1 | ###########################################################################
|
---|
2 | #
|
---|
3 | # metadataaction.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 redistr te it and/or modify
|
---|
11 | # it under the terms of the GNU General Public License as published by
|
---|
12 | # the Free Software Foundation; either version 2 of the License, or
|
---|
13 | # (at your option) any later version.
|
---|
14 | #
|
---|
15 | # This program is distributed in the hope that it will be useful,
|
---|
16 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
17 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
18 | # GNU General Public License for more details.
|
---|
19 | #
|
---|
20 | # You should have received a copy of the GNU General Public License
|
---|
21 | # along with this program; if not, write to the Free Software
|
---|
22 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
23 | #
|
---|
24 | ###########################################################################
|
---|
25 |
|
---|
26 | package metadataaction;
|
---|
27 |
|
---|
28 | use strict;
|
---|
29 |
|
---|
30 | use cgiactions::baseaction;
|
---|
31 |
|
---|
32 | use dbutil;
|
---|
33 | use ghtml;
|
---|
34 |
|
---|
35 | use JSON;
|
---|
36 |
|
---|
37 |
|
---|
38 | BEGIN {
|
---|
39 | # unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan/perl-5.8");
|
---|
40 | require XML::Rules;
|
---|
41 | }
|
---|
42 |
|
---|
43 | @metadataaction::ISA = ('baseaction');
|
---|
44 |
|
---|
45 | my $action_table =
|
---|
46 | {
|
---|
47 | #GET METHODS
|
---|
48 | "get-import-metadata" => {
|
---|
49 | 'compulsory-args' => [ "d", "metaname" ],
|
---|
50 | 'optional-args' => [ "metapos" ] },
|
---|
51 |
|
---|
52 | "get-archives-metadata" => {
|
---|
53 | 'compulsory-args' => [ "d", "metaname" ],
|
---|
54 | 'optional-args' => [ "metapos" ] },
|
---|
55 |
|
---|
56 | "get-index-metadata" => {
|
---|
57 | 'compulsory-args' => [ "d", "metaname" ],
|
---|
58 | 'optional-args' => [ "metapos" ] },
|
---|
59 |
|
---|
60 | "get-metadata" => { # alias for get-index-metadata
|
---|
61 | 'compulsory-args' => [ "d", "metaname" ],
|
---|
62 | 'optional-args' => [ "metapos" ] },
|
---|
63 |
|
---|
64 | "get-live-metadata" => {
|
---|
65 | 'compulsory-args' => [ "d", "metaname" ],
|
---|
66 | 'optional-args' => [ ] },
|
---|
67 |
|
---|
68 | "get-metadata-array" => { # where param can be ONE of: index (default), import, archives, live
|
---|
69 | 'compulsory-args' => [ "json" ],
|
---|
70 | 'optional-args' => [ "where" ],
|
---|
71 | 'help-string' => [
|
---|
72 | 'metadata-server.pl?a=get-metadata-array&c=demo&where=index&json=[{"docid":"HASHc5bce2d6d3e5b04e470ec9","metatable":[{"metaname":"username","metapos":"all"},{"metaname":"usertimestamp","metapos":"all"}, {"metaname":"usercomment","metapos":"all"}]}]'
|
---|
73 | ]},
|
---|
74 |
|
---|
75 | #SET METHODS
|
---|
76 | "set-live-metadata" => {
|
---|
77 | 'compulsory-args' => [ "d", "metaname", "metavalue" ],
|
---|
78 | 'optional-args' => [ ] },
|
---|
79 |
|
---|
80 | "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
|
---|
81 | 'compulsory-args' => [ "metaname", "metavalue" ],
|
---|
82 | 'optional-args' => [ "where", "metapos", "metamode", "prevmetavalue", "d", "f" ] },
|
---|
83 |
|
---|
84 | "set-index-metadata" => {
|
---|
85 | 'compulsory-args' => [ "d", "metaname", "metavalue" ],
|
---|
86 | 'optional-args' => [ "metapos", "metamode" ] },
|
---|
87 |
|
---|
88 | "set-archives-metadata" => {
|
---|
89 | 'compulsory-args' => [ "d", "metaname", "metavalue" ],
|
---|
90 | 'optional-args' => [ "metapos", "metamode", "prevmetavalue" ] }, # metamode can be "accumulate", "override",
|
---|
91 |
|
---|
92 | "set-import-metadata" => {
|
---|
93 | 'compulsory-args' => [ "metaname", "metavalue" ],
|
---|
94 | '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
|
---|
95 |
|
---|
96 | #SET METHODS (ARRAY)
|
---|
97 | "set-metadata-array" => {
|
---|
98 | 'compulsory-args' => [ "where", "json" ],
|
---|
99 | 'optional-args' => [ ],
|
---|
100 | 'help-string' => [
|
---|
101 | '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"}]',
|
---|
102 |
|
---|
103 | '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"}]' ] },
|
---|
104 |
|
---|
105 | # The same examples rewritten for when running the metadata-server.pl script from the commandline:
|
---|
106 |
|
---|
107 | # 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\"}]",
|
---|
108 |
|
---|
109 | # 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\"}]"
|
---|
110 |
|
---|
111 | "set-archives-metadata-array" => {
|
---|
112 | 'compulsory-args' => [ "json" ],
|
---|
113 | 'optional-args' => [ ] },
|
---|
114 |
|
---|
115 | "set-import-metadata-array" => {
|
---|
116 | 'compulsory-args' => [ "json" ],
|
---|
117 | 'optional-args' => [ ] },
|
---|
118 |
|
---|
119 | "set-index-metadata-array" => {
|
---|
120 | 'compulsory-args' => [ "json" ],
|
---|
121 | 'optional-args' => [ ] },
|
---|
122 |
|
---|
123 | "set-live-metadata-array" => {
|
---|
124 | 'compulsory-args' => [ "json" ],
|
---|
125 | 'optional-args' => [ ] },
|
---|
126 |
|
---|
127 | #REMOVE METHODS
|
---|
128 | "remove-import-metadata" => {
|
---|
129 | 'compulsory-args' => [ "d", "metaname", "metavalue" ], #TODO: add f argument
|
---|
130 | 'optional-args' => [ "metapos" ] }, # only provide metapos arg for SIMPLE collections
|
---|
131 |
|
---|
132 | "remove-archives-metadata" => {
|
---|
133 | 'compulsory-args' => [ "d", "metaname" ], #TODO: add f argument
|
---|
134 | 'optional-args' => [ "metapos", "metavalue" ] },
|
---|
135 |
|
---|
136 | "remove-live-metadata" => {
|
---|
137 | 'compulsory-args' => [ "d", "metaname" ],
|
---|
138 | 'optional-args' => [ ] },
|
---|
139 |
|
---|
140 | "remove-index-metadata" => {
|
---|
141 | 'compulsory-args' => [ "d", "metaname" ],
|
---|
142 | 'optional-args' => [ "metapos", "metavalue" ] },
|
---|
143 |
|
---|
144 | "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
|
---|
145 | 'compulsory-args' => [ "d", "metaname" ],
|
---|
146 | 'optional-args' => [ "where", "metapos", "metavalue" ] },
|
---|
147 |
|
---|
148 | #INSERT METHODS
|
---|
149 | "insert-metadata" => {
|
---|
150 | 'compulsory-args' => [ "d", "metaname", "metavalue" ],
|
---|
151 | 'optional-args' => [ ] }
|
---|
152 | };
|
---|
153 |
|
---|
154 |
|
---|
155 | sub new
|
---|
156 | {
|
---|
157 | my $class = shift (@_);
|
---|
158 | my ($gsdl_cgi,$iis6_mode) = @_;
|
---|
159 |
|
---|
160 | # Treat metavalue specially. To transmit this through a GET request
|
---|
161 | # the Javascript side has url-encoded it, so here we need to decode
|
---|
162 | # it before proceeding
|
---|
163 |
|
---|
164 | my $url_encoded_metavalue = $gsdl_cgi->param("metavalue");
|
---|
165 | my $url_decoded_metavalue = &unicode::url_decode($url_encoded_metavalue,1);
|
---|
166 | my $unicode_array = &unicode::utf82unicode($url_decoded_metavalue);
|
---|
167 |
|
---|
168 | $url_decoded_metavalue = join("",map(chr($_),@$unicode_array));
|
---|
169 | $gsdl_cgi->param("metavalue",$url_decoded_metavalue);
|
---|
170 |
|
---|
171 | my $self = new baseaction($action_table,$gsdl_cgi,$iis6_mode);
|
---|
172 |
|
---|
173 | return bless $self, $class;
|
---|
174 | }
|
---|
175 |
|
---|
176 |
|
---|
177 | sub get_live_metadata
|
---|
178 | {
|
---|
179 | my $self = shift @_;
|
---|
180 |
|
---|
181 | my $username = $self->{'username'};
|
---|
182 | my $collect = $self->{'collect'};
|
---|
183 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
184 | my $gsdlhome = $self->{'gsdlhome'};
|
---|
185 | my $infodbtype = $self->{'infodbtype'};
|
---|
186 |
|
---|
187 | # live metadata gets/saves value scoped (prefixed) by the current usename
|
---|
188 | # so (for now) let's not bother to enforce authentication
|
---|
189 |
|
---|
190 | # Obtain the collect dir
|
---|
191 | my $site = $self->{'site'};
|
---|
192 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
193 | ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
194 |
|
---|
195 | # No locking collection when getting metadata, only when setting it
|
---|
196 | # $self->lock_collection($username, $collect); # Make sure the collection isn't locked by someone else
|
---|
197 |
|
---|
198 | # look up additional args
|
---|
199 | my $docid = $self->{'d'};
|
---|
200 | if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
|
---|
201 | $gsdl_cgi->generate_error("No docid (d=...) specified.");
|
---|
202 | }
|
---|
203 |
|
---|
204 | # Generate the dbkey
|
---|
205 | my $metaname = $self->{'metaname'};
|
---|
206 | my $dbkey = "$docid.$metaname";
|
---|
207 |
|
---|
208 | # To people who know $collect_tail please add some comments
|
---|
209 | # Obtain path to the database
|
---|
210 | my $collect_tail = $collect;
|
---|
211 | $collect_tail =~ s/^.*[\/|\\]//;
|
---|
212 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
---|
213 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, "live-$collect_tail", $index_text_directory);
|
---|
214 |
|
---|
215 | # Obtain the content of the key
|
---|
216 | my $cmd = "gdbmget $infodb_file_path $dbkey";
|
---|
217 | if (open(GIN,"$cmd |") == 0) {
|
---|
218 | # Catch error if gdbmget failed
|
---|
219 | my $mess = "Failed to get metadata key: $metaname\n";
|
---|
220 | $mess .= "$!\n";
|
---|
221 |
|
---|
222 | $gsdl_cgi->generate_error($mess);
|
---|
223 | }
|
---|
224 | else {
|
---|
225 | binmode(GIN,":utf8");
|
---|
226 | # Read everything in and concatenate them into $metavalue
|
---|
227 | my $metavalue = "";
|
---|
228 | my $line;
|
---|
229 | while (defined ($line=<GIN>)) {
|
---|
230 | $metavalue .= $line;
|
---|
231 | }
|
---|
232 | close(GIN);
|
---|
233 | chomp($metavalue); # Get rid off the tailing newlines
|
---|
234 | $gsdl_cgi->generate_ok_message("$metavalue");
|
---|
235 | }
|
---|
236 |
|
---|
237 | # Release the lock once it is done
|
---|
238 | # $self->unlock_collection($username, $collect);
|
---|
239 | }
|
---|
240 |
|
---|
241 | # just calls the index version
|
---|
242 | sub get_metadata
|
---|
243 | {
|
---|
244 | my $self = shift @_;
|
---|
245 | $self->get_index_metadata(@_);
|
---|
246 | }
|
---|
247 |
|
---|
248 | # JSON version that will get the requested metadata values
|
---|
249 | # from the requested source (index, import, archives or live)
|
---|
250 | # One of the params is a JSON string and the return value is JSON too
|
---|
251 | # http://forums.asp.net/t/1844684.aspx/1 - Web api method return json in string
|
---|
252 | sub get_metadata_array
|
---|
253 | {
|
---|
254 | my $self = shift @_;
|
---|
255 |
|
---|
256 | my $where = $self->{'where'};
|
---|
257 | if (!$where) {
|
---|
258 | $where = "index"; # default behaviour is to get the values from index
|
---|
259 | }
|
---|
260 |
|
---|
261 | # Only when setting metadata do we perform authentication and do we lock the collection,
|
---|
262 | # not when getting metadata
|
---|
263 |
|
---|
264 | # for get_meta_array, the where param can only be ONE of import, archives, index, live
|
---|
265 | if($where =~ m/index/) {
|
---|
266 | $self->_get_index_metadata_array(@_);
|
---|
267 | }
|
---|
268 | elsif($where =~ m/archives/) {
|
---|
269 | $self->_get_archives_metadata_array(@_);
|
---|
270 | }
|
---|
271 | elsif($where =~ m/import/) {
|
---|
272 | $self->_get_import_metadata_array(@_);
|
---|
273 | }
|
---|
274 | elsif($where =~ m/live/) {
|
---|
275 | $self->_get_live_metadata_array(@_);
|
---|
276 | }
|
---|
277 | }
|
---|
278 |
|
---|
279 | # Unused at present. Added for completion. Tested.
|
---|
280 | sub _get_import_metadata_array {
|
---|
281 |
|
---|
282 | my $self = shift @_;
|
---|
283 |
|
---|
284 | my $collect = $self->{'collect'};
|
---|
285 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
286 | my $site = $self->{'site'};
|
---|
287 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
288 |
|
---|
289 | # look up additional args
|
---|
290 | my $infodbtype = $self->{'infodbtype'};
|
---|
291 |
|
---|
292 | my $archive_dir = &util::filename_cat($collect_dir, $collect, "archives");
|
---|
293 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
---|
294 | my $json_str = $self->{'json'};
|
---|
295 | my $doc_array = decode_json $json_str;
|
---|
296 |
|
---|
297 | my $json_result_str = "[";
|
---|
298 | my $first_doc_rec = 1;
|
---|
299 | foreach my $doc_array_rec ( @$doc_array ) {
|
---|
300 |
|
---|
301 | my $docid = $doc_array_rec->{'docid'}; # no subsection metadata support in metadata.xml, only toplevel meta
|
---|
302 |
|
---|
303 | if($first_doc_rec) {
|
---|
304 | $first_doc_rec = 0;
|
---|
305 | } else {
|
---|
306 | $json_result_str .= ",";
|
---|
307 | }
|
---|
308 | $json_result_str = $json_result_str . "{\"docid\":\"" . $docid . "\"";
|
---|
309 |
|
---|
310 | my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
|
---|
311 | $json_result_str = $json_result_str . ",\"metatable\":[";
|
---|
312 |
|
---|
313 | my $first_rec = 1;
|
---|
314 | foreach my $metatable_rec ( @$metatable ) { # the subarray metatable is an array of hashmaps
|
---|
315 |
|
---|
316 | # Read the docid entry
|
---|
317 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
---|
318 | # This now stores the full pathname
|
---|
319 | my $import_filename = $doc_rec->{'src-file'}->[0];
|
---|
320 |
|
---|
321 | # figure out correct metadata.xml file [?]
|
---|
322 | # Assuming the metadata.xml file is next to the source file
|
---|
323 | # Note: This will not work if it is using the inherited metadata from the parent folder
|
---|
324 | my ($import_tailname, $import_dirname) = File::Basename::fileparse($import_filename);
|
---|
325 | my $metadata_xml_filename = &util::filename_cat($import_dirname, "metadata.xml");
|
---|
326 |
|
---|
327 |
|
---|
328 | if($first_rec) {
|
---|
329 | $first_rec = 0;
|
---|
330 | } else {
|
---|
331 | $json_result_str .= ",";
|
---|
332 | }
|
---|
333 |
|
---|
334 | my $metaname = $metatable_rec->{'metaname'};
|
---|
335 | $json_result_str .= "{\"metaname\":\"$metaname\",\"metavals\":[";
|
---|
336 |
|
---|
337 | my $metapos = $metatable_rec->{'metapos'}; # 0... 1|all|undefined
|
---|
338 | if(!defined $metapos) {
|
---|
339 | $metapos = 0;
|
---|
340 | }
|
---|
341 |
|
---|
342 | # Obtain the specified metadata value(s)
|
---|
343 | my $metavalue;
|
---|
344 |
|
---|
345 | if($metapos ne "all") { # get the value at a single metapos
|
---|
346 | $metavalue = $self->get_metadata_from_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $import_tailname);
|
---|
347 |
|
---|
348 | #print STDERR "**** Metafilename, metaname, metapos, sec_num: $metadata_xml_filename, $metaname, $metapos, $import_tailname\n";
|
---|
349 |
|
---|
350 | $json_result_str .= "{\"metapos\":\"$metapos\",\"metavalue\":\"$metavalue\"}";
|
---|
351 |
|
---|
352 | } else {
|
---|
353 | my $first_metaval = 1;
|
---|
354 | $metapos = 0;
|
---|
355 | $metavalue = $self->get_metadata_from_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $import_tailname);
|
---|
356 |
|
---|
357 | while (defined $metavalue && $metavalue ne "") {
|
---|
358 | if($first_metaval) {
|
---|
359 | $first_metaval = 0;
|
---|
360 | } else {
|
---|
361 | $json_result_str .= ",";
|
---|
362 | }
|
---|
363 |
|
---|
364 | $json_result_str .= "{\"metapos\":\"$metapos\",\"metavalue\":\"$metavalue\"}";
|
---|
365 |
|
---|
366 | $metapos++;
|
---|
367 | $metavalue = $self->get_metadata_from_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $import_tailname);
|
---|
368 | }
|
---|
369 | }
|
---|
370 |
|
---|
371 | $json_result_str .= "]}"; # close metavals array and metatable record
|
---|
372 | }
|
---|
373 |
|
---|
374 | $json_result_str .= "]}"; # close metatable array and docid record
|
---|
375 | }
|
---|
376 |
|
---|
377 | $json_result_str .= "]"; # close array of docids
|
---|
378 | $gsdl_cgi->generate_ok_message($json_result_str."\n");
|
---|
379 | }
|
---|
380 |
|
---|
381 | # Unused method, but included for completion. Tested, works. Takes a JSON string and returns a JSON string.
|
---|
382 | # For more information on the format of the output, see _get_index_metadata_array, which is in use.
|
---|
383 | sub _get_archives_metadata_array {
|
---|
384 |
|
---|
385 | my $self = shift @_;
|
---|
386 |
|
---|
387 | my $collect = $self->{'collect'};
|
---|
388 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
389 | my $site = $self->{'site'};
|
---|
390 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
391 |
|
---|
392 | # look up additional args
|
---|
393 | my $infodbtype = $self->{'infodbtype'};
|
---|
394 |
|
---|
395 | my $archive_dir = &util::filename_cat($collect_dir, $collect, "archives");
|
---|
396 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
---|
397 |
|
---|
398 | my $json_str = $self->{'json'};
|
---|
399 | my $doc_array = decode_json $json_str;
|
---|
400 |
|
---|
401 | my $json_result_str = "[";
|
---|
402 | my $first_doc_rec = 1;
|
---|
403 | foreach my $doc_array_rec ( @$doc_array ) {
|
---|
404 |
|
---|
405 | my $docid = $doc_array_rec->{'docid'};
|
---|
406 |
|
---|
407 | if($first_doc_rec) {
|
---|
408 | $first_doc_rec = 0;
|
---|
409 | } else {
|
---|
410 | $json_result_str .= ",";
|
---|
411 | }
|
---|
412 | $json_result_str = $json_result_str . "{\"docid\":\"" . $docid . "\"";
|
---|
413 |
|
---|
414 | my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
|
---|
415 | $json_result_str = $json_result_str . ",\"metatable\":[";
|
---|
416 |
|
---|
417 | my $first_rec = 1;
|
---|
418 | foreach my $metatable_rec ( @$metatable ) { # the subarray metatable is an array of hashmaps
|
---|
419 |
|
---|
420 | my ($docid, $docid_secnum) = ($doc_array_rec->{'docid'} =~ m/^(.*?)(\..*)?$/);
|
---|
421 | $docid_secnum = "" if (!defined $docid_secnum);
|
---|
422 |
|
---|
423 | # Read the docid entry
|
---|
424 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
---|
425 | # This now stores the full pathname
|
---|
426 | my $doc_filename = $doc_rec->{'doc-file'}->[0];
|
---|
427 | $doc_filename = &util::filename_cat($archive_dir, $doc_filename);
|
---|
428 |
|
---|
429 | if($first_rec) {
|
---|
430 | $first_rec = 0;
|
---|
431 | } else {
|
---|
432 | $json_result_str .= ",";
|
---|
433 | }
|
---|
434 |
|
---|
435 | my $metaname = $metatable_rec->{'metaname'};
|
---|
436 | $json_result_str .= "{\"metaname\":\"$metaname\",\"metavals\":[";
|
---|
437 |
|
---|
438 | my $metapos = $metatable_rec->{'metapos'}; # 0... 1|all|undefined
|
---|
439 | if(!defined $metapos) {
|
---|
440 | $metapos = 0;
|
---|
441 | }
|
---|
442 |
|
---|
443 |
|
---|
444 | # Obtain the specified metadata value(s)
|
---|
445 | my $metavalue;
|
---|
446 |
|
---|
447 | if($metapos ne "all") { # get the value at a single metapos
|
---|
448 |
|
---|
449 | $metavalue = $self->get_metadata_from_archive_xml($gsdl_cgi, $doc_filename, $metaname, $metapos, $docid_secnum);
|
---|
450 | #print STDERR "**** Docname, metaname, metapos, sec_num: $doc_filename, $metaname, $metapos, $docid_secnum\n";
|
---|
451 |
|
---|
452 | $json_result_str .= "{\"metapos\":\"$metapos\",\"metavalue\":\"$metavalue\"}";
|
---|
453 |
|
---|
454 | } else {
|
---|
455 | my $first_metaval = 1;
|
---|
456 | $metapos = 0;
|
---|
457 | $metavalue = $self->get_metadata_from_archive_xml($gsdl_cgi, $doc_filename, $metaname, $metapos, $docid_secnum);
|
---|
458 |
|
---|
459 | while (defined $metavalue && $metavalue ne "") {
|
---|
460 | if($first_metaval) {
|
---|
461 | $first_metaval = 0;
|
---|
462 | } else {
|
---|
463 | $json_result_str .= ",";
|
---|
464 | }
|
---|
465 |
|
---|
466 | $json_result_str .= "{\"metapos\":\"$metapos\",\"metavalue\":\"$metavalue\"}";
|
---|
467 |
|
---|
468 | $metapos++;
|
---|
469 | $metavalue = $self->get_metadata_from_archive_xml($gsdl_cgi, $doc_filename, $metaname, $metapos, $docid_secnum);
|
---|
470 | }
|
---|
471 | }
|
---|
472 |
|
---|
473 | $json_result_str .= "]}"; # close metavals array and metatable record
|
---|
474 | }
|
---|
475 |
|
---|
476 | $json_result_str .= "]}"; # close metatable array and docid record
|
---|
477 | }
|
---|
478 |
|
---|
479 | $json_result_str .= "]"; # close array of docids
|
---|
480 | $gsdl_cgi->generate_ok_message($json_result_str."\n");
|
---|
481 | }
|
---|
482 |
|
---|
483 |
|
---|
484 | # Unused at present. Added for completion. Tested, but not sure if it retrieves metadata in the manner it's expected to.
