source: main/trunk/greenstone3/src/java/org/greenstone/gsdl3/service/OAIPMH.java@ 31580

Last change on this file since 31580 was 31580, checked in by ak19, 7 years ago

Accidentally committed a file I didn't want to.
File size: 34.5 KB
Line 
1/*
2 * OAIPMH.java
3 * Copyright (C) 2010 New Zealand Digital Library, http://www.nzdl.org
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
18 */
19package org.greenstone.gsdl3.service;
20
21// Greenstone classes
22import org.greenstone.gsdl3.core.GSException;
23import org.greenstone.gsdl3.util.GSXML;
24import org.greenstone.gsdl3.util.OAIXML;
25import org.greenstone.gsdl3.util.OID;
26import org.greenstone.gsdl3.util.GSFile;
27import org.greenstone.gsdl3.util.XMLConverter;
28
29import org.greenstone.gsdl3.util.SimpleCollectionDatabase;
30import org.greenstone.gsdl3.util.DBInfo;
31// XML classes
32import org.w3c.dom.Document;
33import org.w3c.dom.Element;
34import org.w3c.dom.NodeList;
35
36// General Java classes
37import java.io.File;
38import java.util.StringTokenizer;
39import java.util.Vector;
40import java.util.Set;
41import java.util.Iterator;
42import java.util.ArrayList;
43import java.util.Date;
44import java.util.HashMap;
45import java.util.HashSet;
46import java.util.Map.Entry;
47
48import org.apache.log4j.Logger;
49
50/** Implements the oai metadata retrieval service for GS3 collections.
51 * Dig into each collection's database and retrieve the metadata
52 *
53 */
54
55public class OAIPMH extends ServiceRack {
56
57 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.OAIPMH.class.getName());
58
59 protected SimpleCollectionDatabase coll_db = null;
60 protected SimpleCollectionDatabase oaiinf_db = null;
61
62 protected String site_name = "";
63 protected String coll_name = "";
64
65 // set this up during configure
66 protected Element list_sets_response = null;
67
68 protected Element meta_formats_definition = null;
69 protected HashMap<String, HashSet<String>> format_elements_map = null;
70 protected HashMap<String, Element> format_response_map = null;
71 /** constructor */
72 public OAIPMH() {
73
74 }
75
76 public void cleanUp() {
77 super.cleanUp();//??
78 this.coll_db.closeDatabase();
79 if (this.oaiinf_db != null){
80 this.oaiinf_db.closeDatabase();
81 }
82
83 }
84 /** configure this service
85 info is the OAIPMH service rack from collectionConfig.xml, and
86 extra_info is buildConfig.xml */
87 public boolean configure(Element info, Element extra_info) {
88 if (!super.configure(info, extra_info)){
89 logger.info("Configuring ServiceRack.java returns false.");
90 return false;
91 }
92
93 //get the names from ServiceRack.java
94 this.site_name = this.router.getSiteName();
95 this.coll_name = this.cluster_name;
96
97 logger.info("Configuring OAIPMH...");
98
99 this.config_info = info;
100
101 // the index stem is either specified in the buildConfig.xml file (extra_info) or uses the collection name
102 Element metadata_list = (Element) GSXML.getChildByTagName(extra_info, GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
103 String index_stem = "";
104 String infodb_type = "";
105 if (metadata_list != null) {
106
107 Element index_stem_elem = (Element) GSXML.getNamedElement(metadata_list, GSXML.METADATA_ELEM, GSXML.NAME_ATT, "indexStem");
108
109 if (index_stem_elem != null) {
110 index_stem = GSXML.getNodeText(index_stem_elem);
111 }
112
113 Element infodb_type_elem = (Element) GSXML.getNamedElement(metadata_list, GSXML.METADATA_ELEM, GSXML.NAME_ATT, "infodbType");
114 if (infodb_type_elem != null) {
115 infodb_type = GSXML.getNodeText(infodb_type_elem);
116 }
117
118 }
119
120 if (index_stem == null || index_stem.equals("")) {
121 index_stem = this.cluster_name; // index_stem is the name of the db in indext/text, it is <colname>.<db>
122 }
123 if (infodb_type == null || infodb_type.equals("")) {
124 infodb_type = "gdbm"; // the default
125 }
126
127 coll_db = new SimpleCollectionDatabase(infodb_type);
128 if (!coll_db.databaseOK()) {
129 logger.error("Couldn't create the collection database of type "+infodb_type);
130 return false;
131 }
132
133 oaiinf_db = new SimpleCollectionDatabase(infodb_type);
134 if (!oaiinf_db.databaseOK()) {
135 logger.error("Couldn't create the oai-inf database of type "+infodb_type);
136 return false;
137 }
138
139
140 // Open databases for querying
141 String coll_db_file = GSFile.collectionDatabaseFile(this.site_home, this.cluster_name, index_stem, infodb_type);
142 if (!this.coll_db.openDatabase(coll_db_file, SimpleCollectionDatabase.READ)) {
143 logger.error("Could not open collection database!");
144 return false;
145 }
146 // the oaiinf_db is called oai-inf.<infodb_type_extension>
147 String oaiinf_db_file = GSFile.OAIInfoDatabaseFile(this.site_home, this.cluster_name, "oai-inf", infodb_type);
148 File oaiinfFile = new File(oaiinf_db_file);
149
150 if(!oaiinfFile.exists()) {
151 logger.warn("oai-inf database for collection + " + this.cluster_name + " does not exist.");
152 oaiinf_db = null;
153 } else if (!this.oaiinf_db.openDatabase(oaiinf_db_file, SimpleCollectionDatabase.READ)) {
154 logger.warn("Could not open oai-inf database for collection + " + this.cluster_name + "!");
155 oaiinf_db = null;
156 }
157
158 // work out what sets this collection has. Will usually contain the collection itself, optional super collection, and maybe subcolls if appropriate classifiers are present.
