source: main/trunk/greenstone2/perllib/cgiactions/modmetadataaction.pm@ 32782

Last change on this file since 32782 was 32782, checked in by ak19, 5 years ago

Fix for Zeddy: empty lines and annoying spaces are now removed when metadata gets removed from doc xml files. Haven't tested the metadata xml variants, but changed them in symmetry.
File size: 93.8 KB
Line 
1##########################################################################
2#
3# modmetadataaction.pm --
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) 2009 New Zealand Digital Library Project
9#
10# This program is free software; you can redistribute it and/or modify
11# it under the terms of the GNU General Public License as published by
12# the Free Software Foundation; either version 2 of the License, or
13# (at your option) any later version.
14#
15# This program is distributed in the hope that it will be useful,
16# but WITHOUT ANY WARRANTY; without even the implied warranty of
17# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18# GNU General Public License for more details.
19#
20# You should have received a copy of the GNU General Public License
21# along with this program; if not, write to the Free Software
22# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
23#
24###########################################################################
25
26# This is a submodule of metadataaction.pm. It is loaded conditionally
27# upon having the permissions to modify (not just get) metadata.
28# With sufficient permissions, these additional subroutines should be made
29# available to metadataaction.pm
30
31# See http://www.perlmonks.org/?node_id=881761 for splitting module into multiple files
32# and how variables declared with 'our' are used there.
33
34package metadataaction;
35
36use strict;
37
38use cgiactions::baseaction;
39
40use dbutil;
41use ghtml;
42
43use JSON;
44
45
46$metadataaction::modmeta_action_table = #OR: our $modmeta_action_table =
47{
48 #SET METHODS
49 "set-live-metadata" => {
50 'compulsory-args' => [ "d", "metaname", "metavalue" ],
51 'optional-args' => [ ] },
52
53 "set-metadata" => { # generic set-meta function. The 'where' param can be set to any combination of index|archives|import|live. docid d is still compulsory for setting index, archives and live meta
54 'compulsory-args' => [ "metaname", "metavalue" ],
55 'optional-args' => [ "where", "metapos", "metamode", "prevmetavalue", "d", "f" ] },
56
57 "set-index-metadata" => {
58 'compulsory-args' => [ "d", "metaname", "metavalue" ],
59 'optional-args' => [ "metapos", "metamode" ] },
60
61 "set-archives-metadata" => {
62 'compulsory-args' => [ "d", "metaname", "metavalue" ],
63 'optional-args' => [ "metapos", "metamode", "prevmetavalue" ] }, # metamode can be "accumulate", "override",
64
65 "set-import-metadata" => {
66 'compulsory-args' => [ "metaname", "metavalue" ],
67 'optional-args' => [ "d", "f", "metamode", "metapos", "prevmetavalue" ] }, # metamode can be "accumulate", "override", or "unique-id". Also need to add the ability to specify a previous metadata value to overwrite (because we can't use metapos). Metapos now supported, but assumes you are working with a Simple (instead of Complex) collection
68
69 #SET METHODS (ARRAY)
70 "set-metadata-array" => {
71 'compulsory-args' => [ "where", "json" ],
72 'optional-args' => [ ],
73 'help-string' => [
74 'A simple example: metadata-server.pl?a=set-metadata-array&where=archives|index|import&c=demo&json=[{"docid":"HASHc5bce2d6d3e5b04e470ec9","metaname":"Title","metavalue":"Tralalala","metamode":"accumulate"},{"docid":"HASHbe483fa4df4e096335d1c8","metaname":"Title","metavalue":"Lala was here","metapos":0, "metamode":"override"}]',
75
76 'A more complex example: metadata-server.pl?a=set-metadata-array&where=archives|index&c=demo&json=[{"docid":"HASHc5bce2d6d3e5b04e470ec9.1","metatable":[{"metaname":"Title","metavals":["Transformers","Robots in disguise","Autobots"]}],"metamode":"override"},{"docid":"HASHbe483fa4df4e096335d1c8.2","metaname":"Title","metavalue":"Pinky was here","metamode":"accumulate"}]' ] },
77
78# The same examples rewritten for when running the metadata-server.pl script from the commandline:
79
80# the simple example: metadata-server.pl a="set-metadata-array" where="archives|index|import" c="demo" json="[{\"docid\":\"HASHc5bce2d6d3e5b04e470ec9\",\"metaname\":\"Title\",\"metavalue\":\"Tralalala\",\"metamode\":\"accumulate\"},{\"docid\":\"HASHbe483fa4df4e096335d1c8\",\"metaname\":\"Title\",\"metavalue\":\"Lala was here\",\"metapos\":0, \"metamode\":\"override\"}]",
81
82# the more complex example: metadata-server.pl a="set-metadata-array" where="archives|index" c="demo" json="[{\"docid\":\"HASHc5bce2d6d3e5b04e470ec9.1\",\"metatable\":[{\"metaname\":\"Title\",\"metavals\":[\"Transformers\",\"Robots in disguise\",\"Autobots\"]}],\"metamode\":\"override\"},{\"docid\":\"HASHbe483fa4df4e096335d1c8.2\",\"metaname\":\"Title\",\"metavalue\":\"Pinky was here\",\"metamode\":\"accumulate\"}]"
83
84 "set-archives-metadata-array" => {
85 'compulsory-args' => [ "json" ],
86 'optional-args' => [ ] },
87
88 "set-import-metadata-array" => {
89 'compulsory-args' => [ "json" ],
90 'optional-args' => [ ] },
91
92 "set-index-metadata-array" => {
93 'compulsory-args' => [ "json" ],
94 'optional-args' => [ ] },
95
96 "set-live-metadata-array" => {
97 'compulsory-args' => [ "json" ],
98 'optional-args' => [ ] },
99
100 #REMOVE METHODS
101 "remove-import-metadata" => {
102 'compulsory-args' => [ "d", "metaname" ], #TODO: add f argument
103 'optional-args' => [ "metapos", "metavalue", "metamode" ] }, # only provide metapos arg for SIMPLE collections.
104# Metavalue is now an optional arg for remove_import_metadata() based on what the implementation did, which allowed metavalue to be undefined, and if so, used metapos.
105
106 "remove-archives-metadata" => {
107 'compulsory-args' => [ "d", "metaname" ], #TODO: add f argument
108 'optional-args' => [ "metapos", "metavalue", "metamode" ] },
109
110 "remove-live-metadata" => {
111 'compulsory-args' => [ "d", "metaname" ],
112 'optional-args' => [ ] },
113
114 "remove-index-metadata" => {
115 'compulsory-args' => [ "d", "metaname" ],
116 'optional-args' => [ "metapos", "metavalue" ] },
117
118 "remove-metadata" => { # generic remove-meta function. The 'where' param can be set to any combination of index|archives|import|live. docid d is still compulsory for setting index, archives and live meta
119 'compulsory-args' => [ "d", "metaname" ],
120 'optional-args' => [ "where", "metapos", "metavalue", "metamode" ] }, # metamode is optional since remove-metadata can call any of remove_import_meta and remove_archives_meta, remove_index_meta, of which the first two accept metamode as an optional param
121
122 #INSERT METHODS
123 "insert-metadata" => {
124 'compulsory-args' => [ "d", "metaname", "metavalue" ],
125 'optional-args' => [ ] }
126};
127
128
129sub _set_live_metadata
130{
131 my $self = shift @_;
132
133 my $collect = $self->{'collect'};
134 my $gsdl_cgi = $self->{'gsdl_cgi'};
135 #my $gsdlhome = $self->{'gsdlhome'};
136 my $infodbtype = $self->{'infodbtype'};
137
138 # Obtain the collect dir
139 my $site = $self->{'site'};
140 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
141 ##my $collect_dir = &util::filename_cat($gsdlhome, "collect");
142
143
144 # look up additional args
145 my $docid = $self->{'d'};
146 if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
147 $gsdl_cgi->generate_error("No docid (d=...) specified."); # generates error and dies
148 }
149 my $metavalue = $self->{'metavalue'};
150
151 # Generate the dbkey
152 my $metaname = $self->{'metaname'};
153 my $dbkey = "$docid.$metaname";
154
155 # To people who know $collect_tail please add some comments
156 # Obtain path to the database
157 my $collect_tail = $collect;
158 $collect_tail =~ s/^.*[\/|\\]//;
159 my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
160 my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, "live-$collect_tail", $index_text_directory);
161
162 # Set the new value
163 my $cmd = "gdbmset \"$infodb_file_path\" \"$dbkey\" \"$metavalue\"";
164 my $status = system($cmd);
165 if ($status != 0) {
166 # Catch error if gdbmget failed
167 my $mess = "Failed to set metadata key: $dbkey\n";
168
169 $mess .= "PATH: $ENV{'PATH'}\n";
170 $mess .= "cmd = $cmd\n";
171 $mess .= "Exit status: $status\n";
172 $mess .= "System Error Message: $!\n";
173
174 $gsdl_cgi->generate_error($mess);
175 }
176 else {
177 $gsdl_cgi->generate_ok_message("set-live-metadata successful: Key[$metaname]=$metavalue");
178 }
179
180 #return $status; # in case calling functions have any further use for this
181}
182
183sub set_live_metadata
184{
185 my $self = shift @_;
186
187 my $username = $self->{'username'};
188 my $collect = $self->{'collect'};
189 my $gsdl_cgi = $self->{'gsdl_cgi'};
190
191 if ($baseaction::authentication_enabled) {
192 # Ensure the user is allowed to edit this collection
193 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
194 }
195
196 # Make sure the collection isn't locked by someone else
197 $self->lock_collection($username, $collect);
198
199 $self->_set_live_metadata(@_);
200
201 # Release the lock once it is done
202 $self->unlock_collection($username, $collect);
203}
204
205sub set_index_metadata_entry
206{
207 my $self = shift @_;
208 my ($collect_dir,$collect,$infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode,$prevmetavalue) = @_;
209
210 # To people who know $collect_tail please add some comments
211 # Obtain path to the database
212 my $collect_tail = $collect;
213 $collect_tail =~ s/^.*[\/|\\]//;
214 my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
215 my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, $collect_tail, $index_text_directory);
216
217# print STDERR "**** infodb file path = $infodb_file_path\n";
218# print STDERR "***** infodb type = $infodbtype\n";
219
220 # Read the docid entry
221 my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $infodb_file_path, $docid);
222
223 # Set the metadata value
224 if (defined $metapos) {
225 # if metamode=accumulate AND metapos, warn user and then use metapos
226 if (defined $metamode && $metamode eq "accumulate") {
227 print STDERR "**** Warning: metamode is set to accumulate yet metapos is also provided for $docid\n";
228 print STDERR "**** Proceeding by using metapos\n";
229 }
230 $doc_rec->{$metaname}->[$metapos] = $metavalue;
231 }
232 elsif (defined $prevmetavalue) {
233 my $array = $doc_rec->{$metaname};
234 my $length = @$array;
235
236 my $found = 0;
237 for (my $i = 0; $i < $length; $i++){
238 if(defined $doc_rec->{$metaname}->[$i] && $doc_rec->{$metaname}->[$i] eq $prevmetavalue){
239 $doc_rec->{$metaname}->[$i] = $metavalue;
240 $found = 1;
241 last;
242 }
243 }
244
245 if($found == 0){
246 $doc_rec->{$metaname} = [ $metavalue ];
247 }
248 }
249 elsif (defined $metamode && $metamode eq "override") {
250 $doc_rec->{$metaname} = [ $metavalue ];
251 }
252 else { # default for index was to override, but because accumulate is less destructive,
253 # and because accumulate is the default for archives and import, that's the new default for index too
254 if(defined $doc_rec->{$metaname}) {
255 push(@{$doc_rec->{$metaname}}, $metavalue); # accumulate the value for that metaname
256 } else {
257 $doc_rec->{$metaname} = [ $metavalue ];
258 }
259 }
260
261 my $status = &dbutil::set_infodb_entry($infodbtype, $infodb_file_path,$docid,$doc_rec);
262
263 return $status;
264
265}
266
267sub _set_import_metadata
268{
269 my $self = shift @_;
270
271 my $collect = $self->{'collect'};
272 my $gsdl_cgi = $self->{'gsdl_cgi'};
273 my $infodbtype = $self->{'infodbtype'};
274# my $gsdlhome = $self->{'gsdlhome'};
275
276 # Obtain the collect and archive dir
277 my $site = $self->{'site'};
278 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
279 ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
280 my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
281
282 # look up additional args
283 # want either d= or f=
284 my $docid = $self->{'d'};
285 my ($docid_root,$docid_secnum);
286 if(defined $docid) {
287 ($docid_root,$docid_secnum) = ($docid =~ m/^(.*?)(\..*)?$/);
288 # as yet no support for setting subsection metadata in metadata.xml
289 if ((defined $docid_secnum) && ($docid_secnum !~ m/^\s*$/)) {
290 $gsdl_cgi->generate_message("*** No support yet for setting import metadata at subsections level.\n");
291 return;
292 }
293 }
294
295 my $import_file = $self->{'f'};
