source: main/trunk/greenstone2/perllib/inexport.pm@ 27698

Last change on this file since 27698 was 27698, checked in by ak19, 11 years ago

import.pl/export.pl now issues a reminder that sortmeta needs to be paired with -sort or -reversesort on ArchivesInfPlugin.
  • Property svn:executable set to *
File size: 44.6 KB
Line 
1###########################################################################
2#
3# inexport.pm -- useful class to support import.pl and export.pl
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
26package inexport;
27
28use strict;
29
30no strict 'refs'; # allow filehandles to be variables and vice versa
31no strict 'subs'; # allow barewords (eg STDERR) as function arguments
32
33use arcinfo;
34use colcfg;
35use dbutil;
36use doc;
37use plugin;
38use plugout;
39use manifest;
40use inexport;
41use util;
42use scriptutil;
43use FileHandle;
44use gsprintf 'gsprintf';
45use printusage;
46use parse2;
47
48use File::Basename;
49
50sub new
51{
52 my $class = shift (@_);
53 my ($mode,$argv,$options,$opt_listall_options) = @_;
54
55 my $self = { 'xml' => 0, 'mode' => $mode };
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
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
86 if ($self->{'xml'}) {
87 &PrintUsage::print_xml_usage($options);
88 print "\n";
89 return bless $self, $class;
90 }
91
92 if ($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
100
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);
112 $out = 'inexport::OUT';
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
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;
126
127 return bless $self, $class;
128}
129
130# Simplified version of the contstructor for use with CGI scripts
131sub newCGI
132{
133 my $class = shift (@_);
134 my ($mode,$collect,$gsdl_cgi,$opt_site) = @_;
135
136 my $self = { 'xml' => 0, 'mode' => $mode };
137
138 $self->{'out'} = STDERR;
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'} = "";
147 $self->{'collectdir'} = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"collect");
148 }
149 $self->{'faillog'} = "";
150
151 $self->{'collection'} = $collect;
152
153 return bless $self, $class;
154}
155sub get_collection
156{
157 my $self = shift @_;
158
159 return $self->{'collection'};
160}
161
162
163sub 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 = $self->{'out'};
171
172 if (($collection = &colcfg::use_collection($site, $collection, $collectdir)) eq "") {
173 &PrintUsage::print_txt_usage($options, "{import.params}");
174 die "\n";
175 }
176
177 # set gs_version 2/3
178 $self->{'gs_version'} = "2";
179 if ((defined $site) && ($site ne "")) {
180 # gs3
181 $self->{'gs_version'} = "3";
182 }
183
184 # add collection's perllib dir into include path in
185 # case we have collection specific modules
186 &util::augmentINC(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, 'perllib'));
187
188 # check that we can open the faillog
189 my $faillog = $self->{'faillog'};
190 if ($faillog eq "") {
191 $faillog = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "etc", "fail.log");
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;
202 $self->{'close_faillog'} = 1;
203
204 # Read in the collection configuration file.
205 my $gs_mode = "gs".$self->{'gs_version'}; #gs2 or gs3
206 my $config_filename = &colcfg::get_collect_cfg_name($out, $gs_mode);
207 my $collectcfg = &colcfg::read_collection_cfg ($config_filename, $gs_mode);
208
209 return ($config_filename,$collectcfg);
210}
211
212sub set_collection_options
213{
214 my $self = shift @_;
215 my ($collectcfg) = @_;
216
217 my $inexport_mode = $self->{'mode'};
218
219 my $verbosity = $self->{'verbosity'};
220 my $debug = $self->{'debug'};
221 my $importdir = $self->{'importdir'};
222 my $archivedir = $self->{'archivedir'} || $self->{'exportdir'} || "";
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 $collectcfg->{'infodbtype'} = &dbutil::get_default_infodb_type();
229 }
230 if ($collectcfg->{'infodbtype'} eq "gdbm-txtgz") {
231 # we can't use the text version for archives dbs.
232 $collectcfg->{'infodbtype'} = "gdbm";
233 }
234
235 if (defined $collectcfg->{'importdir'} && $importdir eq "") {
236 $importdir = $collectcfg->{'importdir'};
237 }
238 if (defined $collectcfg->{'archivedir'} && $archivedir eq "") {
239 $archivedir = $collectcfg->{'archivedir'};
240 }
241 # fill in the default import and archives directories if none
242 # were supplied, turn all \ into / and remove trailing /
243 if ($importdir eq "")
244 {
245 $importdir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "import");
246 }
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 }
260 $self->{'importdir'} = $importdir;
261
262 if ($archivedir eq "") {
263 if ($inexport_mode eq "import") {
264 $archivedir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "archives");
265 }
266 elsif ($inexport_mode eq "export") {
267 $archivedir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "export");
268 }
269 else {
270 print STDERR "Warning: Unrecognized import/export mode '$inexport_mode'\n";
271 print STDERR " Defaulting to 'archives' for file output\n";
272 $archivedir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "archives");
273 }
274 }
275
276 $archivedir = &FileUtils::sanitizePath($archivedir);
277 #$archivedir =~ s/[\\\/]+/\//g;
278 #$archivedir =~ s/\/$//;
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 }
288 $self->{'verbosity'} = $verbosity;
289