159 configureSetInfo();
160 // the short_service_info is used by the message router to find the method names,
161
162 Element list_records = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
163 list_records.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_RECORDS);
164 list_records.setAttribute(GSXML.TYPE_ATT, "oai");
165 this.short_service_info.appendChild(list_records);
166
167 Element list_identifiers = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
168 list_identifiers.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_IDENTIFIERS);
169 list_identifiers.setAttribute(GSXML.TYPE_ATT, "oai");
170 this.short_service_info.appendChild(list_identifiers);
171
172 Element list_sets = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
173 list_sets.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_SETS);
174 list_sets.setAttribute(GSXML.TYPE_ATT, "oai");
175 this.short_service_info.appendChild(list_sets);
176
177 Element list_metadata_formats = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
178 list_metadata_formats.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_METADATA_FORMATS);
179 list_metadata_formats.setAttribute(GSXML.TYPE_ATT, "oai");
180 this.short_service_info.appendChild(list_metadata_formats);
181
182 Element get_record = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
183 get_record.setAttribute(GSXML.NAME_ATT, OAIXML.GET_RECORD);
184 get_record.setAttribute(GSXML.TYPE_ATT, "oai");
185 this.short_service_info.appendChild(get_record);
186
187 return true;
188 }
189
190 public boolean configureOAI(Element oai_config_elem) {
191 this.meta_formats_definition = this.desc_doc.createElement(OAIXML.LIST_METADATA_FORMATS);
192 this.format_response_map = new HashMap<String, Element>();
193 this.format_elements_map = new HashMap<String, HashSet<String>>();
194
195 // for now, all we want is the metadata prefix description and the mapping list
196 Element main_lmf_elem = (Element) GSXML.getChildByTagName(oai_config_elem, OAIXML.LIST_METADATA_FORMATS);
197 if (main_lmf_elem == null) {
198 logger.error("No listMetadataFormats element found in OAIConfig.xml");
199 return false;
200 }
201 NodeList meta_formats_list = this.config_info.getElementsByTagName(OAIXML.METADATA_FORMAT);
202 if (meta_formats_list.getLength() == 0) {
203 logger.error("no metadataFormat elements found in OAIPMH serviceRack element");
204 return false;
205 }
206 boolean found_meta_format = false;
207 for(int i=0; i<meta_formats_list.getLength(); i++) {
208 Element mf = (Element) meta_formats_list.item(i);
209 String prefix = mf.getAttribute(OAIXML.METADATA_PREFIX);
210 if (prefix.equals("")) {
211 logger.error("metadataFormat element had no metadataPrefix attribute");
212 continue;
213 }
214 // get the right format from OAICOnfig
215 Element meta_format = findNamedMetadataFormat(main_lmf_elem, prefix);
216 if (meta_format == null) {
217 logger.error("Couldn't find metadataFormat named "+prefix+" in OAIConfig.xml");
218 continue;
219 }
220 // copy the format definition into our stored Element
221 Element collection_version_format = (Element) this.desc_doc.importNode(meta_format, true);
222 collection_version_format.setAttribute(GSXML.NAME_ATT, prefix); // for convenience
223 this.meta_formats_definition.appendChild(collection_version_format);
224 // set up the response element for this format
225 format_response_map.put(prefix, OAIXML.getMetadataFormatShort(this.desc_doc, collection_version_format));
226 // add in collection specific mappings
227 addCollectionMappings(collection_version_format, mf);
228 // now set up a list of all collection elements for reverse lookup of the mapping
229 format_elements_map.put(prefix, getAllCollectionElements(collection_version_format));
230
231 }
232 return true;
233 }
234
235 protected Element findNamedMetadataFormat(Element list_meta_formats, String prefix) {
236 NodeList formats = list_meta_formats.getElementsByTagName(OAIXML.METADATA_FORMAT);
237 for (int i=0; i<formats.getLength(); i++) {
238 Element format = (Element)formats.item(i);
239 String meta_name = GSXML.getNodeText((Element)GSXML.getChildByTagName(format, OAIXML.METADATA_PREFIX));
240 if (prefix.equals(meta_name)) {
241 return format;
242 }
243 }
244 return null;
245 }
246
247 /** goes through the mappings from the collection one, and replaces existing ones in the main one */
248 protected void addCollectionMappings(Element main_meta_format, Element coll_meta_format) {
249
250 Element element_list = (Element)GSXML.getChildByTagName(main_meta_format, OAIXML.ELEMENT+GSXML.LIST_MODIFIER);