296 if ((!defined $docid || $docid =~ m/^\s*$/) && (!defined $import_file || $import_file =~ m/^\s*$/)) {
297 $gsdl_cgi->generate_error("No docid (d=...) or import file (f=) specified."); # at least d or f must be specified
298 }
299
300 # Get the parameters and set default mode to "accumulate"
301 my $metaname = $self->{'metaname'};
302 my $metavalue = $self->{'metavalue'};
303## $metavalue =~ s/&amp;lt;(.*?)&amp;gt;/<$1>/g;
304 $metavalue =~ s/&lt;(.*?)&gt;/<$1>/g;
305
306 my $metamode = $self->{'metamode'};
307 if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
308 # make "accumulate" the default (less destructive, as it won't actually
309 # delete any existing values)
310 $metamode = "accumulate";
311 }
312
313 # adding metapos and prevmetavalue support to import_metadata subroutines
314 my $metapos = $self->{'metapos'}; # don't force undef to 0. Undef has meaning when metamode=override
315 my $prevmetavalue = $self->{'prevmetavalue'};
316 $metapos = undef if(defined $metapos && ($metapos =~ m/^\s*$/));
317 $prevmetavalue = undef if(defined $prevmetavalue && ($prevmetavalue =~ m/^\s*$/));
318
319 my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
320 my $metadata_xml_filename = $self->set_import_metadata_entry($gsdl_cgi, $arcinfo_doc_filename, $infodbtype, $docid_root, $metaname, $metapos,$metavalue, $metamode,$prevmetavalue, $collect, $collect_dir); # at this point, docid_root = docid
321
322 my $mess = "set-import-metadata successful: Key[$docid] -> $metadata_xml_filename\n";
323 $mess .= " $metaname";
324 $mess .= " = $metavalue";
325 $mess .= " ($metamode)\n";
326
327 $gsdl_cgi->generate_ok_message($mess);
328
329 #return $status; # in case calling functions have any further use for this
330}
331
332# the version of set_archives_meta that doesn't do authentication
333sub _set_archives_metadata
334{
335 my $self = shift @_;
336
337 my $collect = $self->{'collect'};
338 my $gsdl_cgi = $self->{'gsdl_cgi'};
339 my $infodbtype = $self->{'infodbtype'};
340
341 # Obtain the collect and archive dir
342 my $site = $self->{'site'};
343 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
344 my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
345
346 # look up additional args
347 my $docid = $self->{'d'};
348 my $metaname = $self->{'metaname'};
349 my $metavalue = $self->{'metavalue'};
350 my $prevmetavalue = $self->{'prevmetavalue'};
351
352 my $metapos = $self->{'metapos'}; # don't force undef to 0. Undef has meaning when metamode=override
353 # Don't append "|| undef", since if metapos=0 it will then be set to undef
354
355 $metapos = undef if(defined $metapos && ($metapos =~ m/^\s*$/));
356 $prevmetavalue = undef if(defined $prevmetavalue && ($prevmetavalue =~ m/^\s*$/));
357
358 my $metamode = $self->{'metamode'};
359 if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
360 # make "accumulate" the default (less destructive, as it won't actually
361 # delete any existing values)
362 $metamode = "accumulate";
363 }
364
365 my $status = $self->set_archives_metadata_entry($gsdl_cgi,$archive_dir, $collect_dir,$collect, $infodbtype,$docid,
366 $metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
367
368 if ($status == 0) {
369 my $mess = "set-archives-metadata successful: Key[$docid]\n";
370 $mess .= " $metaname";
371 $mess .= "->[$metapos]" if (defined $metapos);
372 $mess .= " = $metavalue";
373 $mess .= " ($metamode)\n";
374
375 $gsdl_cgi->generate_ok_message($mess);
376 }
377 else {
378 my $mess .= "Failed to set archives metadata key: $docid\n";
379 $mess .= "Exit status: $status\n";
380 if(defined $self->{'error_msg'}) {
381 $mess .= "Error Message: $self->{'error_msg'}\n";
382 } else {
383 $mess .= "System Error Message: $!\n";
384 }
385 $mess .= "-" x 20 . "\n";
386
387 $gsdl_cgi->generate_error($mess);
388 }
389
390 #return $status; # in case calling functions have any further use for this
391}
392
393
394# the version of set_index_meta that doesn't do authentication
395sub _set_index_metadata
396{
397 print STDERR "START SET INDEX METADATA\n";
398 my $self = shift @_;
399
400 my $collect = $self->{'collect'};
401 my $gsdl_cgi = $self->{'gsdl_cgi'};
402
403 my $site = $self->{'site'};
404 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
405
406 # look up additional args
407 my $docid = $self->{'d'};
408 my $metaname = $self->{'metaname'};
409 my $metapos = $self->{'metapos'}; # undef has meaning
410 my $metavalue = $self->{'metavalue'};
411 my $infodbtype = $self->{'infodbtype'};
412 my $metamode = $self->{'metamode'};
413 my $prevmetavalue = $self->{'prevmetavalue'};
414
415 $metapos = undef if(defined $metapos && ($metapos =~ m/^\s*$/));
416 $prevmetavalue = undef if(defined $prevmetavalue && ($prevmetavalue =~ m/^\s*$/));
417
418 print STDERR "SETTING INDEX METADATA ENTRY\n";
419 my $status = $self->set_index_metadata_entry($collect_dir,$collect,$infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
420 print STDERR "DONE SETTING INDEX METADATA ENTRY\n";
421 if ($status != 0) {
422 # Catch error if set infodb entry failed
423 my $mess = "Failed to set metadata key: $docid\n";
424
425 $mess .= "PATH: $ENV{'PATH'}\n";
426 $mess .= "Exit status: $status\n";
427 $mess .= "System Error Message: $!\n";
428
429 $gsdl_cgi->generate_error($mess);
430 }
431 else {
432 my $mess = "set-index-metadata successful: Key[$docid]\n";
433 $mess .= " $metaname";
434 $mess .= "->[$metapos]" if (defined $metapos);
435 $mess .= " = $metavalue\n";
436
437 $gsdl_cgi->generate_ok_message($mess);
438 }
439
440 print STDERR "END SET INDEX METADATA\n";
441 #return $status; # in case calling functions have any further use for this
442}
443
444sub set_index_metadata
445{
446 my $self = shift @_;
447
448 my $username = $self->{'username'};
449 my $collect = $self->{'collect'};
450 my $gsdl_cgi = $self->{'gsdl_cgi'};
451 #my $gsdlhome = $self->{'gsdlhome'};
452
453 if ($baseaction::authentication_enabled) {
454 # Ensure the user is allowed to edit this collection
455 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
456 }
457
458 my $site = $self->{'site'};
459 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
460
461 $gsdl_cgi->checked_chdir($collect_dir);
462
463 # Obtain the collect dir
464 ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
465
466 # Make sure the collection isn't locked by someone else
467 $self->lock_collection($username, $collect);
468
469 $self->_set_index_metadata(@_);
470
471 # Release the lock once it is done
472 $self->unlock_collection($username, $collect);
473}
474
475# call this to set the metadata for a combination of dirs archives, import or index, or live
476# if none specified, defaults to index which was the original behaviour of set_metadata.
477sub set_metadata
478{
479 my $self = shift @_;
480
481 # Testing that not defining a variable, setting it to "" or to " " all return false
482 # >perl -e 'my $whichdirs=""; if($whichdirs) {print "$whichdirs\n"};'
483
484 my $where = $self->{'where'};
485 if(!$where || ($where =~ m/^\s*$/)) {
486 $self->set_index_metadata(@_); # call the full version of set_index_meta for the default behaviour
487 return;
488 }
489
490 # authenticate and lock collection once, even if processing multiple dirs
491 my $username = $self->{'username'};
492 my $collect = $self->{'collect'};
493 my $gsdl_cgi = $self->{'gsdl_cgi'};
494
495 if ($baseaction::authentication_enabled) {
496 # Ensure the user is allowed to edit this collection
497 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
498 }
499
500 if($where =~ m/index/) {
501 my $site = $self->{'site'};
502 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
503 $gsdl_cgi->checked_chdir($collect_dir);
504 }
505
506 # Make sure the collection isn't locked by someone else
507 $self->lock_collection($username, $collect);
508
509
510 # now at last can set the metadata. $where can specify multiple
511 # $where is of the form: import|archives|index, or a subset thereof
512
513 #my @whichdirs = split('\|', $where);
514
515 # just check whether $where contains import/archives/index/live in turn, and
516 # for each case, process it accordingly
517 if($where =~ m/import/) {
518 $self->_set_import_metadata(@_);
519 }
520
521 if($where =~ m/archives/) {
522
523 # look up docID arg which is optional to set_metadata because it's optional
524 # to set_import, but which is compulsory to set_archives_metadata
525 my $docid = $self->{'d'};
526 if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
527 $gsdl_cgi->generate_error("No docid (d=...) specified."); # generates error and dies
528 }
529 # we have a docid, so can set archives meta
530 $self->_set_archives_metadata(@_);
531 }
532
533 if($where =~ m/index/) {
534
535 # look up docID arg which is optional to set_metadata because it's optional
536 # to set_import, but which is compulsory to set_archives_metadata
537 my $docid = $self->{'d'};
538 if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
539 $gsdl_cgi->generate_error("No docid (d=...) specified.");
540 }
541 # we have a docid, so can set index meta
542 $self->_set_index_metadata(@_);
543 }
544
545 if($where =~ m/live/) {
546 $self->_set_live_metadata(@_); # docid param, d, is compulsory, but is checked for in subroutine
547 }
548
549 # Release the lock once it is done
550 $self->unlock_collection($username, $collect);
551}
552
553sub set_metadata_array
554{
555 my $self = shift @_;
556
557 my $where = $self->{'where'};
558 if(!$where || ($where =~ m/^\s*$/)) {
559 $self->set_index_metadata_array(@_); # default behaviour is the full version of set_index_meta_array
560 return;
561 }
562
563 my $username = $self->{'username'};
564 my $collect = $self->{'collect'};
565 my $gsdl_cgi = $self->{'gsdl_cgi'};
566
567 if ($baseaction::authentication_enabled) {
568 # Ensure the user is allowed to edit this collection
569 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
570 }
571
572 # Not sure if the checked_chdir is necessary, since lock_collection also does a chdir
573 # But including the stmt during this code reorganisation to preserve as-is what used to happen
574 my $site = $self->{'site'};
575 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
576 $gsdl_cgi->checked_chdir($collect_dir);
577
578 # Make sure the collection isn't locked by someone else
579 $self->lock_collection($username, $collect);
580
581 if($where =~ m/import/) {
582 $self->_set_import_metadata_array(@_);
583 }
584 if($where =~ m/archives/) {
585 $self->_set_archives_metadata_array(@_);
586 }
587 if($where =~ m/index/) {
588 $self->_set_index_metadata_array(@_);
589 }
590 if($where =~ m/live/) {
591 $self->_set_live_metadata_array(@_);
592 }
593
594 # Release the lock once it is done
595 $self->unlock_collection($username, $collect);
596}
597
598sub _set_index_metadata_array
599{
600 my $self = shift @_;
601
602 my $collect = $self->{'collect'};
603 my $gsdl_cgi = $self->{'gsdl_cgi'};
604 my $site = $self->{'site'};
605 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
606
607
608 # look up additional args
609
610 my $infodbtype = $self->{'infodbtype'};
611
612 my $json_str = $self->{'json'};
613 my $doc_array = decode_json $json_str;
614
615
616 my $global_status = 0;
617 my $global_mess = "";
618
619 my @all_docids = ();
620
621 foreach my $doc_array_rec ( @$doc_array ) {
622
623 my $status = -1;
624 my $docid = $doc_array_rec->{'docid'};
625
626 push(@all_docids,$docid);
627
628 my $metaname = $doc_array_rec->{'metaname'};
629 if(defined $metaname) {
630 my $metapos = $doc_array_rec->{'metapos'}; # can legitimately be undef
631 my $metavalue = $doc_array_rec->{'metavalue'};
632 my $metamode = $doc_array_rec->{'metamode'} || $self->{'metamode'};
633
634 $status = $self->set_index_metadata_entry($collect_dir,$collect,$infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode);
635 } elsif (defined $doc_array_rec->{'metatable'}) { # if no metaname, we expect a metatable
636 my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
637
638 foreach my $metatable_rec ( @$metatable ) { # the subarray metatable is an array of hashmaps
639 $metaname = $metatable_rec->{'metaname'};
640 my $metamode = $metatable_rec->{'metamode'} || $doc_array_rec->{'metamode'} || $self->{'metamode'};
641 my $metapos = undef;
642 my $metavals = $metatable_rec->{'metavals'}; # a sub-subarray
643
644 foreach my $metavalue ( @$metavals ) { # metavals is an array
645 $status = $self->set_index_metadata_entry($collect_dir,$collect,$infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode); # how do we use metamode in set_meta_entry?