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 }
307
308 if ((defined $self->{'groupsize'}) && ($self->{'groupsize'} == 1)) {
309 if (defined $collectcfg->{'groupsize'} && $collectcfg->{'groupsize'} =~ /\d+/) {
310 $self->{'groupsize'} = $collectcfg->{'groupsize'};
311 }
312 }
313
314 if (!defined $self->{'OIDtype'}
315 || ($self->{'OIDtype'} !~ /^(hash|hash_on_full_filename|incremental|assigned|filename|dirname|full_filename)$/ )) {
316 # OIDtype was either not defined on the command-line, or if it was not one of the recognized values
317 if (defined $collectcfg->{'OIDtype'}
318 && $collectcfg->{'OIDtype'} =~ /^(hash|hash_on_full_filename|incremental|assigned|filename|dirname|full_filename)$/) {
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 if (defined $sortmeta) {
344 &gsprintf($out, "{import.sortmeta_paired_with_ArchivesInfPlugin}\n\n");
345 }
346 $self->{'sortmeta'} = $sortmeta;
347
348 if (defined $collectcfg->{'removeprefix'} && $self->{'removeprefix'} eq "") {
349 $self->{'removeprefix'} = $collectcfg->{'removeprefix'};
350 }
351
352 if (defined $collectcfg->{'removesuffix'} && $self->{'removesuffix'} eq "") {
353 $self->{'removesuffix'} = $collectcfg->{'removesuffix'};
354 }
355 if (defined $collectcfg->{'debug'} && $collectcfg->{'debug'} =~ /^true$/i) {
356 $self->{'debug'} = 1;
357 }
358 if (defined $collectcfg->{'gli'} && $collectcfg->{'gli'} =~ /^true$/i) {
359 $self->{'gli'} = 1;
360 }
361 $self->{'gli'} = 0 unless defined $self->{'gli'};
362
363 # check keepold and removeold
364 my $checkdir = ($inexport_mode eq "import") ? "archives" : "export";
365
366 my ($removeold, $keepold, $incremental, $incremental_mode)
367 = &scriptutil::check_removeold_and_keepold($self->{'removeold'}, $self->{'keepold'},
368 $self->{'incremental'}, $checkdir,
369 $collectcfg);
370
371 $self->{'removeold'} = $removeold;
372 $self->{'keepold'} = $keepold;
373 $self->{'incremental'} = $incremental;
374 $self->{'incremental_mode'} = $incremental_mode;
375
376 # Since this wasted my morning, let's at least warn a user that manifest
377 # files now *only* work if keepold is set [jmt12]
378 if ($self->{'manifest'} && !$self->{'keepold'})
379 {
380 print STDERR "Warning: -manifest flag should not be specified without also setting -keepold or -incremental\n";
381 }
382}
383
384sub process_files
385{
386 my $self = shift @_;
387 my ($config_filename,$collectcfg) = @_;
388
389 my $inexport_mode = $self->{'mode'};
390
391 my $verbosity = $self->{'verbosity'};
392 my $debug = $self->{'debug'};
393
394 my $importdir = $self->{'importdir'};
395 my $archivedir = $self->{'archivedir'} || $self->{'exportdir'};
396
397 my $incremental = $self->{'incremental'};
398 my $incremental_mode = $self->{'incremental_mode'};
399
400 my $gs_version = $self->{'gs_version'};
401
402 my $removeold = $self->{'removeold'};
403 my $keepold = $self->{'keepold'};
404
405 my $saveas = $self->{'saveas'};
406 my $OIDtype = $self->{'OIDtype'};
407 my $OIDmetadata = $self->{'OIDmetadata'};
408
409 my $out = $self->{'out'};
410 my $faillog = $self->{'faillog'};
411
412 my $maxdocs = $self->{'maxdocs'};
413 my $gzip = $self->{'gzip'};
414 my $groupsize = $self->{'groupsize'};
415 my $sortmeta = $self->{'sortmeta'};
416
417 my $removeprefix = $self->{'removeprefix'};
418 my $removesuffix = $self->{'removesuffix'};
419
420 my $gli = $self->{'gli'};
421
422 # related to export
423 my $xsltfile = $self->{'xsltfile'};
424 my $group_marc = $self->{'group_marc'};
425 my $mapping_file = $self->{'mapping_file'};
426 my $xslt_mets = $self->{'xslt_mets'};
427 my $xslt_txt = $self->{'xslt_txt'};
428 my $fedora_namespace = $self->{'fedora_namespace'};
429 my $metadata_prefix = $self->{'metadata_prefix'};
430
431 if ($inexport_mode eq "import") {
432 print STDERR "<Import>\n" if $gli;
433 }
434 else {
435 print STDERR "<export>\n" if $gli;
436 }
437
438 my $manifest_lookup = new manifest($collectcfg->{'infodbtype'},$archivedir);
439 if ($self->{'manifest'} ne "") {
440 my $manifest_filename = $self->{'manifest'};
441
442 if (!&FileUtils::isFilenameAbsolute($manifest_filename)) {
443 $manifest_filename = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, $manifest_filename);
444 }
445
446 $self->{'manifest'} = &FileUtils::sanitizePath($self->{'manifest'});
447 #$self->{'manifest'} =~ s/[\\\/]+/\//g;
448 #$self->{'manifest'} =~ s/\/$//;
449
450 $manifest_lookup->parse($manifest_filename);
451
452 # manifests may now include a version number [jmt12]
453 $self->{'manifest_version'} = $manifest_lookup->get_version();
454 }
455
456 my $manifest = $self->{'manifest'};
457
458 # load all the plugins
459 my $plugins = [];
460 if (defined $collectcfg->{'plugin'}) {
461 $plugins = $collectcfg->{'plugin'};
462 }
463
464 my $plugin_incr_mode = $incremental_mode;
465 if ($manifest ne "") {
466 # if we have a manifest file, then we pretend we are fully incremental for plugins
467 $plugin_incr_mode = "all";
468 }
469 #some global options for the plugins
470 my @global_opts = ();
471
472 my $pluginfo = &plugin::load_plugins ($plugins, $verbosity, $out, $faillog, \@global_opts, $plugin_incr_mode, $gs_version);
473 if (scalar(@$pluginfo) == 0) {
474 &gsprintf($out, "{import.no_plugins_loaded}\n");
475 die "\n";
476 }
477
478 # remove the old contents of the archives directory (and tmp
479 # directory) if needed
480
481 if ($removeold) {
482 if (&FileUtils::directoryExists($archivedir)) {