251 Document doc = element_list.getOwnerDocument();
252 NodeList coll_elements = coll_meta_format.getElementsByTagName(OAIXML.ELEMENT);
253 if (coll_elements.getLength()==0) {
254 // no mappings to include
255 return;
256 }
257 for (int i=0; i<coll_elements.getLength(); i++) {
258 Element e = (Element)coll_elements.item(i);
259 String elem_name = e.getAttribute(GSXML.NAME_ATT);
260 Element main_elem = GSXML.getNamedElement(element_list, OAIXML.ELEMENT, GSXML.NAME_ATT, elem_name);
261 if (main_elem == null) {
262 logger.error(elem_name+" not found in meta format, not using it");
263 } else {
264 element_list.replaceChild(doc.importNode(e, true),main_elem );
265 }
266 }
267 }
268
269 /** goes through all the mappings and makes a set of all collection
270 metadata names that could become an oai meta element - acts as
271 a reverse lookup for the mappings */
272 protected HashSet<String> getAllCollectionElements(Element meta_format) {
273 HashSet<String> meta_name_set = new HashSet<String>();
274 NodeList elements = meta_format.getElementsByTagName(OAIXML.ELEMENT);
275 for (int i=0; i<elements.getLength(); i++) {
276 Element e = (Element)elements.item(i);
277 Element map = (Element)GSXML.getChildByTagName(e, OAIXML.MAPPING);
278 if (map == null) {
279 // there is no mapping, just use the element name
280 meta_name_set.add(e.getAttribute(GSXML.NAME_ATT));
281 } else {
282 String list_of_names = map.getAttribute(OAIXML.ELEMENTS);
283 String[] name_array = list_of_names.split(",");
284 for (int j=0; j<name_array.length; j++) {
285 meta_name_set.add(name_array[j]);
286 }
287 }
288 }
289 return meta_name_set;
290 }
291
292 /** returns a specific service description */
293 public Element getServiceDescription(Document doc, String service_id, String lang, String subset) {
294
295 if (service_id.equals(OAIXML.LIST_RECORDS)) {
296 Element list_records = doc.createElement(GSXML.SERVICE_ELEM);
297 list_records.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_RECORDS);
298 list_records.setAttribute(GSXML.TYPE_ATT, "oai");
299 return list_records;
300 }
301
302 if (service_id.equals(OAIXML.LIST_IDENTIFIERS)) {
303 Element list_identifiers = doc.createElement(GSXML.SERVICE_ELEM);
304 list_identifiers.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_IDENTIFIERS);
305 list_identifiers.setAttribute(GSXML.TYPE_ATT, "oai");
306 return list_identifiers;
307 }
308 if (service_id.equals(OAIXML.LIST_SETS)) {
309 Element list_sets = doc.createElement(GSXML.SERVICE_ELEM);
310 list_sets.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_SETS);
311 list_sets.setAttribute(GSXML.TYPE_ATT, "oai");
312 return list_sets;
313 }
314 if (service_id.equals(OAIXML.LIST_METADATA_FORMATS)) {
315 Element list_metadata_formats = doc.createElement(GSXML.SERVICE_ELEM);
316 list_metadata_formats.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_METADATA_FORMATS);
317 list_metadata_formats.setAttribute(GSXML.TYPE_ATT, "oai");
318 return list_metadata_formats;
319 }
320
321 if (service_id.equals(OAIXML.GET_RECORD)) {
322 Element get_record = doc.createElement(GSXML.SERVICE_ELEM);
323 get_record.setAttribute(GSXML.NAME_ATT, OAIXML.GET_RECORD);
324 get_record.setAttribute(GSXML.TYPE_ATT, "oai");
325 return get_record;
326 }
327
328 return null;
329 }
330
331 /** The list sets service returns all the sets that this collection is/is part of/contains. This is gathered by Receptionist from all collections to answer the OAI ListSets request. */
332 protected Element processListSets(Element req) {
333 return list_sets_response;
334 }
335 /** returns the actual record element used in the OAI GetRecord response */
336 protected Element processGetRecord(Element req) {
337 /** arguments:
338 identifier: required
339 metadataPrefix: required
340 * Exceptions: badArgument; cannotDisseminateFormat; idDoesNotExist
341 */
342 NodeList params = GSXML.getChildrenByTagName(req, GSXML.PARAM_ELEM);
343 HashMap<String, String> param_map = GSXML.getParamMap(params);
344
345 String prefix = param_map.get(OAIXML.METADATA_PREFIX);
346 if (prefix == null || prefix.equals("")) {
347 //Just a double-check
348 logger.error("the value of metadataPrefix att is not present in the request.");
349 return OAIXML.createErrorResponse(OAIXML.CANNOT_DISSEMINATE_FORMAT, "");
350 }
351
352 // check that we support this format
353 if (!format_response_map.containsKey(prefix)) {
354 logger.error("metadata prefix is not supported for collection "+this.coll_name);
355 return OAIXML.createErrorResponse(OAIXML.CANNOT_DISSEMINATE_FORMAT, "");