|
---|
485 | sub _get_live_metadata_array
|
---|
486 | {
|
---|
487 | my $self = shift @_;
|
---|
488 |
|
---|
489 | my $collect = $self->{'collect'};
|
---|
490 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
491 | my $site = $self->{'site'};
|
---|
492 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
493 |
|
---|
494 | # look up additional args
|
---|
495 | my $infodbtype = $self->{'infodbtype'};
|
---|
496 |
|
---|
497 | # To people who know $collect_tail please add some comments
|
---|
498 | # Obtain the path to the database
|
---|
499 | my $collect_tail = $collect;
|
---|
500 | $collect_tail =~ s/^.*[\/|\\]//;
|
---|
501 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
---|
502 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, "live-$collect_tail", $index_text_directory);
|
---|
503 |
|
---|
504 | my $json_str = $self->{'json'};
|
---|
505 | my $doc_array = decode_json $json_str;
|
---|
506 |
|
---|
507 | my $json_result_str = "[";
|
---|
508 | my $first_doc_rec = 1;
|
---|
509 |
|
---|
510 | foreach my $doc_array_rec ( @$doc_array ) {
|
---|
511 |
|
---|
512 | my $docid = $doc_array_rec->{'docid'};
|
---|
513 |
|
---|
514 | if($first_doc_rec) {
|
---|
515 | $first_doc_rec = 0;
|
---|
516 | } else {
|
---|
517 | $json_result_str .= ",";
|
---|
518 | }
|
---|
519 | $json_result_str = $json_result_str . "{\"docid\":\"" . $docid . "\"";
|
---|
520 |
|
---|
521 | my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
|
---|
522 | $json_result_str = $json_result_str . ",\"metatable\":[";
|
---|
523 |
|
---|
524 | my $first_rec = 1;
|
---|
525 | foreach my $metatable_rec ( @$metatable ) { # the subarray metatable is an array of hashmaps
|
---|
526 | if($first_rec) {
|
---|
527 | $first_rec = 0;
|
---|
528 | } else {
|
---|
529 | $json_result_str .= ",";
|
---|
530 | }
|
---|
531 |
|
---|
532 | my $metaname = $metatable_rec->{'metaname'};
|
---|
533 | $json_result_str .= "{\"metaname\":\"$metaname\",\"metavals\":[";
|
---|
534 |
|
---|
535 | # Generate the dbkey
|
---|
536 | my $dbkey = "$docid.$metaname";
|
---|
537 |
|
---|
538 | # metapos for get_live_metadata is always assumed to be "all".
|
---|
539 | # It's always going to get all the lines of metavalues associated with a metaname
|
---|
540 | # (It's the metaname itself that should contain an increment number, if there are to be multiple values).
|
---|
541 | #my $metapos = "all";
|
---|
542 | my $metapos = $metatable_rec->{'metapos'} || 0; # Can be 0... 1|all|undefined. Defaults to 0 if undefined/false
|
---|
543 | my $metavalue = "";
|
---|
544 |
|
---|
545 | # Obtain the content of the key
|
---|
546 | my $cmd = "gdbmget $infodb_file_path $dbkey";
|
---|
547 | if (open(GIN,"$cmd |") != 0) { # Success.
|
---|
548 |
|
---|
549 | binmode(GIN,":utf8");
|
---|
550 | # Read everything in and concatenate them into $metavalue
|
---|
551 | my $line;
|
---|
552 | my $first_metaval = 1;
|
---|
553 | my $pos = 0;
|
---|
554 | while (defined ($line=<GIN>)) {
|
---|
555 | chomp($line); # Get rid off the tailing newlines
|
---|
556 |
|
---|
557 | if($metapos eq "all") {
|
---|
558 | if($first_metaval) {
|
---|
559 | $first_metaval = 0;
|
---|
560 | } else {
|
---|
561 | $json_result_str .= ",";
|
---|
562 | }
|
---|
563 | $metavalue = $line;
|
---|
564 | $json_result_str .= "{\"metapos\":\"$pos\",\"metavalue\":\"$metavalue\"}";
|
---|
565 | } elsif($metapos == $pos) {
|
---|
566 | $metavalue = $line;
|
---|
567 | $json_result_str .= "{\"metapos\":\"$metapos\",\"metavalue\":\"$metavalue\"}";
|
---|
568 | last;
|
---|
569 | } # else, the current $pos is not the required $metapos
|
---|
570 | $pos += 1;
|
---|
571 | }
|
---|
572 | close(GIN);
|
---|
573 | } # else open cmd == 0 (failed) and metavals array will be empty [] for this metaname
|
---|
574 |
|
---|
575 | $json_result_str .= "]}"; # close metavals array and metatable record
|
---|
576 | }
|
---|
577 |
|
---|
578 | $json_result_str .= "]}"; # close metatable array and docid record
|
---|
579 | }
|
---|
580 |
|
---|
581 | $json_result_str .= "]"; # close array of docids
|
---|
582 |
|
---|
583 | $gsdl_cgi->generate_ok_message($json_result_str."\n");
|
---|
584 | }
|
---|
585 |
|
---|
586 |
|
---|
587 | # Takes a JSON string and returns a JSON string
|
---|
588 | # Request string is of the form:
|
---|
589 | # http://localhost:8283/greenstone/cgi-bin/metadata-server.pl?a=get-metadata-array&c=demo&where=index&json=[{"docid":"HASHc5bce2d6d3e5b04e470ec9","metatable":[{"metaname":"username","metapos":"all"},{"metaname":"usertimestamp","metapos":"all"}, {"metaname":"usercomment","metapos":"all"}]}]
|
---|
590 | # Resulting string is of the form:
|
---|
591 | # [{"docid":"HASHc5bce2d6d3e5b04e470ec9","metatable":[{"metaname":"username","metavals":[{"metapos":"0","metavalue":"me"},{"metapos":"1","metavalue":"admin"}]},{"metaname":"usertimestamp","metavals":[{"metapos":"0","metavalue":"1367900586888"},{"metapos":"1","metavalue":"1367900616574"}]},{"metaname":"usercomment","metavals":[{"metapos":"0","metavalue":"Hi"},{"metapos":"1","metavalue":"Hello"}]}]}]
|
---|
592 | sub _get_index_metadata_array
|
---|
593 | {
|
---|
594 | my $self = shift @_;
|
---|
595 |
|
---|
596 | my $collect = $self->{'collect'};
|
---|
597 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
598 | my $site = $self->{'site'};
|
---|
599 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
600 |
|
---|
601 | # look up additional args
|
---|
602 | my $infodbtype = $self->{'infodbtype'};
|
---|
603 |
|
---|
604 | # To people who know $collect_tail please add some comments
|
---|
605 | # Obtain the path to the database
|
---|
606 | my $collect_tail = $collect;
|
---|
607 | $collect_tail =~ s/^.*[\/|\\]//;
|
---|
608 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
---|
609 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, $collect_tail, $index_text_directory);
|
---|
610 |
|
---|
611 | my $json_str = $self->{'json'};
|
---|
612 | my $doc_array = decode_json $json_str;
|
---|
613 |
|
---|
614 | my $json_result_str = "[";
|
---|
615 | my $first_doc_rec = 1;
|
---|
616 |
|
---|
617 | foreach my $doc_array_rec ( @$doc_array ) {
|
---|
618 |
|
---|
619 | my $docid = $doc_array_rec->{'docid'};
|
---|
620 |
|
---|
621 | if($first_doc_rec) {
|
---|
622 | $first_doc_rec = 0;
|
---|
623 | } else {
|
---|
624 | $json_result_str .= ",";
|
---|
625 | }
|
---|
626 | $json_result_str = $json_result_str . "{\"docid\":\"" . $docid . "\"";
|
---|
627 |
|
---|
628 | my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
|
---|
629 | $json_result_str = $json_result_str . ",\"metatable\":[";
|
---|
630 |
|
---|
631 | my $first_rec = 1;
|
---|
632 | foreach my $metatable_rec ( @$metatable ) { # the subarray metatable is an array of hashmaps
|
---|
633 | if($first_rec) {
|
---|
634 | $first_rec = 0;
|
---|
635 | } else {
|
---|
636 | $json_result_str .= ",";
|
---|
637 | }
|
---|
638 |
|
---|
639 | my $metaname = $metatable_rec->{'metaname'};
|
---|
640 | $json_result_str .= "{\"metaname\":\"$metaname\",\"metavals\":[";
|
---|
641 |
|
---|
642 | my $metapos = $metatable_rec->{'metapos'}; # 0... 1|all|undefined
|
---|
643 | if(!defined $metapos) {
|
---|
644 | $metapos = 0;
|
---|
645 | }
|
---|
646 |
|
---|
647 | # Read the docid entry
|
---|
648 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $infodb_file_path, $docid);
|
---|
649 |
|
---|
650 | # Basically loop through and unescape_html the values
|
---|
651 | foreach my $k (keys %$doc_rec) {
|
---|
652 | my @escaped_v = ();
|
---|
653 | foreach my $v (@{$doc_rec->{$k}}) {
|
---|
654 | my $ev = &ghtml::unescape_html($v);
|
---|
655 | push(@escaped_v, $ev);
|
---|
656 | }
|
---|
657 | $doc_rec->{$k} = \@escaped_v;
|
---|
658 | }
|
---|
659 |
|
---|
660 | # Obtain the specified metadata value(s)
|
---|
661 | my $metavalue;
|
---|
662 |
|
---|
663 | if($metapos ne "all") { # get the value at a single metapos
|
---|
664 |
|
---|
665 | $metavalue = $doc_rec->{$metaname}->[$metapos];
|
---|
666 |
|
---|
667 | # protect any double quotes and colons in the metavalue before putting it into JSON
|
---|
668 | $metavalue =~ s/\"/"/g if defined $metavalue;
|
---|
669 | $metavalue =~ s/\:/&58;/g if defined $metavalue;
|
---|
670 |
|
---|
671 | $json_result_str .= "{\"metapos\":\"$metapos\",\"metavalue\":\"$metavalue\"}";
|
---|
672 |
|
---|
673 | } else {
|
---|
674 | my $first_metaval = 1;
|
---|
675 | $metapos = 0;
|
---|
676 | $metavalue = $doc_rec->{$metaname}->[$metapos];
|
---|
677 |
|
---|
678 | while (defined $metavalue) {
|
---|
679 |
|
---|
680 | # protect any double quotes and colons in the metavalue before putting it into JSON
|
---|
681 | $metavalue =~ s/\"/"/g;
|
---|
682 | $metavalue =~ s/\:/&58;/g;
|
---|
683 |
|
---|
684 | if($first_metaval) {
|
---|
685 | $first_metaval = 0;
|
---|
686 | } else {
|
---|
687 | $json_result_str .= ",";
|
---|
688 | }
|
---|
689 |
|
---|
690 | $json_result_str .= "{\"metapos\":\"$metapos\",\"metavalue\":\"$metavalue\"}";
|
---|
691 |
|
---|
692 | $metapos++;
|
---|
693 | $metavalue = $doc_rec->{$metaname}->[$metapos];
|
---|
694 | }
|
---|
695 | }
|
---|
696 |
|
---|
697 | $json_result_str .= "]}"; # close metavals array and metatable record
|
---|
698 | }
|
---|
699 |
|
---|
700 | $json_result_str .= "]}"; # close metatable array and docid record
|
---|
701 | }
|
---|
702 |
|
---|
703 | $json_result_str .= "]"; # close array of docids
|
---|
704 |
|
---|
705 | $gsdl_cgi->generate_ok_message($json_result_str."\n");
|
---|
706 | }
|
---|
707 |
|
---|
708 |
|
---|
709 | sub get_index_metadata
|
---|
710 | {
|
---|
711 | my $self = shift @_;
|
---|
712 |
|
---|
713 | my $username = $self->{'username'};
|
---|
714 | my $collect = $self->{'collect'};
|
---|
715 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
716 | my $gsdlhome = $self->{'gsdlhome'};
|
---|
717 |
|
---|
718 | # Obtain the collect dir
|
---|
719 | my $site = $self->{'site'};
|
---|
720 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
721 | ##my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
722 |
|
---|
723 | # look up additional args
|
---|
724 | my $docid = $self->{'d'};
|
---|
725 | my $metaname = $self->{'metaname'};
|
---|
726 | my $metapos = $self->{'metapos'};
|
---|
727 | my $infodbtype = $self->{'infodbtype'};
|
---|
728 |
|
---|
729 | # To people who know $collect_tail please add some comments
|
---|
730 | # Obtain path to the database
|
---|
731 | my $collect_tail = $collect;
|
---|
732 | $collect_tail =~ s/^.*[\/|\\]//;
|
---|
733 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
---|
734 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, $collect_tail, $index_text_directory);
|
---|
735 |
|
---|
736 | # Read the docid entry
|
---|
737 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $infodb_file_path, $docid);
|
---|
738 |
|
---|
739 | # Basically loop through and unescape_html the values
|
---|
740 | foreach my $k (keys %$doc_rec) {
|
---|
741 | my @escaped_v = ();
|
---|
742 | foreach my $v (@{$doc_rec->{$k}}) {
|
---|
743 | my $ev = &ghtml::unescape_html($v);
|
---|
744 | push(@escaped_v, $ev);
|
---|
745 | }
|
---|
746 | $doc_rec->{$k} = \@escaped_v;
|
---|
747 | }
|
---|
748 |
|
---|
749 | # Obtain the specified metadata value
|
---|
750 | $metapos = 0 if (!defined $metapos);
|
---|
751 | my $metavalue = $doc_rec->{$metaname}->[$metapos];
|
---|
752 | $gsdl_cgi->generate_ok_message("$metavalue");
|
---|
753 |
|
---|
754 | }
|
---|
755 |
|
---|
756 |
|
---|
757 | sub get_import_metadata
|
---|
758 | {
|
---|
759 | my $self = shift @_;
|
---|
760 |
|
---|
761 | my $username = $self->{'username'};
|
---|
762 | my $collect = $self->{'collect'};
|
---|
763 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
764 | my $gsdlhome = $self->{'gsdlhome'};
|
---|
765 |
|
---|
766 | # Obtain the collect dir
|
---|
767 | my $site = $self->{'site'};
|
---|
768 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
769 | ##my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
770 |
|
---|
771 | # look up additional args
|
---|
772 | my $docid = $self->{'d'};
|
---|
773 | my $metaname = $self->{'metaname'};
|
---|
774 | my $metapos = $self->{'metapos'};
|
---|
775 | $metapos = 0 if (!defined $metapos); # gets the first value by default since metapos defaults to 0
|
---|
776 |
|
---|
777 | my $infodbtype = $self->{'infodbtype'};
|
---|
778 | if (!defined $docid)
|
---|
779 | {
|
---|
780 | $gsdl_cgi->generate_error("No docid (d=...) specified.\n");
|
---|
781 | }
|
---|
782 |
|
---|
783 | # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
|
---|
784 | # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
|
---|
785 | my $metadata_xml_file;
|
---|
786 | my $import_filename = undef;
|
---|
787 |
|
---|
788 |
|
---|
789 | my $archive_dir = &util::filename_cat($collect_dir, $collect, "archives");
|
---|
790 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
---|
791 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
---|
792 |
|
---|
793 | # This now stores the full pathname
|
---|
794 | $import_filename = $doc_rec->{'src-file'}->[0];
|
---|
795 |
|
---|
796 | # figure out correct metadata.xml file [?]