483 &gsprintf($out, "{import.removing_archives}\n");
484 &FileUtils::removeFilesRecursive($archivedir);
485 }
486 my $tmpdir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "tmp");
487 $tmpdir =~ s/[\\\/]+/\//g;
488 $tmpdir =~ s/\/$//;
489 if (&FileUtils::directoryExists($tmpdir)) {
490 &gsprintf($out, "{import.removing_tmpdir}\n");
491 &FileUtils::removeFilesRecursive($tmpdir);
492 }
493 }
494
495 # create the archives dir if needed
496 &FileUtils::makeAllDirectories($archivedir);
497
498 # read the archive information file
499
500 # BACKWARDS COMPATIBILITY: Just in case there are old .ldb/.bdb files (won't do anything for other infodbtypes)
501 &util::rename_ldb_or_bdb_file(&FileUtils::filenameConcatenate($archivedir, "archiveinf-doc"));
502 &util::rename_ldb_or_bdb_file(&FileUtils::filenameConcatenate($archivedir, "archiveinf-src"));
503
504 # When we make these initial calls to determine the archive information doc
505 # and src databases we pass through a '1' to indicate this is the first
506 # time we are referring to these databases. When using dynamic dbutils
507 # (available in extensions) this indicates to some database types (for
508 # example, persistent servers) that this is a good time to perform any
509 # one time initialization. The argument has no effect on vanilla dbutils
510 # [jmt12]
511 my $perform_firsttime_init = 1;
512 my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($collectcfg->{'infodbtype'}, "archiveinf-doc", $archivedir, $perform_firsttime_init);
513 my $arcinfo_src_filename = &dbutil::get_infodb_file_path($collectcfg->{'infodbtype'}, "archiveinf-src", $archivedir, $perform_firsttime_init);
514
515 my $archive_info = new arcinfo ($collectcfg->{'infodbtype'});
516 $archive_info->load_info ($arcinfo_doc_filename);
517
518 if ($manifest eq "") {
519 # Load in list of files in import folder from last import (if present)
520 $archive_info->load_prev_import_filelist ($arcinfo_src_filename);
521 }
522
523 ####Use Plugout####
524 my $plugout;
525
526 if ($inexport_mode eq "import") {
527 if (defined $collectcfg->{'plugout'}) {
528 # If a plugout was specified in the collect.cfg file, assume it is sensible
529 # We can't check the name because it could be anything, if it is a custom plugout
530 $plugout = $collectcfg->{'plugout'};
531 }
532 else{
533 if ($saveas !~ /^(GreenstoneXML|GreenstoneMETS)$/) {
534 push @$plugout,"GreenstoneXMLPlugout";
535 }
536 else{
537 push @$plugout,$saveas."Plugout";
538 }
539 }
540 }
541 else {
542 if (defined $collectcfg->{'plugout'} && $collectcfg->{'plugout'} =~ /^(.*METS|DSpace|MARCXML)Plugout/) {
543 $plugout = $collectcfg->{'plugout'};
544 }
545 else{
546 if ($saveas !~ /^(GreenstoneMETS|FedoraMETS|DSpace|MARCXML)$/) {
547 push @$plugout,"GreenstoneMETSPlugout";
548 }
549 else{
550 push @$plugout,$saveas."Plugout";
551 }
552 }
553 }
554
555 my $plugout_name = $plugout->[0];
556
557 push @$plugout,("-output_info",$archive_info) if (defined $archive_info);
558 push @$plugout,("-verbosity",$verbosity) if (defined $verbosity);
559 push @$plugout,("-debug") if ($debug);
560 push @$plugout,("-group_size",$groupsize) if (defined $groupsize);
561 push @$plugout,("-gzip_output") if ($gzip);
562 push @$plugout,("-output_handle",$out) if (defined $out);
563
564 push @$plugout,("-xslt_file",$xsltfile) if (defined $xsltfile && $xsltfile ne "");
565
566 if ($plugout_name =~ m/^MARCXMLPlugout$/) {
567 push @$plugout,("-group") if ($group_marc);
568 push @$plugout,("-mapping_file",$mapping_file) if (defined $mapping_file && $mapping_file ne "");
569 }
570 if ($plugout_name =~ m/^.*METSPlugout$/) {
571 push @$plugout,("-xslt_mets",$xslt_mets) if (defined $xslt_mets && $xslt_mets ne "");
572 push @$plugout,("-xslt_txt",$xslt_txt) if (defined $xslt_txt && $xslt_txt ne "");
573 }
574
575 if ($plugout_name eq "FedoraMETSPlugout") {
576 push @$plugout,("-fedora_namespace",$fedora_namespace) if (defined $fedora_namespace && $fedora_namespace ne "");
577 }
578
579 if ($plugout_name eq "DSpacePlugout") {
580 push @$plugout,("-metadata_prefix",$metadata_prefix) if (defined $metadata_prefix && $metadata_prefix ne "");
581 }
582
583 my $processor = &plugout::load_plugout($plugout);
584 $processor->setoutputdir ($archivedir);
585 $processor->set_sortmeta ($sortmeta, $removeprefix, $removesuffix) if defined $sortmeta;
586
587 $processor->set_OIDtype ($OIDtype, $OIDmetadata);
588
589 &plugin::begin($pluginfo, $importdir, $processor, $maxdocs, $gli);
590
591 if ($removeold) {
592 # occasionally, plugins may want to do something on remove
593 # old, eg pharos image indexing
594 &plugin::remove_all($pluginfo, $importdir, $processor, $maxdocs, $gli);
595 }
596
597 # process the import directory
598 my $block_hash = {};
599 $block_hash->{'new_files'} = {};
600 $block_hash->{'reindex_files'} = {};
601 # all of these are set somewhere else, so it's more readable to define them
602 # here [jmt12]
603 $block_hash->{'all_files'} = {};
604 $block_hash->{'deleted_files'} = {};
605 $block_hash->{'file_blocks'} = {};
606 $block_hash->{'metadata_files'} = {};
607 $block_hash->{'shared_fileroot'} = '';
608 # a new flag so we can tell we had a manifest way down in the plugins
609 # [jmt12]
610 $block_hash->{'manifest'} = 'false';
611 my $metadata = {};
612
613 # global blocking pass may set up some metadata
614 # - when we have a newer manifest file we don't do this -unless- the