356 }
357
358 Document doc = XMLConverter.newDOM();
359
360 String oid = param_map.get(OAIXML.OID); // TODO should this be identifier???
361 boolean OID_is_deleted = false;
362 long millis = -1;
363
364 DBInfo oai_info = null;
365 if(oaiinf_db != null) {
366 oai_info = this.oaiinf_db.getInfo(oid);
367 if (oai_info == null) {
368 logger.warn("OID: " + oid + " is not present in the collection's oai-inf database.");
369 } else {
370 String oaiinf_status = oai_info.getInfo(OAIXML.OAI_INF_STATUS);
371 if(oaiinf_status != null && oaiinf_status.equals(OAIXML.OAI_INF_DELETED)) {
372 OID_is_deleted = true;
373
374 // get the right timestamp for deletion: from oaiinf db
375 String timestamp = oai_info.getInfo(OAIXML.OAI_INF_TIMESTAMP); // in seconds presumably, like oailastmodified in the collection index db
376
377 millis = Long.parseLong(timestamp)*1000; // in milliseconds
378 }
379 }
380 }
381
382 //get a DBInfo object of the identifier; if this identifier is not present in the database,
383 // null is returned.
384 DBInfo info = this.coll_db.getInfo(oid);
385 if (info == null) {
386 logger.error("OID: " + oid + " is not present in the collection database.");
387 //return OAIXML.createErrorResponse(OAIXML.ID_DOES_NOT_EXIST, ""); // not an error: may exist as deleted (marked 'D') in oai-inf db
388 }
389 else if (millis == -1) { // so !OID_is_deleted, get oailastmodified from collection's index db
390 ArrayList<String> keys = new ArrayList<String>(info.getKeys());
391 millis = getDateStampMillis(info);
392 }
393 String oailastmodified = (millis == -1) ? "" : OAIXML.getTime(millis);
394
395
396 Element get_record_response = doc.createElement(GSXML.RESPONSE_ELEM);
397 Element get_record = doc.createElement(OAIXML.GET_RECORD);
398 get_record_response.appendChild(get_record);
399 Element record = doc.createElement(OAIXML.RECORD);
400 //compose the header element
401 record.appendChild(createHeaderElement(doc, oid, oailastmodified, OID_is_deleted));
402 if(!OID_is_deleted) {
403 //compose the metadata element
404 record.appendChild(createMetadataElement(doc, prefix, info));
405 }
406 get_record.appendChild(record);
407 return get_record_response;
408 }
409
410 /** return a list of records in specified set, containing metadata from specified prefix*/
411 protected Element processListRecords(Element req) {
412 return processListIdentifiersOrRecords(req, OAIXML.LIST_RECORDS, true);
413 }
414
415 /** return a list of identifiers in specified set that contain metadata belonging to specified prefix. */
416 protected Element processListIdentifiers(Element req) {
417 return processListIdentifiersOrRecords(req, OAIXML.LIST_IDENTIFIERS, false);
418 }
419
420 // Get a list of records/identifiers that match the parameters.