646 if($metamode eq "override") { # now, having overridden the metavalue for the first,
647 # need to accumulate subsequent metavals for this metaname, else the just-assigned
648 # metavalue for this metaname will be lost
649 $metamode = "accumulate";
650 }
651 }
652 }
653 }
654
655 if ($status != 0) {
656 # Catch error if set infodb entry failed
657 $global_status = $status;
658 $global_mess .= "Failed to set metadata key: $docid\n";
659 $global_mess .= "Exit status: $status\n";
660 $global_mess .= "System Error Message: $!\n";
661 $global_mess .= "-" x 20;
662 }
663 }
664
665 if ($global_status != 0) {
666 $global_mess .= "PATH: $ENV{'PATH'}\n";
667 $gsdl_cgi->generate_error($global_mess);
668 }
669 else {
670 my $mess = "set-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
671 $gsdl_cgi->generate_ok_message($mess);
672 }
673}
674
675sub set_index_metadata_array
676{
677 my $self = shift @_;
678
679 my $username = $self->{'username'};
680 my $collect = $self->{'collect'};
681 my $gsdl_cgi = $self->{'gsdl_cgi'};
682# my $gsdlhome = $self->{'gsdlhome'};
683
684 if ($baseaction::authentication_enabled) {
685 # Ensure the user is allowed to edit this collection
686 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
687 }
688
689 my $site = $self->{'site'};
690 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
691
692 $gsdl_cgi->checked_chdir($collect_dir);
693
694 # Obtain the collect dir
695 ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
696
697 # Make sure the collection isn't locked by someone else
698 $self->lock_collection($username, $collect);
699
700 $self->_set_index_metadata_array(@_);
701
702 # Release the lock once it is done
703 $self->unlock_collection($username, $collect);
704}
705
706# experimental, newly added in and untested
707sub _set_live_metadata_array
708{
709 my $self = shift @_;
710
711 my $collect = $self->{'collect'};
712 my $gsdl_cgi = $self->{'gsdl_cgi'};
713
714 my $site = $self->{'site'};
715 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
716
717
718 # look up additional args
719 my $infodbtype = $self->{'infodbtype'};
720 # To people who know $collect_tail please add some comments
721 # Obtain path to the database
722 my $collect_tail = $collect;
723 $collect_tail =~ s/^.*[\/|\\]//;
724 my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
725 my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, "live-$collect_tail", $index_text_directory);
726
727
728 my $json_str = $self->{'json'};
729 my $doc_array = decode_json $json_str;
730
731
732 my $global_status = 0;
733 my $global_mess = "";
734
735 my @all_docids = ();
736
737
738 foreach my $doc_array_rec ( @$doc_array ) {
739
740 my $status = -1;
741 my $docid = $doc_array_rec->{'docid'};
742
743 push(@all_docids,$docid);
744
745 my $metaname = $doc_array_rec->{'metaname'};
746 if(defined $metaname) {
747 my $dbkey = "$docid.$metaname";
748 my $metavalue = $doc_array_rec->{'metavalue'};
749
750 # Set the new value
751 my $cmd = "gdbmset \"$infodb_file_path\" \"$dbkey\" \"$metavalue\"";
752 $status = system($cmd);
753
754 } elsif (defined $doc_array_rec->{'metatable'}) { # if no metaname, we expect a metatable
755 my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
756 foreach my $metatable_rec ( @$metatable ) {
757 $metaname = $metatable_rec->{'metaname'};
758 my $dbkey = "$docid.$metaname";
759
760 my $metavals = $metatable_rec->{'metavals'}; # a sub-subarray
761 foreach my $metavalue ( @$metavals ) {
762 my $cmd = "gdbmset \"$infodb_file_path\" \"$dbkey\" \"$metavalue\"";
763 $status = system($cmd);
764 }
765 }
766
767 }
768
769 if ($status != 0) {
770 # Catch error if gdbmget failed
771 $global_status = $status;
772 $global_mess .= "Failed to set metadata key: $docid\n"; # $dbkey
773 $global_mess .= "Exit status: $status\n";
774 $global_mess .= "System Error Message: $!\n";
775 $global_mess .= "-" x 20;
776 }
777 }
778
779 if ($global_status != 0) {
780 $global_mess .= "PATH: $ENV{'PATH'}\n";
781 $gsdl_cgi->generate_error($global_mess);
782 }
783 else {
784 my $mess = "set-live-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
785 $gsdl_cgi->generate_ok_message($mess);
786 }
787}
788
789sub set_live_metadata_array
790{
791 my $self = shift @_;
792
793 my $username = $self->{'username'};
794 my $collect = $self->{'collect'};
795 my $gsdl_cgi = $self->{'gsdl_cgi'};
796
797 if ($baseaction::authentication_enabled) {
798 # Ensure the user is allowed to edit this collection
799 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
800 }
801
802 my $site = $self->{'site'};
803 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
804
805 $gsdl_cgi->checked_chdir($collect_dir);
806
807 # Make sure the collection isn't locked by someone else
808 $self->lock_collection($username, $collect);
809
810 $self->_set_live_metadata_array(@_);
811
812 # Release the lock once it is done
813 $self->unlock_collection($username, $collect);
814}
815
816
817sub dxml_metadata
818{
819 my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
820 my $metaname = $parser->{'parameters'}->{'metaname'};
821 my $metamode = $parser->{'parameters'}->{'metamode'};
822
823###!!! print STDERR "**** Processing closing </Metadata> tag\n";
824
825 my $opt_doc_secnum = $parser->{'parameters'}->{'secnum'};
826
827 # Find the right metadata tag and checks if we are going to
828 # override it
829 #
830 # Note: This over writes the first metadata block it
831 # encountered. If there are multiple Sections in the doc.xml, it
832 # might not behave as you would expect
833
834 my $curr_secnum = $parser->{'parameters'}->{'curr_section_num'};
835## print STDERR "**** checking $opt_doc_secnum <=> $curr_secnum\n" if (defined $opt_doc_secnum);
836## print STDERR "**** metamode = $metamode\n";
837
838 if ((!defined $opt_doc_secnum) || ($opt_doc_secnum eq $curr_secnum))
839 {
840 my $name_attr = $attrHash->{'name'};
841 # print STDOUT "*** testing: $name_attr eq $metaname ? and $metamode eq override ?\n";
842 if (($name_attr eq $metaname) && ($metamode eq "override"))
843 {
844 if (!defined $parser->{'parameters'}->{'poscount'})
845 {
846 $parser->{'parameters'}->{'poscount'} = 0;
847 }
848 else
849 {
850 $parser->{'parameters'}->{'poscount'}++;
851 }
852
853 if (defined $parser->{'parameters'}->{'metapos'}) { # if pos defined, ignore any prevmeta if also defined, and just wait for pos match
854
855 if ($parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'}) {
856 ###print STDERR "#### got pos match!!\n";
857 # Get the value and override the current value
858 my $metavalue = $parser->{'parameters'}->{'metavalue'};
859 $attrHash->{'_content'} = $metavalue;
860
861 # Don't want it to wipe out any other pieces of metadata
862 $parser->{'parameters'}->{'metamode'} = "done";
863 }
864 }
865 elsif (defined $parser->{'parameters'}->{'prevmetavalue'}) { # if no pos defined but prevmeta defined, then wait for prevmeta match
866 if ($parser->{'parameters'}->{'prevmetavalue'} eq $attrHash->{'_content'}) {
867 ###print STDERR "### prev meta value matches\n";
868 my $metavalue = $parser->{'parameters'}->{'metavalue'};
869 $attrHash->{'_content'} = $metavalue;
870 $parser->{'parameters'}->{'metamode'} = "done";
871
872 }
873 }
874 # Note that before commit rev 32076, if neither metapos nor prevmeta defined, then code tried to set pos 0 for that metaname to the new value
875 # However that is not how the code works now (and couldn't get it to work that way yet after rev 32076).
876 # What it does now, and what Dr Bainbridge approved of is that:
877 # If no pos or prevmeta defined, then on override, the metavalue set for the specificed metaname becomes the ONLY instance of that metaname
878 # So any earlier instances of that metaname is removed and replaced by the single new instance.
879 }
880 }
881
882 # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
883 return [$tagname => $attrHash];
884}
885
886# This method exists purely for catching invalid section numbers that the client
887# requested to edit. Once the parser has reached the end (the final </Archive> tag),
888# we've seen all the Sections in the doc.xml, and none of their section nums matched
889# if the metamode has not been set to 'done' by then.
890sub dxml_archive
891{
892 my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
893 my $metamode = $parser->{'parameters'}->{'metamode'};
894
895 my $curr_secnum = $parser->{'parameters'}->{'curr_section_num'};
896 my $opt_doc_secnum = $parser->{'parameters'}->{'secnum'};
897
898 # print STDERR "@@@ $tagname Processing a closing </Archive> tag [$curr_secnum|$opt_doc_secnum]\n";
899
900 if ($metamode ne "done" && $curr_secnum ne $opt_doc_secnum) {
901 #print STDERR "@@@ $tagname Finished processing FINAL Section.\n";
902
903 my $metaname = $parser->{'parameters'}->{'metaname'};
904 my $metavalue = $parser->{'parameters'}->{'metavalue'};
905
906 print STDERR "@@@ Requested section number $opt_doc_secnum not found.\n";
907 print STDERR "\t(last seen section number in document was $curr_secnum)\n";
908 print STDERR "\tDiscarded metadata value '$metavalue' for meta '$metaname'\n";
909 print STDERR "\tin section $opt_doc_secnum.\n";
910 $parser->{'custom_err_msg'} = "Requested section number $opt_doc_secnum not found.";
911 }
912
913 # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
914 return [$tagname => $attrHash];
915}
916
917sub dxml_description
918{
919 my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
920 my $metamode = $parser->{'parameters'}->{'metamode'};
921
922 my $curr_secnum = $parser->{'parameters'}->{'curr_section_num'};
923 my $opt_doc_secnum = $parser->{'parameters'}->{'secnum'} || "";
924
925###!!! print STDERR "**** Processing a closing </Description> tag \n";
926 # print STDERR "@@@ $tagname Processing a closing </Description> tag [$curr_secnum|$opt_doc_secnum]\n";
927
928 # Accumulate the metadata
929
930 # We'll be accumulating metadata at this point if we haven't found and therefore
931 # haven't processed the metadata yet.
932 # For subsections, this means that if we're at a matching subsection, but haven't
933 # found the correct metaname to override in that subsection, we accumulate it as new
934 # meta in the subsection by adding it to the current description.
935 # If there's no subsection info for the metadata, it will accumulate at the top level
936 # section description if we hadn't found a matching metaname to override at this point.
937
938 # Both curr_secnum and opt_doc_secnum can be "". In the former case, it means we're now
939 # at the toplevel section. In the latter case, it means we want to process meta in the
940 # toplevel section. So the eq check between the values below will work in all cases.
941
942 # The only time this won't work is if an opt_doc_secnum beyond the section numbers of
943 # this document has been provided. In that case, the metadata for that opt_doc_secnum
944 # won't get attached/accumulated to any part of the doc, not even its top-level section.
945
946 if ($curr_secnum eq $opt_doc_secnum
947 && ($metamode eq "accumulate" || $metamode eq "override")) {
948 # note, metamode would have been set to done if a metadata operation had already succeeded.
949 # If we get here, we still have the metadata to append.
950 if ($metamode eq "override") {
951 print "Got to end of <Description> block. No metadata value to override. Switching 'metamode' to accumulate\n";
952 }
953
954 # If we get to here and metamode is override, this means there
955 # was no existing value to overide => treat as an append operation
956
957 # Tack a new metadata tag on to the end of the <Metadata>+ block
958 my $metaname = $parser->{'parameters'}->{'metaname'};
959 my $metavalue = $parser->{'parameters'}->{'metavalue'};
960
961 my $metadata_attr = {
962 '_content' => $metavalue,
963 'name' => $metaname,
964 'mode' => "accumulate"
965 };
966
967 my $append_metadata = [ "Metadata" => $metadata_attr ];
968 my $description_content = $attrHash->{'_content'};
969 print "Appending metadata to doc.xml\n";
970
971 if (ref($description_content)) {
972 # got some existing interesting nested content
973 push(@$description_content, " ", $append_metadata, "\n ");
974 }
975 else {
976 #description_content is most likely a string such as "\n"
977 $attrHash->{'_content'} = [$description_content, " ", $append_metadata ,"\n" ];
978 }
979
980 $parser->{'parameters'}->{'metamode'} = "done";
981 }
982 else {
983 # metamode most likely "done" signifying that it has already found a position to add the metadata to.
