root/main/trunk/greenstone2/perllib/inexport.pm @ 32586

Revision 32586, 48.1 KB (checked in by ak19, 20 months ago)

Renaming 'site_name' parameter used by GS SQL Plugout and Plugin to 'site' for consistency with existing uses of site.

  • Property svn:executable set to *
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 oaiinfo;
38use plugin;
39use plugout;
40use manifest;
41use inexport;
42use util;
43use scriptutil;
44use FileHandle;
45use gsprintf 'gsprintf';
46use printusage;
47use parse2;
48
49use File::Basename;
50
51my $oidtype_list =
52    [ { 'name' => "hash",
53        'desc' => "{import.OIDtype.hash}" },
54      { 'name' => "hash_on_full_filename",
55        'desc' => "{import.OIDtype.hash_on_full_filename}" },
56      { 'name' => "assigned",
57        'desc' => "{import.OIDtype.assigned}" },
58      { 'name' => "incremental",
59        'desc' => "{import.OIDtype.incremental}" },
60      { 'name' => "filename",
61        'desc' => "{import.OIDtype.filename}" },
62      { 'name' => "dirname",
63        'desc' => "{import.OIDtype.dirname}" },
64      { 'name' => "full_filename",
65        'desc' => "{import.OIDtype.full_filename}" } ];
66
67$inexport::directory_arguments =
68[
69      { 'name' => "importdir",
70    'desc' => "{import.importdir}",
71    'type' => "string",
72    'reqd' => "no",
73    'deft' => "import",
74        'hiddengli' => "yes" },
75      { 'name' => "collectdir",
76    'desc' => "{import.collectdir}",
77    'type' => "string",
78    # parsearg left "" as default
79    #'deft' => &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "collect"),
80    'deft' => "",
81    'reqd' => "no",
82        'hiddengli' => "yes" },
83 
84];
85$inexport::arguments =
86[
87      # don't set the default to hash - want to allow this to come from
88      # entry in collect.cfg but want to override it here
89      { 'name' => "OIDtype",
90    'desc' => "{import.OIDtype}",
91    'type' => "enum",
92    'list' => $oidtype_list,
93    'deft' => "hash_on_full_filename",
94    'reqd' => "no",
95    'modegli' => "2" },
96      { 'name' => "OIDmetadata",
97    'desc' => "{import.OIDmetadata}",
98    'type' => "string",
99    'deft' => "dc.Identifier",
100    'reqd' => "no",
101    'modegli' => "2" },
102      { 'name' => "site",
103    'desc' => "{import.site}",
104    'type' => "string",
105    'deft' => "",
106    'reqd' => "no",
107        'hiddengli' => "yes" },
108      { 'name' => "manifest",
109    'desc' => "{import.manifest}",
110    'type' => "string",
111    'deft' => "",
112    'reqd' => "no",
113        'hiddengli' => "yes" } ,
114     { 'name' => "incremental",
115    'desc' => "{import.incremental}",
116    'type' => "flag",
117    'hiddengli' => "yes" },
118      { 'name' => "keepold",
119    'desc' => "{import.keepold}",
120    'type' => "flag",
121    'reqd' => "no",
122    'hiddengli' => "yes" },
123      { 'name' => "removeold",
124    'desc' => "{import.removeold}",
125    'type' => "flag",
126    'reqd' => "no",
127    'hiddengli' => "yes" },
128      { 'name' => "language",
129    'desc' => "{scripts.language}",
130    'type' => "string",
131    'reqd' => "no",
132    'hiddengli' => "yes" },
133      { 'name' => "maxdocs",
134    'desc' => "{import.maxdocs}",
135    'type' => "int",
136    'reqd' => "no",
137    'deft' => "-1",
138    'range' => "-1,",
139    'modegli' => "1" },
140       { 'name' => "debug",
141    'desc' => "{import.debug}",
142    'type' => "flag",
143    'reqd' => "no",
144        'hiddengli' => "yes" },
145      { 'name' => "faillog",
146    'desc' => "{import.faillog}",
147    'type' => "string",
148    # parsearg left "" as default
149    #'deft' => &FileUtils::filenameConcatenate("<collectdir>", "colname", "etc", "fail.log"),
150    'deft' => "",
151    'reqd' => "no",
152        'modegli' => "3" },
153       { 'name' => "out",
154    'desc' => "{import.out}",
155    'type' => "string",
156    'deft' => "STDERR",
157    'reqd' => "no",
158        'hiddengli' => "yes" },
159      { 'name' => "statsfile",
160    'desc' => "{import.statsfile}",
161    'type' => "string",
162    'deft' => "STDERR",
163    'reqd' => "no",
164        'hiddengli' => "yes" },
165      { 'name' => "verbosity",
166    'desc' => "{import.verbosity}",
167    'type' => "int",
168    'range' => "0,",
169    'deft' => "2",
170    'reqd' => "no",
171    'modegli' => "3" },
172      { 'name' => "gli",
173    'desc' => "{scripts.gli}",
174    'type' => "flag",
175    'reqd' => "no",
176    'hiddengli' => "yes" },
177      { 'name' => "xml",
178    'desc' => "{scripts.xml}",
179    'type' => "flag",
180    'reqd' => "no",
181    'hiddengli' => "yes" },
182
183];
184
185sub new
186{
187    my $class = shift (@_);
188    my ($mode,$argv,$options,$opt_listall_options) = @_;
189
190    my $self = { 'xml' => 0, 'mode' => $mode };
191
192    # general options available to all plugins
193    my $arguments = $options->{'args'};
194    my $intArgLeftinAfterParsing = parse2::parse($argv,$arguments,$self,"allow_extra_options");
195    # Parse returns -1 if something has gone wrong
196    if ($intArgLeftinAfterParsing == -1)
197    {
198    &PrintUsage::print_txt_usage($options, "{import.params}",1);
199    print STDERR "Something went wrong during parsing the arguments. Scroll up for details.\n";
200    die "\n";
201    }
202
203    my $language = $self->{'language'};
204    # If $language has been specified, load the appropriate resource bundle
205    # (Otherwise, the default resource bundle will be loaded automatically)
206    if ($language && $language =~ /\S/) {
207    &gsprintf::load_language_specific_resource_bundle($language);
208    }
209
210    if ($self->{'listall'}) {
211    if ($self->{'xml'}) {