421 protected Element processListIdentifiersOrRecords(Element req, String response_name, boolean include_metadata) {
422 /** arguments:
423 metadataPrefix: required
424 * from: optional
425 * until: optional
426 * set: optional
427 * resumptionToken: exclusive and optional (ignored as it has been handled by OAIReceptionist)
428 * Exceptions: badArgument; cannotDisseminateFormat; idDoesNotExist
429 */
430 NodeList params = GSXML.getChildrenByTagName(req, GSXML.PARAM_ELEM);
431
432 if(params.getLength() == 0) {
433 logger.error("must at least have the metadataPrefix parameter, can't be none");
434 return OAIXML.createErrorResponse(OAIXML.BAD_ARGUMENT, "");
435 }
436
437 HashMap<String, String> param_map = GSXML.getParamMap(params);
438
439 String prefix = "";
440 Date from_date = null;
441 Date until_date = null;
442
443 if(param_map.containsKey(OAIXML.METADATA_PREFIX) == false) {
444 //Just a double-check
445 logger.error("A param element containing the metadataPrefix is not present.");
446 return OAIXML.createErrorResponse(OAIXML.CANNOT_DISSEMINATE_FORMAT, "");
447 }
448 prefix = param_map.get(OAIXML.METADATA_PREFIX);
449 if (prefix == null || prefix.equals("")) {
450 //Just a double-check
451 logger.error("the value of metadataPrefix att is not present in the request.");
452 return OAIXML.createErrorResponse(OAIXML.CANNOT_DISSEMINATE_FORMAT, "");
453 }
454
455 if(param_map.containsKey(OAIXML.FROM)) {
456 String from = param_map.get(OAIXML.FROM);
457 from_date = OAIXML.getDate(from);
458 }
459 if(param_map.containsKey(OAIXML.UNTIL)) {
460 String until = param_map.get(OAIXML.UNTIL);
461 until_date = OAIXML.getDate(until);
462 }
463
464 if (!format_response_map.containsKey(prefix)) {
465 logger.error(prefix + " metadata prefix is not supported for collection "+this.coll_name);
466 return OAIXML.createErrorResponse(OAIXML.CANNOT_DISSEMINATE_FORMAT, "");
467 }
468
469 // get list of oids
470 ArrayList<String> oid_list = null;
471 if(oaiinf_db != null) { // try getting the OIDs from the oaiinf_db
472 oid_list = new ArrayList<String>(oaiinf_db.getAllKeys());
473
474 if(oid_list == null) { // try getting the OIDs from the oai entries in the index db
475 logger.warn("@@@@@@@@@@@@@ NO OIDs in oai-inf db for " + this.cluster_name);
476 oid_list = getChildrenIds(OAIXML.BROWSELIST);
477 }
478 }
479
480 if (oid_list == null) {
481 logger.error("No matched records found in collection: oai-inf and index db's browselist are empty");
482 return OAIXML.createErrorResponse(OAIXML.NO_RECORDS_MATCH, "");
483 }
484 // all validation is done
485
486 // get the list of elements that are in this metadata prefix
487 HashSet<String> set_of_elems = format_elements_map.get(prefix);
488
489 Document doc = XMLConverter.newDOM();
490 Element list_items_response = doc.createElement(GSXML.RESPONSE_ELEM);
491 Element list_items = doc.createElement(response_name);
492 list_items_response.appendChild(list_items);
493
494 for(int i=0; i<oid_list.size(); i++) {
495 String oid = oid_list.get(i);
496 boolean OID_is_deleted = false;
497 long millis = -1;
498
499 DBInfo oai_info = null;
500 if(oaiinf_db != null) {
501 oai_info = this.oaiinf_db.getInfo(oid);
502 if (oai_info == null) {
503 logger.warn("OID: " + oid + " is not present in the collection's oai-inf database.");
504 } else {
505 String oaiinf_status = oai_info.getInfo(OAIXML.OAI_INF_STATUS);
506 if(oaiinf_status != null && oaiinf_status.equals(OAIXML.OAI_INF_DELETED)) {
507 OID_is_deleted = true;
508
509 // get the right timestamp for deletion: from oaiinf db
510 String timestamp = oai_info.getInfo(OAIXML.OAI_INF_TIMESTAMP); // in seconds presumably, like oailastmodified in the collection index db
511
512 millis = Long.parseLong(timestamp)*1000; // in milliseconds
513 }
514 }
515 }
516 DBInfo info = this.coll_db.getInfo(oid);
517 if (info == null) { // can happen if oid was deleted, in which case only oai_info keeps a record of the oid
518 logger.error("Collection database does not contain information about oid: " +oid);
519 }
520 else if (millis == -1) { // so !OID_is_deleted, get oailastmodified from collection's index db
521
522 millis = getDateStampMillis(info);
523 }
524
525 Date this_date = null;
526 if (millis == -1) {
527 if (from_date != null || until_date !=null) {
528 continue; // if this doc doesn't have a date for some reason, and
529 // we are doing a date range, then don't include it.
530 }
531 } else {
532 this_date = new Date(millis);
533 if (from_date != null) {
534 if(this_date.before(from_date)) {
535 continue;
536 }
537 }
538 if (until_date != null) {
539 if (this_date.after(until_date)) {
540 continue;
541 }
542 }
543 }
544
545 //compose the header element, which we'll be appending no matter what
546 Element header = createHeaderElement(doc, oid, OAIXML.getTime(millis), OID_is_deleted);
547
548 if (include_metadata) { // doing ListRecords
549 // compose a record for adding header and metadata
550 Element record = doc.createElement(OAIXML.RECORD);
551 list_items.appendChild(record);
552 //insert the header element
553 record.appendChild(header);
554 //Now check that this id has metadata for the required prefix.