615 # collection configuration indicates this collection contains complex
616 # (inherited) metadata [jmt12]
617 if ($manifest eq '' || (defined $collectcfg->{'complexmeta'} && $collectcfg->{'complexmeta'} eq 'true'))
618 {
619 &plugin::file_block_read($pluginfo, $importdir, "", $block_hash, $metadata, $gli);
620 }
621 else
622 {
623 print STDERR "Skipping global file scan due to manifest and complexmeta configuration\n";
624 }
625
626 if ($manifest ne "") {
627
628 # mark that we are using a manifest - information that might be needed
629 # down in plugins (for instance DirectoryPlugin)
630 $block_hash->{'manifest'} = $self->{'manifest_version'};
631
632 #
633 # 1. Process delete files first
634 #
635 my @deleted_files = keys %{$manifest_lookup->{'delete'}};
636 my @full_deleted_files = ();
637
638 # ensure all filenames are absolute
639 foreach my $df (@deleted_files) {
640 my $full_df =
641 (&FileUtils::isFilenameAbsolute($df))
642 ? $df
643 : &FileUtils::filenameConcatenate($importdir,$df);
644
645 if (-d $full_df) {
646 &add_dir_contents_to_list($full_df, \@full_deleted_files);
647 } else {
648 push(@full_deleted_files,$full_df);
649 }
650 }
651
652 &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@full_deleted_files);
653 mark_docs_for_deletion($archive_info,{},
654 \@full_deleted_files,
655 $archivedir, $verbosity, "delete");
656
657
658 #
659 # 2. Now files for reindexing
660 #
661
662 my @reindex_files = keys %{$manifest_lookup->{'reindex'}};
663 my @full_reindex_files = ();
664 # ensure all filenames are absolute
665 foreach my $rf (@reindex_files) {
666 my $full_rf =
667 (&FileUtils::isFilenameAbsolute($rf))
668 ? $rf
669 : &FileUtils::filenameConcatenate($importdir,$rf);
670
671 if (-d $full_rf) {
672 &add_dir_contents_to_list($full_rf, \@full_reindex_files);
673 } else {
674 push(@full_reindex_files,$full_rf);
675 }
676 }
677
678 &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@full_reindex_files);
679 mark_docs_for_deletion($archive_info,{},\@full_reindex_files, $archivedir,$verbosity, "reindex");
680
681 # And now to ensure the new version of the file processed by
682 # appropriate plugin, we need to add it to block_hash reindex list
683 foreach my $full_rf (@full_reindex_files) {
684 $block_hash->{'reindex_files'}->{$full_rf} = 1;
685 }
686
687
688 #
689 # 3. Now finally any new files - add to block_hash new_files list
690 #
691
692 my @new_files = keys %{$manifest_lookup->{'index'}};
693 my @full_new_files = ();
694
695 foreach my $nf (@new_files) {
696 # ensure filename is absolute
697 my $full_nf =
698 (&FileUtils::isFilenameAbsolute($nf))
699 ? $nf
700 : &FileUtils::filenameConcatenate($importdir,$nf);
701
702 if (-d $full_nf) {
703 &add_dir_contents_to_list($full_nf, \@full_new_files);
704 } else {
705 push(@full_new_files,$full_nf);
706 }
707 }
708
709 my $arcinfo_src_filename = &dbutil::get_infodb_file_path($collectcfg->{'infodbtype'}, "archiveinf-src", $archivedir);
710 # need to check this file exists before trying to read it - in the past
711 # it wasn't possible to have a manifest unless keepold was also set so
712 # you were pretty much guarenteed arcinfo existed
713 # [jmt12]
714 # @todo &FileUtils::fileExists($arcinfo_src_filename) [jmt12]
715 if (-e $arcinfo_src_filename)
716 {
717 my $arcinfodb_map = {};
718 &dbutil::read_infodb_file($collectcfg->{'infodbtype'}, $arcinfo_src_filename, $arcinfodb_map);
719 foreach my $f (@full_new_files) {
720 # check that we haven't seen it already
721 if (defined $arcinfodb_map->{$f}) {
722 # TODO make better warning
723 print STDERR "Warning: $f already in src archive, \n";
724 } else {
725 $block_hash->{'new_files'}->{$f} = 1;
726 }
727 }
728
729 undef $arcinfodb_map;
730 }
731 # no existing files - so we can just add all the files [jmt12]
732 else
733 {
734 foreach my $f (@full_new_files)
735 {
736 $block_hash->{'new_files'}->{$f} = 1;
737 }
738 }
739
740 # If we are not using complex inherited metadata (and thus have skipped
741 # the global file scan) we need to at least check for a matching
742 # metadata.xml for the files being indexed/reindexed
743 # - unless we are using the newer version of Manifests, which are treated
744 # verbatim, and should have a metadata element for metadata files (so
745 # we can explicitly process metadata files other than metadata.xml)
746 # [jmt12]
747 if ($self->{'manifest_version'} < 1 && (!defined $collectcfg->{'complexmeta'} || $collectcfg->{'complexmeta'} ne 'true'))
748 {
749 my @all_files_to_import = (keys %{$block_hash->{'reindex_files'}}, keys %{$block_hash->{'new_files'}});
750 foreach my $file_to_import (@all_files_to_import)
751 {
752 my $metadata_xml_path = $file_to_import;
753 $metadata_xml_path =~ s/[^\\\/]*$/metadata.xml/;
754 if (&FileUtils::fileExists($metadata_xml_path))
755 {
756 &plugin::file_block_read($pluginfo, '', $metadata_xml_path, $block_hash, $metadata, $gli);
757 }
758 }
759 }
760
761 # new version manifest files explicitly list metadata files to be
762 # processed (ignoring complexmeta if set)
763 # [jmt12]
764 if ($self->{'manifest_version'} > 1)
765 {
766 # Process metadata files
767 foreach my $file_to_import (keys %{$block_hash->{'reindex_files'}}, keys %{$block_hash->{'new_files'}})
768 {
769 $self->perform_process_files($manifest, $pluginfo, '', $file_to_import, $block_hash, $metadata, $processor, $maxdocs);
770 }
771 }
772 }
773 else {
774 # if incremental, we read through the import folder to see whats changed.