212        &PrintUsage::print_xml_usage($opt_listall_options);
213    }
214    else
215    {
216        &PrintUsage::print_txt_usage($opt_listall_options,"{export.params}");
217    }
218    die "\n";
219    }
220
221    if ($self->{'xml'}) {
222        &PrintUsage::print_xml_usage($options);
223    print "\n";
224    return bless $self, $class;
225    }
226
227    if ($self->{'gli'}) { # the gli wants strings to be in UTF-8
228    &gsprintf::output_strings_in_UTF8;
229    }
230
231    # If the user specified -h, then we output the usage
232    if (@$argv && $argv->[0] =~ /^\-+h/) {
233    &PrintUsage::print_txt_usage($options, "{import.params}");
234    die "\n";
235    }
236    # now check that we had exactly one leftover arg, which should be
237    # the collection name. We don't want to do this earlier, cos
238    # -xml arg doesn't need a collection name
239
240    if ($intArgLeftinAfterParsing != 1 )
241    {
242    &PrintUsage::print_txt_usage($options, "{import.params}", 1);
243    print STDERR "There should be one argument left after parsing the script args: the collection name.\n";
244    die "\n";
245    }
246
247    $self->{'close_out'} = 0;
248    my $out = $self->{'out'};
249    if ($out !~ /^(STDERR|STDOUT)$/i) {
250    open (OUT, ">$out") ||
251        (&gsprintf(STDERR, "{common.cannot_open_output_file}: $!\n", $out) && die);
252    $out = 'inexport::OUT';
253    $self->{'close_out'} = 1;
254    }
255    $out->autoflush(1);
256    $self->{'out'} = $out;
257
258    my $statsfile = $self->{'statsfile'};
259    if ($statsfile !~ /^(STDERR|STDOUT)$/i) {
260    open (STATSFILE, ">$statsfile") ||
261        (&gsprintf(STDERR, "{common.cannot_open_output_file}: $!\n", $statsfile) && die);
262    $statsfile = 'inexport::STATSFILE';
263    $self->{'close_stats'} = 1;
264    }
265    $statsfile->autoflush(1);
266    $self->{'statsfile'} = $statsfile;
267
268    # @ARGV should be only one item, the name of the collection
269    $self->{'collection'} = shift @$argv;
270
271    # Unless otherwise stated all manifests are considered version 1---where
272    # they act more like an advanced process expression---as compared to newer
273    # manifest files that act as an explicit (and exhaustive) list of files to
274    # process [jmt12]
275    $self->{'manifest_version'} = 1;
276
277    return bless $self, $class;
278}
279
280# Simplified version of the contstructor for use with CGI scripts
281sub newCGI
282{
283    my $class = shift (@_);
284    my ($mode,$collect,$gsdl_cgi,$opt_site) = @_;
285
286    my $self = { 'xml' => 0, 'mode' => $mode };
287
288    $self->{'out'} = STDERR;
289   
290    if (defined $gsdl_cgi) {
291        $self->{'site'} = $opt_site;
292        my $collect_dir = $gsdl_cgi->get_collection_dir($opt_site);
293        $self->{'collectdir'} = $collect_dir;
294    }
295    else { 
296        $self->{'site'} = "";
297        $self->{'collectdir'} = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"collect");
298    }
299    $self->{'faillog'} = "";
300   
301    $self->{'collection'} = $collect;
302
303    return bless $self, $class;
304}
305sub get_collection
306{
307    my $self = shift @_;
308   
309    return $self->{'collection'};
310}
311
312
313sub read_collection_cfg
314{
315    my $self = shift @_;
316    my ($collection,$options) = @_;
317
318    my $collectdir = $self->{'collectdir'};
319    my $site       = $self->{'site'};
320    my $out        = $self->{'out'};
321     
322    if (($collection = &colcfg::use_collection($site, $collection, $collectdir)) eq "") {
323    #&PrintUsage::print_txt_usage($options, "{import.params}", 1);
324    die "\n";
325    }
326
327    # set gs_version 2/3
328    $self->{'gs_version'} = "2";
329    if ((defined $site) && ($site ne "")) {
330    # gs3
331    $self->{'gs_version'} = "3";
332    }
333
334    # add collection's perllib dir into include path in
335    # case we have collection specific modules
336    &util::augmentINC(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, 'perllib'));
337
338    # check that we can open the faillog
339    my $faillog = $self->{'faillog'};
340    if ($faillog eq "") {
341    $faillog = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "etc", "fail.log");
342    }
343    open (FAILLOG, ">$faillog") ||
344    (&gsprintf(STDERR, "{import.cannot_open_fail_log}\n", $faillog) && die);
345
346   
347    my $faillogname = $faillog;
348    $faillog = 'inexport::FAILLOG';
349    $faillog->autoflush(1);
350    $self->{'faillog'} = $faillog;
351    $self->{'faillogname'} = $faillogname;
352    $self->{'close_faillog'} = 1;
353
354    # Read in the collection configuration file.
355    my $gs_mode = "gs".$self->{'gs_version'}; #gs2 or gs3
356    my $config_filename = &colcfg::get_collect_cfg_name($out, $gs_mode);
357
358    # store the config file's name, so oaiinfo object constructor can be instantiated with it
359    $self->{'config_filename'} = $config_filename;
360
361    my $collectcfg = &colcfg::read_collection_cfg ($config_filename, $gs_mode);
362
363    return ($config_filename,$collectcfg);
364}
365
366sub set_collection_options
367{
368    my $self = shift @_;
369    my ($collectcfg) = @_;
370
371    my $inexport_mode = $self->{'mode'};
372
373    my $importdir  = $self->{'importdir'};
374    my $archivedir = $self->{'archivedir'} || $self->{'exportdir'};
375    my $out        = $self->{'out'};
376
377    # If the infodbtype value wasn't defined in the collect.cfg file, use the default
378    if (!defined($collectcfg->{'infodbtype'}))
379    {
380      $collectcfg->{'infodbtype'} = &dbutil::get_default_infodb_type();
381    }
382    if ($collectcfg->{'infodbtype'} eq "gdbm-txtgz") {
383    # we can't use the text version for archives dbs.
