr_file}) {
|
---|
| 1129 | # the new file is the special sort of file greenstone uses
|
---|
| 1130 | # to attach metadata to src documents
|
---|
| 1131 | # i.e metadata.xml
|
---|
| 1132 | # (but note, the filename used is not constrained in
|
---|
| 1133 | # Greenstone to always be this)
|
---|
[18457] | 1134 |
|
---|
[23485] | 1135 | print STDERR "*** Detected *new* metadata file: $full_curr_file\n" if $verbosity >= 2;
|
---|
[20776] | 1136 | $block_hash->{'new_or_modified_metadata_files'}->{$full_curr_file} = 1;
|
---|
[20578] | 1137 | }
|
---|
| 1138 | else {
|
---|
| 1139 | $block_hash->{'new_files'}->{$full_curr_file} = 1;
|
---|
| 1140 | }
|
---|
[18457] | 1141 | }
|
---|
[20578] | 1142 |
|
---|
[18457] | 1143 |
|
---|
| 1144 | delete $block_hash->{'all_files'}->{$curr_file};
|
---|
| 1145 | }
|
---|
| 1146 |
|
---|
[20578] | 1147 |
|
---|
[21306] | 1148 |
|
---|
| 1149 |
|
---|
[20776] | 1150 | # Deal with complication of new or modified metadata files by forcing
|
---|
[20578] | 1151 | # everything from this point down in the file hierarchy to
|
---|
| 1152 | # be freshly imported.
|
---|
| 1153 | #
|
---|
| 1154 | # This may mean files that have not changed are reindexed, but does
|
---|
| 1155 | # guarantee by the end of processing all new metadata is correctly
|
---|
| 1156 | # associated with the relevant document(s).
|
---|
| 1157 |
|
---|
[20776] | 1158 | foreach my $new_mdf (keys %{$block_hash->{'new_or_modified_metadata_files'}}) {
|
---|
[20578] | 1159 | my ($fileroot,$situated_dir,$ext) = fileparse($new_mdf, "\\.[^\\.]+\$");
|
---|
| 1160 |
|
---|
| 1161 | $situated_dir =~ s/[\\\/]+$//; # remove tailing slashes
|
---|
[24829] | 1162 | $situated_dir = &util::filename_to_regex($situated_dir); # need to escape windows slash \ and brackets in regular expression
|
---|
[20769] | 1163 |
|
---|
[20578] | 1164 | # Go through existing_files, and mark anything that is contained
|
---|
| 1165 | # within 'situated_dir' to be reindexed (in case some of the metadata
|
---|
| 1166 | # attaches to one of these files)
|
---|
| 1167 |
|
---|
| 1168 | my $reindex_files = [];
|
---|
| 1169 |
|
---|
| 1170 | foreach my $existing_f (keys %{$block_hash->{'existing_files'}}) {
|
---|
[20769] | 1171 |
|
---|
[20578] | 1172 | if ($existing_f =~ m/^$situated_dir/) {
|
---|
[23485] | 1173 |
|
---|
| 1174 | print STDERR "**** Existing file $existing_f\nis located within\n$situated_dir\n";
|
---|
| 1175 |
|
---|
[20578] | 1176 | push(@$reindex_files,$existing_f);
|
---|
| 1177 | $block_hash->{'reindex_files'}->{$existing_f} = 1;
|
---|
[21306] | 1178 | delete $block_hash->{'existing_files'}->{$existing_f};
|
---|
[20578] | 1179 |
|
---|
| 1180 | }
|
---|
| 1181 | }
|
---|
| 1182 |
|
---|
| 1183 | # metadata file needs to be in new_files list so parsed by MetadataXMLPlug
|
---|
| 1184 | # (or equivalent)
|
---|
| 1185 | $block_hash->{'new_files'}->{$new_mdf} = 1;
|
---|
| 1186 |
|
---|
| 1187 | }
|
---|
| 1188 |
|
---|
[21306] | 1189 | # go through remaining existing files and work out what has changed and needs to be reindexed.
|
---|
| 1190 | my @existing_files = sort keys %{$block_hash->{'existing_files'}};
|
---|
| 1191 |
|
---|
| 1192 | my $reindex_files = [];
|
---|
| 1193 |
|
---|
| 1194 | foreach my $existing_filename (@existing_files) {
|
---|
| 1195 | if (-M $existing_filename < $archiveinf_timestamp) {
|
---|
| 1196 | # file is newer than last build
|
---|
| 1197 |
|
---|
| 1198 | my $existing_file = $existing_filename;
|
---|
[27302] | 1199 | #my $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'});
|
---|
[21306] | 1200 |
|
---|
| 1201 | #my $collectdir_resafe = &util::filename_to_regex($collectdir);
|
---|
| 1202 | #$existing_file =~ s/^$collectdir_resafe(\\|\/)?//;
|
---|
| 1203 |
|
---|
| 1204 | print STDERR "**** Reindexing existing file: $existing_file\n";
|
---|
| 1205 |
|
---|
| 1206 | push(@$reindex_files,$existing_file);
|
---|
| 1207 | $block_hash->{'reindex_files'}->{$existing_filename} = 1;
|
---|
| 1208 | }
|
---|
| 1209 |
|
---|
| 1210 | }
|
---|
| 1211 |
|
---|
[20578] | 1212 |
|
---|
[18469] | 1213 | # By this point full_prev_all_files contains the files
|
---|
| 1214 | # mentioned in archiveinf-src.db but are not in the 'import'
|
---|
| 1215 | # folder (or whatever was specified through -importdir ...)