775
776 if ($incremental || $incremental_mode eq "onlyadd") {
777 prime_doc_oid_count($archivedir);
778
779 # Can now work out which files were new, already existed, and have
780 # been deleted
781
782 new_vs_old_import_diff($archive_info,$block_hash,$importdir,
783 $archivedir,$verbosity,$incremental_mode);
784
785 my @new_files = sort keys %{$block_hash->{'new_files'}};
786 if (scalar(@new_files>0)) {
787 print STDERR "New files and modified metadata files since last import:\n ";
788 print STDERR join("\n ",@new_files), "\n";
789 }
790
791 if ($incremental) {
792 # only look for deletions if we are truely incremental
793 my @deleted_files = sort keys %{$block_hash->{'deleted_files'}};
794 # Filter out any in gsdl/tmp area
795 my @filtered_deleted_files = ();
796 my $gsdl_tmp_area = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "tmp");
797 my $collect_tmp_area = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "tmp");
798 $gsdl_tmp_area = &util::filename_to_regex($gsdl_tmp_area);
799 $collect_tmp_area = &util::filename_to_regex($collect_tmp_area);
800
801 foreach my $df (@deleted_files) {
802 next if ($df =~ m/^$gsdl_tmp_area/);
803 next if ($df =~ m/^$collect_tmp_area/);
804
805 push(@filtered_deleted_files,$df);
806 }
807
808
809 @deleted_files = @filtered_deleted_files;
810
811 if (scalar(@deleted_files)>0) {
812 print STDERR "Files deleted since last import:\n ";
813 print STDERR join("\n ",@deleted_files), "\n";
814
815
816 &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@deleted_files);
817
818 mark_docs_for_deletion($archive_info,$block_hash,\@deleted_files, $archivedir,$verbosity, "delete");
819 }
820
821 my @reindex_files = sort keys %{$block_hash->{'reindex_files'}};
822
823 if (scalar(@reindex_files)>0) {
824 print STDERR "Files to reindex since last import:\n ";
825 print STDERR join("\n ",@reindex_files), "\n";
826 &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@reindex_files);
827 mark_docs_for_deletion($archive_info,$block_hash,\@reindex_files, $archivedir,$verbosity, "reindex");
828 }
829
830 }
831 }
832 }
833
834 # Check for existence of the file that's to contain earliestDateStamp in archivesdir
835 # Do nothing if the file already exists (file exists on incremental build).
836 # If the file doesn't exist, as happens on full build, create it and write out the current datestamp into it
837 # In buildcol, read the file's contents and set the earliestdateStamp in GS2's build.cfg / GS3's buildconfig.xml
838 # In doc.pm have set_oaiLastModified similar to set_lastmodified, and create the doc fields
839 # oailastmodified and oailastmodifieddate
840 my $earliestDatestampFile = &FileUtils::filenameConcatenate($archivedir, "earliestDatestamp");
841 if (!-f $earliestDatestampFile && -d $archivedir) {
842 my $current_time_in_seconds = time; # in seconds
843
844 if(open(FOUT, ">$earliestDatestampFile")) {
845 # || (&gsprintf(STDERR, "{common.cannot_open}: $!\n", $earliestDatestampFile) && die);
846 print FOUT $current_time_in_seconds;
847 close(FOUT);
848 }
849 else {
850 &gsprintf(STDERR, "{import.cannot_write_earliestdatestamp}\n", $earliestDatestampFile);
851 }
852
853 }
854
855
856 $self->perform_process_files($manifest, $pluginfo, $importdir, '', $block_hash, $metadata, $processor, $maxdocs);
857
858 if ($saveas eq "FedoraMETS") {
859 # create collection "doc obj" for Fedora that contains
860 # collection-level metadata
861
862 my $doc_obj = new doc($config_filename,"nonindexed_doc","none");
863 $doc_obj->set_OID("collection");
864
865 my $col_name = undef;
866 my $col_meta = $collectcfg->{'collectionmeta'};
867
868 if (defined $col_meta) {
869 store_collectionmeta($col_meta,"collectionname",$doc_obj); # in GS3 this is a collection's name
870 store_collectionmeta($col_meta,"collectionextra",$doc_obj); # in GS3 this is a collection's description
871 }
872 $processor->process($doc_obj);
873 }
874
875 &plugin::end($pluginfo, $processor);
876
877 &plugin::deinit($pluginfo, $processor);
878
879 # Store the value of OIDCount (used in doc.pm) so it can be
880 # restored correctly to this value on an incremental build
881 # - this OIDcount file should only be generated for numerical oids [jmt12]
882 if ($self->{'OIDtype'} eq 'incremental')
883 {
884 store_doc_oid_count($archivedir);
885 }
886
887 # write out the archive information file
888 $processor->close_file_output() if (defined $groupsize) && ($groupsize > 1);
889 $processor->close_group_output() if $processor->is_group();
890
891 # for backwards compatability with archvies.inf file
892 if ($arcinfo_doc_filename =~ m/(contents)|(\.inf)$/) {
893 $archive_info->save_info($arcinfo_doc_filename);
894 }
895 else {
896 $archive_info->save_revinfo_db($arcinfo_src_filename);
897 }
898
899 return $pluginfo;
900}
901
902# @function perform_process_files()
903# while process_files() above prepares the system to import files this is the
904# function that actually initiates the plugin pipeline to process the files.