384    $collectcfg->{'infodbtype'} = "gdbm";
385    }
386
387    if (defined $self->{'default_importdir'} && defined $collectcfg->{'importdir'}) {
388    $importdir = $collectcfg->{'importdir'};
389    }
390
391    if ($inexport_mode eq "import") {
392    if ( defined $self->{'default_archivedir'} && defined $collectcfg->{'archivedir'}) {
393        $archivedir = $collectcfg->{'archivedir'};
394    }
395    }
396    elsif ($inexport_mode eq "export") {
397    if (defined $self->{'default_exportdir'} && defined $collectcfg->{'exportdir'}) {
398        $archivedir = $collectcfg->{'exportdir'};
399    }
400    }
401    # fill in the default import and archives directories if none
402    # were supplied, turn all \ into / and remove trailing /
403    if (!&FileUtils::isFilenameAbsolute($importdir))
404    {
405      $importdir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, $importdir);
406    }
407    else
408    {
409      # Don't do this - it kills protocol prefixes
410      #$importdir =~ s/[\\\/]+/\//g;
411      #$importdir =~ s/\/$//;
412      # Do this instead
413      &FileUtils::sanitizePath($importdir);
414    }
415    if (!&FileUtils::directoryExists($importdir))
416    {
417      &gsprintf($out, "{import.no_import_dir}\n\n", $importdir);
418      die "\n";
419    }
420    $self->{'importdir'} = $importdir;
421
422    if (!&FileUtils::isFilenameAbsolute($archivedir)) {
423    $archivedir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, $archivedir);
424    }
425    else {
426   
427    $archivedir = &FileUtils::sanitizePath($archivedir);
428    }
429    $self->{'archivedir'} = $archivedir;
430
431    if (defined $self->{'default_verbosity'}) {
432    if (defined $collectcfg->{'verbosity'} && $collectcfg->{'verbosity'} =~ /\d+/) {
433        $self->{'verbosity'} = $collectcfg->{'verbosity'};
434    }
435    }
436 
437    if (defined $collectcfg->{'manifest'} && $self->{'manifest'} eq "") {
438    $self->{'manifest'} = $collectcfg->{'manifest'};
439    }
440
441    if (defined $collectcfg->{'gzip'} && !$self->{'gzip'}) {
442    if ($collectcfg->{'gzip'} =~ /^true$/i) {
443        $self->{'gzip'} = 1;
444    }
445    }
446
447    if (defined $self->{'default_maxdocs'}) {
448    if (defined $collectcfg->{'maxdocs'} && $collectcfg->{'maxdocs'} =~ /\-?\d+/) {
449        $self->{'maxdocs'} = $collectcfg->{'maxdocs'};
450    }
451    }
452
453   
454
455    if (defined $self->{'default_OIDtype'} ) {
456    if (defined $collectcfg->{'OIDtype'}
457        && $collectcfg->{'OIDtype'} =~ /^(hash|hash_on_full_filename|incremental|assigned|filename|dirname|full_filename)$/) {
458        $self->{'OIDtype'} = $collectcfg->{'OIDtype'};
459    }
460    }
461
462    if (defined $self->{'default_OIDmetadata'}) {
463    if (defined $collectcfg->{'OIDmetadata'}) {
464        $self->{'OIDmetadata'} = $collectcfg->{'OIDmetadata'};
465    }
466    }
467
468    if (defined $collectcfg->{'debug'} && $collectcfg->{'debug'} =~ /^true$/i) {
469    $self->{'debug'} = 1;
470    }
471    if (defined $collectcfg->{'gli'} && $collectcfg->{'gli'} =~ /^true$/i) {
472    $self->{'gli'} = 1;
473    }
474    $self->{'gli'} = 0 unless defined $self->{'gli'};
475       
476    # check keepold and removeold
477    my $checkdir = ($inexport_mode eq "import") ? "archives" : "export";
478
479    my ($removeold, $keepold, $incremental, $incremental_mode)
480    = &scriptutil::check_removeold_and_keepold($self->{'removeold'}, $self->{'keepold'},
481                           $self->{'incremental'}, $checkdir,
482                           $collectcfg);
483
484    $self->{'removeold'}        = $removeold;
485    $self->{'keepold'}          = $keepold;
486    $self->{'incremental'}      = $incremental;
487    $self->{'incremental_mode'} = $incremental_mode;
488
489    # Since this wasted my morning, let's at least warn a user that manifest
490    # files now *only* work if keepold is set [jmt12]
491    if ($self->{'manifest'} && (!$keepold || !$incremental))
492    {
493      print STDERR "Warning: -manifest flag should not be specified without also setting -keepold or -incremental\n";
494    }
495    }
496
497sub process_files
498{
499    my $self = shift @_;
500    my ($config_filename,$collectcfg) = @_;
501
502    my $inexport_mode = $self->{'mode'};
503
504    my $verbosity   = $self->{'verbosity'};
505    my $debug       = $self->{'debug'};
506
507    my $importdir   = $self->{'importdir'};
508    my $archivedir = $self->{'archivedir'} || $self->{'exportdir'};
509
510    my $incremental = $self->{'incremental'};
511    my $incremental_mode = $self->{'incremental_mode'};
512
513    my $gs_version = $self->{'gs_version'};
514
515    my $removeold   = $self->{'removeold'};
516    my $keepold     = $self->{'keepold'};
517
518    my $saveas      = $self->{'saveas'};
519    my $saveas_options = $self->{'saveas_options'};
520    my $OIDtype     = $self->{'OIDtype'};
521    my $OIDmetadata = $self->{'OIDmetadata'};
522
523    my $out         = $self->{'out'};
524    my $faillog     = $self->{'faillog'};
525
526    my $maxdocs     = $self->{'maxdocs'};
527    my $gzip        = $self->{'gzip'};
528    my $groupsize   = $self->{'groupsize'};
529    my $sortmeta    = $self->{'sortmeta'};
530
531    my $removeprefix = $self->{'removeprefix'};
532    my $removesuffix = $self->{'removesuffix'};
533
534    my $gli          = $self->{'gli'};
535
536    # related to export
537    my $xsltfile         = $self->{'xsltfile'};
538    my $group_marc       = $self->{'group_marc'};
539    my $mapping_file     = $self->{'mapping_file'};
540    my $xslt_mets        = $self->{'xslt_mets'};
541    my $xslt_txt         = $self->{'xslt_txt'};
542    my $fedora_namespace = $self->{'fedora_namespace'};
543    my $metadata_prefix  = $self->{'metadata_prefix'};
544
545    if ($inexport_mode eq "import") {
546    print STDERR "<Import>\n" if $gli;
547    }
548    else {
549    print STDERR "<export>\n" if $gli;
550    }
551
552    my $manifest_lookup = new manifest($collectcfg->{'infodbtype'},$archivedir);
553    if ($self->{'manifest'} ne "") {
554    my $manifest_filename = $self->{'manifest'};
555
556    if (!&FileUtils::isFilenameAbsolute($manifest_filename)) {
557        $manifest_filename = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, $manifest_filename);
558    }
559        $self->{'manifest'} = &FileUtils::sanitizePath($self->{'manifest'});
560    #$self->{'manifest'} =~ s/[\\\/]+/\//g;
561    #$self->{'manifest'} =~ s/\/$//;
562
563    $manifest_lookup->parse($manifest_filename);
564
565        # manifests may now include a version number [jmt12]
566        $self->{'manifest_version'} = $manifest_lookup->get_version();
567    }
568
569    my $manifest = $self->{'manifest'};
570
571    # load all the plugins
572    my $plugins = [];
573    if (defined $collectcfg->{'plugin'}) {
574    $plugins = $collectcfg->{'plugin'};
575    }
576
577    my $plugin_incr_mode = $incremental_mode;
578    if ($manifest ne "") {
579    # if we have a manifest file, then we pretend we are fully incremental for plugins