555 if (info != null && documentContainsMetadata(info, set_of_elems)) {
556 // YES, it does have some metadata for this prefix
557 //compose the metadata element
558 record.appendChild(createMetadataElement(doc, prefix, info));
559 } // otherwise the oid was 'deleted' and only in the oai-inf db and not in the info (collection index) db
560 } else { // doing ListIdentifiers
561 //append the header element
562 list_items.appendChild(header);
563 }
564
565 }//end of for(int i=0; i<oid_list.size(); i++) of doing thru each record
566
567 return list_items_response;
568
569 }
570
571
572 // have implemented setDescription as an element, instead of a container containing metadata
573 private boolean configureSetInfo() {
574
575 Document doc = XMLConverter.newDOM();
576 this.list_sets_response = doc.createElement(GSXML.RESPONSE_ELEM);
577 Element list_sets_elem = doc.createElement(OAIXML.LIST_SETS);
578 this.list_sets_response.appendChild(list_sets_elem);
579 String set_name = this.coll_name;
580 String set_description = null;
581 Element name_elem = (Element)GSXML.getChildByTagName(this.config_info, OAIXML.SET_NAME);
582 if (name_elem!=null) {
583 set_name = GSXML.getNodeText(name_elem);
584 if (set_name.equals("")) {
585 set_name = this.coll_name; // default to coll name if can't find one
586 }
587 }
588 Element description_elem = (Element)GSXML.getChildByTagName(this.config_info, OAIXML.SET_DESCRIPTION);
589 if (description_elem!=null) {
590 set_description = GSXML.getNodeText(description_elem);
591 if (set_description.equals("")) {
592 set_description = null;
593 }
594 }
595 Element coll_set = OAIXML.createSet(doc, this.coll_name, set_name, set_description);
596 list_sets_elem.appendChild(coll_set);
597
598 // are we part of any super sets?
599 NodeList super_set_list = GSXML.getChildrenByTagName(this.config_info, OAIXML.OAI_SUPER_SET);
600 for (int i=0; i<super_set_list.getLength(); i++) {
601 String super_name = ((Element)super_set_list.item(i)).getAttribute(GSXML.NAME_ATT);
602 if (super_name != null && !super_name.equals("")) {
603 list_sets_elem.appendChild(OAIXML.createSet(doc, super_name, super_name, null));
604 }
605 }
606 return true;
607 }
608
609 /** create the metadata element used when processing ListRecords/GetRecord requests
610 */
611 protected Element createMetadataElement(Document doc, String prefix, DBInfo info) {
612 // the <metadata> element
613 Element metadata = doc.createElement(OAIXML.METADATA);
614 // the <oai:dc namespace...> element
615 Element prfx_str_elem = OAIXML.getMetadataPrefixElement(doc, prefix, OAIXML.oai_version);
616 metadata.appendChild(prfx_str_elem);
617
618 Element meta_format_element = GSXML.getNamedElement(this.meta_formats_definition, OAIXML.METADATA_FORMAT, GSXML.NAME_ATT, prefix);
619 NodeList elements = meta_format_element.getElementsByTagName(OAIXML.ELEMENT);
620 // for each element in the definition
621 for (int i=0; i<elements.getLength(); i++) {
622 Element e = (Element)elements.item(i);
623 Element map = (Element)GSXML.getChildByTagName(e, OAIXML.MAPPING);
624 if (map == null) {
625 // look up the element name
626 addMetadata(prfx_str_elem, e.getAttribute(GSXML.NAME_ATT), info);
627 } else {
628 // we go though the list of names in the mapping
629 addMetadata(prfx_str_elem, e.getAttribute(GSXML.NAME_ATT), map.getAttribute(OAIXML.SELECT), map.getAttribute(OAIXML.ELEMENTS), info);
630 }
631 }
632 // output any metadata that is not just a simple mapping
633 addCustomMetadata(prfx_str_elem, prefix, info);
634 return metadata;
635 }
636
637 /** a simple addMetadata where we look for meta_name metadata, and add as that name*/
638 protected void addMetadata(Element meta_list_elem, String meta_name, DBInfo info) {
639 Vector<String> values = info.getMultiInfo(meta_name);
640 if (values != null && values.size()!=0) {
641 for (int i=0; i<values.size(); i++) {
642 addMetadataElement(meta_list_elem, meta_name, values.get(i));
643 }
644 }
645 }
646
647 /** more complicated addMetadata - can add multiple items. */
648 protected void addMetadata(Element meta_list_elem, String new_meta_name, String select_type, String name_list, DBInfo info) {
649 String[] names = name_list.split(",");
650 for (int i=0; i<names.length; i++) {
651 Vector<String> values = info.getMultiInfo(names[i]);
652 if (values == null || values.size()==0) {
653 continue;
654 }
655 for (int j=0; j<values.size(); j++) {
656 addMetadataElement(meta_list_elem, new_meta_name, values.get(j));
657 if (select_type.equals(OAIXML.SELECT_SINGLE_VALUE)) {
658 return; // only want to add one value
659 }
660 }
661 if (select_type.equals(OAIXML.SELECT_FIRST_VALID_META)) {
662 return; // we have added all values of this meta elem
663 }
664 // otherwise, we will keep going through the list and add them all.