|
---|
797 | # Assuming the metadata.xml file is next to the source file
|
---|
798 | # Note: This will not work if it is using the inherited metadata from the parent folder
|
---|
799 | my ($import_tailname, $import_dirname) = File::Basename::fileparse($import_filename);
|
---|
800 | my $metadata_xml_filename = &util::filename_cat($import_dirname, "metadata.xml");
|
---|
801 |
|
---|
802 | $gsdl_cgi->generate_ok_message($self->get_metadata_from_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $import_tailname));
|
---|
803 |
|
---|
804 | }
|
---|
805 |
|
---|
806 | sub get_metadata_from_metadata_xml
|
---|
807 | {
|
---|
808 | my $self = shift @_;
|
---|
809 | my ($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $src_file) = @_;
|
---|
810 |
|
---|
811 | my @rules =
|
---|
812 | (
|
---|
813 | _default => 'raw',
|
---|
814 | 'Metadata' => \&gfmxml_metadata,
|
---|
815 | 'FileName' => \&mxml_filename
|
---|
816 | );
|
---|
817 |
|
---|
818 | my $parser = XML::Rules->new
|
---|
819 | (
|
---|
820 | rules => \@rules,
|
---|
821 | output_encoding => 'utf8'
|
---|
822 | );
|
---|
823 |
|
---|
824 | my $xml_in = "";
|
---|
825 | if (!open(MIN,"<$metadata_xml_filename"))
|
---|
826 | {
|
---|
827 | $gsdl_cgi->generate_error("Unable to read in $metadata_xml_filename: $!");
|
---|
828 | }
|
---|
829 | else
|
---|
830 | {
|
---|
831 | # Read them in
|
---|
832 | my $line;
|
---|
833 | while (defined ($line=<MIN>)) {
|
---|
834 | $xml_in .= $line;
|
---|
835 | }
|
---|
836 | close(MIN);
|
---|
837 |
|
---|
838 | $parser->parse($xml_in, {metaname => $metaname, metapos => $metapos, src_file => $src_file});
|
---|
839 |
|
---|
840 | if(defined $parser->{'pad'}->{'metavalue'})
|
---|
841 | {
|
---|
842 | return $parser->{'pad'}->{'metavalue'};
|
---|
843 | }
|
---|
844 | else
|
---|
845 | {
|
---|
846 | return "";
|
---|
847 | }
|
---|
848 | }
|
---|
849 | }
|
---|
850 |
|
---|
851 | sub gfmxml_metadata
|
---|
852 | {
|
---|
853 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
854 |
|
---|
855 | # no subsection support yet in metadata.xml
|
---|
856 |
|
---|
857 | if (($parser->{'parameters'}->{'src_file'} eq $parser->{'parameters'}->{'current_file'}) && $parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'})
|
---|
858 | {
|
---|
859 | if (!defined $parser->{'parameters'}->{'poscount'})
|
---|
860 | {
|
---|
861 | $parser->{'parameters'}->{'poscount'} = 0;
|
---|
862 | }
|
---|
863 | else
|
---|
864 | {
|
---|
865 | $parser->{'parameters'}->{'poscount'}++;
|
---|
866 | }
|
---|
867 |
|
---|
868 | # gets the first value by default, since metapos defaults to 0
|
---|
869 | if (($parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'}))
|
---|
870 | {
|
---|
871 | if($parser->{'parameters'}->{'metapos'} > 0) {
|
---|
872 | print STDERR "@@@@ WARNING: non-zero metapos.\n";
|
---|
873 | print STDERR "@@@@ Assuming SIMPLE collection and proceeding to retrieve the import meta at position: ".$parser->{'parameters'}->{'metapos'}.".\n";
|
---|
874 | }
|
---|
875 | $parser->{'pad'}->{'metavalue'} = $attrHash->{'_content'};
|
---|
876 | }
|
---|
877 | }
|
---|
878 | }
|
---|
879 |
|
---|
880 | sub get_archives_metadata
|
---|
881 | {
|
---|
882 | my $self = shift @_;
|
---|
883 |
|
---|
884 | my $username = $self->{'username'};
|
---|
885 | my $collect = $self->{'collect'};
|
---|
886 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
887 | # my $gsdlhome = $self->{'gsdlhome'};
|
---|
888 | my $infodbtype = $self->{'infodbtype'};
|
---|
889 |
|
---|
890 | # Obtain the collect dir
|
---|
891 | my $site = $self->{'site'};
|
---|
892 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
893 |
|
---|
894 | my $archive_dir = &util::filename_cat($collect_dir, $collect, "archives");
|
---|
895 |
|
---|
896 | # look up additional args
|
---|
897 | my ($docid, $docid_secnum) = ($self->{'d'} =~ m/^(.*?)(\..*)?$/);
|
---|
898 | $docid_secnum = "" if (!defined $docid_secnum);
|
---|
899 |
|
---|
900 | my $metaname = $self->{'metaname'};
|
---|
901 | my $metapos = $self->{'metapos'};
|
---|
902 | $metapos = 0 if (!defined $metapos);
|
---|
903 |
|
---|
904 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
---|
905 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
---|
906 |
|
---|
907 | # This now stores the full pathname
|
---|
908 | my $doc_filename = $doc_rec->{'doc-file'}->[0];
|
---|
909 |
|
---|
910 | $gsdl_cgi->generate_ok_message($self->get_metadata_from_archive_xml($gsdl_cgi, &util::filename_cat($archive_dir, $doc_filename), $metaname, $metapos, $docid_secnum));
|
---|
911 |
|
---|
912 | }
|
---|
913 |
|
---|
914 | sub get_metadata_from_archive_xml
|
---|
915 | {
|
---|
916 | my $self = shift @_;
|
---|
917 | my ($gsdl_cgi, $doc_xml_filename, $metaname, $metapos, $secid) = @_;
|
---|
918 |
|
---|
919 | my @start_rules = ('Section' => \&dxml_start_section);
|
---|
920 |
|
---|
921 | my @rules =
|
---|
922 | (
|
---|
923 | _default => 'raw',
|
---|
924 | 'Metadata' => \&gfdxml_metadata
|
---|
925 | );
|
---|
926 |
|
---|
927 | my $parser = XML::Rules->new
|
---|
928 | (
|
---|
929 | start_rules => \@start_rules,
|
---|
930 | rules => \@rules,
|
---|
931 | output_encoding => 'utf8'
|
---|
932 | );
|
---|
933 |
|
---|
934 | my $xml_in = "";
|
---|
935 | if (!open(MIN,"<$doc_xml_filename"))
|
---|
936 | {
|
---|
937 | $gsdl_cgi->generate_error("Unable to read in $doc_xml_filename: $!");
|
---|
938 | }
|
---|
939 | else
|
---|
940 | {
|
---|
941 | # Read them in
|
---|
942 | my $line;
|
---|
943 | while (defined ($line=<MIN>)) {
|
---|
944 | $xml_in .= $line;
|
---|
945 | }
|
---|
946 | close(MIN);
|
---|
947 |
|
---|
948 | $parser->parse($xml_in, {metaname => $metaname, metapos => $metapos, secid => $secid});
|
---|
949 |
|
---|
950 | if(defined $parser->{'pad'}->{'metavalue'})
|
---|
951 | {
|
---|
952 | return $parser->{'pad'}->{'metavalue'};
|
---|
953 | }
|
---|
954 | else
|
---|
955 | {
|
---|
956 | return "";
|
---|
957 | }
|
---|
958 | }
|
---|
959 | }
|
---|
960 |
|
---|
961 | sub gfdxml_metadata
|
---|
962 | {
|
---|
963 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
964 |
|
---|
965 | if(!($parser->{'parameters'}->{'secid'} eq $parser->{'parameters'}->{'curr_section_num'}))
|
---|
966 | {
|
---|
967 | return;
|
---|
968 | }
|
---|
969 |
|
---|
970 | if ($parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'})
|
---|
971 | {
|
---|
972 | if (!defined $parser->{'parameters'}->{'poscount'})
|
---|
973 | {
|
---|
974 | $parser->{'parameters'}->{'poscount'} = 0;
|
---|
975 | }
|
---|
976 | else
|
---|
977 | {
|
---|
978 | $parser->{'parameters'}->{'poscount'}++;
|
---|
979 | }
|
---|
980 | }
|
---|
981 |
|
---|
982 | if (($parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'}) && ($parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'}))
|
---|
983 | {
|
---|
984 | $parser->{'pad'}->{'metavalue'} = $attrHash->{'_content'};
|
---|
985 | }
|
---|
986 | }
|
---|
987 |
|
---|
988 | sub _set_live_metadata
|
---|
989 | {
|
---|
990 | my $self = shift @_;
|
---|
991 |
|
---|
992 | my $collect = $self->{'collect'};
|
---|
993 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
994 | #my $gsdlhome = $self->{'gsdlhome'};
|
---|
995 | my $infodbtype = $self->{'infodbtype'};
|
---|
996 |
|
---|
997 | # Obtain the collect dir
|
---|
998 | my $site = $self->{'site'};
|
---|
999 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
1000 | ##my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
1001 |
|
---|
1002 |
|
---|
1003 | # look up additional args
|
---|
1004 | my $docid = $self->{'d'};
|
---|
1005 | if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
|
---|
1006 | $gsdl_cgi->generate_error("No docid (d=...) specified."); # generates error and dies
|
---|
1007 | }
|
---|
1008 | my $metavalue = $self->{'metavalue'};
|
---|
1009 |
|
---|
1010 | # Generate the dbkey
|
---|
1011 | my $metaname = $self->{'metaname'};
|
---|
1012 | my $dbkey = "$docid.$metaname";
|
---|
1013 |
|
---|
1014 | # To people who know $collect_tail please add some comments
|
---|
1015 | # Obtain path to the database
|
---|
1016 | my $collect_tail = $collect;
|
---|
1017 | $collect_tail =~ s/^.*[\/|\\]//;
|
---|
1018 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
---|
1019 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, "live-$collect_tail", $index_text_directory);
|
---|
1020 |
|
---|
1021 | # Set the new value
|
---|
1022 | my $cmd = "gdbmset \"$infodb_file_path\" \"$dbkey\" \"$metavalue\"";
|
---|
1023 | my $status = system($cmd);
|
---|
1024 | if ($status != 0) {
|
---|
1025 | # Catch error if gdbmget failed
|
---|
1026 | my $mess = "Failed to set metadata key: $dbkey\n";
|
---|
1027 |
|
---|
1028 | $mess .= "PATH: $ENV{'PATH'}\n";
|
---|
1029 | $mess .= "cmd = $cmd\n";
|
---|
1030 | $mess .= "Exit status: $status\n";
|
---|
1031 | $mess .= "System Error Message: $!\n";
|
---|
1032 |
|
---|
1033 | $gsdl_cgi->generate_error($mess);
|
---|
1034 | }
|
---|
1035 | else {
|
---|
1036 | $gsdl_cgi->generate_ok_message("set-live-metadata successful: Key[$metaname]=$metavalue");
|
---|
1037 | }
|
---|
1038 |
|
---|
1039 | #return $status; # in case calling functions have any further use for this
|
---|
1040 | }
|
---|
1041 |
|
---|
1042 | sub set_live_metadata
|
---|
1043 | {
|
---|
1044 | my $self = shift @_;
|
---|
1045 |
|
---|
1046 | my $username = $self->{'username'};
|
---|
1047 | my $collect = $self->{'collect'};
|
---|
1048 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
1049 |
|
---|
1050 | if ($baseaction::authentication_enabled) {
|
---|
1051 | # Ensure the user is allowed to edit this collection
|
---|
1052 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
1053 | }
|
---|
1054 |
|
---|
1055 | # Make sure the collection isn't locked by someone else
|
---|
1056 | $self->lock_collection($username, $collect);
|
---|
1057 |
|
---|
1058 | $self->_set_live_metadata(@_);
|
---|
1059 |
|
---|
1060 | # Release the lock once it is done
|
---|
1061 | $self->unlock_collection($username, $collect);
|
---|
1062 | }
|
---|
1063 |
|
---|
1064 | sub set_index_metadata_entry
|
---|
1065 | {
|
---|
1066 | my $self = shift @_;
|
---|
1067 | my ($collect_dir,$collect,$infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode,$prevmetavalue) = @_;
|
---|
1068 |
|
---|
1069 | # To people who know $collect_tail please add some comments
|
---|
1070 | # Obtain path to the database
|
---|
1071 | my $collect_tail = $collect;
|
---|
1072 | $collect_tail =~ s/^.*[\/|\\]//;
|
---|
1073 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
---|
1074 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, $collect_tail, $index_text_directory);
|
---|
1075 |
|
---|
1076 | # print STDERR "**** infodb file path = $infodb_file_path\n";
|
---|
1077 | # print STDERR "***** infodb type = $infodbtype\n";
|
---|
1078 |
|
---|
1079 | # Read the docid entry
|
---|
1080 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $infodb_file_path, $docid);
|
---|
1081 |
|
---|
1082 | # Set the metadata value
|
---|
1083 | if (defined $metapos) {
|
---|
1084 | # if metamode=accumulate AND metapos, warn user and then use metapos
|
---|
1085 | if (defined $metamode && $metamode eq "accumulate") {
|
---|
1086 | print STDERR "**** Warning: metamode is set to accumulate yet metapos is also provided for $docid\n";
|
---|
1087 | print STDERR "**** Proceeding by using metapos\n";
|
---|
1088 | }
|
---|
1089 | $doc_rec->{$metaname}->[$metapos] = $metavalue;
|
---|
1090 | }
|
---|
1091 | elsif (defined $prevmetavalue) {
|
---|
1092 | my $array = $doc_rec->{$metaname};
|
---|
1093 | my $length = @$array;
|
---|
1094 |
|
---|
1095 | my $found = 0;
|
---|
1096 | for(my $i = 0; $i < $length; $i++){
|
---|
1097 | if(defined $doc_rec->{$metaname}->[$i] && $doc_rec->{$metaname}->[$i] eq $prevmetavalue){
|
---|
1098 | $doc_rec->{$metaname}->[$i] = $metavalue;
|
---|
1099 | $found = 1;
|
---|
1100 | last;
|
---|
1101 | }
|
---|
1102 | }
|
---|
1103 |
|
---|
1104 | if($found == 0){
|
---|
1105 | $doc_rec->{$metaname} = [ $metavalue ];
|
---|
1106 | }
|
---|
1107 | }
|
---|
1108 | elsif (defined $metamode && $metamode eq "override") {
|
---|
1109 | $doc_rec->{$metaname} = [ $metavalue ];
|
---|
1110 | }
|
---|
1111 | else { # default for index was to override, but because accumulate is less destructive,
|
---|
1112 | # and because accumulate is the default for archives and import, that's the new default for index too
|
---|
1113 | if(defined $doc_rec->{$metaname}) {
|
---|
1114 | push(@{$doc_rec->{$metaname}}, $metavalue); # accumulate the value for that metaname
|
---|
1115 | } else {
|
---|
1116 | $doc_rec->{$metaname} = [ $metavalue ];
|
---|
1117 | }
|
---|
1118 | }
|
---|
1119 |
|
---|
1120 | my $status = &dbutil::set_infodb_entry($infodbtype, $infodb_file_path,$docid,$doc_rec);
|
---|
1121 |
|
---|
1122 | return $status;
|
---|
1123 |
|
---|
1124 | }
|
---|
1125 |
|
---|
1126 | sub _set_import_metadata
|
---|
1127 | {
|
---|
1128 | my $self = shift @_;
|
---|
1129 |
|
---|
1130 | my $collect = $self->{'collect'};
|
---|
1131 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
1132 | my $infodbtype = $self->{'infodbtype'};
|
---|
1133 | # my $gsdlhome = $self->{'gsdlhome'};
|
---|
1134 |
|
---|
1135 | # Obtain the collect and archive dir
|
---|
1136 | my $site = $self->{'site'};
|
---|
1137 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
1138 | ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
1139 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
---|
1140 |
|
---|
1141 | # look up additional args
|
---|
1142 | # want either d= or f=
|
---|
1143 | my $docid = $self->{'d'};
|
---|
1144 | my ($docid_root,$docid_secnum);
|
---|
1145 | if(defined $docid) {
|
---|
1146 | ($docid_root,$docid_secnum) = ($docid =~ m/^(.*?)(\..*)?$/);
|
---|
1147 | # as yet no support for setting subsection metadata in metadata.xml
|
---|
1148 | if ((defined $docid_secnum) && ($docid_secnum !~ m/^\s*$/)) {
|
---|
1149 | $gsdl_cgi->generate_message("*** No support yet for setting import metadata at subsections level.\n");
|
---|
1150 | return;
|
---|
1151 | }
|
---|
1152 | }
|
---|
1153 |
|
---|
1154 | my $import_file = $self->{'f'};
|
---|
1155 | if ((!defined $docid) && (!defined $import_file)) {
|
---|
1156 | $gsdl_cgi->generate_error("No docid (d=...) or import file (f=) specified.");
|
---|
1157 | }
|
---|
1158 |
|
---|
1159 | # Get the parameters and set default mode to "accumulate"
|
---|
1160 | my $metaname = $self->{'metaname'};
|
---|
1161 | my $metavalue = $self->{'metavalue'};
|
---|
1162 | ## $metavalue =~ s/&lt;(.*?)&gt;/<$1>/g;
|
---|
1163 | $metavalue =~ s/<(.*?)>/<$1>/g;
|
---|
1164 |
|
---|
1165 | my $metamode = $self->{'metamode'};
|
---|
1166 | if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
|
---|
1167 | # make "accumulate" the default (less destructive, as won't actually
|
---|
1168 | # delete any existing values)
|
---|
1169 | $metamode = "accumulate";
|
---|
1170 | }
|
---|
1171 |
|
---|
1172 | # adding metapos and prevmetavalue support to import_metadata subroutines
|
---|
1173 | my $metapos = $self->{'metapos'}; # don't force undef to 0. Undef has meaning when metamode=override
|
---|
1174 | my $prevmetavalue = $self->{'prevmetavalue'};
|
---|
1175 |
|
---|
1176 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
---|
1177 | 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
|
---|
1178 |
|
---|
1179 | my $mess = "set-import-metadata successful: Key[$docid] -> $metadata_xml_filename\n";
|
---|
1180 | $mess .= " $metaname";
|
---|
1181 | $mess .= " = $metavalue";
|
---|
1182 | $mess .= " ($metamode)\n";
|
---|
1183 |
|
---|
1184 | $gsdl_cgi->generate_ok_message($mess);
|
---|
1185 |
|
---|
1186 | #return $status; # in case calling functions have any further use for this
|
---|
1187 | }
|
---|
1188 |
|
---|
1189 | # the version of set_index_meta that doesn't do authentication
|
---|
1190 | sub _set_archives_metadata
|
---|
1191 | {
|
---|
1192 | my $self = shift @_;
|
---|
1193 |
|
---|
1194 | my $collect = $self->{'collect'};
|
---|
1195 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
1196 | my $infodbtype = $self->{'infodbtype'};
|
---|
1197 |
|
---|
1198 | # Obtain the collect and archive dir
|
---|
1199 | my $site = $self->{'site'};
|
---|
1200 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
1201 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
---|
1202 |
|
---|
1203 | # look up additional args
|
---|
1204 | my $docid = $self->{'d'};
|
---|
1205 | my $metaname = $self->{'metaname'};
|
---|
1206 | my $metavalue = $self->{'metavalue'};
|
---|
1207 | my $prevmetavalue = $self->{'prevmetavalue'};
|
---|
1208 |
|
---|
1209 | my $metapos = $self->{'metapos'}; # don't force undef to 0. Undef has meaning when metamode=override
|
---|
1210 | # Don't append "|| undef", since if metapos=0 it will then be set to undef
|
---|
1211 |
|
---|
1212 | my $metamode = $self->{'metamode'};
|
---|
1213 | if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
|
---|
1214 | # make "accumulate" the default (less destructive, as won't actually
|
---|
1215 | # delete any existing values)
|
---|
1216 | $metamode = "accumulate";
|
---|
1217 | }
|
---|
1218 |
|
---|
1219 | my $status = $self->set_archives_metadata_entry($gsdl_cgi,$archive_dir, $collect_dir,$collect, $infodbtype,$docid,
|
---|
1220 | $metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
|
---|
1221 |
|
---|
1222 | if ($status == 0) {
|
---|
1223 | my $mess = "set-archives-metadata successful: Key[$docid]\n";
|
---|
1224 | $mess .= " $metaname";
|
---|
1225 | $mess .= "->[$metapos]" if (defined $metapos);
|
---|
1226 | $mess .= " = $metavalue";
|
---|
1227 | $mess .= " ($metamode)\n";
|
---|
1228 |
|
---|
1229 | $gsdl_cgi->generate_ok_message($mess);
|
---|
1230 | }
|
---|
1231 | else {
|
---|
1232 | my $mess .= "Failed to set archives metadata key: $docid\n";
|
---|
1233 | $mess .= "Exit status: $status\n";
|
---|
1234 | if(defined $self->{'error_msg'}) {
|
---|
1235 | $mess .= "Error Message: $self->{'error_msg'}\n";
|
---|
1236 | } else {
|
---|
1237 | $mess .= "System Error Message: $!\n";
|
---|
1238 | }
|
---|
1239 | $mess .= "-" x 20 . "\n";
|
---|
1240 |
|
---|
1241 | $gsdl_cgi->generate_error($mess);
|
---|
1242 | }
|
---|
1243 |
|
---|
1244 | #return $status; # in case calling functions have any further use for this
|
---|
1245 | }
|
---|
1246 |
|
---|
1247 |
|
---|
1248 | # the version of set_index_meta that doesn't do authentication
|
---|
1249 | sub _set_index_metadata
|
---|
1250 | {
|
---|
1251 | my $self = shift @_;
|
---|
1252 |
|
---|
1253 | my $collect = $self->{'collect'};
|
---|
1254 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
1255 |
|
---|
1256 | my $site = $self->{'site'};
|
---|
1257 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
1258 |
|
---|
1259 | # look up additional args
|
---|
1260 | my $docid = $self->{'d'};
|
---|
1261 | my $metaname = $self->{'metaname'};
|
---|
1262 | my $metapos = $self->{'metapos'}; # undef has meaning
|
---|
1263 | my $metavalue = $self->{'metavalue'};
|
---|
1264 | my $infodbtype = $self->{'infodbtype'};
|
---|
1265 | my $metamode = $self->{'metamode'};
|
---|
1266 | my $prevmetavalue = $self->{'prevmetavalue'};
|
---|
1267 |
|
---|
1268 | my $status = $self->set_index_metadata_entry($collect_dir,$collect,$infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
|
---|
1269 |
|
---|
1270 | if ($status != 0) {
|
---|
1271 | # Catch error if set infodb entry failed
|
---|
1272 | my $mess = "Failed to set metadata key: $docid\n";
|
---|
1273 |
|
---|
1274 | $mess .= "PATH: $ENV{'PATH'}\n";
|
---|
1275 | $mess .= "Exit status: $status\n";
|
---|
1276 | $mess .= "System Error Message: $!\n";
|
---|
1277 |
|
---|
1278 | $gsdl_cgi->generate_error($mess);
|
---|
1279 | }
|
---|
1280 | else {
|
---|
1281 | my $mess = "set-index-metadata successful: Key[$docid]\n";
|
---|
1282 | $mess .= " $metaname";
|
---|
1283 | $mess .= "->[$metapos]" if (defined $metapos);
|
---|
1284 | $mess .= " = $metavalue\n";
|
---|
1285 |
|
---|
1286 | $gsdl_cgi->generate_ok_message($mess);
|
---|
1287 | }
|
---|
1288 |
|
---|
1289 | #return $status; # in case calling functions have any further use for this
|
---|
1290 | }
|
---|
1291 |
|
---|
1292 | sub set_index_metadata
|
---|
1293 | {
|
---|
1294 | my $self = shift @_;
|
---|
1295 |
|
---|
1296 | my $username = $self->{'username'};
|
---|
1297 | my $collect = $self->{'collect'};
|
---|
1298 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
1299 | #my $gsdlhome = $self->{'gsdlhome'};
|
---|
1300 |
|
---|
1301 | if ($baseaction::authentication_enabled) {
|
---|
1302 | # Ensure the user is allowed to edit this collection
|
---|
1303 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
1304 | }
|
---|
1305 |
|
---|
1306 | my $site = $self->{'site'};
|
---|
1307 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
1308 |
|
---|
1309 | $gsdl_cgi->checked_chdir($collect_dir);
|
---|
1310 |
|
---|
1311 | # Obtain the collect dir
|
---|
1312 | ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
1313 |
|
---|
1314 | # Make sure the collection isn't locked by someone else
|
---|
1315 | $self->lock_collection($username, $collect);
|
---|
1316 |
|
---|
1317 | $self->_set_index_metadata(@_);
|
---|
1318 |
|
---|
1319 | # Release the lock once it is done
|
---|
1320 | $self->unlock_collection($username, $collect);
|
---|
1321 | }
|
---|
1322 |
|
---|
1323 | # call this to set the metadata for a combination of dirs archives, import or index, or live
|
---|
1324 | # if none specified, defaults to index which was the original behaviour of set_metadata.