905# This function the therefore be overridden in subclasses of inexport.pm should
906# they wish to do different or further processing
907# @author jmt12
908sub perform_process_files
909{
910 my $self = shift(@_);
911 my ($manifest, $pluginfo, $importdir, $file_to_import, $block_hash, $metadata, $processor, $maxdocs) = @_;
912 my $gli = $self->{'gli'};
913 # specific file to process - via manifest version 2+
914 if ($file_to_import ne '')
915 {
916 &plugin::read ($pluginfo, '', $file_to_import, $block_hash, $metadata, $processor, $maxdocs, 0, $gli);
917 }
918 # global file scan - if we are using a new version manifest, files would have
919 # been read above. Older manifests use extra settings in the $block_hash to
920 # control what is imported, while non-manifest imports use a regular
921 # $block_hash (so obeying process_exp and block_exp) [jmt12]
922 elsif ($manifest eq '' || $self->{'manifest_version'} < 1)
923 {
924 &plugin::read ($pluginfo, $importdir, '', $block_hash, $metadata, $processor, $maxdocs, 0, $gli);
925 }
926 else
927 {
928 print STDERR "Skipping perform_process_files() due to manifest presence and version\n";
929 }
930}
931# perform_process_files()
932
933# @function generate_statistics()
934sub generate_statistics
935{
936 my $self = shift @_;
937 my ($pluginfo) = @_;
938
939 my $inexport_mode = $self->{'mode'};
940 my $out = $self->{'out'};
941 my $faillogname = $self->{'faillogname'};
942 my $gli = $self->{'gli'};
943
944 &gsprintf($out, "\n");
945 &gsprintf($out, "*********************************************\n");
946 &gsprintf($out, "{$inexport_mode.complete}\n");
947 &gsprintf($out, "*********************************************\n");
948
949 &plugin::write_stats($pluginfo, 'STDERR', $faillogname, $gli);
950}
951# generate_statistics()
952
953
954# @function deinit()
955# Close down any file handles that we opened (and hence are responsible for
956# closing
957sub deinit
958{
959 my $self = shift(@_);
960 close OUT if $self->{'close_out'};
961 close FAILLOG if $self->{'close_faillog'};
962}
963# deinit()
964
965
966sub store_collectionmeta
967{
968 my ($collectionmeta,$field,$doc_obj) = @_;
969
970 my $section = $doc_obj->get_top_section();
971
972 my $field_hash = $collectionmeta->{$field};
973
974 foreach my $k (keys %$field_hash)
975 {
976 my $val = $field_hash->{$k};
977
978 ### print STDERR "*** $k = $field_hash->{$k}\n";
979
980 my $md_label = "ex.$field";
981
982
983 if ($k =~ m/^\[l=(.*?)\]$/)
984 {
985
986 my $md_suffix = $1;
987 $md_label .= "^$md_suffix";
988 }
989
990
991 $doc_obj->add_utf8_metadata($section,$md_label, $val);
992
993 # see collConfigxml.pm: GS2's "collectionextra" is called "description" in GS3,
994 # while "collectionname" in GS2 is called "name" in GS3.
995 # Variable $nameMap variable in collConfigxml.pm maps between GS2 and GS3
996 if (($md_label eq "ex.collectionname^en") || ($md_label eq "ex.collectionname"))
997 {
998 $doc_obj->add_utf8_metadata($section,"dc.Title", $val);
999 }
1000
1001 }
1002}
1003
1004
1005sub oid_count_file {
1006 my ($archivedir) = @_;
1007 return &FileUtils::filenameConcatenate($archivedir, "OIDcount");
1008}
1009
1010
1011sub prime_doc_oid_count
1012{
1013 my ($archivedir) = @_;
1014 my $oid_count_filename = &oid_count_file($archivedir);
1015
1016 if (-e $oid_count_filename) {
1017 if (open(OIDIN,"<$oid_count_filename")) {
1018 my $OIDcount = <OIDIN>;
1019 chomp $OIDcount;
1020 close(OIDIN);
1021
1022 $doc::OIDcount = $OIDcount;
1023 }
1024 else {
1025 &gsprintf(STDERR, "{import.cannot_read_OIDcount}\n", $oid_count_filename);
1026 }
1027 }
1028
1029}
1030
1031sub store_doc_oid_count
1032{
1033 # Use the file "OIDcount" in the archives directory to record
1034 # what value doc.pm got up to
1035
1036 my ($archivedir) = @_;
1037 my $oid_count_filename = &oid_count_file($archivedir);
1038
1039 # @todo $oidout = &FileUtils::openFileDescriptor($oid_count_filename, 'w') [jmt12]
1040 if (open(OIDOUT,">$oid_count_filename")) {
1041 print OIDOUT $doc::OIDcount, "\n";
1042
1043 close(OIDOUT);
1044 }
1045 else {
1046 &gsprintf(STDERR, "{import.cannot_write_OIDcount}\n", $oid_count_filename);
1047 }
1048}
1049
1050
1051
1052sub new_vs_old_import_diff
1053{
1054 my ($archive_info,$block_hash,$importdir,$archivedir,$verbosity,$incremental_mode) = @_;
1055
1056 # Get the infodbtype value for this collection from the arcinfo object
1057 my $infodbtype = $archive_info->{'infodbtype'};
1058
1059 # in this method, we want to know if metadata files are modified or not.