580    $plugin_incr_mode = "all";
581    }
582    #some global options for the plugins
583    my @global_opts = ();
584
585    my $pluginfo = &plugin::load_plugins ($plugins, $verbosity, $out, $faillog, \@global_opts, $plugin_incr_mode, $gs_version, $self->{'site'});
586    if (scalar(@$pluginfo) == 0) {
587    &gsprintf($out, "{import.no_plugins_loaded}\n");
588    die "\n";
589    }
590
591    # remove the old contents of the archives directory (and tmp
592    # directory) if needed
593
594    if ($removeold) {
595    if (&FileUtils::directoryExists($archivedir)) {
596        &gsprintf($out, "{import.removing_archives}\n");
597        &FileUtils::removeFilesRecursive($archivedir);
598    }
599    my $tmpdir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "tmp");
600    $tmpdir =~ s/[\\\/]+/\//g;
601    $tmpdir =~ s/\/$//;
602    if (&FileUtils::directoryExists($tmpdir)) {
603        &gsprintf($out, "{import.removing_tmpdir}\n");
604        &FileUtils::removeFilesRecursive($tmpdir);
605    }
606    }
607
608    # create the archives dir if needed
609    &FileUtils::makeAllDirectories($archivedir);
610
611    # read the archive information file
612
613    # BACKWARDS COMPATIBILITY: Just in case there are old .ldb/.bdb files (won't do anything for other infodbtypes)
614    &util::rename_ldb_or_bdb_file(&FileUtils::filenameConcatenate($archivedir, "archiveinf-doc"));
615    &util::rename_ldb_or_bdb_file(&FileUtils::filenameConcatenate($archivedir, "archiveinf-src"));
616
617    # When we make these initial calls to determine the archive information doc
618    # and src databases we pass through a '1' to indicate this is the first
619    # time we are referring to these databases. When using dynamic dbutils
620    # (available in extensions) this indicates to some database types (for
621    # example, persistent servers) that this is a good time to perform any
622    # one time initialization. The argument has no effect on vanilla dbutils
623    # [jmt12]
624    my $perform_firsttime_init = 1;
625    my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($collectcfg->{'infodbtype'}, "archiveinf-doc", $archivedir, $perform_firsttime_init);
626    my $arcinfo_src_filename = &dbutil::get_infodb_file_path($collectcfg->{'infodbtype'}, "archiveinf-src", $archivedir, $perform_firsttime_init);
627
628    my $archive_info = new arcinfo ($collectcfg->{'infodbtype'});
629    $archive_info->load_info ($arcinfo_doc_filename);
630
631    if ($manifest eq "") {
632    # Load in list of files in import folder from last import (if present)
633    $archive_info->load_prev_import_filelist ($arcinfo_src_filename);
634    }
635
636    ####Use Plugout####
637    my $plugout;
638
639    my $generate_auxiliary_files = 0;
640    if ($inexport_mode eq "import") {
641    $generate_auxiliary_files = 1;
642    }
643    elsif ($self->{'include_auxiliary_database_files'}) {
644    $generate_auxiliary_files = 1;
645    }
646    $self->{'generate_auxiliary_files'} = $generate_auxiliary_files;
647
648    # Option to use user defined plugout
649    if ($inexport_mode eq "import") {
650    if (defined $collectcfg->{'plugout'}) {
651        # If a plugout was specified in the collect.cfg file, assume it is sensible
652        # We can't check the name because it could be anything, if it is a custom plugout
653        print STDERR "Using plugout specified in collect.cfg: ".join(' ', @{$collectcfg->{'plugout'}})."\n";
654        $plugout = $collectcfg->{'plugout'};
655    }
656    else {
657        push @$plugout,$saveas."Plugout";
658    }
659
660    }
661    else {
662    if (defined $collectcfg->{'plugout'} && $collectcfg->{'plugout'} =~ /^(GreenstoneXML|.*METS|DSpace|MARCXML)Plugout/) {
663        $plugout = $collectcfg->{'plugout'};
664        print STDERR "Using plugout specified in collect.cfg: $collectcfg->{'plugout'}\n";
665    }
666    else {
667        push @$plugout,$saveas."Plugout";
668    }
669    }
670
671    my $plugout_name = $plugout->[0];
672
673    if (defined $saveas_options) {
674    my @user_plugout_options = split(" ", $saveas_options);
675    push @$plugout, @user_plugout_options;
676    }
677    push @$plugout,("-output_info",$archive_info)  if (defined $archive_info);
678    push @$plugout,("-verbosity",$verbosity)       if (defined $verbosity);
679    push @$plugout,("-debug")                      if ($debug);
680    push @$plugout,("-gzip_output")                if ($gzip);
681    push @$plugout,("-output_handle",$out)         if (defined $out);
682    push @$plugout,("-site",$self->{'site'})  if (defined $self->{'site'});
683
684    push @$plugout,("-xslt_file",$xsltfile)        if (defined $xsltfile && $xsltfile ne "");
685    push @$plugout, ("-no_auxiliary_databases") if ($generate_auxiliary_files == 0);
686    if ($inexport_mode eq "import") {
687    if ($plugout_name =~ m/^GreenstoneXMLPlugout$/) {
688        push @$plugout,("-group_size",$groupsize)      if (defined $groupsize);
689    }
690    }
691    my $processor = &plugout::load_plugout($plugout);
692    $processor->setoutputdir ($archivedir);
693    $processor->set_sortmeta ($sortmeta, $removeprefix, $removesuffix) if defined $sortmeta;
694    $processor->set_OIDtype ($OIDtype, $OIDmetadata);
695    $processor->begin();
696    &plugin::begin($pluginfo, $importdir, $processor, $maxdocs, $gli);
697   
698    if ($removeold) {
699        # occasionally, plugins may want to do something on remove
700        # old, eg pharos image indexing
701    &plugin::remove_all($pluginfo, $importdir, $processor, $maxdocs, $gli);
702    }
703
704    # process the import directory
705    my $block_hash = {};
706    $block_hash->{'new_files'} = {};
707    $block_hash->{'reindex_files'} = {};
708    # all of these are set somewhere else, so it's more readable to define them
709    # here [jmt12]
710    $block_hash->{'all_files'} = {};
711    $block_hash->{'deleted_files'} = {};
712    $block_hash->{'file_blocks'} = {};
713    $block_hash->{'metadata_files'} = {};
714    $block_hash->{'shared_fileroot'} = '';
715    # a new flag so we can tell we had a manifest way down in the plugins
716    # [jmt12]
717    $block_hash->{'manifest'} = 'false';
718    my $metadata = {};
719   
720    # global blocking pass may set up some metadata
721    # does this set up metadata?????
722    # - when we have a newer manifest file we don't do this -unless- the
723    #   collection configuration indicates this collection contains complex
724    #   (inherited) metadata [jmt12]
725    if ($manifest eq '' || (defined $collectcfg->{'complexmeta'} && $collectcfg->{'complexmeta'} eq 'true'))
726    {
727      &plugin::file_block_read($pluginfo, $importdir, "", $block_hash, $metadata, $gli);
728    }
729    else
730    {
731      print STDERR "Skipping global file scan due to manifest and complexmeta configuration\n";
732    }
733
734
735    # Prepare to work with the <collection>/etc/oai-inf.<db> that keeps track
736    # of the OAI identifiers with their time stamps and deleted status.   
