source: gs3-extensions/iiif-servlet/trunk/src/gsdl-src/java/org/greenstone/gsdl3/service/IIIFPMH.java@ 32842

Last change on this file since 32842 was 32842, checked in by davidb, 5 years ago

Shift from OAI as a template, to separate IIIF based classes

File size: 37.0 KB
Line 
1/*
2 * IIIFPMH.java
3 * Copyright (C) 2019 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.IIIFXML;
25import org.greenstone.gsdl3.util.OAIXML; // ****
26import org.greenstone.gsdl3.util.OID;
27import org.greenstone.gsdl3.util.GSFile;
28import org.greenstone.gsdl3.util.XMLConverter;
29
30import org.greenstone.gsdl3.util.SimpleCollectionDatabase;
31import org.greenstone.gsdl3.util.DBInfo;
32// XML classes
33import org.w3c.dom.Document;
34import org.w3c.dom.Element;
35import org.w3c.dom.NodeList;
36
37// General Java classes
38import java.io.File;
39import java.util.StringTokenizer;
40import java.util.Vector;
41import java.util.Set;
42import java.util.Iterator;
43import java.util.ArrayList;
44import java.util.Date;
45import java.util.HashMap;
46import java.util.HashSet;
47import java.util.Map.Entry;
48
49import org.apache.log4j.Logger;
50
51/** Implements the IIIF service for GS3 collections.
52 * Digs into each collection's database to study the metadata
53 * and from that determine what file in a document's assocdir
54 * can be used to provide a source image for the doc
55 *
56 */
57
58public class IIIFPMH extends ServiceRack {
59
60 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.IIIFPMH.class.getName());
61
62 protected SimpleCollectionDatabase coll_db = null;
63 //protected SimpleCollectionDatabase oaiinf_db = null;
64
65 protected String site_name = "";
66 protected String coll_name = "";
67
68 // set this up during configure
69 protected Element list_sets_response = null;
70
71 protected Element meta_formats_definition = null;
72 protected HashMap<String, HashSet<String>> format_elements_map = null;
73 protected HashMap<String, Element> format_response_map = null;
74 protected HashMap<String, Element> format_meta_elem_map = null;
75
76 protected String index_stem = "";
77 protected String infodb_type = "";
78
79 /** constructor */
80 public IIIFPMH() {
81
82 }
83
84 public void cleanUp() {
85 super.cleanUp();//??
86
87 if(this.coll_db != null) {
88 this.coll_db.closeDatabase();
89 this.coll_db = null;
90 }
91 /*
92 if (this.oaiinf_db != null){
93 this.oaiinf_db.closeDatabase();
94 }*/
95 }
96
97 /** configure this service
98 info is the IIIFPMH service rack from collectionConfig.xml, and
99 extra_info is buildConfig.xml */
100 public boolean configure(Element info, Element extra_info) {
101 if (!super.configure(info, extra_info)){
102 logger.info("Configuring IIIFPMH(extends ServiceRack) returns false.");
103 return false;
104 }
105
106 //get the names from ServiceRack.java
107 this.site_name = this.router.getSiteName();
108 this.coll_name = this.cluster_name;
109
110 logger.info("Configuring IIIFPMH...");
111
112 this.config_info = info;
113
114 // the index stem is either specified in the buildConfig.xml file (extra_info) or uses the collection name
115 Element metadata_list = (Element) GSXML.getChildByTagName(extra_info, GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
116
117 if (metadata_list != null) {
118
119 Element index_stem_elem = (Element) GSXML.getNamedElement(metadata_list, GSXML.METADATA_ELEM, GSXML.NAME_ATT, "indexStem");
120
121 if (index_stem_elem != null) {
122 this.index_stem = GSXML.getNodeText(index_stem_elem);
123 }
124
125 Element infodb_type_elem = (Element) GSXML.getNamedElement(metadata_list, GSXML.METADATA_ELEM, GSXML.NAME_ATT, "infodbType");
126 if (infodb_type_elem != null) {
127 this.infodb_type = GSXML.getNodeText(infodb_type_elem);
128 }
129
130 }
131
132 if (index_stem == null || index_stem.equals("")) {
133 this.index_stem = this.cluster_name; // index_stem is the name of the db in indext/text, it is <colname>.<db>
134 }
135 if (infodb_type == null || infodb_type.equals("")) {
136 this.infodb_type = "gdbm"; // the default
137 }
138
139 /*
140 // Work out what sets this collection has.
141 //
142 // Will usually contain the collection itself, optional super
143 // collection, and maybe subcolls if appropriate classifiers are
144 // present.
145
146 configureSetInfo();*/