984## print STDERR "**** NOT ACCUMULATE?!? \n";
985 }
986 # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
987 return [$tagname => $attrHash];
988}
989
990sub edit_xml_file
991{
992 my $self = shift @_;
993 my ($gsdl_cgi, $filename, $start_rules, $rules, $options) = @_;
994
995 # use XML::Rules to add it in (read in and out again)
996 my $parser = XML::Rules->new(start_rules => $start_rules,
997 rules => $rules,
998 style => 'filter',
999 output_encoding => 'utf8' );
1000
1001 my $xml_in = "";
1002 if (!open(MIN,"<$filename")) {
1003 $gsdl_cgi->generate_error("Unable to read in $filename: $!");
1004 }
1005 else {
1006 # Read all the text in
1007 my $line;
1008 while (defined ($line=<MIN>)) {
1009 $xml_in .= $line if($line !~ m/^\s*$/); # preserve all but empty lines
1010 }
1011 close(MIN);
1012
1013 my $MOUT;
1014 if (!open($MOUT,">$filename")) {
1015 $gsdl_cgi->generate_error("Unable to write out to $filename: $!");
1016 }
1017 else {
1018 # Matched lines will get handled by the call backs
1019## my $xml_out = "";
1020
1021 binmode($MOUT,":utf8");
1022 $parser->filter($xml_in,$MOUT, $options);
1023
1024# binmode(MOUT,":utf8");
1025# print MOUT $xml_out;
1026 close($MOUT);
1027 }
1028 }
1029
1030 # copy across any custom error information that was stored during parsing
1031 $self->{'error_msg'} = $parser->{'custom_err_msg'} if(defined $parser->{'custom_err_msg'});
1032}
1033
1034sub edit_doc_xml
1035{
1036 my $self = shift @_;
1037 my ($gsdl_cgi, $doc_xml_filename, $metaname, $metavalue, $metapos, $metamode, $opt_secnum, $prevmetavalue) = @_;
1038
1039 my $info_mess = <<RAWEND;
1040****************************
1041 edit_doc_xml()
1042****************************
1043RAWEND
1044
1045 $info_mess .= " doc_xml_filename = $doc_xml_filename\n" if defined($doc_xml_filename);
1046 $info_mess .= " metaname = $metaname\n" if defined($metaname);
1047 $info_mess .= " metapos = $metapos\n" if defined($metapos);
1048 $info_mess .= " metavalue = $metavalue\n" if defined($metavalue);
1049 $info_mess .= " metamode = $metamode\n" if defined($metamode);
1050 $info_mess .= " opt_secnum = $opt_secnum\n" if defined($opt_secnum);
1051 $info_mess .= " prevmetaval = $prevmetavalue\n" if defined($prevmetavalue);
1052
1053 $info_mess .= "****************************\n";
1054
1055 $gsdl_cgi->generate_message($info_mess);
1056
1057 # To monitor which section/subsection number we are in
1058 my @start_rules =
1059 ( 'Section' => \&dxml_start_section );
1060
1061 # use XML::Rules to add it in (read in and out again)
1062 # Set the call back functions
1063 my @rules =
1064 ( _default => 'raw',
1065 'Metadata' => \&dxml_metadata,
1066 'Description' => \&dxml_description,
1067 'Archive' => \&dxml_archive); # just for catching errors at end
1068
1069 # Sets the parameters
1070 my $options = { 'metaname' => $metaname,
1071 'metapos' => $metapos,
1072 'metavalue' => $metavalue,
1073 'metamode' => $metamode,
1074 'prevmetavalue' => $prevmetavalue };
1075
1076 if (defined $opt_secnum) {
1077 $options->{'secnum'} = $opt_secnum;
1078 }
1079
1080 $self->edit_xml_file($gsdl_cgi,$doc_xml_filename,\@start_rules,\@rules,$options);
1081}
1082
1083sub set_archives_metadata_entry
1084{
1085 my $self = shift @_;
1086 my ($gsdl_cgi, $archive_dir, $collect_dir, $collect, $infodbtype, $docid, $metaname, $metapos, $metavalue, $metamode, $prevmetavalue) = @_;
1087
1088 my $info_mess = <<RAWEND;
1089****************************
1090 set_archives_metadata_entry()
1091****************************
1092RAWEND
1093
1094 $info_mess .= " archive_dir = $archive_dir\n" if defined($archive_dir);
1095 $info_mess .= " collect_dir = $collect_dir\n" if defined($collect_dir);
1096 $info_mess .= " collect = $collect\n" if defined($collect);
1097 $info_mess .= " infodbtype = $infodbtype\n" if defined($infodbtype);
1098 $info_mess .= " docid = $docid\n" if defined($docid);
1099 $info_mess .= " metaname = $metaname\n" if defined($metaname);
1100 $info_mess .= " metapos = $metapos\n" if defined($metapos);
1101 $info_mess .= " metavalue = $metavalue\n" if defined($metavalue);
1102 $info_mess .= " metamode = $metamode\n" if defined($metamode);
1103 $info_mess .= " prevmetaval = $prevmetavalue\n" if defined($prevmetavalue);
1104
1105 $info_mess .= "****************************\n";
1106
1107 $gsdl_cgi->generate_message($info_mess);
1108
1109 # Obtain the doc.xml path for the specified docID
1110 my ($docid_root,$docid_secnum) = ($docid =~ m/^(.*?)(\..*)?$/);
1111
1112 my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
1113 my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid_root);
1114 my $doc_xml_file = $doc_rec->{'doc-file'}->[0];
1115
1116 # The $doc_xml_file is relative to the archives, and now let's get the full path
1117 my $archives_dir = &util::filename_cat($collect_dir,$collect,"archives");
1118 my $doc_xml_filename = &util::filename_cat($archives_dir,$doc_xml_file);
1119
1120 # If we're overriding everything, then $metamode=override combined with $metapos=undefined and $prevmetavalue=undefined
1121 # in which case, we need to remove all metavalues for the metaname at the given (sub)section
1122 # Thereafter, we will finally be setting the overriding metavalue for this metaname
1123 if (!defined $prevmetavalue && !defined $metapos && $metamode eq "override") {
1124 print "override mode with no pos or prev value set. Removing all existing values for $metaname, and setting mode to accumulate for the new value.\n";
1125 # remove all values of $metaname metadata
1126 $self->remove_from_doc_xml($gsdl_cgi, &util::filename_cat($archive_dir, $doc_xml_file), $metaname, undef, undef, $docid_secnum, $metamode);
1127 $metamode = "accumulate";
1128 }
1129 # Edit the doc.xml file with the specified metadata name, value and position.
1130 # TODO: there is a potential problem here as this edit_doc_xml function
1131 # is assuming the simple doc.xml situation where there is only one Section and no SubSections.
1132 # Running import.pl -groupsize will cause this to have multiple sections in one doc.xml
1133
1134 # dxml_metadata method ignores metapos if metamode anything other than override
1135 $self->edit_doc_xml($gsdl_cgi,$doc_xml_filename,
1136 $metaname,$metavalue,$metapos,$metamode,$docid_secnum,$prevmetavalue);
1137
1138 # return 0; # return 0 for now to indicate no error
1139 return (defined $self->{'error_msg'}) ? 1 : 0;
1140}
1141
1142
1143sub set_archives_metadata
1144{
1145 my $self = shift @_;
1146
1147 my $username = $self->{'username'};
1148 my $collect = $self->{'collect'};
1149 my $gsdl_cgi = $self->{'gsdl_cgi'};
1150
1151 if ($baseaction::authentication_enabled) {
1152 # Ensure the user is allowed to edit this collection
1153 $self->authenticate_user($username, $collect);
1154 }
1155
1156 # Make sure the collection isn't locked by someone else
1157 $self->lock_collection($username, $collect);
1158
1159 $self->_set_archives_metadata(@_);
1160
1161 # Release the lock once it is done
1162 $self->unlock_collection($username, $collect);
1163}
1164
1165sub _set_archives_metadata_array
1166{
1167 my $self = shift @_;
1168
1169 my $collect = $self->{'collect'};
1170 my $gsdl_cgi = $self->{'gsdl_cgi'};
1171 my $site = $self->{'site'};
1172 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
1173
1174 # look up additional args
1175
1176 my $infodbtype = $self->{'infodbtype'};
1177
1178 my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
1179
1180 my $json_str = $self->{'json'};
1181 my $doc_array = decode_json $json_str;
1182
1183
1184 my $global_status = 0;
1185 my $global_mess = "";
1186
1187 my @all_docids = ();
1188
1189 foreach my $doc_array_rec ( @$doc_array ) {
1190 my $status = -1;
1191 my $docid = $doc_array_rec->{'docid'};
1192
1193 push(@all_docids,$docid);
1194
1195 my $metaname = $doc_array_rec->{'metaname'};
1196 if(defined $metaname) {
1197
1198 my $metapos = $doc_array_rec->{'metapos'}; # don't force undef to 0. Undef has meaning when metamode=override
1199
1200 my $metamode = $doc_array_rec->{'metamode'} || $self->{'metamode'};
1201 my $metavalue = $doc_array_rec->{'metavalue'};
1202 my $prevmetavalue = $self->{'prevmetavalue'}; # to make this sub behave as _set_archives_metadata
1203
1204
1205 if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
1206 # make "accumulate" the default (less destructive, as it won't actually
1207 # delete any existing values)
1208 $metamode = "accumulate";
1209 }
1210
1211 $status = $self->set_archives_metadata_entry($gsdl_cgi,$archive_dir, $collect_dir,$collect, $infodbtype,$docid,
1212 $metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
1213 } elsif (defined $doc_array_rec->{'metatable'}) { # if no metaname, we expect a metatable
1214 my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
1215
1216 foreach my $metatable_rec ( @$metatable ) {
1217 $metaname = $metatable_rec->{'metaname'};
1218 my $metamode = $metatable_rec->{'metamode'} || $doc_array_rec->{'metamode'} || $self->{'metamode'};
1219 my $metapos = undef;
1220 my $prevmetavalue = undef;
1221 my $metavals = $metatable_rec->{'metavals'}; # a sub-subarray
1222
1223 foreach my $metavalue ( @$metavals ) {
1224 $status = $self->set_archives_metadata_entry($gsdl_cgi,$archive_dir, $collect_dir,$collect,$infodbtype,
1225 $docid,$metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
1226
1227 if($metamode eq "override") { # now, having overridden the metavalue for the first,
1228 # need to accumulate subsequent metavals for this metaname, else the just-assigned
1229 # metavalue for this metaname will be lost
1230 $metamode = "accumulate";
1231 }
1232 }
1233 }
1234 }
1235
1236 if ($status != 0) {
1237 # Catch error if set infodb entry failed
1238 $global_status = $status;
1239 $global_mess .= "Failed to set metadata key: $docid\n";
1240 $global_mess .= "Exit status: $status\n";
1241 $global_mess .= "System Error Message: $!\n";
1242 $global_mess .= "-" x 20 . "\n";
1243 }
1244 }
1245
1246 if ($global_status != 0) {
1247 $global_mess .= "PATH: $ENV{'PATH'}\n";
1248 $gsdl_cgi->generate_error($global_mess);
1249 }
1250 else {
1251 my $mess = "set-archives-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
1252 $gsdl_cgi->generate_ok_message($mess);
1253 }
1254}
1255
1256sub set_archives_metadata_array
1257{
1258 my $self = shift @_;
1259
1260 my $username = $self->{'username'};
1261 my $collect = $self->{'collect'};
1262 my $gsdl_cgi = $self->{'gsdl_cgi'};
1263# my $gsdlhome = $self->{'gsdlhome'};
1264
1265 if ($baseaction::authentication_enabled) {
1266 # Ensure the user is allowed to edit this collection
1267 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
1268 }
1269
1270 my $site = $self->{'site'};
1271 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
1272
1273 $gsdl_cgi->checked_chdir($collect_dir);
1274
1275 # Obtain the collect dir
1276 ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
1277
1278 # Make sure the collection isn't locked by someone else
1279 $self->lock_collection($username, $collect);
1280
1281 $self->_set_archives_metadata_array(@_);
1282
1283 # Release the lock once it is done
1284 $self->unlock_collection($username, $collect);
1285}
1286
1287sub _remove_archives_metadata
1288{
1289 my $self = shift @_;
1290
1291 my $collect = $self->{'collect'};
1292 my $gsdl_cgi = $self->{'gsdl_cgi'};
1293# my $gsdlhome = $self->{'gsdlhome'};
1294 my $infodbtype = $self->{'infodbtype'};
1295
1296 my $site = $self->{'site'};
1297
1298 # Obtain the collect and archive dir
1299 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
1300
1301 my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
1302
1303 # look up additional args
1304 my ($docid, $docid_secnum) = ($self->{'d'} =~ m/^(.*?)(\..*)?$/);
1305
1306 my $metaname = $self->{'metaname'};
1307 my $metapos = $self->{'metapos'};
1308 my $metavalue = $self->{'metavalue'};
1309
1310 $metapos = undef if(defined $metapos && ($metapos =~ m/^\s*$/));
1311 $metavalue = undef if(defined $metavalue && ($metavalue =~ m/^\s*$/)); # necessary to force fallback to undef here
1312
1313 # if the user hasn't told us what to delete, not having given a metavalue or metapos,
1314 # default to deleting the first metavalue for the given metaname
1315 # Beware that if both metapos AND metavalue are defined, both matches (if any)
1316 # seem to get deleted in one single remove_archives_meta action invocation.