1060 my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archivedir);
1061
1062 my $archiveinf_timestamp = -M $arcinfo_doc_filename;
1063
1064 # First convert all files to absolute form
1065 # This is to support the situation where the import folder is not
1066 # the default
1067
1068 my $prev_all_files = $archive_info->{'prev_import_filelist'};
1069 my $full_prev_all_files = {};
1070
1071 foreach my $prev_file (keys %$prev_all_files) {
1072
1073 if (!&FileUtils::isFilenameAbsolute($prev_file)) {
1074 my $full_prev_file = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},$prev_file);
1075 $full_prev_all_files->{$full_prev_file} = $prev_file;
1076 }
1077 else {
1078 $full_prev_all_files->{$prev_file} = $prev_file;
1079 }
1080 }
1081
1082
1083 # Figure out which are the new files, existing files and so
1084 # by implication the files from the previous import that are not
1085 # there any more => mark them for deletion
1086 foreach my $curr_file (keys %{$block_hash->{'all_files'}}) {
1087
1088 my $full_curr_file = $curr_file;
1089
1090 # entry in 'all_files' is moved to either 'existing_files',
1091 # 'deleted_files', 'new_files', or 'new_or_modified_metadata_files'
1092
1093 if (!&FileUtils::isFilenameAbsolute($curr_file)) {
1094 # add in import dir to make absolute
1095 $full_curr_file = &FileUtils::filenameConcatenate($importdir,$curr_file);
1096 }
1097
1098 # figure out if new file or not
1099 if (defined $full_prev_all_files->{$full_curr_file}) {
1100 # delete it so that only files that need deleting are left
1101 delete $full_prev_all_files->{$full_curr_file};
1102
1103 # had it before. is it a metadata file?
1104 if ($block_hash->{'metadata_files'}->{$full_curr_file}) {
1105
1106 # is it modified??
1107 if (-M $full_curr_file < $archiveinf_timestamp) {
1108 print STDERR "*** Detected a *modified metadata* file: $full_curr_file\n" if $verbosity >= 2;
1109 # its newer than last build
1110 $block_hash->{'new_or_modified_metadata_files'}->{$full_curr_file} = 1;
1111 }
1112 }
1113 else {
1114 if ($incremental_mode eq "all") {
1115
1116 # had it before
1117 $block_hash->{'existing_files'}->{$full_curr_file} = 1;
1118
1119 }
1120 else {
1121 # Warning in "onlyadd" mode, but had it before!
1122 print STDERR "Warning: File $full_curr_file previously imported.\n";
1123 print STDERR " Treating as new file\n";
1124
1125 $block_hash->{'new_files'}->{$full_curr_file} = 1;
1126
1127 }
1128 }
1129 }
1130 else {
1131 if ($block_hash->{'metadata_files'}->{$full_curr_file}) {
1132 # the new file is the special sort of file greenstone uses
1133 # to attach metadata to src documents
1134 # i.e metadata.xml
1135 # (but note, the filename used is not constrained in
1136 # Greenstone to always be this)
1137
1138 print STDERR "*** Detected *new* metadata file: $full_curr_file\n" if $verbosity >= 2;
1139 $block_hash->{'new_or_modified_metadata_files'}->{$full_curr_file} = 1;
1140 }
1141 else {
1142 $block_hash->{'new_files'}->{$full_curr_file} = 1;
1143 }
1144 }
1145
1146
1147 delete $block_hash->{'all_files'}->{$curr_file};
1148 }
1149
1150
1151
1152
1153 # Deal with complication of new or modified metadata files by forcing
1154 # everything from this point down in the file hierarchy to
1155 # be freshly imported.
1156 #
1157 # This may mean files that have not changed are reindexed, but does
1158 # guarantee by the end of processing all new metadata is correctly
1159 # associated with the relevant document(s).
1160
1161 foreach my $new_mdf (keys %{$block_hash->{'new_or_modified_metadata_files'}}) {
1162 my ($fileroot,$situated_dir,$ext) = fileparse($new_mdf, "\\.[^\\.]+\$");
1163
1164 $situated_dir =~ s/[\\\/]+$//; # remove tailing slashes
1165 $situated_dir = &util::filename_to_regex($situated_dir); # need to escape windows slash \ and brackets in regular expression
1166
1167 # Go through existing_files, and mark anything that is contained
1168 # within 'situated_dir' to be reindexed (in case some of the metadata
1169 # attaches to one of these files)
1170
1171 my $reindex_files = [];
1172
1173 foreach my $existing_f (keys %{$block_hash->{'existing_files'}}) {
1174
1175 if ($existing_f =~ m/^$situated_dir/) {
1176
1177 print STDERR "**** Existing file $existing_f\nis located within\n$situated_dir\n";
1178
1179 push(@$reindex_files,$existing_f);
1180 $block_hash->{'reindex_files'}->{$existing_f} = 1;
1181 delete $block_hash->{'existing_files'}->{$existing_f};
1182
1183 }
1184 }
1185
1186 # metadata file needs to be in new_files list so parsed by MetadataXMLPlug
1187 # (or equivalent)
1188 $block_hash->{'new_files'}->{$new_mdf} = 1;
1189
1190 }
1191
1192 # go through remaining existing files and work out what has changed and needs to be reindexed.
1193 my @existing_files = sort keys %{$block_hash->{'existing_files'}};
1194
1195 my $reindex_files = [];
1196
1197 foreach my $existing_filename (@existing_files) {
1198 if (-M $existing_filename < $archiveinf_timestamp) {
1199 # file is newer than last build
1200
1201 my $existing_file = $existing_filename;
1202 #my $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'});
1203
1204 #my $collectdir_resafe = &util::filename_to_regex($collectdir);
1205 #$existing_file =~ s/^$collectdir_resafe(\\|\/)?//;
1206
1207 print STDERR "**** Reindexing existing file: $existing_file\n";
1208
1209 push(@$reindex_files,$existing_file);
1210 $block_hash->{'reindex_files'}->{$existing_filename} = 1;
1211 }
1212
1213 }
1214
1215
1216 # By this point full_prev_all_files contains the files
1217 # mentioned in archiveinf-src.db but are not in the 'import'
1218 # folder (or whatever was specified through -importdir ...)