|
---|
| 1216 |
|
---|
| 1217 | # This list can contain files that were created in the 'tmp' or
|
---|
| 1218 | # 'cache' areas (such as screen-size and thumbnail images).
|
---|
[18457] | 1219 | #
|
---|
[18469] | 1220 | # In building the final list of files to delete, we test to see if
|
---|
[20578] | 1221 | # it exists on the filesystem and if it does (unusual for a "normal"
|
---|
| 1222 | # file in import, but possible in the case of 'tmp' files),
|
---|
| 1223 | # supress it from going into the final list
|
---|
[18469] | 1224 |
|
---|
| 1225 | my $collectdir = $ENV{'GSDLCOLLECTDIR'};
|
---|
| 1226 |
|
---|
[18457] | 1227 | my @deleted_files = values %$full_prev_all_files;
|
---|
[18469] | 1228 | map { my $curr_file = $_;
|
---|
| 1229 | my $full_curr_file = $curr_file;
|
---|
| 1230 |
|
---|
[27302] | 1231 | if (!&FileUtils::isFilenameAbsolute($curr_file)) {
|
---|
[18469] | 1232 | # add in import dir to make absolute
|
---|
| 1233 |
|
---|
[27302] | 1234 | $full_curr_file = &FileUtils::filenameConcatenate($collectdir,$curr_file);
|
---|
[18469] | 1235 | }
|
---|
| 1236 |
|
---|
| 1237 |
|
---|
| 1238 | if (!-e $full_curr_file) {
|
---|
| 1239 | $block_hash->{'deleted_files'}->{$curr_file} = 1;
|
---|
| 1240 | }
|
---|
| 1241 | } @deleted_files;
|
---|
[20578] | 1242 |
|
---|
| 1243 |
|
---|
| 1244 |
|
---|
[18457] | 1245 | }
|
---|
| 1246 |
|
---|
[19498] | 1247 |
|
---|
[20788] | 1248 | # this is used to delete "deleted" docs, and to remove old versions of "changed" docs
|
---|
[21306] | 1249 | # $mode is 'delete' or 'reindex'
|
---|
| 1250 | sub mark_docs_for_deletion
|
---|
[18457] | 1251 | {
|
---|
[21306] | 1252 | my ($archive_info,$block_hash,$deleted_files,$archivedir,$verbosity,$mode) = @_;
|
---|
[18457] | 1253 |
|
---|
[21306] | 1254 | my $mode_text = "deleted from index";
|
---|
| 1255 | if ($mode eq "reindex") {
|
---|
| 1256 | $mode_text = "reindexed";
|
---|
| 1257 | }
|
---|
[18457] | 1258 |
|
---|
[21620] | 1259 | # Get the infodbtype value for this collection from the arcinfo object
|
---|
| 1260 | my $infodbtype = $archive_info->{'infodbtype'};
|
---|
[18457] | 1261 |
|
---|
[21620] | 1262 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archivedir);
|
---|
| 1263 | my $arcinfo_src_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-src", $archivedir);
|
---|
| 1264 |
|
---|
[22010] | 1265 |
|
---|
[18457] | 1266 | # record files marked for deletion in arcinfo
|
---|
[19498] | 1267 | foreach my $file (@$deleted_files) {
|
---|
[21564] | 1268 | # use 'archiveinf-src' info database file to look up all the OIDs
|
---|
[19789] | 1269 | # that this file is used in (note in most cases, it's just one OID)
|
---|
[18457] | 1270 |
|
---|
[23485] | 1271 | my $src_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_src_filename, $file);
|
---|
[18457] | 1272 | my $oids = $src_rec->{'oid'};
|
---|
[20776] | 1273 | my $file_record_deleted = 0;
|
---|
[20788] | 1274 |
|
---|
| 1275 | # delete the src record
|
---|
[22010] | 1276 | my $src_infodb_file_handle = &dbutil::open_infodb_write_handle($infodbtype, $arcinfo_src_filename, "append");
|
---|
[21620] | 1277 | &dbutil::delete_infodb_entry($infodbtype, $src_infodb_file_handle, $file);
|
---|
[22010] | 1278 | &dbutil::close_infodb_write_handle($infodbtype, $src_infodb_file_handle);
|
---|
| 1279 |
|
---|
| 1280 |
|
---|
[18457] | 1281 | foreach my $oid (@$oids) {
|
---|
| 1282 |
|
---|
[20788] | 1283 | # find the source doc (the primary file that becomes this oid)
|
---|
[23485] | 1284 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $oid);
|
---|
[20776] | 1285 | my $doc_source_file = $doc_rec->{'src-file'}->[0];
|
---|