1317 # Similarly, if 2 identical metavalues for a metaname exist and that metavalue is being
1318 # deleted, both get deleted.
1319 if(!defined $metapos && !defined $metavalue) {
1320 $metapos = 0;
1321 }
1322
1323 my $metamode = $self->{'metamode'};
1324 $metamode = undef if(defined $metamode && ($metamode =~ m/^\s*$/));
1325
1326 my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
1327 my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
1328
1329 # This now stores the full pathname
1330 my $doc_filename = $doc_rec->{'doc-file'}->[0];
1331
1332 my $status = $self->remove_from_doc_xml($gsdl_cgi, &util::filename_cat($archive_dir, $doc_filename), $metaname, $metapos, $metavalue, $docid_secnum, $metamode);
1333# my $status = $self->remove_from_doc_xml($gsdl_cgi, &util::filename_cat($archive_dir, $doc_filename), $metaname, $metapos, undef, $docid_secnum);
1334
1335 if ($status == 0)
1336 {
1337 my $mess = "\nremove-archives-metadata successful: \nKey[$docid]\n";
1338 $mess .= " $metaname";
1339 $mess .= "->[$metapos]" if (defined $metapos);
1340 $mess .= " ($metavalue)" if (defined $metavalue);
1341 $gsdl_cgi->generate_ok_message($mess);
1342 }
1343 else
1344 {
1345 my $mess .= "Failed to remove archives metadata key: $docid\n";
1346 $mess .= "Exit status: $status\n";
1347 $mess .= "System Error Message: $!\n";
1348 $mess .= "-" x 20 . "\n";
1349
1350 $gsdl_cgi->generate_error($mess);
1351 }
1352
1353 #return $status; # in case calling functions have a use for this
1354}
1355
1356sub remove_archives_metadata
1357{
1358 my $self = shift @_;
1359
1360 my $username = $self->{'username'};
1361 my $collect = $self->{'collect'};
1362 my $gsdl_cgi = $self->{'gsdl_cgi'};
1363
1364 if ($baseaction::authentication_enabled)
1365 {
1366 # Ensure the user is allowed to edit this collection
1367 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
1368 }
1369
1370 # Make sure the collection isn't locked by someone else
1371 $self->lock_collection($username, $collect);
1372
1373 $self->_remove_archives_metadata(@_);
1374
1375 # Release the lock once it is done
1376 $self->unlock_collection($username, $collect);
1377}
1378
1379sub remove_from_doc_xml
1380{
1381 my $self = shift @_;
1382 my ($gsdl_cgi, $doc_xml_filename, $metaname, $metapos, $metavalue, $secid, $metamode) = @_;
1383
1384 my @start_rules = ('Section' => \&dxml_start_section);
1385
1386 # Set the call-back functions for the metadata tags
1387 my @rules =
1388 (
1389 _default => 'raw',
1390 'Metadata' => \&rfdxml_metadata
1391 );
1392
1393 my $parser = XML::Rules->new
1394 (
1395 start_rules => \@start_rules,
1396 rules => \@rules,
1397 style => 'filter',
1398 output_encoding => 'utf8',
1399# normalisespaces => 1, # http://search.cpan.org/~jenda/XML-Rules-1.16/lib/XML/Rules.pm
1400 stripspaces => 2|0|0 # ineffectual
1401 );
1402
1403 my $status = 0;
1404 my $xml_in = "";
1405 if (!open(MIN,"<$doc_xml_filename"))
1406 {
1407 $gsdl_cgi->generate_error("Unable to read in $doc_xml_filename: $!");
1408 $status = 1;
1409 }
1410 else
1411 {
1412 # Read them in
1413 my $line;
1414 while (defined ($line=<MIN>)) {
1415 $xml_in .= $line if($line !~ m/^\s*$/); # preserve all but empty lines
1416 }
1417 close(MIN);
1418
1419 # Filter with the call-back functions
1420 my $xml_out = "";
1421
1422 my $MOUT;
1423 if (!open($MOUT,">$doc_xml_filename")) {
1424 $gsdl_cgi->generate_error("Unable to write out to $doc_xml_filename: $!");
1425 $status = 1;
1426 }
1427 else {
1428 binmode($MOUT,":utf8");
1429 $parser->filter($xml_in, $MOUT, {metaname => $metaname, metapos => $metapos, metavalue => $metavalue, secid => $secid, metamode => $metamode});
1430 close($MOUT);
1431 }
1432 }
1433 return $status;
1434}
1435
1436sub rfdxml_metadata
1437{
1438 my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
1439
1440 # For comparisons, toplevel section is indicated by ""
1441 my $curr_sec_num = $parser->{'parameters'}->{'curr_section_num'} || "";
1442 my $secid = $parser->{'parameters'}->{'secid'} || "";
1443
1444 if (!($secid eq $curr_sec_num))
1445 {
1446 # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
1447 return [$tagname => $attrHash];
1448 }
1449
1450 if ($parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'})
1451 {
1452 if (!defined $parser->{'parameters'}->{'poscount'})
1453 {
1454 $parser->{'parameters'}->{'poscount'} = 0;
1455 }
1456 else
1457 {
1458 $parser->{'parameters'}->{'poscount'}++;
1459 }
1460
1461 # if overriding (for set-meta) but no metapos, then clear all the meta for this metaname
1462 if ((defined $parser->{'parameters'}->{'metamode'}) && ($parser->{'parameters'}->{'metamode'} eq "override") && (!defined $parser->{'parameters'}->{'metapos'}) &&(!defined $parser->{'parameters'}->{'metavalue'})) {
1463 return [];
1464 }
1465
1466 if ((defined $parser->{'parameters'}->{'metapos'}) && ($parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'}))
1467 {
1468 return [];
1469 }
1470
1471 if ((defined $parser->{'parameters'}->{'metavalue'}) && ($parser->{'parameters'}->{'metavalue'} eq $attrHash->{'_content'}))
1472 {
1473 return [];
1474 }
1475 }
1476
1477 # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
1478 return [$tagname => $attrHash];
1479}
1480
1481sub mxml_metadata
1482{
1483 my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
1484 my $metaname = $parser->{'parameters'}->{'metaname'};
1485 my $metamode = $parser->{'parameters'}->{'metamode'};
1486
1487 # Report error if we don't see FileName tag before this
1488 die "Fatal Error: Unexpected metadata.xml structure. Undefined current_file, possibly encountered Description before FileName" if (!defined($parser->{'parameters'}->{'current_file'}));
1489
1490 # Don't do anything if we are not in the right FileSet
1491 my $file_regexp = $parser->{'parameters'}->{'current_file'};
1492 if ($file_regexp =~ /\.\*/) {
1493 # Only interested in a file_regexp if it specifies precisely one
1494 # file.
1495 # So, skip anything with a .* in it as it is too general
1496## print STDERR "@@@@ Skipping entry in metadata.xml where FileName=.* as it is too general\n";
1497 return [$tagname => $attrHash];
1498 }
1499 my $src_file = $parser->{'parameters'}->{'src_file'};
1500 if (!($src_file =~ /$file_regexp/)) {
1501 return [$tagname => $attrHash];
1502 }
1503## print STDERR "*** mxl metamode = $metamode\n";
1504
1505 # Find the right metadata tag and checks if we are going to override it
1506 my $name_attr = $attrHash->{'name'};
1507 if (($name_attr eq $metaname) && ($metamode eq "override")) {
1508
1509 # now metadata.xml functions need to keep track of metapos
1510 if (!defined $parser->{'parameters'}->{'poscount'})
1511 {
1512 $parser->{'parameters'}->{'poscount'} = 0;
1513 }
1514 else
1515 {
1516 $parser->{'parameters'}->{'poscount'}++;
1517 }
1518
1519 # If either the metapos or prevmetavalue is set,
1520 # get the value and override the current value
1521 my $metavalue = $parser->{'parameters'}->{'metavalue'};
1522
1523 if(defined $parser->{'parameters'}->{'prevmetavalue'} && $parser->{'parameters'}->{'prevmetavalue'} eq $attrHash->{'_content'})
1524 {
1525 $attrHash->{'_content'} = $metavalue;
1526
1527 ## print STDERR "**** overriding metadata.xml\n";
1528
1529 # Don't want it to wipe out any other pieces of metadata
1530 $parser->{'parameters'}->{'metamode'} = "done";
1531 }
1532 elsif(defined $parser->{'parameters'}->{'metapos'} && $parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'})
1533 {
1534 $attrHash->{'_content'} = $metavalue;
1535 $parser->{'parameters'}->{'metamode'} = "done";
1536 }
1537 }
1538
1539 # mxml_description will process the metadata if metadata is accumulate,
1540 # or if we haven't found the metadata to override
1541
1542 # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
1543 return [$tagname => $attrHash];
1544}
1545
1546
1547sub mxml_description
1548{
1549 my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
1550 my $metamode = $parser->{'parameters'}->{'metamode'};
1551
1552 # Failed... Report error if we don't see FileName tag before this
1553 die "Fatal Error: Unexpected metadata.xml structure. Undefind current_file, possiblely encountered Description before FileName" if (!defined($parser->{'parameters'}->{'current_file'}));
1554
1555 # Don't do anything if we are not in the right FileSet
1556 my $file_regexp = $parser->{'parameters'}->{'current_file'};
1557 if ($file_regexp =~ m/\.\*/) {
1558 # Only interested in a file_regexp if it specifies precisely one
1559 # file.
1560 # So, skip anything with a .* in it as it is too general
1561 return [$tagname => $attrHash];
1562 }
1563 my $src_file = $parser->{'parameters'}->{'src_file'};
1564
1565 if (!($src_file =~ m/$file_regexp/)) {
1566 return [$tagname => $attrHash];
1567 }
1568
1569 # Accumulate the metadata block to the end of the description block
1570 # Note: This adds metadata block to all description blocks, so if there are
1571 # multiple FileSets, it will add to all of them
1572 if (($metamode eq "accumulate") || ($metamode eq "override")) {
1573
1574 # if metamode was "override" but get to here then it failed to
1575 # find an item to override, in which case it should append its
1576 # value to the end, just like the "accumulate" mode
1577
1578 if ($metamode eq "override") {
1579 print "No metadata value to override. Switching 'metamode' to accumulate\n";
1580 }
1581
1582 # tack a new metadata tag on to the end of the <Metadata>+ block
1583 my $metaname = $parser->{'parameters'}->{'metaname'};
1584 my $metavalue = $parser->{'parameters'}->{'metavalue'};
1585
1586 my $metadata_attr = { '_content' => $metavalue,
1587 'name' => $metaname,
1588 'mode' => "accumulate" };
1589
1590 my $append_metadata = [ "Metadata" => $metadata_attr ];
1591 my $description_content = $attrHash->{'_content'};
1592
1593## print STDERR "*** appending to metadata.xml\n";
1594
1595 # append the new metadata element to the end of the current
1596 # content contained inside this tag
1597 if (ref($description_content) eq "") {
1598 # => string or numeric literal
1599 # this is caused by a <Description> block has no <Metadata> child elements
1600 # => set up an empty array in '_content'
1601 $attrHash->{'_content'} = [ "\n" ];
1602 $description_content = $attrHash->{'_content'};
1603 }
1604
1605 push(@$description_content, " ", $append_metadata, "\n ");
1606 $parser->{'parameters'}->{'metamode'} = "done";
1607 }
1608
1609 # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
1610 return [$tagname => $attrHash];
1611}
1612
1613
1614sub mxml_fileset
1615{
1616 my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
1617
1618 # Initilise the current_file
1619 # Note: According to http://greenstone.org/dtd/DirectoryMetadata/1.0/DirectoryMetadata.dtd
1620 # FileName tag must come before Description tag
1621 $parser->{'parameters'}->{'current_file'} = "";
1622
1623 # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
1624 return [$tagname => $attrHash];
1625}
1626
1627sub mxml_directorymetadata
1628{
1629 my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
1630
1631 # if we haven't processed the metadata when we reach the end of metadata.xml
1632 # it's because there's no particular FileSet element whose FileName matched
1633 # In which case, add a new FileSet for this FileName
1634 my $metamode = $parser->{'parameters'}->{'metamode'};
1635 if($metamode ne "done") {
1636
1637 if ($metamode eq "override") {
1638 print "No metadata value to override. Switching 'metamode' to accumulate\n";
1639 }
1640
1641 # If we get to here and metamode is override, this means there
1642 # was no existing value to overide => treat as an append operation
1643
1644 # Create a new FileSet element and append to DirectoryMetadata
1645 # <FileSet>
1646 # <FileName>src_file</FileName>
1647 # <Description>
1648 # <Metadata mode="" name="">metavalue</Metadata>
1649 # </Description>
1650 # </FileSet>
1651 my $src_file = $parser->{'parameters'}->{'src_file'};
1652 my $metaname = $parser->{'parameters'}->{'metaname'};
1653 my $metavalue = $parser->{'parameters'}->{'metavalue'};
1654 my $metadata_attr = {
1655 '_content' => $metavalue,
1656 'name' => $metaname,
1657 'mode' => "accumulate"
1658 };
1659 my $append_metadata = [ "Metadata" => $metadata_attr ];
1660 my $description_attr = { '_content' => [ "\n\t\t ", $append_metadata, "\n\t\t"] };
1661 my $description_element = [ "Description" => $description_attr ];
1662
1663 #_content is not an attribute, it's special and holds the children of this element
1664 # including the textnode value embedded in this element if any.