737    my $oai_info = new oaiinfo($self->{'config_filename'}, $collectcfg->{'infodbtype'}, $verbosity);
738    my $have_manifest = ($manifest eq '') ? 0 : 1;   
739    $oai_info->import_stage($removeold, $have_manifest);
740
741
742    if ($manifest ne "") {
743
744      # mark that we are using a manifest - information that might be needed
745      # down in plugins (for instance DirectoryPlugin)
746      $block_hash->{'manifest'} = $self->{'manifest_version'};
747
748    #
749    # 1. Process delete files first
750    #
751    my @deleted_files = keys %{$manifest_lookup->{'delete'}};
752    my @full_deleted_files = ();
753
754    # ensure all filenames are absolute
755    foreach my $df (@deleted_files) {
756        my $full_df =
757        (&FileUtils::isFilenameAbsolute($df))
758        ? $df
759        : &FileUtils::filenameConcatenate($importdir,$df);
760
761        if (-d $full_df) {
762        &add_dir_contents_to_list($full_df, \@full_deleted_files);
763        } else {
764        push(@full_deleted_files,$full_df);
765        }
766    }
767   
768    &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@full_deleted_files);
769    mark_docs_for_deletion($archive_info,{},
770                   \@full_deleted_files,
771                   $archivedir, $verbosity, "delete");
772
773
774    #
775    # 2. Now files for reindexing
776    #
777
778    my @reindex_files = keys %{$manifest_lookup->{'reindex'}};
779    my @full_reindex_files = ();
780    # ensure all filenames are absolute
781    foreach my $rf (@reindex_files) {       
782        my $full_rf =
783        (&FileUtils::isFilenameAbsolute($rf))
784        ? $rf
785        : &FileUtils::filenameConcatenate($importdir,$rf);
786
787        if (-d $full_rf) {
788        &add_dir_contents_to_list($full_rf, \@full_reindex_files);
789        } else {
790        push(@full_reindex_files,$full_rf);
791        }
792    }
793   
794    &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@full_reindex_files);
795    mark_docs_for_deletion($archive_info,{},\@full_reindex_files, $archivedir,$verbosity, "reindex");
796
797    # And now to ensure the new version of the file processed by
798    # appropriate plugin, we need to add it to block_hash reindex list
799    foreach my $full_rf (@full_reindex_files) {
800        $block_hash->{'reindex_files'}->{$full_rf} = 1;
801    }
802
803
804    #
805    # 3. Now finally any new files - add to block_hash new_files list
806    #
807
808    my @new_files = keys %{$manifest_lookup->{'index'}};
809    my @full_new_files = ();
810
811    foreach my $nf (@new_files) {
812        # ensure filename is absolute
813        my $full_nf =
814        (&FileUtils::isFilenameAbsolute($nf))
815        ? $nf
816        : &FileUtils::filenameConcatenate($importdir,$nf);
817
818        if (-d $full_nf) {
819        &add_dir_contents_to_list($full_nf, \@full_new_files);
820        } else {
821        push(@full_new_files,$full_nf);
822        }
823    }
824
825    my $arcinfo_src_filename = &dbutil::get_infodb_file_path($collectcfg->{'infodbtype'}, "archiveinf-src", $archivedir);
826      # need to check this file exists before trying to read it - in the past
827      # it wasn't possible to have a manifest unless keepold was also set so
828      # you were pretty much guaranteed arcinfo existed
829      # [jmt12]
830      # @todo &FileUtils::fileExists($arcinfo_src_filename) [jmt12]
831      if (-e $arcinfo_src_filename)
832      {
833    my $arcinfodb_map = {};
834    &dbutil::read_infodb_file($collectcfg->{'infodbtype'}, $arcinfo_src_filename, $arcinfodb_map);
835    foreach my $f (@full_new_files) {
836        my $rel_f = &util::abspath_to_placeholders($f);
837
838        # check that we haven't seen it already
839        if (defined $arcinfodb_map->{$rel_f}) {
840        # TODO make better warning
841        print STDERR "Warning: $f ($rel_f) already in src archive, \n";
842        } else {
843        $block_hash->{'new_files'}->{$f} = 1;
844        }
845    }
846
847    undef $arcinfodb_map;
848      }
849      # no existing files - so we can just add all the files [jmt12]
850      else
851      {
852        foreach my $f (@full_new_files)
853        {
854          $block_hash->{'new_files'}->{$f} = 1;
855        }
856      }
857
858      # If we are not using complex inherited metadata (and thus have skipped
859      # the global file scan) we need to at least check for a matching
860      # metadata.xml for the files being indexed/reindexed
861      # - unless we are using the newer version of Manifests, which are treated
862      #   verbatim, and should have a metadata element for metadata files (so
863      #   we can explicitly process metadata files other than metadata.xml)
864      # [jmt12]
865      if ($self->{'manifest_version'} == 1 && (!defined $collectcfg->{'complexmeta'} || $collectcfg->{'complexmeta'} ne 'true'))
866      {
867        my @all_files_to_import = (keys %{$block_hash->{'reindex_files'}}, keys %{$block_hash->{'new_files'}});
868        foreach my $file_to_import (@all_files_to_import)
869        {
870          my $metadata_xml_path = $file_to_import;
871          $metadata_xml_path =~ s/[^\\\/]*$/metadata.xml/;
872          if (&FileUtils::fileExists($metadata_xml_path))
873          {
874            &plugin::file_block_read($pluginfo, '', $metadata_xml_path, $block_hash, $metadata, $gli);
875          }
876        }
877      }
878
879      # new version manifest files explicitly list metadata files to be
880      # processed (ignoring complexmeta if set)
881      # [jmt12]
882      if ($self->{'manifest_version'} > 1)
883      {
884        # Process metadata files
885        foreach my $file_to_import (keys %{$block_hash->{'reindex_files'}}, keys %{$block_hash->{'new_files'}})
886        {
887          $self->perform_process_files($manifest, $pluginfo, '', $file_to_import, $block_hash, $metadata, $processor, $maxdocs);
888        }
889      }
890    } # end if (manifest ne "")
891    else {
892    # if incremental, we read through the import folder to see whats changed.
893
894    if ($incremental || $incremental_mode eq "onlyadd") {
895        prime_doc_oid_count($archivedir);
896
897        # Can now work out which files were new, already existed, and have
898        # been deleted
899       
900        new_vs_old_import_diff($archive_info,$block_hash,$importdir,
901                   $archivedir,$verbosity,$incremental_mode);
902       
903        my @new_files = sort keys %{$block_hash->{'new_files'}};
904        if (scalar(@new_files>0)) {
905        print STDERR "New files and modified metadata files since last import:\n  ";
906        print STDERR join("\n  ",@new_files), "\n";
907        }
908
909        if ($incremental) {
910               # only look for deletions if we are truely incremental
911        my @deleted_files = sort keys %{$block_hash->{'deleted_files'}};
912        # Filter out any in gsdl/tmp area
913        my @filtered_deleted_files = ();
914        my $gsdl_tmp_area = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "tmp");
915        my $collect_tmp_area = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "tmp");
916        $gsdl_tmp_area = &util::filename_to_regex($gsdl_tmp_area);
917        $collect_tmp_area = &util::filename_to_regex($collect_tmp_area);
918                 
919        foreach my $df (@deleted_files) {
920            next if ($df =~ m/^$gsdl_tmp_area/);
921            next if ($df =~ m/^$collect_tmp_area/);
922           
923            push(@filtered_deleted_files,$df);
924        }       
925       
926
927        @deleted_files = @filtered_deleted_files;
928       
929        if (scalar(@deleted_files)>0) {
930            print STDERR "Files deleted since last import:\n  ";
931            print STDERR join("\n  ",@deleted_files), "\n";
932       
933       
934            &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@deleted_files);
935           
936            mark_docs_for_deletion($archive_info,$block_hash,\@deleted_files, $archivedir,$verbosity, "delete");
937        }
938       
939        my @reindex_files = sort keys %{$block_hash->{'reindex_files'}};
940       
941        if (scalar(@reindex_files)>0) {
942            print STDERR "Files to reindex since last import:\n  ";
943            print STDERR join("\n  ",@reindex_files), "\n";
944            &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@reindex_files);
945            mark_docs_for_deletion($archive_info,$block_hash,\@reindex_files, $archivedir,$verbosity, "reindex");
946        }
947               
948        }       
949    } # end if incremental/only_add mode
950    # else no manifest AND not incremental
951    } # end if else block of manifest ne "" else eq ""
952
953    # Check for existence of the file that's to contain earliestDateStamp in archivesdir
954    # Do nothing if the file already exists (file exists on incremental build).