|
---|
1325 | sub set_metadata
|
---|
1326 | {
|
---|
1327 | my $self = shift @_;
|
---|
1328 |
|
---|
1329 | # Testing that not defining a variable, setting it to "" or to " " all return false
|
---|
1330 | # >perl -e 'my $whichdirs=""; if($whichdirs) {print "$whichdirs\n"};'
|
---|
1331 |
|
---|
1332 | my $where = $self->{'where'};
|
---|
1333 | if(!$where) {
|
---|
1334 | $self->set_index_metadata(@_); # call the full version of set_index_meta for the default behaviour
|
---|
1335 | return;
|
---|
1336 | }
|
---|
1337 |
|
---|
1338 | # authenticate and lock collection once, even if processing multiple dirs
|
---|
1339 | my $username = $self->{'username'};
|
---|
1340 | my $collect = $self->{'collect'};
|
---|
1341 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
1342 |
|
---|
1343 | if ($baseaction::authentication_enabled) {
|
---|
1344 | # Ensure the user is allowed to edit this collection
|
---|
1345 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
1346 | }
|
---|
1347 |
|
---|
1348 | if($where =~ m/index/) {
|
---|
1349 | my $site = $self->{'site'};
|
---|
1350 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
1351 | $gsdl_cgi->checked_chdir($collect_dir);
|
---|
1352 | }
|
---|
1353 |
|
---|
1354 | # Make sure the collection isn't locked by someone else
|
---|
1355 | $self->lock_collection($username, $collect);
|
---|
1356 |
|
---|
1357 |
|
---|
1358 | # now at last can set the metadata. $where can specify multiple
|
---|
1359 | # $where is of the form: import|archives|index, or a subset thereof
|
---|
1360 |
|
---|
1361 | #my @whichdirs = split('\|', $where);
|
---|
1362 |
|
---|
1363 | # just check whether $where contains import/archives/index/live in turn, and
|
---|
1364 | # for each case, process it accordingly
|
---|
1365 | if($where =~ m/import/) {
|
---|
1366 | $self->_set_import_metadata(@_);
|
---|
1367 | }
|
---|
1368 |
|
---|
1369 | if($where =~ m/archives/) {
|
---|
1370 |
|
---|
1371 | # look up docID arg which is optional to set_metadata because it's optional
|
---|
1372 | # to set_import, but which is compulsory to set_archives_metadata
|
---|
1373 | my $docid = $self->{'d'};
|
---|
1374 | if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
|
---|
1375 | $gsdl_cgi->generate_error("No docid (d=...) specified."); # generates error and dies
|
---|
1376 | }
|
---|
1377 | # we have a docid, so can set archives meta
|
---|
1378 | $self->_set_archives_metadata(@_);
|
---|
1379 | }
|
---|
1380 |
|
---|
1381 | if($where =~ m/index/) {
|
---|
1382 |
|
---|
1383 | # look up docID arg which is optional to set_metadata because it's optional
|
---|
1384 | # to set_import, but which is compulsory to set_archives_metadata
|
---|
1385 | my $docid = $self->{'d'};
|
---|
1386 | if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
|
---|
1387 | $gsdl_cgi->generate_error("No docid (d=...) specified.");
|
---|
1388 | }
|
---|
1389 | # we have a docid, so can set index meta
|
---|
1390 | $self->_set_index_metadata(@_);
|
---|
1391 | }
|
---|
1392 |
|
---|
1393 | if($where =~ m/live/) {
|
---|
1394 | $self->_set_live_metadata(@_); # docid param, d, is compulsory, but is checked for in subroutine
|
---|
1395 | }
|
---|
1396 |
|
---|
1397 | # Release the lock once it is done
|
---|
1398 | $self->unlock_collection($username, $collect);
|
---|
1399 | }
|
---|
1400 |
|
---|
1401 | sub set_metadata_array
|
---|
1402 | {
|
---|
1403 | my $self = shift @_;
|
---|
1404 |
|
---|
1405 | my $where = $self->{'where'};
|
---|
1406 | if(!$where) {
|
---|
1407 | $self->set_index_metadata_array(@_); # default behaviour is the full version of set_index_meta_array
|
---|
1408 | return;
|
---|
1409 | }
|
---|
1410 |
|
---|
1411 | my $username = $self->{'username'};
|
---|
1412 | my $collect = $self->{'collect'};
|
---|
1413 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
1414 |
|
---|
1415 | if ($baseaction::authentication_enabled) {
|
---|
1416 | # Ensure the user is allowed to edit this collection
|
---|
1417 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
1418 | }
|
---|
1419 |
|
---|
1420 | # Not sure if the checked_chdir is necessary, since lock_collection also does a chdir
|
---|
1421 | # But including the stmt during this code reorganisation to preserve as-is what used to happen
|
---|
1422 | my $site = $self->{'site'};
|
---|
1423 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
1424 | $gsdl_cgi->checked_chdir($collect_dir);
|
---|
1425 |
|
---|
1426 | # Make sure the collection isn't locked by someone else
|
---|
1427 | $self->lock_collection($username, $collect);
|
---|
1428 |
|
---|
1429 | if($where =~ m/import/) {
|
---|
1430 | $self->_set_import_metadata_array(@_);
|
---|
1431 | }
|
---|
1432 | if($where =~ m/archives/) {
|
---|
1433 | $self->_set_archives_metadata_array(@_);
|
---|
1434 | }
|
---|
1435 | if($where =~ m/index/) {
|
---|
1436 | $self->_set_index_metadata_array(@_);
|
---|
1437 | }
|
---|
1438 | if($where =~ m/live/) {
|
---|
1439 | $self->_set_live_metadata_array(@_);
|
---|
1440 | }
|
---|
1441 |
|
---|
1442 | # Release the lock once it is done
|
---|
1443 | $self->unlock_collection($username, $collect);
|
---|
1444 | }
|
---|
1445 |
|
---|
1446 | sub _set_index_metadata_array
|
---|
1447 | {
|
---|
1448 | my $self = shift @_;
|
---|
1449 |
|
---|
1450 | my $collect = $self->{'collect'};
|
---|
1451 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
1452 | my $site = $self->{'site'};
|
---|
1453 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
1454 |
|
---|
1455 |
|
---|
1456 | # look up additional args
|
---|
1457 |
|
---|
1458 | my $infodbtype = $self->{'infodbtype'};
|
---|
1459 |
|
---|
1460 | my $json_str = $self->{'json'};
|
---|
1461 | my $doc_array = decode_json $json_str;
|
---|
1462 |
|
---|
1463 |
|
---|
1464 | my $global_status = 0;
|
---|
1465 | my $global_mess = "";
|
---|
1466 |
|
---|
1467 | my @all_docids = ();
|
---|
1468 |
|
---|
1469 | foreach my $doc_array_rec ( @$doc_array ) {
|
---|
1470 |
|
---|
1471 | my $status = -1;
|
---|
1472 | my $docid = $doc_array_rec->{'docid'};
|
---|
1473 |
|
---|
1474 | push(@all_docids,$docid);
|
---|
1475 |
|
---|
1476 | my $metaname = $doc_array_rec->{'metaname'};
|
---|
1477 | if(defined $metaname) {
|
---|
1478 | my $metapos = $doc_array_rec->{'metapos'}; # can legitimately be undef
|
---|
1479 | my $metavalue = $doc_array_rec->{'metavalue'};
|
---|
1480 | my $metamode = $doc_array_rec->{'metamode'} || $self->{'metamode'};
|
---|
1481 |
|
---|
1482 | $status = $self->set_index_metadata_entry($collect_dir,$collect,$infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode);
|
---|
1483 | } elsif (defined $doc_array_rec->{'metatable'}) { # if no metaname, we expect a metatable
|
---|
1484 | my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
|
---|
1485 |
|
---|
1486 | foreach my $metatable_rec ( @$metatable ) { # the subarray metatable is an array of hashmaps
|
---|
1487 | $metaname = $metatable_rec->{'metaname'};
|
---|
1488 | my $metamode = $metatable_rec->{'metamode'} || $doc_array_rec->{'metamode'} || $self->{'metamode'};
|
---|
1489 | my $metapos = undef;
|
---|
1490 | my $metavals = $metatable_rec->{'metavals'}; # a sub-subarray
|
---|
1491 |
|
---|
1492 | foreach my $metavalue ( @$metavals ) { # metavals is an array
|
---|
1493 | $status = $self->set_index_metadata_entry($collect_dir,$collect,$infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode); # how do we use metamode in set_meta_entry?
|
---|
1494 | if($metamode eq "override") { # now, having overridden the metavalue for the first,
|
---|
1495 | # need to accumulate subsequent metavals for this metaname, else the just-assigned
|
---|
1496 | # metavalue for this metaname will be lost
|
---|
1497 | $metamode = "accumulate";
|
---|
1498 | }
|
---|
1499 | }
|
---|
1500 | }
|
---|
1501 | }
|
---|
1502 |
|
---|
1503 | if ($status != 0) {
|
---|
1504 | # Catch error if set infodb entry failed
|
---|
1505 | $global_status = $status;
|
---|
1506 | $global_mess .= "Failed to set metadata key: $docid\n";
|
---|
1507 | $global_mess .= "Exit status: $status\n";
|
---|
1508 | $global_mess .= "System Error Message: $!\n";
|
---|
1509 | $global_mess .= "-" x 20;
|
---|
1510 | }
|
---|
1511 | }
|
---|
1512 |
|
---|
1513 | if ($global_status != 0) {
|
---|
1514 | $global_mess .= "PATH: $ENV{'PATH'}\n";
|
---|
1515 | $gsdl_cgi->generate_error($global_mess);
|
---|
1516 | }
|
---|
1517 | else {
|
---|
1518 | my $mess = "set-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
|
---|
1519 | $gsdl_cgi->generate_ok_message($mess);
|
---|
1520 | }
|
---|
1521 | }
|
---|
1522 |
|
---|
1523 | sub set_index_metadata_array
|
---|
1524 | {
|
---|
1525 | my $self = shift @_;
|
---|
1526 |
|
---|
1527 | my $username = $self->{'username'};
|
---|
1528 | my $collect = $self->{'collect'};
|
---|
1529 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
1530 | # my $gsdlhome = $self->{'gsdlhome'};
|
---|
1531 |
|
---|
1532 | if ($baseaction::authentication_enabled) {
|
---|
1533 | # Ensure the user is allowed to edit this collection
|
---|
1534 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
1535 | }
|
---|
1536 |
|
---|
1537 | my $site = $self->{'site'};
|
---|
1538 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
1539 |
|
---|
1540 | $gsdl_cgi->checked_chdir($collect_dir);
|
---|
1541 |
|
---|
1542 | # Obtain the collect dir
|
---|
1543 | ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
1544 |
|
---|
1545 | # Make sure the collection isn't locked by someone else
|
---|
1546 | $self->lock_collection($username, $collect);
|
---|
1547 |
|
---|
1548 | $self->_set_index_metadata_array(@_);
|
---|
1549 |
|
---|
1550 | # Release the lock once it is done
|
---|
1551 | $self->unlock_collection($username, $collect);
|
---|
1552 | }
|
---|
1553 |
|
---|
1554 | # experimental, newly added in and untested
|
---|
1555 | sub _set_live_metadata_array
|
---|
1556 | {
|
---|
1557 | my $self = shift @_;
|
---|
1558 |
|
---|
1559 | my $collect = $self->{'collect'};
|
---|
1560 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
1561 |
|
---|
1562 | my $site = $self->{'site'};
|
---|
1563 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
1564 |
|
---|
1565 |
|
---|
1566 | # look up additional args
|
---|
1567 | my $infodbtype = $self->{'infodbtype'};
|
---|
1568 | # To people who know $collect_tail please add some comments
|
---|
1569 | # Obtain path to the database
|
---|
1570 | my $collect_tail = $collect;
|
---|
1571 | $collect_tail =~ s/^.*[\/|\\]//;
|
---|
1572 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
---|
1573 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, "live-$collect_tail", $index_text_directory);
|
---|
1574 |
|
---|
1575 |
|
---|
1576 | my $json_str = $self->{'json'};
|
---|
1577 | my $doc_array = decode_json $json_str;
|
---|
1578 |
|
---|
1579 |
|
---|
1580 | my $global_status = 0;
|
---|
1581 | my $global_mess = "";
|
---|
1582 |
|
---|
1583 | my @all_docids = ();
|
---|
1584 |
|
---|
1585 |
|
---|
1586 | foreach my $doc_array_rec ( @$doc_array ) {
|
---|
1587 |
|
---|
1588 | my $status = -1;
|
---|
1589 | my $docid = $doc_array_rec->{'docid'};
|
---|
1590 |
|
---|
1591 | push(@all_docids,$docid);
|
---|
1592 |
|
---|
1593 | my $metaname = $doc_array_rec->{'metaname'};
|
---|
1594 | if(defined $metaname) {
|
---|
1595 | my $dbkey = "$docid.$metaname";
|
---|
1596 | my $metavalue = $doc_array_rec->{'metavalue'};
|
---|
1597 |
|
---|
1598 | # Set the new value
|
---|
1599 | my $cmd = "gdbmset \"$infodb_file_path\" \"$dbkey\" \"$metavalue\"";
|
---|
1600 | $status = system($cmd);
|
---|
1601 |
|
---|
1602 | } elsif (defined $doc_array_rec->{'metatable'}) { # if no metaname, we expect a metatable
|
---|
1603 | my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
|
---|
1604 | foreach my $metatable_rec ( @$metatable ) {
|
---|
1605 | $metaname = $metatable_rec->{'metaname'};
|
---|
1606 | my $dbkey = "$docid.$metaname";
|
---|
1607 |
|
---|
1608 | my $metavals = $metatable_rec->{'metavals'}; # a sub-subarray
|
---|
1609 | foreach my $metavalue ( @$metavals ) {
|
---|
1610 | my $cmd = "gdbmset \"$infodb_file_path\" \"$dbkey\" \"$metavalue\"";
|
---|
1611 | $status = system($cmd);
|
---|
1612 | }
|
---|
1613 | }
|
---|
1614 |
|
---|
1615 | }
|
---|
1616 |
|
---|
1617 | if ($status != 0) {
|
---|
1618 | # Catch error if gdbmget failed
|
---|
1619 | $global_status = $status;
|
---|
1620 | $global_mess .= "Failed to set metadata key: $docid\n"; # $dbkey
|
---|
1621 | $global_mess .= "Exit status: $status\n";
|
---|
1622 | $global_mess .= "System Error Message: $!\n";
|
---|
1623 | $global_mess .= "-" x 20;
|
---|
1624 | }
|
---|
1625 | }
|
---|
1626 |
|
---|
1627 | if ($global_status != 0) {
|
---|
1628 | $global_mess .= "PATH: $ENV{'PATH'}\n";
|
---|
1629 | $gsdl_cgi->generate_error($global_mess);
|
---|
1630 | }
|
---|
1631 | else {
|
---|
1632 | my $mess = "set-live-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
|
---|
1633 | $gsdl_cgi->generate_ok_message($mess);
|
---|
1634 | }
|
---|
1635 | }
|
---|
1636 |
|
---|
1637 | sub set_live_metadata_array
|
---|
1638 | {
|
---|
1639 | my $self = shift @_;
|
---|
1640 |
|
---|
1641 | my $username = $self->{'username'};
|
---|
1642 | my $collect = $self->{'collect'};
|
---|
1643 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
1644 |
|
---|
1645 | if ($baseaction::authentication_enabled) {
|
---|
1646 | # Ensure the user is allowed to edit this collection
|
---|
1647 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
1648 | }
|
---|
1649 |
|
---|
1650 | my $site = $self->{'site'};
|
---|
1651 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
1652 |
|
---|
1653 | $gsdl_cgi->checked_chdir($collect_dir);
|
---|
1654 |
|
---|
1655 | # Make sure the collection isn't locked by someone else
|
---|
1656 | $self->lock_collection($username, $collect);
|
---|
1657 |
|
---|
1658 | $self->_set_live_metadata_array(@_);
|
---|
1659 |
|
---|
1660 | # Release the lock once it is done
|
---|
1661 | $self->unlock_collection($username, $collect);
|
---|
1662 | }
|
---|
1663 |
|
---|
1664 |
|
---|
1665 | sub dxml_metadata
|
---|
1666 | {
|
---|
1667 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
1668 | my $metaname = $parser->{'parameters'}->{'metaname'};
|
---|
1669 | my $metamode = $parser->{'parameters'}->{'metamode'};
|
---|
1670 |
|
---|
1671 | print STDERR "**** Processing closing </Metadata> tag\n";
|
---|
1672 |
|
---|
1673 | my $opt_doc_secnum = $parser->{'parameters'}->{'secnum'};
|
---|
1674 |
|
---|
1675 | # Find the right metadata tag and checks if we are going to
|
---|
1676 | # override it
|
---|
1677 | #
|
---|
1678 | # Note: This over writes the first metadata block it
|
---|
1679 | # encountered. If there are multiple Sections in the doc.xml, it
|
---|
1680 | # might not behave as you would expect
|
---|
1681 |
|
---|
1682 | my $curr_secnum = $parser->{'parameters'}->{'curr_section_num'};
|
---|
1683 | ## print STDERR "**** checking $opt_doc_secnum <=> $curr_secnum\n";
|
---|
1684 | ## print STDERR "**** metamode = $metamode\n";
|
---|
1685 |
|
---|
1686 | if ((!defined $opt_doc_secnum) || ($opt_doc_secnum eq $curr_secnum))
|
---|
1687 | {
|
---|
1688 | my $name_attr = $attrHash->{'name'};
|
---|
1689 | if (($name_attr eq $metaname) && ($metamode eq "override"))
|
---|
1690 | {
|
---|
1691 | if (!defined $parser->{'parameters'}->{'poscount'})
|
---|
1692 | {
|
---|
1693 | $parser->{'parameters'}->{'poscount'} = 0;
|
---|
1694 | }
|
---|
1695 | else
|
---|
1696 | {
|
---|
1697 | $parser->{'parameters'}->{'poscount'}++;
|
---|
1698 | }
|
---|
1699 |
|
---|
1700 | if(defined $parser->{'parameters'}->{'metapos'} && $parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'})
|
---|
1701 | {
|
---|
1702 | ##print STDERR "#### got match!!\n";
|
---|
1703 | # Get the value and override the current value
|
---|
1704 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
---|
1705 | $attrHash->{'_content'} = $metavalue;
|
---|
1706 |
|
---|
1707 | # Don't want it to wipe out any other pieces of metadata
|
---|
1708 | $parser->{'parameters'}->{'metamode'} = "done";
|
---|
1709 | }
|
---|
1710 | elsif(defined $parser->{'parameters'}->{'prevmetavalue'} && $parser->{'parameters'}->{'prevmetavalue'} eq $attrHash->{'_content'})
|
---|
1711 | {
|
---|
1712 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
---|
1713 | $attrHash->{'_content'} = $metavalue;
|
---|
1714 | $parser->{'parameters'}->{'metamode'} = "done";
|
---|
1715 | }
|
---|
1716 | }
|
---|
1717 | }
|
---|
1718 |
|
---|
1719 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
---|
1720 | return [$tagname => $attrHash];
|
---|
1721 | }
|
---|
1722 |
|
---|
1723 | # This method exists purely for catching invalid section numbers that the client
|
---|
1724 | # requested to edit. Once the parser has reached the end (the final </Archive> tag),
|
---|
1725 | # we've seen all the Sections in the doc.xml, and none of their section nums matched
|
---|
1726 | # if the metamode has not been set to 'done' by then.
|
---|
1727 | sub dxml_archive
|
---|
1728 | {
|
---|
1729 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
1730 | my $metamode = $parser->{'parameters'}->{'metamode'};
|
---|
1731 |
|
---|
1732 | my $curr_secnum = $parser->{'parameters'}->{'curr_section_num'};
|
---|
1733 | my $opt_doc_secnum = $parser->{'parameters'}->{'secnum'};
|
---|
1734 |
|
---|
1735 | # print STDERR "@@@ $tagname Processing a closing </Archive> tag [$curr_secnum|$opt_doc_secnum]\n";
|
---|
1736 |
|
---|
1737 | if ($metamode ne "done" && $curr_secnum ne $opt_doc_secnum) {
|
---|
1738 | print STDERR "@@@ $tagname Finished processing FINAL Section.\n";
|
---|
1739 |
|
---|
1740 | my $metaname = $parser->{'parameters'}->{'metaname'};
|
---|
1741 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
---|
1742 |
|
---|
1743 | print STDERR "@@@ Requested section number $opt_doc_secnum not found.\n";
|
---|
1744 | print STDERR "\t(last seen section number in document was $curr_secnum)\n";
|
---|
1745 | print STDERR "\tDiscarded metadata value '$metavalue' for meta '$metaname'\n";
|
---|
1746 | print STDERR "\tin section $opt_doc_secnum.\n";
|
---|
1747 | $parser->{'custom_err_msg'} = "Requested section number $opt_doc_secnum not found.";
|
---|
1748 | }
|
---|
1749 |
|
---|
1750 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
---|
1751 | return [$tagname => $attrHash];
|
---|
1752 | }
|
---|
1753 |
|
---|
1754 | sub dxml_description
|
---|
1755 | {
|
---|
1756 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
1757 | my $metamode = $parser->{'parameters'}->{'metamode'};
|
---|
1758 |
|
---|
1759 | my $curr_secnum = $parser->{'parameters'}->{'curr_section_num'};
|
---|
1760 | my $opt_doc_secnum = $parser->{'parameters'}->{'secnum'} || "";
|
---|
1761 |
|
---|
1762 | print STDERR "**** Processing a closing </Description> tag \n";
|
---|
1763 | # print STDERR "@@@ $tagname Processing a closing </Description> tag [$curr_secnum|$opt_doc_secnum]\n";
|
---|
1764 |
|
---|
1765 | # Accumulate the metadata
|
---|
1766 |
|
---|
1767 | # We'll be accumulating metadata at this point if we haven't found and therefore
|
---|
1768 | # haven't processed the metadata yet.
|
---|
1769 | # For subsections, this means that if we're at a matching subsection, but haven't
|
---|
1770 | # found the correct metaname to override in that subsection, we accumulate it as new
|
---|
1771 | # meta in the subsection by adding it to the current description.
|
---|
1772 | # If there's no subsection info for the metadata, it will accumulate at the top level
|
---|
1773 | # section description if we hadn't found a matching metaname to override at this point.
|
---|
1774 |
|
---|
1775 | # Both curr_secnum and opt_doc_secnum can be "". In the former case, it means we're now
|
---|
1776 | # at the toplevel section. In the latter case, it means we want to process meta in the
|
---|
1777 | # toplevel section. So the eq check between the values below will work in all cases.
|
---|
1778 |
|
---|
1779 | # The only time this won't work is if an opt_doc_secnum beyond the section numbers of
|
---|
1780 | # this document has been provided. In that case, the metadata for that opt_doc_secnum
|
---|
1781 | # won't get attached/accumulated to any part of the doc, not even its top-level section.