665 }
666 }
667
668 // specific metadata formats might need to do some custom metadata that is not
669 //just a standard mapping. eg oai_dc outputting an identifier that is a link
670 protected void addCustomMetadata(Element meta_list_elem, String prefix, DBInfo info) {
671
672 if (prefix.equals(OAIXML.META_FORMAT_DC)) {
673 // we want to add in another dc:identifier element with a link to the resource if possible
674 // try gs.OAIResourceURL first, then srclinkFile, then GS version of documnet
675 String gsURL = info.getInfo(OAIXML.GS_OAI_RESOURCE_URL);
676 if (gsURL.equals("")) {
677 String base_url = OAIXML.getBaseURL(); // e.g. e.g. http://host:port/greenstone3/oaiserver
678 // try srclinkFile
679 gsURL = info.getInfo("srclinkFile");
680 if (!gsURL.equals("")) {
681 // make up the link to the file
682 gsURL = base_url.replace("oaiserver", "") + "sites/" + this.site_name
683 + "/collect/" + this.coll_name + "/index/assoc/"
684 + info.getInfo("assocfilepath") + "/" + gsURL;
685 } else {
686 // no srclink file, lets provide a link to the greenstone doc
687 gsURL = base_url.replace("oaiserver", "library") + "/collection/" + this.coll_name + "/document/" + info.getInfo("Identifier");
688 }
689 }
690 // now we have the url link, add as metadata
691 addMetadataElement(meta_list_elem, "dc:identifier", gsURL);
692 }
693 }
694
695 /** create the actual metadata element for the list */
696 protected void addMetadataElement(Element meta_list_elem, String name, String value) {
697
698 Element meta = GSXML.createTextElement(meta_list_elem.getOwnerDocument(), name, value);
699 meta_list_elem.appendChild(meta);
700 }
701
702
703 /** create a header element used when processing requests like ListRecords/GetRecord/ListIdentifiers
704 */
705 protected Element createHeaderElement(Document doc, String oid, String oailastmodified, boolean deleted) {
706
707 Element header = doc.createElement(OAIXML.HEADER);
708
709 // if deleted, get the date and change oailastmodified to timestamp in oaiinfo
710 if(deleted) {
711 header.setAttribute(OAIXML.OAI_INF_STATUS, OAIXML.HEADER_STATUS_ATTR_DELETED); // set the header status to deleted
712 // then the timestamp for deletion will be from oai-inf database
713 }
714
715 Element identifier = doc.createElement(OAIXML.IDENTIFIER);
716 GSXML.setNodeText(identifier, coll_name + ":" + oid);
717 header.appendChild(identifier);
718 Element set_spec = doc.createElement(OAIXML.SET_SPEC);
719 GSXML.setNodeText(set_spec, coll_name);
720 header.appendChild(set_spec);
721 Element datestamp = doc.createElement(OAIXML.DATESTAMP);
722 GSXML.setNodeText(datestamp, oailastmodified);
723 header.appendChild(datestamp);
724 return header;
725 }
726
727 /** return the metadata information */
728 protected Element processListMetadataFormats(Element req) {
729 // the request sent here must contain an OID. see doListMetadataFormats() in OAIReceptionist
730 Element param = GSXML.getNamedElement(req, GSXML.PARAM_ELEM, GSXML.NAME_ATT, OAIXML.OID);
731 if (param == null) {
732 logger.error("An element containing the OID attribute not is present.");
733 return OAIXML.createErrorResponse(OAIXML.ID_DOES_NOT_EXIST, "");
734 }
735 String oid = param.getAttribute(GSXML.VALUE_ATT);
736 if (oid == null || oid.equals("")) {
737 logger.error("No OID is present in the request.");
738 return OAIXML.createErrorResponse(OAIXML.ID_DOES_NOT_EXIST, "");
739 }
740
741 /*
742 ArrayList<String> oid_list = null;
743 if(oaiinf_db != null) { // try getting the OIDs from the oaiinf_db
744 oid_list = new ArrayList<String>(oaiinf_db.getAllKeys());
745
746 if(oid_list == null) { // try getting the OIDs from the oai entries in the index db
747 oid_list = getChildrenIds(OAIXML.BROWSELIST);
748 }
749 }
750 */
751 // assume meta formats are only for OIDs that have not been deleted
752 // so don't need to check oai-inf db, and can just check collection's index db for list of OIDs