955    # If the file doesn't exist, as happens on full build, create it and write out the current datestamp into it
956    # In buildcol, read the file's contents and set the earliestdateStamp in GS2's build.cfg / GS3's buildconfig.xml
957    # In doc.pm have set_oaiLastModified similar to set_lastmodified, and create the doc fields
958    # oailastmodified and oailastmodifieddate
959    my $earliestDatestampFile = &FileUtils::filenameConcatenate($archivedir, "earliestDatestamp");
960    if ($self->{'generate_auxiliary_files'}) {
961    if (!-f $earliestDatestampFile && -d $archivedir) {
962    my $current_time_in_seconds = time; # in seconds
963
964    if(open(FOUT, ">$earliestDatestampFile")) {
965        # || (&gsprintf(STDERR, "{common.cannot_open}: $!\n", $earliestDatestampFile) && die);
966        print FOUT $current_time_in_seconds;
967        close(FOUT);
968    }
969    else {
970        &gsprintf(STDERR, "{import.cannot_write_earliestdatestamp}\n", $earliestDatestampFile);
971    }
972
973    }
974    }
975   
976    $self->perform_process_files($manifest, $pluginfo, $importdir, '', $block_hash, $metadata, $processor, $maxdocs);
977
978    if ($saveas eq "FedoraMETS") {
979    # create collection "doc obj" for Fedora that contains
980    # collection-level metadata
981   
982    my $doc_obj = new doc($config_filename,"nonindexed_doc","none");
983    $doc_obj->set_OID("collection");
984   
985    my $col_name = undef;
986    my $col_meta = $collectcfg->{'collectionmeta'};
987   
988    if (defined $col_meta) {       
989        store_collectionmeta($col_meta,"collectionname",$doc_obj); # in GS3 this is a collection's name
990        store_collectionmeta($col_meta,"collectionextra",$doc_obj); # in GS3 this is a collection's description     
991    }
992    $processor->process($doc_obj);
993    }
994
995    &plugin::end($pluginfo, $processor);
996
997    &plugin::deinit($pluginfo, $processor);
998
999    # Store the value of OIDCount (used in doc.pm) so it can be
1000    # restored correctly to this value on an incremental build
1001    # - this OIDcount file should only be generated for numerical oids [jmt12]
1002    if ($self->{'OIDtype'} eq 'incremental')
1003    {
1004      store_doc_oid_count($archivedir);
1005    }
1006
1007    # signal to the processor (plugout) that we have finished processing - if we are group processing, then the final output file needs closing.
1008    $processor->close_group_output() if $processor->is_group();
1009    $processor->end();
1010   
1011#    if ($inexport_mode eq "import") {
1012    if ($self->{'generate_auxiliary_files'}) {
1013    # write out the archive information file
1014    # for backwards compatability with archvies.inf file
1015    if ($arcinfo_doc_filename =~ m/(contents)|(\.inf)$/) {
1016        $archive_info->save_info($arcinfo_doc_filename);
1017    }
1018    else {
1019        $archive_info->save_revinfo_db($arcinfo_src_filename);
1020    }
1021    }
1022    return $pluginfo;
1023}
1024
1025# @function perform_process_files()
1026# while process_files() above prepares the system to import files this is the
1027# function that actually initiates the plugin pipeline to process the files.
1028# This function should therefore be overridden in subclasses of inexport.pm should
1029# they wish to do different or further processing
1030# @author jmt12
1031sub perform_process_files
1032{
1033  my $self = shift(@_);
1034  my ($manifest, $pluginfo, $importdir, $file_to_import, $block_hash, $metadata, $processor, $maxdocs) = @_;
1035  my $gli = $self->{'gli'};
1036  # specific file to process - via manifest version 2+
1037  if ($file_to_import ne '')
1038  {
1039    &plugin::read ($pluginfo, '', $file_to_import, $block_hash, $metadata, $processor, $maxdocs, 0, $gli);
1040  }
1041  # global file scan - if we are using a new version manifest, files would have
1042  # been read above. Older manifests use extra settings in the $block_hash to
1043  # control what is imported, while non-manifest imports use a regular
1044  # $block_hash (so obeying process_exp and block_exp) [jmt12]
1045  elsif ($manifest eq '' || $self->{'manifest_version'} == 1)
1046  {
1047    &plugin::read ($pluginfo, $importdir, '', $block_hash, $metadata, $processor, $maxdocs, 0, $gli);
1048  }
1049  else
1050  {
1051    print STDERR "Skipping perform_process_files() due to manifest presence and version\n";
1052  }
1053}
1054# perform_process_files()
1055
1056# @function generate_statistics()
1057sub generate_statistics
1058{
1059  my $self = shift @_;
1060  my ($pluginfo) = @_;
1061
1062  my $inexport_mode = $self->{'mode'};
1063  my $out           = $self->{'out'};
1064  my $faillogname   = $self->{'faillogname'};
1065  my $statsfile     = $self->{'statsfile'};
1066  my $gli           = $self->{'gli'};
1067
1068  &gsprintf($out, "\n");
1069  &gsprintf($out, "*********************************************\n");
1070  &gsprintf($out, "{$inexport_mode.complete}\n");
1071  &gsprintf($out, "*********************************************\n");
1072
1073  &plugin::write_stats($pluginfo, $statsfile, $faillogname, $gli);
1074}
1075# generate_statistics()
1076
1077
1078# @function deinit()
1079# Close down any file handles that we opened (and hence are responsible for
1080# closing
1081sub deinit
1082{
1083  my $self = shift(@_);
1084  close OUT if $self->{'close_out'};
1085  close FAILLOG if $self->{'close_faillog'};
1086  close STATSFILE if $self->{'close_statsfile'};
1087}
1088# deinit()
1089
1090
1091sub store_collectionmeta
1092{
1093    my ($collectionmeta,$field,$doc_obj) = @_;
1094   
1095    my $section = $doc_obj->get_top_section();
1096   
1097    my $field_hash = $collectionmeta->{$field};
1098   
1099    foreach my $k (keys %$field_hash)
1100    {
1101    my $val = $field_hash->{$k};
1102   
1103    ### print STDERR "*** $k = $field_hash->{$k}\n";
1104   
1105    my $md_label = "ex.$field";
1106   
1107   
1108    if ($k =~ m/^\[l=(.*?)\]$/)
1109    {
1110       
1111        my $md_suffix = $1;
1112        $md_label .= "^$md_suffix";
1113    }
1114   
1115   
1116    $doc_obj->add_utf8_metadata($section,$md_label, $val);
1117   
1118    # see collConfigxml.pm: GS2's "collectionextra" is called "description" in GS3,
1119    # while "collectionname" in GS2 is called "name" in GS3.