1665 my $filename_attr = {'_content' => $src_file};
1666 my $filename_element = [ "FileName" => $filename_attr ];
1667
1668 my $fileset_attr = {};
1669 $fileset_attr->{'_content'} = [ "\n\t\t", $filename_element,"\n\t\t",$description_element ,"\n\t" ];
1670 my $fileset = [ "FileSet" => $fileset_attr ]; #my $fileset = [ "FileSet" => {} ];
1671
1672
1673 # get children of dirmeta, and push the new FileSet element onto it
1674 print "Appending metadata to metadata.xml\n";
1675 my $dirmeta_content = $attrHash->{'_content'};
1676 if (ref($dirmeta_content)) {
1677 # got some existing interesting nested content
1678 #push(@$dirmeta_content, " ", $fileset ,"\n ");
1679 push(@$dirmeta_content, "\t", $fileset ,"\n");
1680 }
1681 else {
1682 #description_content is most likely a string such as "\n"
1683 #$attrHash->{'_content'} = [$dirmeta_content, " ", $fileset ,"\n" ];
1684 $attrHash->{'_content'} = [$dirmeta_content, "\t", $fileset ,"\n" ];
1685 }
1686
1687 $parser->{'parameters'}->{'metamode'} = "done";
1688 }
1689 # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
1690 return [$tagname => $attrHash];
1691}
1692
1693
1694sub edit_metadata_xml
1695{
1696 my $self = shift @_;
1697 my ($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $metavalue, $metamode, $src_file, $prevmetavalue) = @_;
1698
1699 # Set the call-back functions for the metadata tags
1700 my @rules =
1701 ( _default => 'raw',
1702 'FileName' => \&mxml_filename,
1703 'Metadata' => \&mxml_metadata,
1704 'Description' => \&mxml_description,
1705 'FileSet' => \&mxml_fileset,
1706 'DirectoryMetadata' => \&mxml_directorymetadata);
1707
1708 # use XML::Rules to add it in (read in and out again)
1709 my $parser = XML::Rules->new(rules => \@rules,
1710 style => 'filter',
1711 output_encoding => 'utf8',
1712 stripspaces => 2|0|0); # http://search.cpan.org/~jenda/XML-Rules-1.16/lib/XML/Rules.pm
1713
1714 if (!-e $metadata_xml_filename) {
1715
1716 if (open(MOUT,">$metadata_xml_filename")) {
1717
1718 my $src_file_re = &util::filename_to_regex($src_file);
1719 # shouldn't the following also be in the above utility routine??
1720 # $src_file_re =~ s/\./\\./g;
1721
1722 print MOUT "<?xml version=\"1.0\"?>\n";
1723 print MOUT "<DirectoryMetadata>\n";
1724 print MOUT " <FileSet>\n";
1725 print MOUT " <FileName>$src_file_re</FileName>\n";
1726 print MOUT " <Description>\n";
1727 print MOUT " </Description>\n";
1728 print MOUT " </FileSet>\n";
1729 print MOUT "</DirectoryMetadata>\n";
1730
1731 close(MOUT);
1732 }
1733 else {
1734 $gsdl_cgi->generate_error("Unable to create $metadata_xml_filename: $!");
1735 }
1736 }
1737
1738
1739 my $xml_in = "";
1740 if (!open(MIN,"<$metadata_xml_filename")) {
1741 $gsdl_cgi->generate_error("Unable to read in $metadata_xml_filename: $!");
1742 }
1743 else {
1744 # Read them in
1745 my $line;
1746 while (defined ($line=<MIN>)) {
1747 $xml_in .= $line if($line !~ m/^\s*$/); # preserve all but empty lines
1748 }
1749 close(MIN);
1750
1751 # Filter with the call-back functions
1752 my $xml_out = "";
1753
1754 my $MOUT;
1755 if (!open($MOUT,">$metadata_xml_filename")) {
1756 $gsdl_cgi->generate_error("Unable to write out to $metadata_xml_filename: $!");
1757 }
1758 else {
1759 binmode($MOUT,":utf8");
1760
1761 # Some wise person please find out how to keep the DTD and encode lines in after it gets filtered by this XML::Rules
1762 # At the moment, I will just hack it!
1763 #my $header_with_utf8_dtd = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
1764 #$header_with_utf8_dtd .= "<!DOCTYPE DirectoryMetadata SYSTEM \"http://greenstone.org/dtd/DirectoryMetadata/1.0/DirectoryMetadata.dtd\">";
1765 #$xml_out =~ s/\<\?xml\sversion\=\"1.0\"\?\>/$header_with_utf8_dtd/;
1766 #print MOUT $xml_out;
1767
1768 $parser->filter($xml_in, $MOUT, { metaname => $metaname,
1769 metapos => $metapos,
1770 metavalue => $metavalue,
1771 metamode => $metamode,
1772 src_file => $src_file,
1773 prevmetavalue => $prevmetavalue,
1774 current_file => undef} );
1775 close($MOUT);
1776 }
1777 }
1778}
1779
1780
1781sub set_import_metadata
1782{
1783 my $self = shift @_;
1784
1785 my $username = $self->{'username'};
1786 my $collect = $self->{'collect'};
1787 my $gsdl_cgi = $self->{'gsdl_cgi'};
1788
1789 if ($baseaction::authentication_enabled) {
1790 # Ensure the user is allowed to edit this collection
1791 $self->authenticate_user($username, $collect);
1792 }
1793
1794 # Make sure the collection isn't locked by someone else
1795 $self->lock_collection($username, $collect);
1796
1797 $self->_set_import_metadata(@_);
1798
1799 # Release the lock once it is done
1800 $self->unlock_collection($username, $collect);
1801
1802}
1803
1804sub set_import_metadata_array
1805{
1806 my $self = shift @_;
1807
1808 my $username = $self->{'username'};
1809 my $collect = $self->{'collect'};
1810 my $gsdl_cgi = $self->{'gsdl_cgi'};
1811# my $gsdlhome = $self->{'gsdlhome'};
1812
1813 if ($baseaction::authentication_enabled) {
1814 # Ensure the user is allowed to edit this collection
1815 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
1816 }
1817
1818 my $site = $self->{'site'};
1819 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
1820
1821 $gsdl_cgi->checked_chdir($collect_dir);
1822
1823 # Make sure the collection isn't locked by someone else
1824 $self->lock_collection($username, $collect);
1825
1826 $self->_set_import_metadata_array(@_);
1827
1828 # Release the lock once it is done
1829 $self->unlock_collection($username, $collect);
1830
1831}
1832
1833
1834sub _set_import_metadata_array
1835{
1836 my $self = shift @_;
1837
1838 my $collect = $self->{'collect'};
1839 my $gsdl_cgi = $self->{'gsdl_cgi'};
1840
1841 my $site = $self->{'site'};
1842 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
1843
1844 # look up additional args
1845
1846 my $infodbtype = $self->{'infodbtype'};
1847
1848 my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
1849 my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
1850
1851 my $json_str = $self->{'json'};
1852 my $doc_array = decode_json $json_str;
1853
1854 my $global_status = 0;
1855 my $global_mess = "";
1856
1857 my @all_docids = ();
1858
1859 foreach my $doc_array_rec ( @$doc_array )
1860 {
1861 my $status = -1;
1862 my $docid = $doc_array_rec->{'docid'};
1863
1864 my ($docid_root,$docid_secnum);
1865 if(defined $docid) {
1866 ($docid_root,$docid_secnum) = ($docid =~ m/^(.*?)(\..*)?$/);
1867 # as yet no support for setting subsection metadata in metadata.xml
1868 if ((defined $docid_secnum) && ($docid_secnum !~ m/^\s*$/)) {
1869 $gsdl_cgi->generate_message("*** docid: $docid. No support yet for setting import metadata at subsections level.\n");
1870 next; # skip this docid in for loop
1871 }
1872 }
1873
1874 push(@all_docids,$docid); # docid_root rather
1875
1876 my $metaname = $doc_array_rec->{'metaname'};
1877 if (defined $metaname) {
1878 my $metamode = $doc_array_rec->{'metamode'} || $self->{'metamode'};
1879 my $metavalue = $doc_array_rec->{'metavalue'};
1880 $metavalue =~ s/&lt;(.*?)&gt;/<$1>/g;
1881
1882 if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
1883 # make "accumulate" the default (less destructive, as it won't actually
1884 # delete any existing values)
1885 $metamode = "accumulate";
1886 }
1887
1888 # adding metapos and prevmetavalue support to import_metadata subroutines
1889 my $metapos = $doc_array_rec->{'metapos'}; # don't force undef to 0. Undef has meaning when metamode=override
1890 my $prevmetavalue = $self->{'prevmetavalue'};
1891
1892 $self->set_import_metadata_entry($gsdl_cgi, $arcinfo_doc_filename, $infodbtype, $docid_root, $metaname, $metapos, $metavalue, $metamode, $prevmetavalue, $collect, $collect_dir); # at this point, docid_root = docid
1893
1894 } elsif (defined $doc_array_rec->{'metatable'}) { # if no metaname, we expect a metatable
1895 my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
1896
1897 foreach my $metatable_rec ( @$metatable ) {
1898 $metaname = $metatable_rec->{'metaname'};
1899 my $metamode = $metatable_rec->{'metamode'} || $doc_array_rec->{'metamode'} || $self->{'metamode'};
1900 if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
1901 # make "accumulate" the default (less destructive, as it won't actually
1902 # delete any existing values)
1903 $metamode = "accumulate";
1904 }
1905
1906 # No support for metapos and prevmetavalue in the JSON metatable substructure
1907 my $metapos = undef;
1908 my $prevmetavalue = undef;
1909 my $metavals = $metatable_rec->{'metavals'}; # a sub-subarray
1910
1911 foreach my $metavalue ( @$metavals ) {
1912 $metavalue =~ s/&lt;(.*?)&gt;/<$1>/g;
1913
1914 $self->set_import_metadata_entry($gsdl_cgi, $arcinfo_doc_filename, $infodbtype, $docid_root, $metaname, $metapos, $metavalue, $metamode, $prevmetavalue, $collect, $collect_dir); # at this point, docid_root = docid
1915 if($metamode eq "override") { # now, having overridden the first metavalue of the metaname,
1916 # need to accumulate subsequent metavals for this metaname, else the just-assigned
1917 # metavalue for this metaname will be lost
1918 $metamode = "accumulate";
1919 }
1920 }
1921 }
1922 }
1923 }
1924
1925 # always a success message
1926 my $mess = "set-archives-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
1927 $gsdl_cgi->generate_ok_message($mess);
1928}
1929
1930# always returns true (1)
1931sub set_import_metadata_entry
1932{
1933 my $self = shift @_;
1934 my ($gsdl_cgi, $arcinfo_doc_filename, $infodbtype, $docid, $metaname, $metapos, $metavalue, $metamode, $prevmetavalue, $collect, $collect_dir) = @_;
1935
1936 my $info_mess = <<RAWEND;
1937****************************
1938 set_import_metadata_entry()
1939****************************
1940RAWEND
1941
1942 $info_mess .= " collect_dir = $collect_dir\n" if defined($collect_dir);
1943 $info_mess .= " collect = $collect\n" if defined($collect);
1944 $info_mess .= " infodbtype = $infodbtype\n" if defined($infodbtype);
1945 $info_mess .= " arcinfo_doc_filename = $arcinfo_doc_filename\n" if defined($arcinfo_doc_filename);
1946 $info_mess .= " docid = $docid\n" if defined($docid);
1947 $info_mess .= " metaname = $metaname\n" if defined($metaname);
1948 $info_mess .= " metapos = $metapos\n" if defined($metapos);
1949 $info_mess .= " metavalue = $metavalue\n" if defined($metavalue);
1950 $info_mess .= " metamode = $metamode\n" if defined($metamode);
1951 $info_mess .= " prevmetaval = $prevmetavalue\n" if defined($prevmetavalue);
1952
1953 $info_mess .= "****************************\n";
1954
1955 $gsdl_cgi->generate_message($info_mess);
1956
1957 # import works with metadata.xml which can have inherited metadata
1958 # so setting or removing at a metapos can have unintended effects for a COMPLEX collection
1959 # (a collection that has or can have inherited metadata). Metapos has expected behaviour for
1960 # a SIMPLE collection, which is one that doesn't have inherited metadata. Assume caller knows
1961 # what they're doing if they provide a metapos.