|
---|
1782 |
|
---|
1783 | if ($curr_secnum eq $opt_doc_secnum
|
---|
1784 | && ($metamode eq "accumulate" || $metamode eq "override")) {
|
---|
1785 | if ($metamode eq "override") {
|
---|
1786 | print "No metadata value to override. Switching 'metamode' to accumulate\n";
|
---|
1787 | }
|
---|
1788 |
|
---|
1789 | # If we get to here and metamode is override, this means there
|
---|
1790 | # was no existing value to overide => treat as an append operation
|
---|
1791 |
|
---|
1792 | # Tack a new metadata tag on to the end of the <Metadata>+ block
|
---|
1793 | my $metaname = $parser->{'parameters'}->{'metaname'};
|
---|
1794 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
---|
1795 |
|
---|
1796 | my $metadata_attr = {
|
---|
1797 | '_content' => $metavalue,
|
---|
1798 | 'name' => $metaname,
|
---|
1799 | 'mode' => "accumulate"
|
---|
1800 | };
|
---|
1801 |
|
---|
1802 | my $append_metadata = [ "Metadata" => $metadata_attr ];
|
---|
1803 | my $description_content = $attrHash->{'_content'};
|
---|
1804 |
|
---|
1805 | print "Appending metadata to doc.xml\n";
|
---|
1806 |
|
---|
1807 | if (ref($description_content)) {
|
---|
1808 | # got some existing interesting nested content
|
---|
1809 | push(@$description_content, " ", $append_metadata ,"\n ");
|
---|
1810 | }
|
---|
1811 | else {
|
---|
1812 | #description_content is most likely a string such as "\n"
|
---|
1813 | $attrHash->{'_content'} = [$description_content, " ", $append_metadata ,"\n" ];
|
---|
1814 | }
|
---|
1815 |
|
---|
1816 | $parser->{'parameters'}->{'metamode'} = "done";
|
---|
1817 | }
|
---|
1818 | else {
|
---|
1819 | # metamode most likely "done" signifying that it has already found a position to add the metadata to.
|
---|
1820 | ## print STDERR "**** NOT ACCUMULATE?!? \n";
|
---|
1821 | }
|
---|
1822 |
|
---|
1823 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
---|
1824 | return [$tagname => $attrHash];
|
---|
1825 | }
|
---|
1826 |
|
---|
1827 |
|
---|
1828 | sub dxml_start_section
|
---|
1829 | {
|
---|
1830 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
1831 |
|
---|
1832 | my $new_depth = scalar(@$contextArray);
|
---|
1833 |
|
---|
1834 | # print STDERR "**** START SECTION \n";
|
---|
1835 |
|
---|
1836 | if ($new_depth == 1) {
|
---|
1837 | $parser->{'parameters'}->{'curr_section_depth'} = 1;
|
---|
1838 | $parser->{'parameters'}->{'curr_section_num'} = "";
|
---|
1839 | }
|
---|
1840 |
|
---|
1841 | my $old_depth = $parser->{'parameters'}->{'curr_section_depth'};
|
---|
1842 | my $old_secnum = $parser->{'parameters'}->{'curr_section_num'};
|
---|
1843 |
|
---|
1844 | my $new_secnum;
|
---|
1845 |
|
---|
1846 | if ($new_depth > $old_depth) {
|
---|
1847 | # child subsection
|
---|
1848 | $new_secnum = "$old_secnum.1";
|
---|
1849 | }
|
---|
1850 | elsif ($new_depth == $old_depth) {
|
---|
1851 | # sibling section => increase it's value by 1
|
---|
1852 | my ($tail_num) = ($old_secnum =~ m/\.(\d+)$/);
|
---|
1853 | $tail_num++;
|
---|
1854 | $new_secnum = $old_secnum;
|
---|
1855 | $new_secnum =~ s/\.(\d+)$/\.$tail_num/;
|
---|
1856 | }
|
---|
1857 | else {
|
---|
1858 | # back up to parent section => lopp off tail
|
---|
1859 | $new_secnum = $old_secnum;
|
---|
1860 | $new_secnum =~ s/\.\d+$//;
|
---|
1861 | }
|
---|
1862 |
|
---|
1863 | $parser->{'parameters'}->{'curr_section_depth'} = $new_depth;
|
---|
1864 | $parser->{'parameters'}->{'curr_section_num'} = $new_secnum;
|
---|
1865 |
|
---|
1866 | 1;
|
---|
1867 | }
|
---|
1868 |
|
---|
1869 | sub edit_xml_file
|
---|
1870 | {
|
---|
1871 | my $self = shift @_;
|
---|
1872 | my ($gsdl_cgi, $filename, $start_rules, $rules, $options) = @_;
|
---|
1873 |
|
---|
1874 | # use XML::Rules to add it in (read in and out again)
|
---|
1875 | my $parser = XML::Rules->new(start_rules => $start_rules,
|
---|
1876 | rules => $rules,
|
---|
1877 | style => 'filter',
|
---|
1878 | output_encoding => 'utf8' );
|
---|
1879 |
|
---|
1880 | my $xml_in = "";
|
---|
1881 | if (!open(MIN,"<$filename")) {
|
---|
1882 | $gsdl_cgi->generate_error("Unable to read in $filename: $!");
|
---|
1883 | }
|
---|
1884 | else {
|
---|
1885 | # Read all the text in
|
---|
1886 | my $line;
|
---|
1887 | while (defined ($line=<MIN>)) {
|
---|
1888 | $xml_in .= $line;
|
---|
1889 | }
|
---|
1890 | close(MIN);
|
---|
1891 |
|
---|
1892 | my $MOUT;
|
---|
1893 | if (!open($MOUT,">$filename")) {
|
---|
1894 | $gsdl_cgi->generate_error("Unable to write out to $filename: $!");
|
---|
1895 | }
|
---|
1896 | else {
|
---|
1897 | # Matched lines will get handled by the call backs
|
---|
1898 | ## my $xml_out = "";
|
---|
1899 |
|
---|
1900 | binmode($MOUT,":utf8");
|
---|
1901 | $parser->filter($xml_in,$MOUT, $options);
|
---|
1902 |
|
---|
1903 | # binmode(MOUT,":utf8");
|
---|
1904 | # print MOUT $xml_out;
|
---|
1905 | close($MOUT);
|
---|
1906 | }
|
---|
1907 | }
|
---|
1908 |
|
---|
1909 | # copy across any custom error information that was stored during parsing
|
---|
1910 | $self->{'error_msg'} = $parser->{'custom_err_msg'} if(defined $parser->{'custom_err_msg'});
|
---|
1911 | }
|
---|
1912 |
|
---|
1913 | sub edit_doc_xml
|
---|
1914 | {
|
---|
1915 | my $self = shift @_;
|
---|
1916 | my ($gsdl_cgi, $doc_xml_filename, $metaname, $metavalue, $metapos, $metamode, $opt_secnum, $prevmetavalue) = @_;
|
---|
1917 |
|
---|
1918 | my $info_mess = <<RAWEND;
|
---|
1919 | ****************************
|
---|
1920 | edit_doc_xml()
|
---|
1921 | ****************************
|
---|
1922 | doc_xml_filename = $doc_xml_filename
|
---|
1923 | metaname = $metaname
|
---|
1924 | metavalue = $metavalue
|
---|
1925 | metapos = $metapos
|
---|
1926 | metamode = $metamode
|
---|
1927 | opt_secnum = $opt_secnum
|
---|
1928 | prevmetavalue = $prevmetavalue
|
---|
1929 | ****************************
|
---|
1930 | RAWEND
|
---|
1931 |
|
---|
1932 | $gsdl_cgi->generate_message($info_mess);
|
---|
1933 |
|
---|
1934 | # To monitor which section/subsection number we are in
|
---|
1935 | my @start_rules =
|
---|
1936 | ( 'Section' => \&dxml_start_section );
|
---|
1937 |
|
---|
1938 | # use XML::Rules to add it in (read in and out again)
|
---|
1939 | # Set the call back functions
|
---|
1940 | my @rules =
|
---|
1941 | ( _default => 'raw',
|
---|
1942 | 'Metadata' => \&dxml_metadata,
|
---|
1943 | 'Description' => \&dxml_description,
|
---|
1944 | 'Archive' => \&dxml_archive); # just for catching errors at end
|
---|
1945 |
|
---|
1946 | # Sets the parameters
|
---|
1947 | my $options = { 'metaname' => $metaname,
|
---|
1948 | 'metapos' => $metapos,
|
---|
1949 | 'metavalue' => $metavalue,
|
---|
1950 | 'metamode' => $metamode,
|
---|
1951 | 'prevmetavalue' => $prevmetavalue };
|
---|
1952 |
|
---|
1953 | if (defined $opt_secnum) {
|
---|
1954 | $options->{'secnum'} = $opt_secnum;
|
---|
1955 | }
|
---|
1956 |
|
---|
1957 | $self->edit_xml_file($gsdl_cgi,$doc_xml_filename,\@start_rules,\@rules,$options);
|
---|
1958 | }
|
---|
1959 |
|
---|
1960 | sub set_archives_metadata_entry
|
---|
1961 | {
|
---|
1962 | my $self = shift @_;
|
---|
1963 | my ($gsdl_cgi, $archive_dir, $collect_dir, $collect, $infodbtype, $docid, $metaname, $metapos, $metavalue, $metamode, $prevmetavalue) = @_;
|
---|
1964 |
|
---|
1965 | my $info_mess = <<RAWEND;
|
---|
1966 | ****************************
|
---|
1967 | set_archives_metadata_entry()
|
---|
1968 | ****************************
|
---|
1969 | archive_dir = $archive_dir
|
---|
1970 | collect_dir = $collect_dir
|
---|
1971 | collect = $collect
|
---|
1972 | infodbtype = $infodbtype
|
---|
1973 | docid = $docid
|
---|
1974 | metaname = $metaname
|
---|
1975 | metapos = $metapos
|
---|
1976 | metavalue = $metavalue
|
---|
1977 | metamode = $metamode
|
---|
1978 | prevmetavalue = $prevmetavalue
|
---|
1979 | ****************************
|
---|
1980 | RAWEND
|
---|
1981 |
|
---|
1982 | $gsdl_cgi->generate_message($info_mess);
|
---|
1983 |
|
---|
1984 | # Obtain the doc.xml path for the specified docID
|
---|
1985 | my ($docid_root,$docid_secnum) = ($docid =~ m/^(.*?)(\..*)?$/);
|
---|
1986 |
|
---|
1987 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
---|
1988 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid_root);
|
---|
1989 | my $doc_xml_file = $doc_rec->{'doc-file'}->[0];
|
---|
1990 |
|
---|
1991 | # The $doc_xml_file is relative to the archives, and now let's get the full path
|
---|
1992 | my $archives_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
---|
1993 | my $doc_xml_filename = &util::filename_cat($archives_dir,$doc_xml_file);
|
---|
1994 |
|
---|
1995 | # If we're overriding everything, then $metamode=override combined with $metapos=undefined
|
---|
1996 | # in which case, we need to remove all metavalues for the metaname at the given (sub)section
|
---|
1997 | # Thereafter, we will finally be setting the overriding metavalue for this metaname
|
---|
1998 | if(!defined $metapos && $metamode eq "override") {
|
---|
1999 | $self->remove_from_doc_xml($gsdl_cgi, &util::filename_cat($archive_dir, $doc_xml_file), $metaname, $metapos, undef, $docid_secnum, $metamode);
|
---|
2000 | }
|
---|
2001 |
|
---|
2002 | # Edit the doc.xml file with the specified metadata name, value and position.
|
---|
2003 | # TODO: there is a potential problem here as this edit_doc_xml function
|
---|
2004 | # is assuming the simple doc.xml situation where there is only one Section and no SubSections.
|
---|
2005 | # Running import.pl -groupsize will cause this to have multiple sections in one doc.xml
|
---|
2006 |
|
---|
2007 | # dxml_metadata method ignores metapos if metamode anything other than override
|
---|
2008 | $self->edit_doc_xml($gsdl_cgi,$doc_xml_filename,
|
---|
2009 | $metaname,$metavalue,$metapos,$metamode,$docid_secnum,$prevmetavalue);
|
---|
2010 |
|
---|
2011 | # return 0; # return 0 for now to indicate no error
|
---|
2012 | return (defined $self->{'error_msg'}) ? 1 : 0;
|
---|
2013 | }
|
---|
2014 |
|
---|
2015 |
|
---|
2016 | sub set_archives_metadata
|
---|
2017 | {
|
---|
2018 | my $self = shift @_;
|
---|
2019 |
|
---|
2020 | my $username = $self->{'username'};
|
---|
2021 | my $collect = $self->{'collect'};
|
---|
2022 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
2023 |
|
---|
2024 | if ($baseaction::authentication_enabled) {
|
---|
2025 | # Ensure the user is allowed to edit this collection
|
---|
2026 | $self->authenticate_user($username, $collect);
|
---|
2027 | }
|
---|
2028 |
|
---|
2029 | # Make sure the collection isn't locked by someone else
|
---|
2030 | $self->lock_collection($username, $collect);
|
---|
2031 |
|
---|
2032 | $self->_set_archives_metadata(@_);
|
---|
2033 |
|
---|
2034 | # Release the lock once it is done
|
---|
2035 | $self->unlock_collection($username, $collect);
|
---|
2036 | }
|
---|
2037 |
|
---|
2038 | sub _set_archives_metadata_array
|
---|
2039 | {
|
---|
2040 | my $self = shift @_;
|
---|
2041 |
|
---|
2042 | my $collect = $self->{'collect'};
|
---|
2043 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
2044 | my $site = $self->{'site'};
|
---|
2045 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
2046 |
|
---|
2047 | # look up additional args
|
---|
2048 |
|
---|
2049 | my $infodbtype = $self->{'infodbtype'};
|
---|
2050 |
|
---|
2051 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
---|
2052 |
|
---|
2053 | my $json_str = $self->{'json'};
|
---|
2054 | my $doc_array = decode_json $json_str;
|
---|
2055 |
|
---|
2056 |
|
---|
2057 | my $global_status = 0;
|
---|
2058 | my $global_mess = "";
|
---|
2059 |
|
---|
2060 | my @all_docids = ();
|
---|
2061 |
|
---|
2062 | foreach my $doc_array_rec ( @$doc_array ) {
|
---|
2063 | my $status = -1;
|
---|
2064 | my $docid = $doc_array_rec->{'docid'};
|
---|
2065 |
|
---|
2066 | push(@all_docids,$docid);
|
---|
2067 |
|
---|
2068 | my $metaname = $doc_array_rec->{'metaname'};
|
---|
2069 | if(defined $metaname) {
|
---|
2070 |
|
---|
2071 | my $metapos = $doc_array_rec->{'metapos'}; # don't force undef to 0. Undef has meaning when metamode=override
|
---|
2072 |
|
---|
2073 | my $metamode = $doc_array_rec->{'metamode'} || $self->{'metamode'};
|
---|
2074 | my $metavalue = $doc_array_rec->{'metavalue'};
|
---|
2075 | my $prevmetavalue = $self->{'prevmetavalue'}; # to make this sub behave as _set_archives_metadata
|
---|
2076 |
|
---|
2077 |
|
---|
2078 | if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
|
---|
2079 | # make "accumulate" the default (less destructive, as it won't actually
|
---|
2080 | # delete any existing values)
|
---|
2081 | $metamode = "accumulate";
|
---|
2082 | }
|
---|
2083 |
|
---|
2084 | $status = $self->set_archives_metadata_entry($gsdl_cgi,$archive_dir, $collect_dir,$collect, $infodbtype,$docid,
|
---|
2085 | $metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
|
---|
2086 | } elsif (defined $doc_array_rec->{'metatable'}) { # if no metaname, we expect a metatable
|
---|
2087 | my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
|
---|
2088 |
|
---|
2089 | foreach my $metatable_rec ( @$metatable ) {
|
---|
2090 | $metaname = $metatable_rec->{'metaname'};
|
---|
2091 | my $metamode = $metatable_rec->{'metamode'} || $doc_array_rec->{'metamode'} || $self->{'metamode'};
|
---|
2092 | my $metapos = undef;
|
---|
2093 | my $prevmetavalue = undef;
|
---|
2094 | my $metavals = $metatable_rec->{'metavals'}; # a sub-subarray
|
---|
2095 |
|
---|
2096 | foreach my $metavalue ( @$metavals ) {
|
---|
2097 | $status = $self->set_archives_metadata_entry($gsdl_cgi,$archive_dir, $collect_dir,$collect,$infodbtype,
|
---|
2098 | $docid,$metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
|
---|
2099 |
|
---|
2100 | if($metamode eq "override") { # now, having overridden the metavalue for the first,
|
---|
2101 | # need to accumulate subsequent metavals for this metaname, else the just-assigned
|
---|
2102 | # metavalue for this metaname will be lost
|
---|
2103 | $metamode = "accumulate";
|
---|
2104 | }
|
---|
2105 | }
|
---|
2106 | }
|
---|
2107 | }
|
---|
2108 |
|
---|
2109 | if ($status != 0) {
|
---|
2110 | # Catch error if set infodb entry failed
|
---|
2111 | $global_status = $status;
|
---|
2112 | $global_mess .= "Failed to set metadata key: $docid\n";
|
---|
2113 | $global_mess .= "Exit status: $status\n";
|
---|
2114 | $global_mess .= "System Error Message: $!\n";
|
---|
2115 | $global_mess .= "-" x 20 . "\n";
|
---|
2116 | }
|
---|
2117 | }
|
---|
2118 |
|
---|
2119 | if ($global_status != 0) {
|
---|
2120 | $global_mess .= "PATH: $ENV{'PATH'}\n";
|
---|
2121 | $gsdl_cgi->generate_error($global_mess);
|
---|
2122 | }
|
---|
2123 | else {
|
---|
2124 | my $mess = "set-archives-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
|
---|
2125 | $gsdl_cgi->generate_ok_message($mess);
|
---|
2126 | }
|
---|
2127 | }
|
---|
2128 |
|
---|
2129 | sub set_archives_metadata_array
|
---|
2130 | {
|
---|
2131 | my $self = shift @_;
|
---|
2132 |
|
---|
2133 | my $username = $self->{'username'};
|
---|
2134 | my $collect = $self->{'collect'};
|
---|
2135 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
2136 | # my $gsdlhome = $self->{'gsdlhome'};
|
---|
2137 |
|
---|
2138 | if ($baseaction::authentication_enabled) {
|
---|
2139 | # Ensure the user is allowed to edit this collection
|
---|
2140 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
2141 | }
|
---|
2142 |
|
---|
2143 | my $site = $self->{'site'};
|
---|
2144 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
2145 |
|
---|
2146 | $gsdl_cgi->checked_chdir($collect_dir);
|
---|
2147 |
|
---|
2148 | # Obtain the collect dir
|
---|
2149 | ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
2150 |
|
---|
2151 | # Make sure the collection isn't locked by someone else
|
---|
2152 | $self->lock_collection($username, $collect);
|
---|
2153 |
|
---|
2154 | $self->_set_archives_metadata_array(@_);
|
---|
2155 |
|
---|
2156 | # Release the lock once it is done
|
---|
2157 | $self->unlock_collection($username, $collect);
|
---|
2158 | }
|
---|
2159 |
|
---|
2160 | sub _remove_archives_metadata
|
---|
2161 | {
|
---|
2162 | my $self = shift @_;
|
---|
2163 |
|
---|
2164 | my $collect = $self->{'collect'};
|
---|
2165 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
2166 | # my $gsdlhome = $self->{'gsdlhome'};
|
---|
2167 | my $infodbtype = $self->{'infodbtype'};
|
---|
2168 |
|
---|
2169 | my $site = $self->{'site'};
|
---|
2170 |
|
---|
2171 | # Obtain the collect and archive dir
|
---|
2172 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
2173 |
|
---|
2174 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
---|
2175 |
|
---|
2176 | # look up additional args
|
---|
2177 | my ($docid, $docid_secnum) = ($self->{'d'} =~ m/^(.*?)(\..*)?$/);
|
---|
2178 |
|
---|
2179 | my $metaname = $self->{'metaname'};
|
---|
2180 | my $metapos = $self->{'metapos'};
|
---|
2181 |
|
---|
2182 | my $metavalue = $self->{'metavalue'} || undef; # necessary to force fallback to undef here
|
---|
2183 |
|
---|
2184 | # if the user hasn't told us what to delete, not having given a metavalue or metapos,
|
---|
2185 | # default to deleting the first metavalue for the given metaname
|
---|
2186 | # Beware that if both metapos AND metavalue are defined, both matches (if any)
|
---|
2187 | # seem to get deleted in one single remove_archives_meta action invocation.
|
---|
2188 | # Similarly, if 2 identical metavalues for a metaname exist and that metavalue is being
|
---|
2189 | # deleted, both get deleted.