1219
1220 # This list can contain files that were created in the 'tmp' or
1221 # 'cache' areas (such as screen-size and thumbnail images).
1222 #
1223 # In building the final list of files to delete, we test to see if
1224 # it exists on the filesystem and if it does (unusual for a "normal"
1225 # file in import, but possible in the case of 'tmp' files),
1226 # supress it from going into the final list
1227
1228 my $collectdir = $ENV{'GSDLCOLLECTDIR'};
1229
1230 my @deleted_files = values %$full_prev_all_files;
1231 map { my $curr_file = $_;
1232 my $full_curr_file = $curr_file;
1233
1234 if (!&FileUtils::isFilenameAbsolute($curr_file)) {
1235 # add in import dir to make absolute
1236
1237 $full_curr_file = &FileUtils::filenameConcatenate($collectdir,$curr_file);
1238 }
1239
1240
1241 if (!-e $full_curr_file) {
1242 $block_hash->{'deleted_files'}->{$curr_file} = 1;
1243 }
1244 } @deleted_files;
1245
1246
1247
1248}
1249
1250
1251# this is used to delete "deleted" docs, and to remove old versions of "changed" docs
1252# $mode is 'delete' or 'reindex'
1253sub mark_docs_for_deletion
1254{
1255 my ($archive_info,$block_hash,$deleted_files,$archivedir,$verbosity,$mode) = @_;
1256
1257 my $mode_text = "deleted from index";
1258 if ($mode eq "reindex") {
1259 $mode_text = "reindexed";
1260 }
1261
1262 # Get the infodbtype value for this collection from the arcinfo object
1263 my $infodbtype = $archive_info->{'infodbtype'};
1264
1265 my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archivedir);
1266 my $arcinfo_src_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-src", $archivedir);
1267
1268
1269 # record files marked for deletion in arcinfo
1270 foreach my $file (@$deleted_files) {
1271 # use 'archiveinf-src' info database file to look up all the OIDs
1272 # that this file is used in (note in most cases, it's just one OID)
1273
1274 my $src_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_src_filename, $file);
1275 my $oids = $src_rec->{'oid'};
1276 my $file_record_deleted = 0;
1277
1278 # delete the src record
1279 my $src_infodb_file_handle = &dbutil::open_infodb_write_handle($infodbtype, $arcinfo_src_filename, "append");
1280 &dbutil::delete_infodb_entry($infodbtype, $src_infodb_file_handle, $file);
1281 &dbutil::close_infodb_write_handle($infodbtype, $src_infodb_file_handle);
1282
1283
1284 foreach my $oid (@$oids) {
1285
1286 # find the source doc (the primary file that becomes this oid)
1287 my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $oid);
1288 my $doc_source_file = $doc_rec->{'src-file'}->[0];
1289 if (!&util::filename_is_absolute($doc_source_file)) {
1290 $doc_source_file = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},$doc_source_file);
1291 }
1292
1293 if ($doc_source_file ne $file) {
1294 # its an associated or metadata file
1295
1296 # mark source doc for reimport as one of its assoc files has changed or deleted
1297 $block_hash->{'reindex_files'}->{$doc_source_file} = 1;
1298
1299 }
1300 my $curr_status = $archive_info->get_status_info($oid);
1301 if (defined($curr_status) && (($curr_status ne "D"))) {
1302 if ($verbosity>1) {
1303 print STDERR "$oid ($doc_source_file) marked to be $mode_text on next buildcol.pl\n";
1304 }
1305 # mark oid for deletion (it will be deleted or reimported)
1306 $archive_info->set_status_info($oid,"D");
1307 my $val = &dbutil::read_infodb_rawentry($infodbtype, $arcinfo_doc_filename, $oid);
1308 $val =~ s/^<index-status>(.*)$/<index-status>D/m;
1309
1310 my $val_rec = &dbutil::convert_infodb_string_to_hash($val);
1311 my $doc_infodb_file_handle = &dbutil::open_infodb_write_handle($infodbtype, $arcinfo_doc_filename, "append");
1312
1313 &dbutil::write_infodb_entry($infodbtype, $doc_infodb_file_handle, $oid, $val_rec);
1314 &dbutil::close_infodb_write_handle($infodbtype, $doc_infodb_file_handle);
1315 }
1316 }
1317
1318 }
1319
1320 # now go through and check that we haven't marked any primary
1321 # files for reindex (because their associated files have
1322 # changed/deleted) when they have been deleted themselves. only in
1323 # delete mode.
1324
1325 if ($mode eq "delete") {
1326 foreach my $file (@$deleted_files) {
1327 if (defined $block_hash->{'reindex_files'}->{$file}) {
1328 delete $block_hash->{'reindex_files'}->{$file};
1329 }
1330 }
1331 }
1332
1333
1334}
1335
1336sub add_dir_contents_to_list {
1337
1338 my ($dirname, $list) = @_;
1339
1340 # Recur over directory contents.
1341 my (@dir, $subfile);
1342
1343 # find all the files in the directory
1344 if (!opendir (DIR, $dirname)) {
1345 print STDERR "inexport: WARNING - couldn't read directory $dirname\n";
1346 return -1; # error in processing
1347 }
1348 @dir = readdir (DIR);
1349 closedir (DIR);
1350
1351 for (my $i = 0; $i < scalar(@dir); $i++) {
1352 my $subfile = $dir[$i];
1353 next if ($subfile =~ m/^\.\.?$/);
1354 next if ($subfile =~ /^\.svn$/);
1355 my $full_file = &FileUtils::filenameConcatenate($dirname, $subfile);
1356 if (-d $full_file) {
1357 &add_dir_contents_to_list($full_file, $list);
1358 } else {
1359 push (@$list, $full_file);
1360 }
1361 }
1362
1363}
1364
1365
13661;
Note: See TracBrowser for help on using the repository browser.