753 ArrayList<String> oid_list = getChildrenIds(OAIXML.BROWSELIST);
754 if (oid_list == null || oid_list.contains(oid) == false) {
755 logger.error("OID: " + oid + " is not present in the database.");
756 Element e= OAIXML.createErrorResponse(OAIXML.ID_DOES_NOT_EXIST, "");
757// logger.error((new XMLConverter()).getPrettyString (e));
758 return e;
759 }
760
761 DBInfo info = null;
762 info = this.coll_db.getInfo(oid);
763 if (info == null) { //just double check
764 return OAIXML.createErrorResponse(OAIXML.OAI_SERVICE_UNAVAILABLE, "");
765 }
766
767 Document doc = XMLConverter.newDOM();
768 Element list_metadata_formats_response = doc.createElement(GSXML.RESPONSE_ELEM);
769
770 Element list_metadata_formats = doc.createElement(OAIXML.LIST_METADATA_FORMATS);
771 list_metadata_formats_response.appendChild(list_metadata_formats);
772 boolean has_meta_format = false;
773
774 // for each format in format_elements_map
775 Iterator<String> it = format_elements_map.keySet().iterator();
776 while (it.hasNext()) {
777 String format = it.next();
778 HashSet<String> set_of_elems = format_elements_map.get(format);
779 if (documentContainsMetadata(info, set_of_elems)) {
780 // add this format into the response
781 has_meta_format = true;
782 list_metadata_formats.appendChild(doc.importNode(format_response_map.get(format), true));
783 }
784 }
785
786 if (has_meta_format == false) {
787 logger.error("Specified metadata names are not contained in the database.");
788 return OAIXML.createErrorResponse(OAIXML.NO_METADATA_FORMATS, "");
789 } else {
790 return list_metadata_formats_response;
791 }
792 }
793
794 protected boolean documentContainsMetadata(DBInfo info, HashSet<String> set_of_elems) {
795 if (set_of_elems.size() == 0) {
796 return false;
797 }
798 Iterator<String> i = set_of_elems.iterator();
799 while (i.hasNext()) {
800 if (!info.getInfo(i.next()).equals("")) {
801 return true;
802 }
803 }
804 return false;
805 }
806
807 /** returns a list of the child ids in order, null if no children */
808 protected ArrayList<String> getChildrenIds(String node_id) {
809 DBInfo info = this.coll_db.getInfo(node_id);
810 if (info == null) {
811 return null;
812 }
813
814 String contains = info.getInfo("contains");
815 if (contains.equals("")) {
816 return null;
817 }
818 ArrayList<String> children = new ArrayList<String>();
819 StringTokenizer st = new StringTokenizer(contains, ";");
820 while (st.hasMoreTokens()) {
821 String child_id = st.nextToken().replaceAll("\"", node_id);
822 children.add(child_id);
823 }
824 return children;
825 }
826 /**method to check whether any of the 'metadata_names' is contained in the 'info'.
827 * The name may be in the form: <name>,<mapped name>, in which the mapped name is
828 * optional. The mapped name is looked up in the DBInfo; if not present, use the first
829 * name which is mandatory.
830 */
831 protected boolean containsMetadata(DBInfo info, String[] metadata_names) {
832 if (metadata_names == null) return false;
833 logger.info("checking metadata names in db.");
834 for(int i=0; i<metadata_names.length; i++) {
835 int index = metadata_names[i].indexOf(",");
836 String meta_name = (index == -1) ? metadata_names[i] :
837 metadata_names[i].substring(index + 1);
838
839 if(info.getInfo(meta_name).equals("") == false) {
840 return true;
841 }
842 }
843 return false;
844 }
845
846 protected long getDateStampMillis(DBInfo info) {
847 // gs.OAIDateStamp is in YYYY-MM-DD
848 String time_stamp = info.getInfo(OAIXML.GS_OAI_DATE_STAMP);
849 long millis = -1;
850 if (!time_stamp.equals("")) {
851 millis = OAIXML.getTime(time_stamp);
852 }
853 if (millis == -1) {
854 // oailastmodified is in seconds
855 time_stamp = info.getInfo(OAIXML.OAI_LASTMODIFIED);
856 if (!time_stamp.equals("")) {
857 millis = Long.parseLong(time_stamp)*1000;
858 }
859 }
860 return millis;
861
862
863 }
864}
865
866
Note: See TracBrowser for help on using the repository browser.