|
---|
2190 | if(!defined $metapos && !defined $metavalue) {
|
---|
2191 | $metapos = 0;
|
---|
2192 | }
|
---|
2193 |
|
---|
2194 | my $metamode = $self->{'metamode'} || undef;
|
---|
2195 |
|
---|
2196 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
---|
2197 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
---|
2198 |
|
---|
2199 | # This now stores the full pathname
|
---|
2200 | my $doc_filename = $doc_rec->{'doc-file'}->[0];
|
---|
2201 |
|
---|
2202 | my $status = $self->remove_from_doc_xml($gsdl_cgi, &util::filename_cat($archive_dir, $doc_filename), $metaname, $metapos, $metavalue, $docid_secnum, $metamode);
|
---|
2203 | # my $status = $self->remove_from_doc_xml($gsdl_cgi, &util::filename_cat($archive_dir, $doc_filename), $metaname, $metapos, undef, $docid_secnum);
|
---|
2204 |
|
---|
2205 | if ($status == 0)
|
---|
2206 | {
|
---|
2207 | my $mess = "remove-archives-metadata successful: Key[$docid]\n";
|
---|
2208 | $mess .= " $metaname";
|
---|
2209 | $mess .= "->[$metapos]" if (defined $metapos);
|
---|
2210 |
|
---|
2211 | $gsdl_cgi->generate_ok_message($mess);
|
---|
2212 | }
|
---|
2213 | else
|
---|
2214 | {
|
---|
2215 | my $mess .= "Failed to remove archives metadata key: $docid\n";
|
---|
2216 | $mess .= "Exit status: $status\n";
|
---|
2217 | $mess .= "System Error Message: $!\n";
|
---|
2218 | $mess .= "-" x 20 . "\n";
|
---|
2219 |
|
---|
2220 | $gsdl_cgi->generate_error($mess);
|
---|
2221 | }
|
---|
2222 |
|
---|
2223 | #return $status; # in case calling functions have a use for this
|
---|
2224 | }
|
---|
2225 |
|
---|
2226 | sub remove_archives_metadata
|
---|
2227 | {
|
---|
2228 | my $self = shift @_;
|
---|
2229 |
|
---|
2230 | my $username = $self->{'username'};
|
---|
2231 | my $collect = $self->{'collect'};
|
---|
2232 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
2233 |
|
---|
2234 | if ($baseaction::authentication_enabled)
|
---|
2235 | {
|
---|
2236 | # Ensure the user is allowed to edit this collection
|
---|
2237 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
2238 | }
|
---|
2239 |
|
---|
2240 | # Make sure the collection isn't locked by someone else
|
---|
2241 | $self->lock_collection($username, $collect);
|
---|
2242 |
|
---|
2243 | $self->_remove_archives_metadata(@_);
|
---|
2244 |
|
---|
2245 | # Release the lock once it is done
|
---|
2246 | $self->unlock_collection($username, $collect);
|
---|
2247 | }
|
---|
2248 |
|
---|
2249 | sub remove_from_doc_xml
|
---|
2250 | {
|
---|
2251 | my $self = shift @_;
|
---|
2252 | my ($gsdl_cgi, $doc_xml_filename, $metaname, $metapos, $metavalue, $secid, $metamode) = @_;
|
---|
2253 |
|
---|
2254 | my @start_rules = ('Section' => \&dxml_start_section);
|
---|
2255 |
|
---|
2256 | # Set the call-back functions for the metadata tags
|
---|
2257 | my @rules =
|
---|
2258 | (
|
---|
2259 | _default => 'raw',
|
---|
2260 | 'Metadata' => \&rfdxml_metadata
|
---|
2261 | );
|
---|
2262 |
|
---|
2263 | my $parser = XML::Rules->new
|
---|
2264 | (
|
---|
2265 | start_rules => \@start_rules,
|
---|
2266 | rules => \@rules,
|
---|
2267 | style => 'filter',
|
---|
2268 | output_encoding => 'utf8',
|
---|
2269 | # normalisespaces => 1, # http://search.cpan.org/~jenda/XML-Rules-1.16/lib/XML/Rules.pm
|
---|
2270 | stripspaces => 2|0|0 # ineffectual
|
---|
2271 | );
|
---|
2272 |
|
---|
2273 | my $status = 0;
|
---|
2274 | my $xml_in = "";
|
---|
2275 | if (!open(MIN,"<$doc_xml_filename"))
|
---|
2276 | {
|
---|
2277 | $gsdl_cgi->generate_error("Unable to read in $doc_xml_filename: $!");
|
---|
2278 | $status = 1;
|
---|
2279 | }
|
---|
2280 | else
|
---|
2281 | {
|
---|
2282 | # Read them in
|
---|
2283 | my $line;
|
---|
2284 | while (defined ($line=<MIN>)) {
|
---|
2285 | $xml_in .= $line;
|
---|
2286 | }
|
---|
2287 | close(MIN);
|
---|
2288 |
|
---|
2289 | # Filter with the call-back functions
|
---|
2290 | my $xml_out = "";
|
---|
2291 |
|
---|
2292 | my $MOUT;
|
---|
2293 | if (!open($MOUT,">$doc_xml_filename")) {
|
---|
2294 | $gsdl_cgi->generate_error("Unable to write out to $doc_xml_filename: $!");
|
---|
2295 | $status = 1;
|
---|
2296 | }
|
---|
2297 | else {
|
---|
2298 | binmode($MOUT,":utf8");
|
---|
2299 | $parser->filter($xml_in, $MOUT, {metaname => $metaname, metapos => $metapos, metavalue => $metavalue, secid => $secid, metamode => $metamode});
|
---|
2300 | close($MOUT);
|
---|
2301 | }
|
---|
2302 | }
|
---|
2303 | return $status;
|
---|
2304 | }
|
---|
2305 |
|
---|
2306 | sub rfdxml_metadata
|
---|
2307 | {
|
---|
2308 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
2309 |
|
---|
2310 | # For comparisons, toplevel section is indicated by ""
|
---|
2311 | my $curr_sec_num = $parser->{'parameters'}->{'curr_section_num'} || "";
|
---|
2312 | my $secid = $parser->{'parameters'}->{'secid'} || "";
|
---|
2313 |
|
---|
2314 | if (!($secid eq $curr_sec_num))
|
---|
2315 | {
|
---|
2316 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
---|
2317 | return [$tagname => $attrHash];
|
---|
2318 | }
|
---|
2319 |
|
---|
2320 | if ($parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'})
|
---|
2321 | {
|
---|
2322 | if (!defined $parser->{'parameters'}->{'poscount'})
|
---|
2323 | {
|
---|
2324 | $parser->{'parameters'}->{'poscount'} = 0;
|
---|
2325 | }
|
---|
2326 | else
|
---|
2327 | {
|
---|
2328 | $parser->{'parameters'}->{'poscount'}++;
|
---|
2329 | }
|
---|
2330 |
|
---|
2331 | # if overriding (for set-meta) but no metapos, then clear all the meta for this metaname
|
---|
2332 | if ((defined $parser->{'parameters'}->{'metamode'}) && ($parser->{'parameters'}->{'metamode'} eq "override") && (!defined $parser->{'parameters'}->{'metapos'})) {
|
---|
2333 | return [];
|
---|
2334 | }
|
---|
2335 | }
|
---|
2336 |
|
---|
2337 | if ((defined $parser->{'parameters'}->{'metapos'}) && ($parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'}) && ($parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'}))
|
---|
2338 | {
|
---|
2339 | return [];
|
---|
2340 | }
|
---|
2341 |
|
---|
2342 | if ((defined $parser->{'parameters'}->{'metavalue'}) && ($parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'}) && ($parser->{'parameters'}->{'metavalue'} eq $attrHash->{'_content'}))
|
---|
2343 | {
|
---|
2344 | return [];
|
---|
2345 | }
|
---|
2346 |
|
---|
2347 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
---|
2348 | return [$tagname => $attrHash];
|
---|
2349 | }
|
---|
2350 |
|
---|
2351 | sub mxml_metadata
|
---|
2352 | {
|
---|
2353 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
2354 | my $metaname = $parser->{'parameters'}->{'metaname'};
|
---|
2355 | my $metamode = $parser->{'parameters'}->{'metamode'};
|
---|
2356 |
|
---|
2357 | # Report error if we don't see FileName tag before this
|
---|
2358 | die "Fatal Error: Unexpected metadata.xml structure. Undefined current_file, possibly encountered Description before FileName" if (!defined($parser->{'parameters'}->{'current_file'}));
|
---|
2359 |
|
---|
2360 | # Don't do anything if we are not in the right FileSet
|
---|
2361 | my $file_regexp = $parser->{'parameters'}->{'current_file'};
|
---|
2362 | if ($file_regexp =~ /\.\*/) {
|
---|
2363 | # Only interested in a file_regexp if it specifies precisely one
|
---|
2364 | # file.
|
---|
2365 | # So, skip anything with a .* in it as it is too general
|
---|
2366 | ## print STDERR "@@@@ Skipping entry in metadata.xml where FileName=.* as it is too general\n";
|
---|
2367 | return [$tagname => $attrHash];
|
---|
2368 | }
|
---|
2369 | my $src_file = $parser->{'parameters'}->{'src_file'};
|
---|
2370 | if (!($src_file =~ /$file_regexp/)) {
|
---|
2371 | return [$tagname => $attrHash];
|
---|
2372 | }
|
---|
2373 | ## print STDERR "*** mxl metamode = $metamode\n";
|
---|
2374 |
|
---|
2375 | # Find the right metadata tag and checks if we are going to override it
|
---|
2376 | my $name_attr = $attrHash->{'name'};
|
---|
2377 | if (($name_attr eq $metaname) && ($metamode eq "override")) {
|
---|
2378 |
|
---|
2379 | # now metadata.xml functions need to keep track of metapos
|
---|
2380 | if (!defined $parser->{'parameters'}->{'poscount'})
|
---|
2381 | {
|
---|
2382 | $parser->{'parameters'}->{'poscount'} = 0;
|
---|
2383 | }
|
---|
2384 | else
|
---|
2385 | {
|
---|
2386 | $parser->{'parameters'}->{'poscount'}++;
|
---|
2387 | }
|
---|
2388 |
|
---|
2389 | # If either the metapos or prevmetavalue is set,
|
---|
2390 | # get the value and override the current value
|
---|
2391 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
---|
2392 |
|
---|
2393 | if(defined $parser->{'parameters'}->{'prevmetavalue'} && $parser->{'parameters'}->{'prevmetavalue'} eq $attrHash->{'_content'})
|
---|
2394 | {
|
---|
2395 | $attrHash->{'_content'} = $metavalue;
|
---|
2396 |
|
---|
2397 | ## print STDERR "**** overriding metadata.xml\n";
|
---|
2398 |
|
---|
2399 | # Don't want it to wipe out any other pieces of metadata
|
---|
2400 | $parser->{'parameters'}->{'metamode'} = "done";
|
---|
2401 | }
|
---|
2402 | elsif(defined $parser->{'parameters'}->{'metapos'} && $parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'})
|
---|
2403 | {
|
---|
2404 | $attrHash->{'_content'} = $metavalue;
|
---|
2405 | $parser->{'parameters'}->{'metamode'} = "done";
|
---|
2406 | }
|
---|
2407 | }
|
---|
2408 |
|
---|
2409 | # mxml_description will process the metadata if metadata is accumulate,
|
---|
2410 | # or if we haven't found the metadata to override
|
---|
2411 |
|
---|
2412 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
---|
2413 | return [$tagname => $attrHash];
|
---|
2414 | }
|
---|
2415 |
|
---|
2416 |
|
---|
2417 | sub mxml_description
|
---|
2418 | {
|
---|
2419 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
2420 | my $metamode = $parser->{'parameters'}->{'metamode'};
|
---|
2421 |
|
---|
2422 | # Failed... Report error if we don't see FileName tag before this
|
---|
2423 | die "Fatal Error: Unexpected metadata.xml structure. Undefind current_file, possiblely encountered Description before FileName" if (!defined($parser->{'parameters'}->{'current_file'}));
|
---|
2424 |
|
---|
2425 | # Don't do anything if we are not in the right FileSet
|
---|
2426 | my $file_regexp = $parser->{'parameters'}->{'current_file'};
|
---|
2427 | if ($file_regexp =~ m/\.\*/) {
|
---|
2428 | # Only interested in a file_regexp if it specifies precisely one
|
---|
2429 | # file.
|
---|
2430 | # So, skip anything with a .* in it as it is too general
|
---|
2431 | return [$tagname => $attrHash];
|
---|
2432 | }
|
---|
2433 | my $src_file = $parser->{'parameters'}->{'src_file'};
|
---|
2434 |
|
---|
2435 | if (!($src_file =~ m/$file_regexp/)) {
|
---|
2436 | return [$tagname => $attrHash];
|
---|
2437 | }
|
---|
2438 |
|
---|
2439 | # Accumulate the metadata block to the end of the description block
|
---|
2440 | # Note: This adds metadata block to all description blocks, so if there are
|
---|
2441 | # multiple FileSets, it will add to all of them
|
---|
2442 | if (($metamode eq "accumulate") || ($metamode eq "override")) {
|
---|
2443 |
|
---|
2444 | # if metamode was "override" but get to here then it failed to
|
---|
2445 | # find an item to override, in which case it should append its
|
---|
2446 | # value to the end, just like the "accumulate" mode
|
---|
2447 |
|
---|
2448 | if ($metamode eq "override") {
|
---|
2449 | print "No metadata value to override. Switching 'metamode' to accumulate\n";
|
---|
2450 | }
|
---|
2451 |
|
---|
2452 | # tack a new metadata tag on to the end of the <Metadata>+ block
|
---|
2453 | my $metaname = $parser->{'parameters'}->{'metaname'};
|
---|
2454 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
---|
2455 |
|
---|
2456 | my $metadata_attr = { '_content' => $metavalue,
|
---|
2457 | 'name' => $metaname,
|
---|
2458 | 'mode' => "accumulate" };
|
---|
2459 |
|
---|
2460 | my $append_metadata = [ "Metadata" => $metadata_attr ];
|
---|
2461 | my $description_content = $attrHash->{'_content'};
|
---|
2462 |
|
---|
2463 | ## print STDERR "*** appending to metadata.xml\n";
|
---|
2464 |
|
---|
2465 | # append the new metadata element to the end of the current
|
---|
2466 | # content contained inside this tag
|
---|
2467 | if (ref($description_content) eq "") {
|
---|
2468 | # => string or numeric literal
|
---|
2469 | # this is caused by a <Description> block has no <Metadata> child elements
|
---|
2470 | # => set up an empty array in '_content'
|
---|
2471 | $attrHash->{'_content'} = [ "\n" ];
|
---|
2472 | $description_content = $attrHash->{'_content'};
|
---|
2473 | }
|
---|
2474 |
|
---|
2475 | push(@$description_content," ", $append_metadata ,"\n ");
|
---|
2476 | $parser->{'parameters'}->{'metamode'} = "done";
|
---|
2477 | }
|
---|
2478 |
|
---|
2479 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
---|
2480 | return [$tagname => $attrHash];
|
---|
2481 | }
|
---|
2482 |
|
---|
2483 |
|
---|
2484 | sub mxml_filename
|
---|
2485 | {
|
---|
2486 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
2487 |
|
---|
2488 | # Store the filename of the Current Fileset
|
---|
2489 | # Note: According to http://greenstone.org/dtd/DirectoryMetadata/1.0/DirectoryMetadata.dtd
|
---|
2490 | # FileName tag must come before Description tag
|
---|
2491 | $parser->{'parameters'}->{'current_file'} = $attrHash->{'_content'};
|
---|
2492 |
|
---|
2493 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
---|
2494 | return [$tagname => $attrHash];
|
---|
2495 | }
|
---|
2496 |
|
---|
2497 |
|
---|
2498 | sub mxml_fileset
|
---|
2499 | {
|
---|
2500 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
2501 |
|
---|
2502 | # Initilise the current_file
|
---|
2503 | # Note: According to http://greenstone.org/dtd/DirectoryMetadata/1.0/DirectoryMetadata.dtd
|
---|
2504 | # FileName tag must come before Description tag
|
---|
2505 | $parser->{'parameters'}->{'current_file'} = "";
|
---|
2506 |
|
---|
2507 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
---|
2508 | return [$tagname => $attrHash];
|
---|
2509 | }
|
---|
2510 |
|
---|
2511 | sub mxml_directorymetadata
|
---|
2512 | {
|
---|
2513 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
2514 |
|
---|
2515 | # if we haven't processed the metadata when we reach the end of metadata.xml
|
---|
2516 | # it's because there's no particular FileSet element whose FileName matched
|
---|
2517 | # In which case, add a new FileSet for this FileName
|
---|
2518 | my $metamode = $parser->{'parameters'}->{'metamode'};
|
---|
2519 | if($metamode ne "done") {
|
---|
2520 |
|
---|
2521 | if ($metamode eq "override") {
|
---|
2522 | print "No metadata value to override. Switching 'metamode' to accumulate\n";
|
---|
2523 | }
|
---|
2524 |
|
---|
2525 | # If we get to here and metamode is override, this means there
|
---|
2526 | # was no existing value to overide => treat as an append operation
|
---|
2527 |
|
---|
2528 | # Create a new FileSet element and append to DirectoryMetadata
|
---|
2529 | # <FileSet>
|
---|
2530 | # <FileName>src_file</FileName>
|
---|
2531 | # <Description>
|
---|
2532 | # <Metadata mode="" name="">metavalue</Metadata>
|
---|
2533 | # </Description>
|
---|
2534 | # </FileSet>
|
---|
2535 | my $src_file = $parser->{'parameters'}->{'src_file'};
|
---|
2536 | my $metaname = $parser->{'parameters'}->{'metaname'};
|
---|
2537 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
---|
2538 | my $metadata_attr = {
|
---|
2539 | '_content' => $metavalue,
|
---|
2540 | 'name' => $metaname,
|
---|
2541 | 'mode' => "accumulate"
|
---|
2542 | };
|
---|
2543 | my $append_metadata = [ "Metadata" => $metadata_attr ];
|
---|
2544 | my $description_attr->{'_content'} = [ "\n\t\t ", $append_metadata, "\n\t\t"];
|
---|
2545 | my $description_element = [ "Description" => $description_attr ];
|
---|
2546 |
|
---|
2547 | #_content is not an attribute, it's special and holds the children of this element
|
---|
2548 | # including the textnode value embedded in this element if any.
|
---|
2549 | my $filename_attr = {'_content' => $src_file};
|
---|
2550 | my $filename_element = [ "FileName" => $filename_attr ];
|
---|
2551 |
|
---|
2552 | my $fileset_attr = {};
|
---|
2553 | $fileset_attr->{'_content'} = [ "\n\t\t", $filename_element,"\n\t\t",$description_element ,"\n\t" ];
|
---|
2554 | my $fileset = [ "FileSet" => $fileset_attr ]; #my $fileset = [ "FileSet" => {} ];
|
---|
2555 |
|
---|
2556 |
|
---|
2557 | # get children of dirmeta, and push the new FileSet element onto it
|
---|
2558 | print "Appending metadata to metadata.xml\n";
|
---|
2559 | my $dirmeta_content = $attrHash->{'_content'};
|
---|
2560 | if (ref($dirmeta_content)) {
|
---|
2561 | # got some existing interesting nested content
|
---|
2562 | #push(@$dirmeta_content, " ", $fileset ,"\n ");
|
---|
2563 | push(@$dirmeta_content, "\t", $fileset ,"\n");
|
---|
2564 | }
|
---|
2565 | else {
|
---|
2566 | #description_content is most likely a string such as "\n"
|
---|
2567 | #$attrHash->{'_content'} = [$dirmeta_content, " ", $fileset ,"\n" ];
|
---|
2568 | $attrHash->{'_content'} = [$dirmeta_content, "\t", $fileset ,"\n" ];
|
---|
2569 | }
|
---|
2570 |
|
---|
2571 | $parser->{'parameters'}->{'metamode'} = "done";
|
---|
2572 | }
|
---|
2573 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
---|
2574 | return [$tagname => $attrHash];
|
---|
2575 | }
|
---|
2576 |
|
---|
2577 |
|
---|
2578 | sub edit_metadata_xml
|
---|
2579 | {
|
---|
2580 | my $self = shift @_;
|
---|
2581 | my ($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $metavalue, $metamode, $src_file, $prevmetavalue) = @_;
|
---|
2582 |
|
---|
2583 | # Set the call-back functions for the metadata tags
|
---|
2584 | my @rules =
|
---|
2585 | ( _default => 'raw',
|
---|
2586 | 'FileName' => \&mxml_filename,
|
---|
2587 | 'Metadata' => \&mxml_metadata,
|
---|
2588 | 'Description' => \&mxml_description,
|
---|
2589 | 'FileSet' => \&mxml_fileset,
|
---|
2590 | 'DirectoryMetadata' => \&mxml_directorymetadata);
|
---|
2591 |
|
---|
2592 | # use XML::Rules to add it in (read in and out again)
|
---|
2593 | my $parser = XML::Rules->new(rules => \@rules,
|
---|
2594 | style => 'filter',
|
---|
2595 | output_encoding => 'utf8',
|
---|
2596 | stripspaces => 2|0|0); # http://search.cpan.org/~jenda/XML-Rules-1.16/lib/XML/Rules.pm
|
---|
2597 |
|
---|
2598 | if (!-e $metadata_xml_filename) {
|
---|
2599 |
|
---|
2600 | if (open(MOUT,">$metadata_xml_filename")) {
|
---|
2601 |
|
---|
2602 | my $src_file_re = &util::filename_to_regex($src_file);
|
---|
2603 | # shouldn't the following also be in the above utility routine??