1962 if(defined $metapos) {
1963 print STDERR "@@@@ WARNING: metapos defined.\n";
1964 print STDERR "@@@@ Assuming SIMPLE collection and proceeding to modify the import meta at $metapos.\n";
1965 }
1966
1967 # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
1968 # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
1969 my $metadata_xml_file;
1970 my $import_filename = undef;
1971
1972 if (defined $docid) {
1973 # my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
1974 my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
1975
1976 # This now stores the full pathname
1977 $import_filename = $doc_rec->{'src-file'}->[0];
1978 $import_filename = &util::placeholders_to_abspath($import_filename);
1979
1980 } else { # only for set_import_meta, not the case when calling method is set_import_metadata_array
1981 # as the array version of the method doesn't support the -f parameter yet
1982 my $import_file = $self->{'f'};
1983 $import_filename = &util::filename_cat($collect_dir,$collect,$import_file);
1984 }
1985
1986 # figure out correct metadata.xml file [?]
1987 # Assuming the metadata.xml file is next to the source file
1988 # Note: This will not work if it is using the inherited metadata from the parent folder
1989 my ($import_tailname, $import_dirname) = File::Basename::fileparse($import_filename);
1990 my $metadata_xml_filename = &util::filename_cat($import_dirname,"metadata.xml");
1991
1992 # If we're overriding everything, then $prevmetavalue=undefined and
1993 # $metamode=override combined with $metapos=undefined
1994 # in which case we need to remove all metavalues for the metaname at the given (sub)section
1995 # Thereafter, we will finally be able to set the overriding metavalue for this metaname
1996 if(!defined $prevmetavalue && !defined $metapos && $metamode eq "override") {
1997## print STDERR "@@@ REMOVING all import metadata for $metaname\n";
1998 $self->remove_from_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, undef, $import_tailname, $metamode); # we're removing all values, so metavalue=undef
1999
2000 }
2001
2002 # Edit the metadata.xml
2003 # Modified by Jeffrey from DL Consulting
2004 # Handle the case where there is one metadata.xml file for multiple FileSets
2005 # The XML filter needs to know whether it is in the right FileSet
2006 # TODO: This doesn't fix the problem where the metadata.xml is not next to the src file.
2007 # TODO: This doesn't handle the common metadata (where FileName doesn't point to a single file)
2008 $self->edit_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname,
2009 $metapos, $metavalue, $metamode, $import_tailname, $prevmetavalue);
2010 #return 0;
2011 return $metadata_xml_filename;
2012}
2013
2014sub _remove_import_metadata
2015{
2016 my $self = shift @_;
2017
2018 my $collect = $self->{'collect'};
2019 my $gsdl_cgi = $self->{'gsdl_cgi'};
2020# my $gsdlhome = $self->{'gsdlhome'};
2021 my $infodbtype = $self->{'infodbtype'};
2022
2023 # Obtain the collect dir
2024 ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
2025 my $site = $self->{'site'};
2026 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
2027
2028 ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
2029 my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
2030
2031 # look up additional args
2032 my $docid = $self->{'d'};
2033 if ((!defined $docid) || ($docid =~ m/^\s*$/))
2034 {
2035 $gsdl_cgi->generate_error("No docid (d=...) specified.\n");
2036 }
2037
2038 my $metaname = $self->{'metaname'};
2039 my $metapos = $self->{'metapos'};
2040 my $metavalue = $self->{'metavalue'};
2041
2042 $metapos = undef if(defined $metapos && ($metapos =~ m/^\s*$/));
2043 $metavalue = undef if(defined $metavalue && ($metavalue =~ m/^\s*$/));
2044
2045 if(defined $metavalue) { # metavalue is now a compulsory arg for remove_import_metadata()
2046 $metavalue =~ s/&lt;(.*?)&gt;/<$1>/g;
2047 } elsif (!defined $metapos) { # if given no metavalue or metapos to delete, default to deleting the 1st
2048 $metapos = 0;
2049 }
2050 my $metamode = $self->{'metamode'};
2051 $metamode = undef if(defined $metamode && ($metamode =~ m/^\s*$/));
2052
2053 # import works with metadata.xml which can have inherited metadata
2054 # so setting or removing at a metapos can have unintended effects for a COMPLEX collection
2055 # (a collection that has or can have inherited metadata). Metapos has expected behaviour for
2056 # a SIMPLE collection, which is one that doesn't have inherited metadata. Assume caller knows
2057 # what they're doing if they provide a metapos.
2058 if(defined $metapos) {
2059 print STDERR "@@@@ WARNING: metapos defined.\n";
2060 print STDERR "@@@@ Assuming SIMPLE collection and proceeding to modify the import meta at $metapos.\n";
2061 }
2062
2063 # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
2064 # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
2065 my $metadata_xml_file;
2066 my $import_filename = undef;
2067 if (defined $docid)
2068 {
2069 my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
2070 my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
2071
2072 # This now stores the full pathname
2073 $import_filename = $doc_rec->{'src-file'}->[0];
2074 $import_filename = &util::placeholders_to_abspath($import_filename);
2075 }
2076
2077 if((!defined $import_filename) || ($import_filename =~ m/^\s*$/))
2078 {
2079 $gsdl_cgi->generate_error("There is no metadata\n");
2080 }
2081
2082 # figure out correct metadata.xml file [?]
2083 # Assuming the metadata.xml file is next to the source file
2084 # Note: This will not work if it is using the inherited metadata from the parent folder
2085 my ($import_tailname, $import_dirname) = File::Basename::fileparse($import_filename);
2086 my $metadata_xml_filename = &util::filename_cat($import_dirname,"metadata.xml");
2087
2088 $self->remove_from_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $metavalue, $import_tailname, $metamode); # metamode has no meaning for removing meta, but is used by set_meta when overriding All
2089
2090 my $mess = "remove-import-metadata successful: Key[$docid] -> $metadata_xml_filename\n";
2091 $mess .= " $metaname";
2092 $mess .= " = $metavalue\n";
2093
2094 $gsdl_cgi->generate_ok_message($mess);
2095
2096 #return $status; # in case calling functions have a use for this
2097}
2098
2099sub remove_import_metadata
2100{
2101 my $self = shift @_;
2102
2103 my $username = $self->{'username'};
2104 my $collect = $self->{'collect'};
2105 my $gsdl_cgi = $self->{'gsdl_cgi'};
2106
2107 if ($baseaction::authentication_enabled) {
2108 # Ensure the user is allowed to edit this collection
2109 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
2110 }
2111
2112 # Make sure the collection isn't locked by someone else
2113 $self->lock_collection($username, $collect);
2114
2115 $self->_remove_import_metadata(@_);
2116
2117 # Release the lock once it is done
2118 $self->unlock_collection($username, $collect);
2119
2120}
2121
2122sub remove_from_metadata_xml
2123{
2124 my $self = shift @_;
2125 my ($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $metavalue, $src_file, $metamode) = @_;
2126 # metamode generally has no meaning for removing meta, but is used by set_meta
2127 # when overriding all metavals for a metaname, in which case remove_meta is called with metamode
2128
2129 # Set the call-back functions for the metadata tags
2130 my @rules =
2131 (
2132 _default => 'raw',
2133 'Metadata' => \&rfmxml_metadata,
2134 'FileName' => \&mxml_filename
2135 );
2136
2137 my $parser = XML::Rules->new
2138 (
2139 rules => \@rules,
2140 style => 'filter',
2141 output_encoding => 'utf8',
2142 #normalisespaces => 1,
2143 stripspaces => 2|0|0 # ineffectual
2144 );
2145
2146 my $xml_in = "";
2147 if (!open(MIN,"<$metadata_xml_filename"))
2148 {
2149 $gsdl_cgi->generate_error("Unable to read in $metadata_xml_filename: $!");
2150 }
2151 else
2152 {
2153 # Read them in
2154 my $line;
2155 while (defined ($line=<MIN>)) {
2156 $xml_in .= $line if($line !~ m/^\s*$/); # preserve all but empty lines
2157 }
2158 close(MIN);
2159
2160 # Filter with the call-back functions
2161 my $xml_out = "";
2162
2163 my $MOUT;
2164 if (!open($MOUT,">$metadata_xml_filename")) {
2165 $gsdl_cgi->generate_error("Unable to write out to $metadata_xml_filename: $!");
2166 }
2167 else {
2168 binmode($MOUT,":utf8");
2169 $parser->filter($xml_in, $MOUT, {metaname => $metaname, metapos => $metapos, metavalue => $metavalue, src_file => $src_file, metamode => $metamode, current_file => undef});
2170 close($MOUT);
2171 }
2172 }
2173}
2174
2175sub rfmxml_metadata
2176{
2177 my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
2178
2179 # metadata.xml does not handle subsections
2180
2181 # since metadata.xml now has to deal with metapos, we keep track of the metadata position
2182 if (($parser->{'parameters'}->{'src_file'} eq $parser->{'parameters'}->{'current_file'})
2183 && $parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'})
2184 {
2185 if (!defined $parser->{'parameters'}->{'poscount'})
2186 {
2187 $parser->{'parameters'}->{'poscount'} = 0;
2188 }
2189 else
2190 {
2191 $parser->{'parameters'}->{'poscount'}++;
2192 }
2193
2194 # if overriding but no metapos, then clear all the meta for this metaname
2195 if ((defined $parser->{'parameters'}->{'metamode'}) && ($parser->{'parameters'}->{'metamode'} eq "override") && (!defined $parser->{'parameters'}->{'metapos'}) && (!defined $parser->{'parameters'}->{'metavalue'})) {
2196 return [];
2197 }
2198
2199 if ((defined $parser->{'parameters'}->{'metapos'}) && ($parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'}))
2200 {
2201 return [];
2202 }
2203 if ((defined $parser->{'parameters'}->{'metavalue'}) && ($attrHash->{'_content'} eq $parser->{'parameters'}->{'metavalue'}))
2204 {
2205 return [];
2206 }
2207 }
2208
2209 # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
2210 return [$tagname => $attrHash];
2211}
2212
2213sub _remove_live_metadata
2214{
2215 my $self = shift @_;
2216
2217 my $collect = $self->{'collect'};
2218 my $gsdl_cgi = $self->{'gsdl_cgi'};
2219# my $gsdlhome = $self->{'gsdlhome'};
2220 my $infodbtype = $self->{'infodbtype'};
2221
2222 # Obtain the collect dir
2223 ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
2224 my $site = $self->{'site'};
2225 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
2226
2227
2228 # look up additional args
2229 my $docid = $self->{'d'};
2230 if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
2231 $gsdl_cgi->generate_error("No docid (d=...) specified.");
2232 }
2233
2234 # Generate the dbkey
2235 my $metaname = $self->{'metaname'};
2236 my $dbkey = "$docid.$metaname";
2237
2238 # To people who know $collect_tail please add some comments
2239 # Obtain the live gdbm_db path
2240 my $collect_tail = $collect;
2241 $collect_tail =~ s/^.*[\/|\\]//;
2242 my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
2243 my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, "live-$collect_tail", $index_text_directory);
2244
2245 # Remove the key
2246 my $cmd = "gdbmdel \"$infodb_file_path\" \"$dbkey\"";
2247 my $status = system($cmd);
2248 if ($status != 0) {
2249 # Catch error if gdbmdel failed
2250 my $mess = "Failed to set metadata key: $dbkey\n";
2251
2252 $mess .= "PATH: $ENV{'PATH'}\n";
2253 $mess .= "cmd = $cmd\n";
2254 $mess .= "Exit status: $status\n";
2255 $mess .= "System Error Message: $!\n";
2256
2257 $gsdl_cgi->generate_error($mess);
2258 }
2259 else {
2260 $gsdl_cgi->generate_ok_message("DB remove successful: Key[$metaname]");
2261 }
2262
2263}
2264
2265sub remove_live_metadata
2266{
2267 my $self = shift @_;
2268
2269 my $username = $self->{'username'};
2270 my $collect = $self->{'collect'};
2271 my $gsdl_cgi = $self->{'gsdl_cgi'};
2272 my $gsdlhome = $self->{'gsdlhome'};
2273
2274 if ($baseaction::authentication_enabled) {
2275 # Ensure the user is allowed to edit this collection
2276 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
2277 }
2278
2279 # Make sure the collection isn't locked by someone else
2280 $self->lock_collection($username, $collect);
2281
2282 $self->_remove_live_metadata(@_);
2283
2284 $self->unlock_collection($username, $collect);
2285}
2286
2287sub remove_metadata
2288{
2289 my $self = shift @_;
2290
2291 my $where = $self->{'where'};
2292 if(!$where || ($where =~ m/^\s*$/)) {
2293 $self->remove_index_metadata(@_); # call the full version of set_index_meta for the default behaviour
2294 return;
2295 }
2296
2297 my $username = $self->{'username'};
2298 my $collect = $self->{'collect'};
2299 my $gsdl_cgi = $self->{'gsdl_cgi'};
2300
2301 if ($baseaction::authentication_enabled) {
2302 # Ensure the user is allowed to edit this collection
2303 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
2304 }
2305
2306 # Make sure the collection isn't locked by someone else
2307 $self->lock_collection($username, $collect);
2308
2309 # check which directories need to be processed, specified in $where as
2310 # any combination of import|archives|index|live
2311 if($where =~ m/import/) {
2312 $self->_remove_import_metadata(@_);
2313 }
2314 if($where =~ m/archives/) {
2315 $self->_remove_archives_metadata(@_);
2316 }
2317 if($where =~ m/index/) {
2318 $self->_remove_index_metadata(@_);
2319 }
2320
2321 # Release the lock once it is done
2322 $self->unlock_collection($username, $collect);
2323}
2324
2325# the internal version, without authentication
2326sub _remove_index_metadata
2327{
2328 my $self = shift @_;
2329
2330 my $collect = $self->{'collect'};
2331 my $gsdl_cgi = $self->{'gsdl_cgi'};
2332# my $gsdlhome = $self->{'gsdlhome'};
2333 my $infodbtype = $self->{'infodbtype'};
2334
2335 # Obtain the collect dir
2336 my $site = $self->{'site'};
2337 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
2338 ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
2339
2340 # look up additional args
2341 my $docid = $self->{'d'};
2342 if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
2343 $gsdl_cgi->generate_error("No docid (d=...) specified.");
2344 }
2345 my $metaname = $self->{'metaname'};
2346 my $metapos = $self->{'metapos'};
2347 my $metavalue = $self->{'metavalue'};
2348
2349 $metapos = undef if(defined $metapos && ($metapos =~ m/^\s*$/));
2350 $metavalue = undef if(defined $metavalue && ($metavalue =~ m/^\s*$/)); # necessary to force fallback to undef here
2351
2352 # To people who know $collect_tail please add some comments
2353 # -> In collection groups, I think collect_tailname is the subcollection name,
2354 # e.g. colgroup-name/col-tail-name
2355 # Obtain the path to the database
2356 my $collect_tail = $collect;
2357 $collect_tail =~ s/^.*[\/|\\]//;
2358 my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
2359 my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, $collect_tail, $index_text_directory);
2360
2361 # Read the docid entry
2362 my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $infodb_file_path, $docid);
2363
2364 # Check to make sure the key does exist
2365 if (!defined ($doc_rec->{$metaname})) {
2366 $gsdl_cgi->generate_error("No metadata field \"" . $metaname . "\" in the specified document: [" . $docid . "]");
2367 }
2368
2369 # Obtain the specified metadata pos
2370 # if no metavalue or metapos to delete, default to deleting the 1st value for the metaname
2371 if(!defined $metapos && !defined $metavalue) {
2372 $metapos = 0;
2373 }
2374
2375
2376 # consider check key is defined before deleting?