| 1286 | if (!&util::filename_is_absolute($doc_source_file)) {
|
---|
[27302] | 1287 | $doc_source_file = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},$doc_source_file);
|
---|
[20776] | 1288 | }
|
---|
[18457] | 1289 |
|
---|
[20788] | 1290 | if ($doc_source_file ne $file) {
|
---|
| 1291 | # its an associated or metadata file
|
---|
| 1292 |
|
---|
[20776] | 1293 | # mark source doc for reimport as one of its assoc files has changed or deleted
|
---|
| 1294 | $block_hash->{'reindex_files'}->{$doc_source_file} = 1;
|
---|
[20788] | 1295 |
|
---|
[18457] | 1296 | }
|
---|
[20788] | 1297 | my $curr_status = $archive_info->get_status_info($oid);
|
---|
| 1298 | if (defined($curr_status) && (($curr_status ne "D"))) {
|
---|
| 1299 | if ($verbosity>1) {
|
---|
| 1300 | print STDERR "$oid ($doc_source_file) marked to be $mode_text on next buildcol.pl\n";
|
---|
[19498] | 1301 | }
|
---|
[20788] | 1302 | # mark oid for deletion (it will be deleted or reimported)
|
---|
| 1303 | $archive_info->set_status_info($oid,"D");
|
---|
[23485] | 1304 | my $val = &dbutil::read_infodb_rawentry($infodbtype, $arcinfo_doc_filename, $oid);
|
---|
[20788] | 1305 | $val =~ s/^<index-status>(.*)$/<index-status>D/m;
|
---|
[21557] | 1306 |
|
---|
| 1307 | my $val_rec = &dbutil::convert_infodb_string_to_hash($val);
|
---|
[22010] | 1308 | my $doc_infodb_file_handle = &dbutil::open_infodb_write_handle($infodbtype, $arcinfo_doc_filename, "append");
|
---|
| 1309 |
|
---|
[21620] | 1310 | &dbutil::write_infodb_entry($infodbtype, $doc_infodb_file_handle, $oid, $val_rec);
|
---|
[22010] | 1311 | &dbutil::close_infodb_write_handle($infodbtype, $doc_infodb_file_handle);
|
---|
[19498] | 1312 | }
|
---|
[18457] | 1313 | }
|
---|
[22327] | 1314 |
|
---|
[18457] | 1315 | }
|
---|
[24344] | 1316 |
|
---|
| 1317 | # now go through and check that we haven't marked any primary
|
---|
| 1318 | # files for reindex (because their associated files have
|
---|
| 1319 | # changed/deleted) when they have been deleted themselves. only in
|
---|
| 1320 | # delete mode.
|
---|
| 1321 |
|
---|
[22567] | 1322 | if ($mode eq "delete") {
|
---|
| 1323 | foreach my $file (@$deleted_files) {
|
---|
| 1324 | if (defined $block_hash->{'reindex_files'}->{$file}) {
|
---|
| 1325 | delete $block_hash->{'reindex_files'}->{$file};
|
---|
| 1326 | }
|
---|
[22327] | 1327 | }
|
---|
| 1328 | }
|
---|
[21560] | 1329 |
|
---|
[22010] | 1330 |
|
---|
[18457] | 1331 | }
|
---|
| 1332 |
|
---|
[23053] | 1333 | sub add_dir_contents_to_list {
|
---|
[18457] | 1334 |
|
---|
[23053] | 1335 | my ($dirname, $list) = @_;
|
---|
| 1336 |
|
---|
| 1337 | # Recur over directory contents.
|
---|
| 1338 | my (@dir, $subfile);
|
---|
| 1339 |
|
---|
| 1340 | # find all the files in the directory
|
---|
| 1341 | if (!opendir (DIR, $dirname)) {
|
---|
| 1342 | print STDERR "inexport: WARNING - couldn't read directory $dirname\n";
|
---|
| 1343 | return -1; # error in processing
|
---|
| 1344 | }
|
---|
| 1345 | @dir = readdir (DIR);
|
---|
| 1346 | closedir (DIR);
|
---|
| 1347 |
|
---|
| 1348 | for (my $i = 0; $i < scalar(@dir); $i++) {
|
---|
| 1349 | my $subfile = $dir[$i];
|
---|
| 1350 | next if ($subfile =~ m/^\.\.?$/);
|
---|
[23119] | 1351 | next if ($subfile =~ /^\.svn$/);
|
---|
[27302] | 1352 | my $full_file = &FileUtils::filenameConcatenate($dirname, $subfile);
|
---|
[23053] | 1353 | if (-d $full_file) {
|
---|
| 1354 | &add_dir_contents_to_list($full_file, $list);
|
---|
| 1355 | } else {
|
---|
| 1356 | push (@$list, $full_file);
|
---|
| 1357 | }
|
---|
| 1358 | }
|
---|
| 1359 |
|
---|
| 1360 | }
|
---|
[18554] | 1361 |
|
---|
[23053] | 1362 |
|
---|
[18457] | 1363 | 1;
|
---|