|
---|
2604 | # $src_file_re =~ s/\./\\./g;
|
---|
2605 |
|
---|
2606 | print MOUT "<?xml version=\"1.0\"?>\n";
|
---|
2607 | print MOUT "<DirectoryMetadata>\n";
|
---|
2608 | print MOUT " <FileSet>\n";
|
---|
2609 | print MOUT " <FileName>$src_file_re</FileName>\n";
|
---|
2610 | print MOUT " <Description>\n";
|
---|
2611 | print MOUT " </Description>\n";
|
---|
2612 | print MOUT " </FileSet>\n";
|
---|
2613 | print MOUT "</DirectoryMetadata>\n";
|
---|
2614 |
|
---|
2615 | close(MOUT);
|
---|
2616 | }
|
---|
2617 | else {
|
---|
2618 | $gsdl_cgi->generate_error("Unable to create $metadata_xml_filename: $!");
|
---|
2619 | }
|
---|
2620 | }
|
---|
2621 |
|
---|
2622 |
|
---|
2623 | my $xml_in = "";
|
---|
2624 | if (!open(MIN,"<$metadata_xml_filename")) {
|
---|
2625 | $gsdl_cgi->generate_error("Unable to read in $metadata_xml_filename: $!");
|
---|
2626 | }
|
---|
2627 | else {
|
---|
2628 | # Read them in
|
---|
2629 | my $line;
|
---|
2630 | while (defined ($line=<MIN>)) {
|
---|
2631 | $xml_in .= $line;
|
---|
2632 | }
|
---|
2633 | close(MIN);
|
---|
2634 |
|
---|
2635 | # Filter with the call-back functions
|
---|
2636 | my $xml_out = "";
|
---|
2637 |
|
---|
2638 | my $MOUT;
|
---|
2639 | if (!open($MOUT,">$metadata_xml_filename")) {
|
---|
2640 | $gsdl_cgi->generate_error("Unable to write out to $metadata_xml_filename: $!");
|
---|
2641 | }
|
---|
2642 | else {
|
---|
2643 | binmode($MOUT,":utf8");
|
---|
2644 |
|
---|
2645 | # Some wise person please find out how to keep the DTD and encode lines in after it gets filtered by this XML::Rules
|
---|
2646 | # At the moment, I will just hack it!
|
---|
2647 | #my $header_with_utf8_dtd = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
|
---|
2648 | #$header_with_utf8_dtd .= "<!DOCTYPE DirectoryMetadata SYSTEM \"http://greenstone.org/dtd/DirectoryMetadata/1.0/DirectoryMetadata.dtd\">";
|
---|
2649 | #$xml_out =~ s/\<\?xml\sversion\=\"1.0\"\?\>/$header_with_utf8_dtd/;
|
---|
2650 | #print MOUT $xml_out;
|
---|
2651 |
|
---|
2652 | $parser->filter($xml_in, $MOUT, { metaname => $metaname,
|
---|
2653 | metapos => $metapos,
|
---|
2654 | metavalue => $metavalue,
|
---|
2655 | metamode => $metamode,
|
---|
2656 | src_file => $src_file,
|
---|
2657 | prevmetavalue => $prevmetavalue,
|
---|
2658 | current_file => undef} );
|
---|
2659 | close($MOUT);
|
---|
2660 | }
|
---|
2661 | }
|
---|
2662 | }
|
---|
2663 |
|
---|
2664 |
|
---|
2665 | sub set_import_metadata
|
---|
2666 | {
|
---|
2667 | my $self = shift @_;
|
---|
2668 |
|
---|
2669 | my $username = $self->{'username'};
|
---|
2670 | my $collect = $self->{'collect'};
|
---|
2671 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
2672 |
|
---|
2673 | if ($baseaction::authentication_enabled) {
|
---|
2674 | # Ensure the user is allowed to edit this collection
|
---|
2675 | $self->authenticate_user($username, $collect);
|
---|
2676 | }
|
---|
2677 |
|
---|
2678 | # Make sure the collection isn't locked by someone else
|
---|
2679 | $self->lock_collection($username, $collect);
|
---|
2680 |
|
---|
2681 | $self->_set_import_metadata(@_);
|
---|
2682 |
|
---|
2683 | # Release the lock once it is done
|
---|
2684 | $self->unlock_collection($username, $collect);
|
---|
2685 |
|
---|
2686 | }
|
---|
2687 |
|
---|
2688 | sub set_import_metadata_array
|
---|
2689 | {
|
---|
2690 | my $self = shift @_;
|
---|
2691 |
|
---|
2692 | my $username = $self->{'username'};
|
---|
2693 | my $collect = $self->{'collect'};
|
---|
2694 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
2695 | # my $gsdlhome = $self->{'gsdlhome'};
|
---|
2696 |
|
---|
2697 | if ($baseaction::authentication_enabled) {
|
---|
2698 | # Ensure the user is allowed to edit this collection
|
---|
2699 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
2700 | }
|
---|
2701 |
|
---|
2702 | my $site = $self->{'site'};
|
---|
2703 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
2704 |
|
---|
2705 | $gsdl_cgi->checked_chdir($collect_dir);
|
---|
2706 |
|
---|
2707 | # Make sure the collection isn't locked by someone else
|
---|
2708 | $self->lock_collection($username, $collect);
|
---|
2709 |
|
---|
2710 | $self->_set_import_metadata_array(@_);
|
---|
2711 |
|
---|
2712 | # Release the lock once it is done
|
---|
2713 | $self->unlock_collection($username, $collect);
|
---|
2714 |
|
---|
2715 | }
|
---|
2716 |
|
---|
2717 |
|
---|
2718 | sub _set_import_metadata_array
|
---|
2719 | {
|
---|
2720 | my $self = shift @_;
|
---|
2721 |
|
---|
2722 | my $collect = $self->{'collect'};
|
---|
2723 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
2724 |
|
---|
2725 | my $site = $self->{'site'};
|
---|
2726 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
2727 |
|
---|
2728 | # look up additional args
|
---|
2729 |
|
---|
2730 | my $infodbtype = $self->{'infodbtype'};
|
---|
2731 |
|
---|
2732 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
---|
2733 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
---|
2734 |
|
---|
2735 | my $json_str = $self->{'json'};
|
---|
2736 | my $doc_array = decode_json $json_str;
|
---|
2737 |
|
---|
2738 | my $global_status = 0;
|
---|
2739 | my $global_mess = "";
|
---|
2740 |
|
---|
2741 | my @all_docids = ();
|
---|
2742 |
|
---|
2743 | foreach my $doc_array_rec ( @$doc_array )
|
---|
2744 | {
|
---|
2745 | my $status = -1;
|
---|
2746 | my $docid = $doc_array_rec->{'docid'};
|
---|
2747 |
|
---|
2748 | my ($docid_root,$docid_secnum);
|
---|
2749 | if(defined $docid) {
|
---|
2750 | ($docid_root,$docid_secnum) = ($docid =~ m/^(.*?)(\..*)?$/);
|
---|
2751 | # as yet no support for setting subsection metadata in metadata.xml
|
---|
2752 | if ((defined $docid_secnum) && ($docid_secnum !~ m/^\s*$/)) {
|
---|
2753 | $gsdl_cgi->generate_message("*** docid: $docid. No support yet for setting import metadata at subsections level.\n");
|
---|
2754 | next; # skip this docid in for loop
|
---|
2755 | }
|
---|
2756 | }
|
---|
2757 |
|
---|
2758 | push(@all_docids,$docid); # docid_root rather
|
---|
2759 |
|
---|
2760 | my $metaname = $doc_array_rec->{'metaname'};
|
---|
2761 | if (defined $metaname) {
|
---|
2762 | my $metamode = $doc_array_rec->{'metamode'} || $self->{'metamode'};
|
---|
2763 | my $metavalue = $doc_array_rec->{'metavalue'};
|
---|
2764 | $metavalue =~ s/<(.*?)>/<$1>/g;
|
---|
2765 |
|
---|
2766 | if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
|
---|
2767 | # make "accumulate" the default (less destructive, as won't actually
|
---|
2768 | # delete any existing values)
|
---|
2769 | $metamode = "accumulate";
|
---|
2770 | }
|
---|
2771 |
|
---|
2772 | # adding metapos and prevmetavalue support to import_metadata subroutines
|
---|
2773 | my $metapos = $doc_array_rec->{'metapos'}; # don't force undef to 0. Undef has meaning when metamode=override
|
---|
2774 | my $prevmetavalue = $self->{'prevmetavalue'};
|
---|
2775 |
|
---|
2776 | $self->set_import_metadata_entry($gsdl_cgi, $arcinfo_doc_filename, $infodbtype, $docid_root, $metaname, $metapos, $metavalue, $metamode, $prevmetavalue, $collect, $collect_dir); # at this point, docid_root = docid
|
---|
2777 |
|
---|
2778 | } elsif (defined $doc_array_rec->{'metatable'}) { # if no metaname, we expect a metatable
|
---|
2779 | my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
|
---|
2780 |
|
---|
2781 | foreach my $metatable_rec ( @$metatable ) {
|
---|
2782 | $metaname = $metatable_rec->{'metaname'};
|
---|
2783 | my $metamode = $metatable_rec->{'metamode'} || $doc_array_rec->{'metamode'} || $self->{'metamode'};
|
---|
2784 | if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
|
---|
2785 | # make "accumulate" the default (less destructive, as won't actually
|
---|
2786 | # delete any existing values)
|
---|
2787 | $metamode = "accumulate";
|
---|
2788 | }
|
---|
2789 |
|
---|
2790 | # No support for metapos and prevmetavalue in the JSON metatable substructure
|
---|
2791 | my $metapos = undef;
|
---|
2792 | my $prevmetavalue = undef;
|
---|
2793 | my $metavals = $metatable_rec->{'metavals'}; # a sub-subarray
|
---|
2794 |
|
---|
2795 | foreach my $metavalue ( @$metavals ) {
|
---|
2796 | $metavalue =~ s/<(.*?)>/<$1>/g;
|
---|
2797 |
|
---|
2798 | $self->set_import_metadata_entry($gsdl_cgi, $arcinfo_doc_filename, $infodbtype, $docid_root, $metaname, $metapos, $metavalue, $metamode, $prevmetavalue, $collect, $collect_dir); # at this point, docid_root = docid
|
---|
2799 | if($metamode eq "override") { # now, having overridden the first metavalue of the metaname,
|
---|
2800 | # need to accumulate subsequent metavals for this metaname, else the just-assigned
|
---|
2801 | # metavalue for this metaname will be lost
|
---|
2802 | $metamode = "accumulate";
|
---|
2803 | }
|
---|
2804 | }
|
---|
2805 | }
|
---|
2806 | }
|
---|
2807 | }
|
---|
2808 |
|
---|
2809 | # always a success message
|
---|
2810 | my $mess = "set-archives-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
|
---|
2811 | $gsdl_cgi->generate_ok_message($mess);
|
---|
2812 | }
|
---|
2813 |
|
---|
2814 | # always returns true (1)
|
---|
2815 | sub set_import_metadata_entry
|
---|
2816 | {
|
---|
2817 | my $self = shift @_;
|
---|
2818 | my ($gsdl_cgi, $arcinfo_doc_filename, $infodbtype, $docid, $metaname, $metapos, $metavalue, $metamode, $prevmetavalue, $collect, $collect_dir) = @_;
|
---|
2819 |
|
---|
2820 | my $info_mess = <<RAWEND;
|
---|
2821 | ****************************
|
---|
2822 | set_import_metadata_entry()
|
---|
2823 | ****************************
|
---|
2824 | collect = $collect
|
---|
2825 | collect_dir = $collect_dir
|
---|
2826 | infodbtype = $infodbtype
|
---|
2827 | arcinfo_doc_filename = $arcinfo_doc_filename
|
---|
2828 | docid = $docid
|
---|
2829 | metaname = $metaname
|
---|
2830 | metapos = $metapos
|
---|
2831 | metavalue = $metavalue
|
---|
2832 | metamode = $metamode
|
---|
2833 | prevmetavalue = $prevmetavalue
|
---|
2834 | ****************************
|
---|
2835 | RAWEND
|
---|
2836 |
|
---|
2837 | $gsdl_cgi->generate_message($info_mess);
|
---|
2838 |
|
---|
2839 | # import works with metadata.xml which can have inherited metadata
|
---|
2840 | # so setting or removing at a metapos can have unintended effects for a COMPLEX collection
|
---|
2841 | # (a collection that has or can have inherited metadata). Metapos has expected behaviour for
|
---|
2842 | # a SIMPLE collection, which is one that doesn't have inherited metadata. Assume caller knows
|
---|
2843 | # what they're doing if they provide a metapos.
|
---|
2844 | if(defined $metapos) {
|
---|
2845 | print STDERR "@@@@ WARNING: metapos defined.\n";
|
---|
2846 | print STDERR "@@@@ Assuming SIMPLE collection and proceeding to modify the import meta at $metapos.\n";
|
---|
2847 | }
|
---|
2848 |
|
---|
2849 | # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
|
---|
2850 | # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
|
---|
2851 | my $metadata_xml_file;
|
---|
2852 | my $import_filename = undef;
|
---|
2853 |
|
---|
2854 | if (defined $docid) {
|
---|
2855 | # my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
---|
2856 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
---|
2857 |
|
---|
2858 | # This now stores the full pathname
|
---|
2859 | $import_filename = $doc_rec->{'src-file'}->[0];
|
---|
2860 | } else { # only for set_import_meta, not the case when calling method is set_import_metadata_array
|
---|
2861 | # as the array version of the method doesn't support the -f parameter yet
|
---|
2862 | my $import_file = $self->{'f'};
|
---|
2863 | $import_filename = &util::filename_cat($collect_dir,$collect,$import_file);
|
---|
2864 | }
|
---|
2865 |
|
---|
2866 | # figure out correct metadata.xml file [?]
|
---|
2867 | # Assuming the metadata.xml file is next to the source file
|
---|
2868 | # Note: This will not work if it is using the inherited metadata from the parent folder
|
---|
2869 | my ($import_tailname, $import_dirname) = File::Basename::fileparse($import_filename);
|
---|
2870 | my $metadata_xml_filename = &util::filename_cat($import_dirname,"metadata.xml");
|
---|
2871 |
|
---|
2872 | # If we're overriding everything, then $metamode=override combined with $metapos=undefined
|
---|
2873 | # in which case, we need to remove all metavalues for the metaname at the given (sub)section
|
---|
2874 | # Thereafter, we will finally be able to set the overriding metavalue for this metaname
|
---|
2875 | if(!defined $metapos && $metamode eq "override") {
|
---|
2876 | ## print STDERR "@@@ REMOVING all import metadata for $metaname\n";
|
---|
2877 | $self->remove_from_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, undef, $import_tailname, $metamode); # we're removing all values, so metavalue=undef
|
---|
2878 |
|
---|
2879 | }
|
---|
2880 |
|
---|
2881 | # Edit the metadata.xml
|
---|
2882 | # Modified by Jeffrey from DL Consulting
|
---|
2883 | # Handle the case where there is one metadata.xml file for multiple FileSets
|
---|
2884 | # The XML filter needs to know whether it is in the right FileSet
|
---|
2885 | # TODO: This doesn't fix the problem where the metadata.xml is not next to the src file.
|
---|
2886 | # TODO: This doesn't handle the common metadata (where FileName doesn't point to a single file)
|
---|
2887 | $self->edit_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname,
|
---|
2888 | $metapos, $metavalue, $metamode, $import_tailname, $prevmetavalue);
|
---|
2889 | #return 0;
|
---|
2890 | return $metadata_xml_filename;
|
---|
2891 | }
|
---|
2892 |
|
---|
2893 | sub _remove_import_metadata
|
---|
2894 | {
|
---|
2895 | my $self = shift @_;
|
---|
2896 |
|
---|
2897 | my $collect = $self->{'collect'};
|
---|
2898 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
2899 | # my $gsdlhome = $self->{'gsdlhome'};
|
---|
2900 | my $infodbtype = $self->{'infodbtype'};
|
---|
2901 |
|
---|
2902 | # Obtain the collect dir
|
---|
2903 | ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
2904 | my $site = $self->{'site'};
|
---|
2905 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
2906 |
|
---|
2907 | ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
2908 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
---|
2909 |
|
---|
2910 | # look up additional args
|
---|
2911 | my $docid = $self->{'d'};
|
---|
2912 | if ((!defined $docid) || ($docid =~ m/^\s*$/))
|
---|
2913 | {
|
---|
2914 | $gsdl_cgi->generate_error("No docid (d=...) specified.\n");
|
---|
2915 | }
|
---|
2916 |
|
---|
2917 | my $metaname = $self->{'metaname'};
|
---|
2918 | my $metapos = $self->{'metapos'};
|
---|
2919 | my $metavalue = $self->{'metavalue'};
|
---|
2920 | if(defined $metavalue) {
|
---|
2921 | $metavalue =~ s/<(.*?)>/<$1>/g;
|
---|
2922 | } elsif (!defined $metapos) { # if given no metavalue or metapos to delete, default to deleting the 1st
|
---|
2923 | $metapos = 0;
|
---|
2924 | }
|
---|
2925 | my $metamode = $self->{'metamode'} || undef;
|
---|
2926 |
|
---|
2927 | # import works with metadata.xml which can have inherited metadata
|
---|
2928 | # so setting or removing at a metapos can have unintended effects for a COMPLEX collection
|
---|
2929 | # (a collection that has or can have inherited metadata). Metapos has expected behaviour for
|
---|
2930 | # a SIMPLE collection, which is one that doesn't have inherited metadata. Assume caller knows
|
---|
2931 | # what they're doing if they provide a metapos.
|
---|
2932 | if(defined $metapos) {
|
---|
2933 | print STDERR "@@@@ WARNING: metapos defined.\n";
|
---|
2934 | print STDERR "@@@@ Assuming SIMPLE collection and proceeding to modify the import meta at $metapos.\n";
|
---|
2935 | }
|
---|
2936 |
|
---|
2937 | # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
|
---|
2938 | # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
|
---|
2939 | my $metadata_xml_file;
|
---|
2940 | my $import_filename = undef;
|
---|
2941 | if (defined $docid)
|
---|
2942 | {
|
---|
2943 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
---|
2944 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
---|
2945 |
|
---|
2946 | # This now stores the full pathname
|
---|
2947 | $import_filename = $doc_rec->{'src-file'}->[0];
|
---|
2948 | }
|
---|
2949 |
|
---|
2950 | if((!defined $import_filename) || ($import_filename =~ m/^\s*$/))
|
---|
2951 | {
|
---|
2952 | $gsdl_cgi->generate_error("There is no metadata\n");
|
---|
2953 | }
|
---|
2954 |
|
---|
2955 | # figure out correct metadata.xml file [?]