2377 # Loop through the metadata array and ignore the specified position
2378 my $filtered_metadata = [];
2379 my $num_metadata_vals = scalar(@{$doc_rec->{$metaname}});
2380 for (my $i=0; $i<$num_metadata_vals; $i++) {
2381 my $metaval = shift(@{$doc_rec->{$metaname}});
2382
2383 if (!defined $metavalue && $i != $metapos) {
2384 push(@$filtered_metadata,$metaval);
2385 }
2386
2387 if(defined $metavalue && !($metavalue eq $metaval))
2388 {
2389 push(@$filtered_metadata,$metaval);
2390 }
2391 }
2392 $doc_rec->{$metaname} = $filtered_metadata;
2393
2394 ## Use the dbutil set_entry method instead of assuming the database is gdbm
2395 my $status = &dbutil::set_infodb_entry($infodbtype, $infodb_file_path, $docid, $doc_rec);
2396
2397 if ($status != 0) {
2398 my $mess = "Failed to set metadata key: $docid\n";
2399
2400 $mess .= "PATH: $ENV{'PATH'}\n";
2401 $mess .= "Exit status: $status\n";
2402 $mess .= "System Error Message: $!\n";
2403
2404 $gsdl_cgi->generate_error($mess);
2405 }
2406 else {
2407 my $mess = "DB set (with item deleted) successful: Key[$docid]\n";
2408 $mess .= " $metaname";
2409 $mess .= "->[$metapos]" if (defined $metapos);
2410 $mess .= " ($metavalue)" if (defined $metavalue);
2411
2412 $gsdl_cgi->generate_ok_message($mess);
2413 }
2414
2415 #return $status; # in case calling functions have a use for this
2416}
2417
2418sub remove_index_metadata
2419{
2420 my $self = shift @_;
2421
2422 my $username = $self->{'username'};
2423 my $collect = $self->{'collect'};
2424 my $gsdl_cgi = $self->{'gsdl_cgi'};
2425# my $gsdlhome = $self->{'gsdlhome'};
2426
2427 if ($baseaction::authentication_enabled) {
2428 # Ensure the user is allowed to edit this collection
2429 $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
2430 }
2431
2432 # Obtain the collect dir
2433 my $site = $self->{'site'};
2434 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
2435 ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
2436
2437 # Make sure the collection isn't locked by someone else
2438 $self->lock_collection($username, $collect);
2439
2440 $self->_remove_index_metadata(@_);
2441
2442 # Release the lock once it is done
2443 $self->unlock_collection($username, $collect);
2444}
2445
2446
2447# Was trying to reused the codes, but the functions need to be broken
2448# down more before they can be reused, otherwise there will be too
2449# much overhead and duplicate process...
2450sub insert_metadata
2451{
2452 my $self = shift @_;
2453
2454 my $username = $self->{'username'};
2455 my $collect = $self->{'collect'};
2456 my $gsdl_cgi = $self->{'gsdl_cgi'};
2457 my $gsdlhome = $self->{'gsdlhome'};
2458 my $infodbtype = $self->{'infodbtype'};
2459
2460 # If the import metadata and gdbm database have been updated, we
2461 # need to insert some notification to warn user that the the text
2462 # they see at the moment is not indexed and require a rebuild.
2463 my $rebuild_pending_macro = "_rebuildpendingmessage_";
2464
2465 if ($baseaction::authentication_enabled) {
2466 # Ensure the user is allowed to edit this collection
2467 $self->authenticate_user($username, $collect);
2468 }
2469
2470 # Obtain the collect and archive dir
2471 my $site = $self->{'site'};
2472 my $collect_dir = $gsdl_cgi->get_collection_dir($site);
2473 ##my $collect_dir = &util::filename_cat($gsdlhome, "collect");
2474 my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
2475
2476 # Make sure the collection isn't locked by someone else
2477 $self->lock_collection($username, $collect);
2478
2479 # Check additional args
2480 my $docid = $self->{'d'};
2481 if (!defined($docid) || ($docid =~ m/^\s*$/)) {
2482 $gsdl_cgi->generate_error("No document id is specified: d=...");
2483 }
2484 my $metaname = $self->{'metaname'};
2485 if (!defined($metaname) || ($metaname =~ m/^\s*$/)) {
2486 $gsdl_cgi->generate_error("No metaname is specified: metadataname=...");
2487 }
2488 my $metavalue = $self->{'metavalue'};
2489 if (!defined($metavalue) || ($metavalue =~ m/^\s*$/)) {
2490 $gsdl_cgi->generate_error("No metavalue or empty metavalue is specified: metadataname=...");
2491 }
2492 # make "accumulate" the default (less destructive, as it won't actually
2493 # delete any existing values)
2494 my $metamode = "accumulate";
2495
2496 # metapos/prevmetavalue were never before used in this subroutine, so set them to undefined
2497 my $metapos = undef;
2498 my $prevmetavalue = undef;
2499
2500 #=======================================================================#
2501 # set_import_metadata [START]
2502 #=======================================================================#
2503 # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
2504 # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
2505 my $metadata_xml_file;
2506 my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
2507 my $archive_doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
2508
2509 # This now stores the full pathname
2510 my $import_filename = $archive_doc_rec->{'src-file'}->[0];
2511 $import_filename = &util::placeholders_to_abspath($import_filename);
2512
2513 # figure out correct metadata.xml file [?]
2514 # Assuming the metadata.xml file is next to the source file
2515 # Note: This will not work if it is using the inherited metadata from the parent folder
2516 my ($import_tailname, $import_dirname)
2517 = File::Basename::fileparse($import_filename);
2518 my $metadata_xml_filename = &util::filename_cat($import_dirname,"metadata.xml");
2519
2520 # Shane's escape characters
2521 $metavalue = pack "U0C*", unpack "C*", $metavalue;
2522 $metavalue =~ s/\,/&#44;/g;
2523 $metavalue =~ s/\:/&#58;/g;
2524 $metavalue =~ s/\|/&#124;/g;
2525 $metavalue =~ s/\(/&#40;/g;
2526 $metavalue =~ s/\)/&#41;/g;
2527 $metavalue =~ s/\[/&#91;/g;
2528 $metavalue =~ s/\\/&#92;/g;
2529 $metavalue =~ s/\]/&#93;/g;
2530 $metavalue =~ s/\{/&#123;/g;
2531 $metavalue =~ s/\}/&#125;/g;
2532 $metavalue =~ s/\"/&#34;/g;
2533 $metavalue =~ s/\`/&#96;/g;
2534 $metavalue =~ s/\n/_newline_/g;
2535
2536 # Edit the metadata.xml
2537 # Modified by Jeffrey from DL Consulting
2538 # Handle the case where there is one metadata.xml file for multiple FileSets
2539 # The XML filter needs to know whether it is in the right FileSet
2540 # TODO: This doesn't fix the problem where the metadata.xml is not next to the src file.
2541 # TODO: This doesn't handle the common metadata (where FileName doesn't point to a single file)
2542 $self->edit_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname,
2543 $metapos, $metavalue, $metamode, $import_tailname, $prevmetavalue);
2544 #=======================================================================#
2545 # set_import_metadata [END]
2546 #=======================================================================#
2547
2548
2549 #=======================================================================#
2550 # set_metadata (accumulate version) [START]
2551 #=======================================================================#
2552 # To people who know $collect_tail please add some comments
2553 # Obtain path to the database
2554 my $collect_tail = $collect;
2555 $collect_tail =~ s/^.*[\/|\\]//;
2556 my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
2557 my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, $collect_tail, $index_text_directory);
2558
2559 # Read the docid entry
2560 my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $infodb_file_path, $docid);
2561
2562 # Protect the quotes
2563 $metavalue =~ s/\"/\\\"/g;
2564
2565 # Adds the pending macro
2566 my $macro_metavalue = $rebuild_pending_macro . $metavalue;
2567
2568 # If the metadata doesn't exist, create a new one
2569 if (!defined($doc_rec->{$metaname})){
2570 $doc_rec->{$metaname} = [ $macro_metavalue ];
2571 }
2572 # Else, let's acculumate the values
2573 else {
2574 push(@{$doc_rec->{$metaname}},$macro_metavalue);
2575 }
2576
2577 ## Use the dbutil set_entry method instead of assuming the database is gdbm
2578 my $status = &dbutil::set_infodb_entry($infodbtype, $infodb_file_path, $docid, $doc_rec);
2579
2580 if ($status != 0) {
2581 # Catch error if gdbmget failed
2582 my $mess = "Failed to set metadata key: $docid\n";
2583
2584 $mess .= "PATH: $ENV{'PATH'}\n";
2585 $mess .= "Exit status: $status\n";
2586 $mess .= "System Error Message: $!\n";
2587
2588 $gsdl_cgi->generate_error($mess);
2589 }
2590 else {
2591 my $mess = "insert-metadata successful: Key[$docid]\n";
2592 $mess .= " [In metadata.xml] $metaname";
2593 $mess .= " = $metavalue\n";
2594 $mess .= " [In database] $metaname";
2595 $mess .= " = $macro_metavalue\n";
2596 $mess .= " The new text has not been indexed, rebuilding collection is required\n";
2597 $gsdl_cgi->generate_ok_message($mess);
2598 }
2599 #=======================================================================#
2600 # set_metadata (accumulate version) [END]
2601 #=======================================================================#
2602
2603 # Release the lock once it is done
2604 $self->unlock_collection($username, $collect);
2605}
2606
2607# not returning 1; here since this file is conditionally included by metadataction.pm
2608# and not otherwise meant to be used on its own
2609
Note: See TracBrowser for help on using the repository browser.