1120    # Variable $nameMap variable in collConfigxml.pm maps between GS2 and GS3
1121    if (($md_label eq "ex.collectionname^en") || ($md_label eq "ex.collectionname"))
1122    {
1123        $doc_obj->add_utf8_metadata($section,"dc.Title", $val);
1124    }
1125   
1126    }
1127}
1128
1129
1130sub oid_count_file {
1131    my ($archivedir) = @_;
1132    return &FileUtils::filenameConcatenate($archivedir, "OIDcount");
1133}
1134
1135
1136sub prime_doc_oid_count
1137{
1138    my ($archivedir) = @_;
1139    my $oid_count_filename = &oid_count_file($archivedir);
1140
1141    if (-e $oid_count_filename) {
1142    if (open(OIDIN,"<$oid_count_filename")) {
1143        my $OIDcount = <OIDIN>;
1144        chomp $OIDcount;       
1145        close(OIDIN);
1146
1147        $doc::OIDcount = $OIDcount;     
1148    }
1149    else {     
1150        &gsprintf(STDERR, "{import.cannot_read_OIDcount}\n", $oid_count_filename);
1151    }
1152    }
1153   
1154}
1155
1156sub store_doc_oid_count
1157{
1158    # Use the file "OIDcount" in the archives directory to record
1159    # what value doc.pm got up to
1160
1161    my ($archivedir) = @_;
1162    my $oid_count_filename = &oid_count_file($archivedir);
1163
1164    # @todo $oidout = &FileUtils::openFileDescriptor($oid_count_filename, 'w') [jmt12]
1165    if (open(OIDOUT,">$oid_count_filename")) {
1166    print OIDOUT $doc::OIDcount, "\n";
1167       
1168    close(OIDOUT);
1169    }
1170    else {
1171    &gsprintf(STDERR, "{import.cannot_write_OIDcount}\n", $oid_count_filename);
1172    }
1173}
1174
1175
1176
1177sub new_vs_old_import_diff
1178{
1179    my ($archive_info,$block_hash,$importdir,$archivedir,$verbosity,$incremental_mode) = @_;
1180
1181    # Get the infodbtype value for this collection from the arcinfo object
1182    my $infodbtype = $archive_info->{'infodbtype'};
1183
1184    # in this method, we want to know if metadata files are modified or not.
1185    my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archivedir);
1186
1187    my $archiveinf_timestamp = -M $arcinfo_doc_filename;
1188
1189    # First convert all files to absolute form
1190    # This is to support the situation where the import folder is not
1191    # the default
1192   
1193    my $prev_all_files = $archive_info->{'prev_import_filelist'};
1194    my $full_prev_all_files = {};
1195
1196    foreach my $prev_file (keys %$prev_all_files) {
1197
1198    if (!&FileUtils::isFilenameAbsolute($prev_file)) {
1199        my $full_prev_file = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},$prev_file);
1200        $full_prev_all_files->{$full_prev_file} = $prev_file;
1201    }
1202    else {
1203        $full_prev_all_files->{$prev_file} = $prev_file;
1204    }
1205    }
1206
1207
1208    # Figure out which are the new files, existing files and so
1209    # by implication the files from the previous import that are not
1210    # there any more => mark them for deletion
1211    foreach my $curr_file (keys %{$block_hash->{'all_files'}}) {
1212   
1213    my $full_curr_file = $curr_file;
1214
1215    # entry in 'all_files' is moved to either 'existing_files',
1216    # 'deleted_files', 'new_files', or 'new_or_modified_metadata_files'
1217
1218    if (!&FileUtils::isFilenameAbsolute($curr_file)) {
1219        # add in import dir to make absolute
1220        $full_curr_file = &FileUtils::filenameConcatenate($importdir,$curr_file);
1221    }
1222
1223    # figure out if new file or not
1224    if (defined $full_prev_all_files->{$full_curr_file}) {
1225        # delete it so that only files that need deleting are left
1226        delete $full_prev_all_files->{$full_curr_file};
1227       
1228        # had it before. is it a metadata file?
1229        if ($block_hash->{'metadata_files'}->{$full_curr_file}) {
1230       
1231        # is it modified??
1232        if (-M $full_curr_file < $archiveinf_timestamp) {
1233            print STDERR "*** Detected a *modified metadata* file: $full_curr_file\n" if $verbosity >= 2;
1234            # its newer than last build
1235            $block_hash->{'new_or_modified_metadata_files'}->{$full_curr_file} = 1;
1236        }
1237        }
1238        else {
1239        if ($incremental_mode eq "all") {
1240           
1241            # had it before
1242            $block_hash->{'existing_files'}->{$full_curr_file} = 1;
1243           
1244        }
1245        else {
1246            # Warning in "onlyadd" mode, but had it before!
1247            print STDERR "Warning: File $full_curr_file previously imported.\n";
1248            print STDERR "         Treating as new file\n";
1249           
1250            $block_hash->{'new_files'}->{$full_curr_file} = 1;
1251           
1252        }
1253        }
1254    }
1255    else {
1256        if ($block_hash->{'metadata_files'}->{$full_curr_file}) {
1257        # the new file is the special sort of file greenstone uses
1258        # to attach metadata to src documents
1259        # i.e metadata.xml
1260        # (but note, the filename used is not constrained in
1261        # Greenstone to always be this)
1262
1263        print STDERR "*** Detected *new* metadata file: $full_curr_file\n" if $verbosity >= 2;
1264        $block_hash->{'new_or_modified_metadata_files'}->{$full_curr_file} = 1;
1265        }
1266        else {
1267        $block_hash->{'new_files'}->{$full_curr_file} = 1;
1268        }
1269    }
1270
1271   
1272    delete $block_hash->{'all_files'}->{$curr_file};
1273    }
1274
1275
1276
1277
1278    # Deal with complication of new or modified metadata files by forcing
1279    # everything from this point down in the file hierarchy to
1280    # be freshly imported. 
1281    #
1282    # This may mean files that have not changed are reindexed, but does
1283    # guarantee by the end of processing all new metadata is correctly
1284    # associated with the relevant document(s).
1285
1286    foreach my $new_mdf (keys %{$block_hash->{'new_or_modified_metadata_files'}}) {
1287    my ($fileroot,$situated_dir,$ext) = fileparse($new_mdf, "\\.[^\\.]+\$");
1288
1289    $situated_dir =~ s/[\\\/]+$//; # remove tailing slashes
1290    $situated_dir = &util::filename_to_regex($situated_dir); # need to escape windows slash \ and brackets in regular expression
1291   
1292    # Go through existing_files, and mark anything that is contained
1293    # within 'situated_dir' to be reindexed (in case some of the metadata
1294    # attaches to one of these files)
1295
1296    my $reindex_files = [];
1297
1298    foreach my $existing_f (keys %{$block_hash->{'existing_files'}}) {
1299   
1300        if ($existing_f =~ m/^$situated_dir/) {
1301
1302#       print STDERR "**** Existing file $existing_f\nis located within\n$situated_dir\n";
1303
1304        push(@$reindex_files,$existing_f);
1305        $block_hash->{'reindex_files'}->{$existing_f} = 1;
1306        delete $block_hash->{'existing_files'}->{$existing_f};
1307
1308        }
1309    }
1310   
1311    # metadata file needs to be in new_files list so parsed by MetadataXMLPlug
1312    # (or equivalent)
1313    $block_hash->{'new_files'}->{$new_mdf} = 1;
1314
1315    }
1316
1317    # go through remaining existing files and work out what has changed and needs to be reindexed.