|
---|
2956 | # Assuming the metadata.xml file is next to the source file
|
---|
2957 | # Note: This will not work if it is using the inherited metadata from the parent folder
|
---|
2958 | my ($import_tailname, $import_dirname) = File::Basename::fileparse($import_filename);
|
---|
2959 | my $metadata_xml_filename = &util::filename_cat($import_dirname,"metadata.xml");
|
---|
2960 |
|
---|
2961 | $self->remove_from_metadata_xml($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
|
---|
2962 |
|
---|
2963 | my $mess = "remove-import-metadata successful: Key[$docid] -> $metadata_xml_filename\n";
|
---|
2964 | $mess .= " $metaname";
|
---|
2965 | $mess .= " = $metavalue\n";
|
---|
2966 |
|
---|
2967 | $gsdl_cgi->generate_ok_message($mess);
|
---|
2968 |
|
---|
2969 | #return $status; # in case calling functions have a use for this
|
---|
2970 | }
|
---|
2971 |
|
---|
2972 | sub remove_import_metadata
|
---|
2973 | {
|
---|
2974 | my $self = shift @_;
|
---|
2975 |
|
---|
2976 | my $username = $self->{'username'};
|
---|
2977 | my $collect = $self->{'collect'};
|
---|
2978 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
2979 |
|
---|
2980 | if ($baseaction::authentication_enabled) {
|
---|
2981 | # Ensure the user is allowed to edit this collection
|
---|
2982 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
2983 | }
|
---|
2984 |
|
---|
2985 | # Make sure the collection isn't locked by someone else
|
---|
2986 | $self->lock_collection($username, $collect);
|
---|
2987 |
|
---|
2988 | $self->_remove_import_metadata(@_);
|
---|
2989 |
|
---|
2990 | # Release the lock once it is done
|
---|
2991 | $self->unlock_collection($username, $collect);
|
---|
2992 |
|
---|
2993 | }
|
---|
2994 |
|
---|
2995 | sub remove_from_metadata_xml
|
---|
2996 | {
|
---|
2997 | my $self = shift @_;
|
---|
2998 | my ($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $metavalue, $src_file, $metamode) = @_;
|
---|
2999 | # metamode generally has no meaning for removing meta, but is used by set_meta
|
---|
3000 | # when overriding all metavals for a metaname, in which case remove_meta is called with metamode
|
---|
3001 |
|
---|
3002 | # Set the call-back functions for the metadata tags
|
---|
3003 | my @rules =
|
---|
3004 | (
|
---|
3005 | _default => 'raw',
|
---|
3006 | 'Metadata' => \&rfmxml_metadata,
|
---|
3007 | 'FileName' => \&mxml_filename
|
---|
3008 | );
|
---|
3009 |
|
---|
3010 | my $parser = XML::Rules->new
|
---|
3011 | (
|
---|
3012 | rules => \@rules,
|
---|
3013 | style => 'filter',
|
---|
3014 | output_encoding => 'utf8',
|
---|
3015 | #normalisespaces => 1,
|
---|
3016 | stripspaces => 2|0|0 # ineffectual
|
---|
3017 | );
|
---|
3018 |
|
---|
3019 | my $xml_in = "";
|
---|
3020 | if (!open(MIN,"<$metadata_xml_filename"))
|
---|
3021 | {
|
---|
3022 | $gsdl_cgi->generate_error("Unable to read in $metadata_xml_filename: $!");
|
---|
3023 | }
|
---|
3024 | else
|
---|
3025 | {
|
---|
3026 | # Read them in
|
---|
3027 | my $line;
|
---|
3028 | while (defined ($line=<MIN>)) {
|
---|
3029 | $xml_in .= $line;
|
---|
3030 | }
|
---|
3031 | close(MIN);
|
---|
3032 |
|
---|
3033 | # Filter with the call-back functions
|
---|
3034 | my $xml_out = "";
|
---|
3035 |
|
---|
3036 | my $MOUT;
|
---|
3037 | if (!open($MOUT,">$metadata_xml_filename")) {
|
---|
3038 | $gsdl_cgi->generate_error("Unable to write out to $metadata_xml_filename: $!");
|
---|
3039 | }
|
---|
3040 | else {
|
---|
3041 | binmode($MOUT,":utf8");
|
---|
3042 | $parser->filter($xml_in, $MOUT, {metaname => $metaname, metapos => $metapos, metavalue => $metavalue, src_file => $src_file, metamode => $metamode, current_file => undef});
|
---|
3043 | close($MOUT);
|
---|
3044 | }
|
---|
3045 | }
|
---|
3046 | }
|
---|
3047 |
|
---|
3048 | sub rfmxml_metadata
|
---|
3049 | {
|
---|
3050 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
---|
3051 |
|
---|
3052 | # metadata.xml does not handle subsections
|
---|
3053 |
|
---|
3054 | # since metadata.xml now has to deal with metapos, we keep track of the metadata position
|
---|
3055 | if (($parser->{'parameters'}->{'src_file'} eq $parser->{'parameters'}->{'current_file'})
|
---|
3056 | && $parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'})
|
---|
3057 | {
|
---|
3058 | if (!defined $parser->{'parameters'}->{'poscount'})
|
---|
3059 | {
|
---|
3060 | $parser->{'parameters'}->{'poscount'} = 0;
|
---|
3061 | }
|
---|
3062 | else
|
---|
3063 | {
|
---|
3064 | $parser->{'parameters'}->{'poscount'}++;
|
---|
3065 | }
|
---|
3066 |
|
---|
3067 | # if overriding but no metapos, then clear all the meta for this metaname
|
---|
3068 | if ((defined $parser->{'parameters'}->{'metamode'}) && ($parser->{'parameters'}->{'metamode'} eq "override") && (!defined $parser->{'parameters'}->{'metapos'})) {
|
---|
3069 | return [];
|
---|
3070 | }
|
---|
3071 |
|
---|
3072 | if ((defined $parser->{'parameters'}->{'metapos'}) && ($parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'}))
|
---|
3073 | {
|
---|
3074 | return [];
|
---|
3075 | }
|
---|
3076 |
|
---|
3077 | if ((defined $parser->{'parameters'}->{'metavalue'}) && ($attrHash->{'_content'} eq $parser->{'parameters'}->{'metavalue'}))
|
---|
3078 | {
|
---|
3079 | return [];
|
---|
3080 | }
|
---|
3081 | }
|
---|
3082 |
|
---|
3083 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
---|
3084 | return [$tagname => $attrHash];
|
---|
3085 | }
|
---|
3086 |
|
---|
3087 | sub _remove_live_metadata
|
---|
3088 | {
|
---|
3089 | my $self = shift @_;
|
---|
3090 |
|
---|
3091 | my $collect = $self->{'collect'};
|
---|
3092 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
3093 | # my $gsdlhome = $self->{'gsdlhome'};
|
---|
3094 | my $infodbtype = $self->{'infodbtype'};
|
---|
3095 |
|
---|
3096 | # Obtain the collect dir
|
---|
3097 | ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
3098 | my $site = $self->{'site'};
|
---|
3099 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
3100 |
|
---|
3101 |
|
---|
3102 | # look up additional args
|
---|
3103 | my $docid = $self->{'d'};
|
---|
3104 | if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
|
---|
3105 | $gsdl_cgi->generate_error("No docid (d=...) specified.");
|
---|
3106 | }
|
---|
3107 |
|
---|
3108 | # Generate the dbkey
|
---|
3109 | my $metaname = $self->{'metaname'};
|
---|
3110 | my $dbkey = "$docid.$metaname";
|
---|
3111 |
|
---|
3112 | # To people who know $collect_tail please add some comments
|
---|
3113 | # Obtain the live gdbm_db path
|
---|
3114 | my $collect_tail = $collect;
|
---|
3115 | $collect_tail =~ s/^.*[\/|\\]//;
|
---|
3116 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
---|
3117 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, "live-$collect_tail", $index_text_directory);
|
---|
3118 |
|
---|
3119 | # Remove the key
|
---|
3120 | my $cmd = "gdbmdel \"$infodb_file_path\" \"$dbkey\"";
|
---|
3121 | my $status = system($cmd);
|
---|
3122 | if ($status != 0) {
|
---|
3123 | # Catch error if gdbmdel failed
|
---|
3124 | my $mess = "Failed to set metadata key: $dbkey\n";
|
---|
3125 |
|
---|
3126 | $mess .= "PATH: $ENV{'PATH'}\n";
|
---|
3127 | $mess .= "cmd = $cmd\n";
|
---|
3128 | $mess .= "Exit status: $status\n";
|
---|
3129 | $mess .= "System Error Message: $!\n";
|
---|
3130 |
|
---|
3131 | $gsdl_cgi->generate_error($mess);
|
---|
3132 | }
|
---|
3133 | else {
|
---|
3134 | $gsdl_cgi->generate_ok_message("DB remove successful: Key[$metaname]");
|
---|
3135 | }
|
---|
3136 |
|
---|
3137 | }
|
---|
3138 |
|
---|
3139 | sub remove_live_metadata
|
---|
3140 | {
|
---|
3141 | my $self = shift @_;
|
---|
3142 |
|
---|
3143 | my $username = $self->{'username'};
|
---|
3144 | my $collect = $self->{'collect'};
|
---|
3145 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
3146 | my $gsdlhome = $self->{'gsdlhome'};
|
---|
3147 |
|
---|
3148 | if ($baseaction::authentication_enabled) {
|
---|
3149 | # Ensure the user is allowed to edit this collection
|
---|
3150 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
3151 | }
|
---|
3152 |
|
---|
3153 | # Make sure the collection isn't locked by someone else
|
---|
3154 | $self->lock_collection($username, $collect);
|
---|
3155 |
|
---|
3156 | $self->_remove_live_metadata(@_);
|
---|
3157 |
|
---|
3158 | $self->unlock_collection($username, $collect);
|
---|
3159 | }
|
---|
3160 |
|
---|
3161 | sub remove_metadata
|
---|
3162 | {
|
---|
3163 | my $self = shift @_;
|
---|
3164 |
|
---|
3165 | my $where = $self->{'where'};
|
---|
3166 | if(!$where) {
|
---|
3167 | $self->remove_index_metadata(@_); # call the full version of set_index_meta for the default behaviour
|
---|
3168 | return;
|
---|
3169 | }
|
---|
3170 |
|
---|
3171 | my $username = $self->{'username'};
|
---|
3172 | my $collect = $self->{'collect'};
|
---|
3173 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
3174 |
|
---|
3175 | if ($baseaction::authentication_enabled) {
|
---|
3176 | # Ensure the user is allowed to edit this collection
|
---|
3177 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
3178 | }
|
---|
3179 |
|
---|
3180 | # Make sure the collection isn't locked by someone else
|
---|
3181 | $self->lock_collection($username, $collect);
|
---|
3182 |
|
---|
3183 | # check which directories need to be processed, specified in $where as
|
---|
3184 | # any combination of import|archives|index|live
|
---|
3185 | if($where =~ m/import/) {
|
---|
3186 | $self->_remove_import_metadata(@_);
|
---|
3187 | }
|
---|
3188 | if($where =~ m/archives/) {
|
---|
3189 | $self->_remove_archives_metadata(@_);
|
---|
3190 | }
|
---|
3191 | if($where =~ m/index/) {
|
---|
3192 | $self->_remove_index_metadata(@_);
|
---|
3193 | }
|
---|
3194 |
|
---|
3195 | # Release the lock once it is done
|
---|
3196 | $self->unlock_collection($username, $collect);
|
---|
3197 | }
|
---|
3198 |
|
---|
3199 | # the internal version, without authentication
|
---|
3200 | sub _remove_index_metadata
|
---|
3201 | {
|
---|
3202 | my $self = shift @_;
|
---|
3203 |
|
---|
3204 | my $collect = $self->{'collect'};
|
---|
3205 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
3206 | # my $gsdlhome = $self->{'gsdlhome'};
|
---|
3207 | my $infodbtype = $self->{'infodbtype'};
|
---|
3208 |
|
---|
3209 | # Obtain the collect dir
|
---|
3210 | my $site = $self->{'site'};
|
---|
3211 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
3212 | ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
3213 |
|
---|
3214 | # look up additional args
|
---|
3215 | my $docid = $self->{'d'};
|
---|
3216 | if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
|
---|
3217 | $gsdl_cgi->generate_error("No docid (d=...) specified.");
|
---|
3218 | }
|
---|
3219 | my $metaname = $self->{'metaname'};
|
---|
3220 | my $metapos = $self->{'metapos'};
|
---|
3221 | my $metavalue = $self->{'metavalue'} || undef; # necessary to force fallback to undef here
|
---|
3222 |
|
---|
3223 | # To people who know $collect_tail please add some comments
|
---|
3224 | # Obtain the path to the database
|
---|
3225 | my $collect_tail = $collect;
|
---|
3226 | $collect_tail =~ s/^.*[\/|\\]//;
|
---|
3227 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
---|
3228 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, $collect_tail, $index_text_directory);
|
---|
3229 |
|
---|
3230 | # Read the docid entry
|
---|
3231 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $infodb_file_path, $docid);
|
---|
3232 |
|
---|
3233 | # Check to make sure the key does exist
|
---|
3234 | if (!defined ($doc_rec->{$metaname})) {
|
---|
3235 | $gsdl_cgi->generate_error("No metadata field \"" . $metaname . "\" in the specified document: [" . $docid . "]");
|
---|
3236 | }
|
---|
3237 |
|
---|
3238 | # Obtain the specified metadata pos
|
---|
3239 | # if no metavalue or metapos to delete, default to deleting the 1st value for the metaname
|
---|
3240 | if(!defined $metapos && !defined $metavalue) {
|
---|
3241 | $metapos = 0;
|
---|
3242 | }
|
---|
3243 |
|
---|
3244 |
|
---|
3245 | # consider check key is defined before deleting?
|
---|
3246 | # Loop through the metadata array and ignore the specified position
|
---|
3247 | my $filtered_metadata = [];
|
---|
3248 | my $num_metadata_vals = scalar(@{$doc_rec->{$metaname}});
|
---|
3249 | for (my $i=0; $i<$num_metadata_vals; $i++) {
|
---|
3250 | my $metaval = shift(@{$doc_rec->{$metaname}});
|
---|
3251 |
|
---|
3252 | if (!defined $metavalue && $i != $metapos) {
|
---|
3253 | push(@$filtered_metadata,$metaval);
|
---|
3254 | }
|
---|
3255 |
|
---|
3256 | if(defined $metavalue && !($metavalue eq $metaval))
|
---|
3257 | {
|
---|
3258 | push(@$filtered_metadata,$metaval);
|
---|
3259 | }
|
---|
3260 | }
|
---|
3261 | $doc_rec->{$metaname} = $filtered_metadata;
|
---|
3262 |
|
---|
3263 | ## Use the dbutil set_entry method instead of assuming the database is gdbm
|
---|
3264 | my $status = &dbutil::set_infodb_entry($infodbtype, $infodb_file_path, $docid, $doc_rec);
|
---|
3265 |
|
---|
3266 | if ($status != 0) {
|
---|
3267 | my $mess = "Failed to set metadata key: $docid\n";
|
---|
3268 |
|
---|
3269 | $mess .= "PATH: $ENV{'PATH'}\n";
|
---|
3270 | $mess .= "Exit status: $status\n";
|
---|
3271 | $mess .= "System Error Message: $!\n";
|
---|
3272 |
|
---|
3273 | $gsdl_cgi->generate_error($mess);
|
---|
3274 | }
|
---|
3275 | else {
|
---|
3276 | my $mess = "DB set (with item deleted) successful: Key[$docid]\n";
|
---|
3277 | $mess .= " $metaname";
|
---|
3278 | $mess .= "->[$metapos]" if (defined $metapos);
|
---|
3279 |
|
---|
3280 | $gsdl_cgi->generate_ok_message($mess);
|
---|
3281 | }
|
---|
3282 |
|
---|
3283 | #return $status; # in case calling functions have a use for this
|
---|
3284 | }
|
---|
3285 |
|
---|
3286 | sub remove_index_metadata
|
---|
3287 | {
|
---|
3288 | my $self = shift @_;
|
---|
3289 |
|
---|
3290 | my $username = $self->{'username'};
|
---|
3291 | my $collect = $self->{'collect'};
|
---|
3292 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
3293 | # my $gsdlhome = $self->{'gsdlhome'};
|
---|
3294 |
|
---|
3295 | if ($baseaction::authentication_enabled) {
|
---|
3296 | # Ensure the user is allowed to edit this collection
|
---|
3297 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
---|
3298 | }
|
---|
3299 |
|
---|
3300 | # Obtain the collect dir
|
---|
3301 | my $site = $self->{'site'};
|
---|
3302 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
3303 | ## my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
3304 |
|
---|
3305 | # Make sure the collection isn't locked by someone else
|
---|
3306 | $self->lock_collection($username, $collect);
|
---|
3307 |
|
---|
3308 | $self->_remove_index_metadata(@_);
|
---|
3309 |
|
---|
3310 | # Release the lock once it is done
|
---|
3311 | $self->unlock_collection($username, $collect);
|
---|
3312 | }
|
---|
3313 |
|
---|
3314 |
|
---|
3315 | # Was trying to reused the codes, but the functions need to be broken
|
---|
3316 | # down more before they can be reused, otherwise there will be too
|
---|
3317 | # much overhead and duplicate process...
|
---|
3318 | sub insert_metadata
|
---|
3319 | {
|
---|
3320 | my $self = shift @_;
|
---|
3321 |
|
---|
3322 | my $username = $self->{'username'};
|
---|
3323 | my $collect = $self->{'collect'};
|
---|
3324 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
3325 | my $gsdlhome = $self->{'gsdlhome'};
|
---|
3326 | my $infodbtype = $self->{'infodbtype'};
|
---|
3327 |
|
---|
3328 | # If the import metadata and gdbm database have been updated, we
|
---|
3329 | # need to insert some notification to warn user that the the text
|
---|
3330 | # they see at the moment is not indexed and require a rebuild.
|
---|
3331 | my $rebuild_pending_macro = "_rebuildpendingmessage_";
|
---|
3332 |
|
---|
3333 | if ($baseaction::authentication_enabled) {
|
---|
3334 | # Ensure the user is allowed to edit this collection
|
---|
3335 | $self->authenticate_user($username, $collect);
|
---|
3336 | }
|
---|
3337 |
|
---|
3338 | # Obtain the collect and archive dir
|
---|
3339 | my $site = $self->{'site'};
|
---|
3340 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
---|
3341 | ##my $collect_dir = &util::filename_cat($gsdlhome, "collect");
|
---|
3342 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
---|
3343 |
|
---|
3344 | # Make sure the collection isn't locked by someone else
|
---|
3345 | $self->lock_collection($username, $collect);
|
---|
3346 |
|
---|
3347 | # Check additional args
|
---|
3348 | my $docid = $self->{'d'};
|
---|
3349 | if (!defined($docid)) {
|
---|
3350 | $gsdl_cgi->generate_error("No document id is specified: d=...");
|
---|
3351 | }
|
---|
3352 | my $metaname = $self->{'metaname'};
|
---|
3353 | if (!defined($metaname)) {
|
---|
3354 | $gsdl_cgi->generate_error("No metaname is specified: metadataname=...");
|
---|
3355 | }
|
---|
3356 | my $metavalue = $self->{'metavalue'};
|
---|
3357 | if (!defined($metavalue) || $metavalue eq "") {
|
---|
3358 | $gsdl_cgi->generate_error("No metavalue or empty metavalue is specified: metadataname=...");
|
---|
3359 | }
|
---|
3360 | # make "accumulate" the default (less destructive, as won't actually
|
---|
3361 | # delete any existing values)
|
---|
3362 | my $metamode = "accumulate";
|
---|
3363 |
|
---|
3364 | # metapos/prevmetavalue were never before used in this subroutine, so set them to undefined
|
---|
3365 | my $metapos = undef;
|
---|
3366 | my $prevmetavalue = undef;
|
---|
3367 |
|
---|
3368 | #=======================================================================#
|
---|
3369 | # set_import_metadata [START]
|
---|
3370 | #=======================================================================#
|
---|
3371 | # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
|
---|
3372 | # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
|
---|
3373 | my $metadata_xml_file;
|
---|
3374 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
---|
3375 | my $archive_doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
---|
3376 |
|
---|
3377 | # This now stores the full pathname
|
---|
3378 | my $import_filename = $archive_doc_rec->{'src-file'}->[0];
|
---|
3379 |
|
---|
3380 | # figure out correct metadata.xml file [?]
|
---|
3381 | # Assuming the metadata.xml file is next to the source file
|
---|
3382 | # Note: This will not work if it is using the inherited metadata from the parent folder
|
---|
3383 | my ($import_tailname, $import_dirname)
|
---|
3384 | = File::Basename::fileparse($import_filename);
|
---|
3385 | my $metadata_xml_filename = &util::filename_cat($import_dirname,"metadata.xml");
|
---|
3386 |
|
---|
3387 | # Shane's escape characters
|
---|
3388 | $metavalue = pack "U0C*", unpack "C*", $metavalue;
|
---|
3389 | $metavalue =~ s/\,/,/g;
|
---|
3390 | $metavalue =~ s/\:/:/g;
|
---|
3391 | $metavalue =~ s/\|/|/g;
|
---|
3392 | $metavalue =~ s/\(/(/g;
|
---|
3393 | $metavalue =~ s/\)/)/g;
|
---|
3394 | $metavalue =~ s/\[/[/g;
|
---|
3395 | $metavalue =~ s/\\/\/g;
|
---|
3396 | $metavalue =~ s/\]/]/g;
|
---|
3397 | $metavalue =~ s/\{/{/g;
|
---|
3398 | $metavalue =~ s/\}/}/g;
|
---|
3399 | $metavalue =~ s/\"/"/g;
|
---|
3400 | $metavalue =~ s/\`/`/g;
|
---|
3401 | $metavalue =~ s/\n/_newline_/g;
|
---|
3402 |
|
---|
3403 | # Edit the metadata.xml
|
---|
3404 | # Modified by Jeffrey from DL Consulting
|
---|
3405 | # Handle the case where there is one metadata.xml file for multiple FileSets
|
---|
3406 | # The XML filter needs to know whether it is in the right FileSet
|
---|
3407 | # TODO: This doesn't fix the problem where the metadata.xml is not next to the src file.
|
---|
3408 | # TODO: This doesn't handle the common metadata (where FileName doesn't point to a single file)
|
---|
3409 | $self->edit_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname,
|
---|
3410 | $metapos, $metavalue, $metamode, $import_tailname, $prevmetavalue);
|
---|
3411 | #=======================================================================#
|
---|
3412 | # set_import_metadata [END]
|
---|
3413 | #=======================================================================#
|
---|
3414 |
|
---|
3415 |
|
---|
3416 | #=======================================================================#
|
---|
3417 | # set_metadata (accumulate version) [START]
|
---|
3418 | #=======================================================================#
|
---|
3419 | # To people who know $collect_tail please add some comments
|
---|
3420 | # Obtain path to the database
|
---|
3421 | my $collect_tail = $collect;
|
---|
3422 | $collect_tail =~ s/^.*[\/|\\]//;
|
---|
3423 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
---|
3424 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, $collect_tail, $index_text_directory);
|
---|
3425 |
|
---|
3426 | # Read the docid entry
|
---|
3427 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $infodb_file_path, $docid);
|
---|
3428 |
|
---|
3429 | # Protect the quotes
|
---|
3430 | $metavalue =~ s/\"/\\\"/g;
|
---|
3431 |
|
---|
3432 | # Adds the pending macro
|
---|
3433 | my $macro_metavalue = $rebuild_pending_macro . $metavalue;
|
---|
3434 |
|
---|
3435 | # If the metadata doesn't exist, create a new one
|
---|
3436 | if (!defined($doc_rec->{$metaname})){
|
---|
3437 | $doc_rec->{$metaname} = [ $macro_metavalue ];
|
---|
3438 | }
|
---|
3439 | # Else, let's acculumate the values
|
---|
3440 | else {
|
---|
3441 | push(@{$doc_rec->{$metaname}},$macro_metavalue);
|
---|
3442 | }
|
---|
3443 |
|
---|
3444 | ## Use the dbutil set_entry method instead of assuming the database is gdbm
|
---|
3445 | my $status = &dbutil::set_infodb_entry($infodbtype, $infodb_file_path, $docid, $doc_rec);
|
---|
3446 |
|
---|
3447 | if ($status != 0) {
|
---|
3448 | # Catch error if gdbmget failed
|
---|
3449 | my $mess = "Failed to set metadata key: $docid\n";
|
---|
3450 |
|
---|
3451 | $mess .= "PATH: $ENV{'PATH'}\n";
|
---|
3452 | $mess .= "Exit status: $status\n";
|
---|
3453 | $mess .= "System Error Message: $!\n";
|
---|
3454 |
|
---|
3455 | $gsdl_cgi->generate_error($mess);
|
---|
3456 | }
|
---|
3457 | else {
|
---|
3458 | my $mess = "insert-metadata successful: Key[$docid]\n";
|
---|
3459 | $mess .= " [In metadata.xml] $metaname";
|
---|
3460 | $mess .= " = $metavalue\n";
|
---|
3461 | $mess .= " [In database] $metaname";
|
---|
3462 | $mess .= " = $macro_metavalue\n";
|
---|
3463 | $mess .= " The new text has not been indexed, rebuilding collection is required\n";
|
---|
3464 | $gsdl_cgi->generate_ok_message($mess);
|
---|
3465 | }
|
---|
3466 | #=======================================================================#
|
---|
3467 | # set_metadata (accumulate version) [END]
|
---|
3468 | #=======================================================================#
|
---|
3469 |
|
---|
3470 | # Release the lock once it is done
|
---|
3471 | $self->unlock_collection($username, $collect);
|
---|
3472 | }
|
---|
3473 |
|
---|
3474 | 1;
|
---|