1318    my @existing_files = sort keys %{$block_hash->{'existing_files'}};
1319
1320    my $reindex_files = [];
1321
1322    foreach my $existing_filename (@existing_files) {
1323    if (-M $existing_filename < $archiveinf_timestamp) {
1324        # file is newer than last build
1325       
1326        my $existing_file = $existing_filename;
1327        #my $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'});
1328
1329        #my $collectdir_resafe = &util::filename_to_regex($collectdir);
1330        #$existing_file =~ s/^$collectdir_resafe(\\|\/)?//;
1331       
1332        print STDERR "**** Reindexing existing file: $existing_file\n";
1333
1334        push(@$reindex_files,$existing_file);
1335        $block_hash->{'reindex_files'}->{$existing_filename} = 1;
1336    }
1337
1338    }
1339
1340   
1341    # By this point full_prev_all_files contains the files
1342    # mentioned in archiveinf-src.db but are not in the 'import'
1343    # folder (or whatever was specified through -importdir ...)
1344
1345    # This list can contain files that were created in the 'tmp' or
1346    # 'cache' areas (such as screen-size and thumbnail images).
1347    #
1348    # In building the final list of files to delete, we test to see if
1349    # it exists on the filesystem and if it does (unusual for a "normal"
1350    # file in import, but possible in the case of 'tmp' files),
1351    # supress it from going into the final list
1352
1353    my $collectdir = $ENV{'GSDLCOLLECTDIR'};
1354
1355    my @deleted_files = values %$full_prev_all_files;
1356    map { my $curr_file = $_;
1357      my $full_curr_file = $curr_file;
1358
1359      if (!&FileUtils::isFilenameAbsolute($curr_file)) {
1360          # add in import dir to make absolute
1361
1362          $full_curr_file = &FileUtils::filenameConcatenate($collectdir,$curr_file);
1363      }
1364
1365
1366      if (!-e $full_curr_file) {
1367          $block_hash->{'deleted_files'}->{$curr_file} = 1;
1368      }
1369      } @deleted_files;
1370
1371
1372
1373}
1374
1375
1376# this is used to delete "deleted" docs, and to remove old versions of "changed" docs
1377# $mode is 'delete' or 'reindex'
1378sub mark_docs_for_deletion
1379{
1380    my ($archive_info,$block_hash,$deleted_files,$archivedir,$verbosity,$mode) = @_;
1381
1382    my $mode_text = "deleted from index";
1383    if ($mode eq "reindex") {
1384    $mode_text = "reindexed";
1385    }
1386
1387    # Get the infodbtype value for this collection from the arcinfo object
1388    my $infodbtype = $archive_info->{'infodbtype'};
1389
1390    my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archivedir);
1391    my $arcinfo_src_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-src", $archivedir);
1392
1393
1394    # record files marked for deletion in arcinfo
1395    foreach my $file (@$deleted_files) {
1396    # use 'archiveinf-src' info database file to look up all the OIDs
1397    # that this file is used in (note in most cases, it's just one OID)
1398   
1399    my $relfile = &util::abspath_to_placeholders($file);
1400
1401    my $src_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_src_filename, $relfile);
1402    my $oids = $src_rec->{'oid'};
1403    my $file_record_deleted = 0;
1404
1405    # delete the src record
1406    my $src_infodb_file_handle = &dbutil::open_infodb_write_handle($infodbtype, $arcinfo_src_filename, "append");
1407    &dbutil::delete_infodb_entry($infodbtype, $src_infodb_file_handle, $relfile);
1408    &dbutil::close_infodb_write_handle($infodbtype, $src_infodb_file_handle);
1409
1410
1411    foreach my $oid (@$oids) {
1412
1413        # find the source doc (the primary file that becomes this oid)
1414        my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $oid);
1415        my $doc_source_file = $doc_rec->{'src-file'}->[0];
1416        $doc_source_file = &util::placeholders_to_abspath($doc_source_file);
1417
1418        if (!&FileUtils::isFilenameAbsolute($doc_source_file)) {
1419        $doc_source_file = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},$doc_source_file);
1420        }
1421
1422        if ($doc_source_file ne $file) {
1423        # its an associated or metadata file
1424       
1425        # mark source doc for reimport as one of its assoc files has changed or deleted
1426        $block_hash->{'reindex_files'}->{$doc_source_file} = 1;
1427       
1428        }
1429        my $curr_status = $archive_info->get_status_info($oid);
1430        if (defined($curr_status) && (($curr_status ne "D"))) {
1431        if ($verbosity>1) {
1432            print STDERR "$oid ($doc_source_file) marked to be $mode_text on next buildcol.pl\n";
1433        }
1434        # mark oid for deletion (it will be deleted or reimported)
1435        $archive_info->set_status_info($oid,"D");
1436        my $val = &dbutil::read_infodb_rawentry($infodbtype, $arcinfo_doc_filename, $oid);
1437        $val =~ s/^<index-status>(.*)$/<index-status>D/m;
1438
1439        my $val_rec = &dbutil::convert_infodb_string_to_hash($infodbtype,$val);
1440        my $doc_infodb_file_handle = &dbutil::open_infodb_write_handle($infodbtype, $arcinfo_doc_filename, "append");
1441
1442        &dbutil::write_infodb_entry($infodbtype, $doc_infodb_file_handle, $oid, $val_rec);
1443        &dbutil::close_infodb_write_handle($infodbtype, $doc_infodb_file_handle);
1444        }
1445    }
1446   
1447    }
1448
1449    # now go through and check that we haven't marked any primary
1450    # files for reindex (because their associated files have
1451    # changed/deleted) when they have been deleted themselves. only in
1452    # delete mode.
1453
1454    if ($mode eq "delete") {
1455    foreach my $file (@$deleted_files) {
1456        if (defined $block_hash->{'reindex_files'}->{$file}) {
1457        delete $block_hash->{'reindex_files'}->{$file};
1458        }
1459    }
1460    }
1461
1462
1463}
1464
1465sub add_dir_contents_to_list {
1466
1467    my ($dirname, $list) = @_;
1468 
1469    # Recur over directory contents.
1470    my (@dir, $subfile);
1471   
1472    # find all the files in the directory
1473    if (!opendir (DIR, $dirname)) {
1474    print STDERR "inexport: WARNING - couldn't read directory $dirname\n";
1475    return -1; # error in processing
1476    }
1477    @dir = readdir (DIR);
1478    closedir (DIR);
1479   
1480    for (my $i = 0; $i < scalar(@dir); $i++) {
1481    my $subfile = $dir[$i];
1482    next if ($subfile =~ m/^\.\.?$/);
1483    next if ($subfile =~ /^\.svn$/);
1484    my $full_file = &FileUtils::filenameConcatenate($dirname, $subfile);
1485    if (-d $full_file) {
1486        &add_dir_contents_to_list($full_file, $list);
1487    } else {
1488        push (@$list, $full_file);
1489    }
1490    }
1491   
1492}
1493
1494   
14951;
Note: See TracBrowser for help on using the browser.