147
148 // the short_service_info is used by the message router to find the method names,
149 /*
150 Element list_records = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
151 list_records.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_RECORDS);
152 list_records.setAttribute(GSXML.TYPE_ATT, "oai");
153 this.short_service_info.appendChild(list_records);
154
155 Element list_identifiers = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
156 list_identifiers.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_IDENTIFIERS);
157 list_identifiers.setAttribute(GSXML.TYPE_ATT, "oai");
158 this.short_service_info.appendChild(list_identifiers);
159
160 Element list_sets = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
161 list_sets.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_SETS);
162 list_sets.setAttribute(GSXML.TYPE_ATT, "oai");
163 this.short_service_info.appendChild(list_sets);
164
165 Element list_metadata_formats = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
166 list_metadata_formats.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_METADATA_FORMATS);
167 list_metadata_formats.setAttribute(GSXML.TYPE_ATT, "oai");
168 this.short_service_info.appendChild(list_metadata_formats);
169 */
170 Element get_record = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
171 get_record.setAttribute(GSXML.NAME_ATT, IIIFXML.GET_RECORD);
172 get_record.setAttribute(GSXML.TYPE_ATT, "oai");
173 this.short_service_info.appendChild(get_record);
174
175 return true;
176 }
177
178 public boolean configureIIIF(Element iiif_config_elem) {
179 /*
180 this.meta_formats_definition = this.desc_doc.createElement(OAIXML.LIST_METADATA_FORMATS);
181 this.format_response_map = new HashMap<String, Element>();
182 this.format_elements_map = new HashMap<String, HashSet<String>>();
183 this.format_meta_elem_map = new HashMap<String, Element>();
184
185 // for now, all we want is the metadata prefix description and the mapping list
186 Element main_lmf_elem = (Element) GSXML.getChildByTagName(oai_config_elem, OAIXML.LIST_METADATA_FORMATS);
187 if (main_lmf_elem == null) {
188 logger.error("No listMetadataFormats element found in OAIConfig.xml");
189 return false;
190 }
191 NodeList meta_formats_list = this.config_info.getElementsByTagName(OAIXML.METADATA_FORMAT);
192 if (meta_formats_list.getLength() == 0) {
193 logger.error("no metadataFormat elements found in OAIPMH serviceRack element");
194 return false;
195 }
196
197 boolean found_meta_format = false;
198 for(int i=0; i<meta_formats_list.getLength(); i++) {
199 Element mf = (Element) meta_formats_list.item(i);
200 String prefix = mf.getAttribute(OAIXML.METADATA_PREFIX);
201 if (prefix.equals("")) {
202 logger.error("metadataFormat element had no metadataPrefix attribute");
203 continue;
204 }
205 // get the right format from OAICOnfig
206 Element meta_format = findNamedMetadataFormat(main_lmf_elem, prefix);
207 if (meta_format == null) {
208 logger.error("Couldn't find metadataFormat named "+prefix+" in OAIConfig.xml");
209 continue;
210 }
211
212 // copy the format definition into our stored Element
213 Element collection_version_format = (Element) this.desc_doc.importNode(meta_format, true);
214 collection_version_format.setAttribute(GSXML.NAME_ATT, prefix); // for convenience
215 this.meta_formats_definition.appendChild(collection_version_format);
216 // set up the response element for this format
217 format_response_map.put(prefix, OAIXML.getMetadataFormatShort(this.desc_doc, collection_version_format));
218 // add in collection specific mappings
219 addCollectionMappings(collection_version_format, mf);
220 // now set up a list of all collection elements for reverse lookup of the mapping
221 format_elements_map.put(prefix, getAllCollectionElements(collection_version_format));
222 format_meta_elem_map.put(prefix, OAIXML.getMetadataPrefixElement(this.desc_doc, prefix, collection_version_format));
223
224 } // end for
225 */
226
227 // Open the coll db db databases and store handles to them
228 coll_db = new SimpleCollectionDatabase(infodb_type);
229 if (!coll_db.databaseOK()) {
230 logger.error("Couldn't create the collection database of type "+infodb_type);
231 return false;
232 }
233 /*
234 oaiinf_db = new SimpleCollectionDatabase(infodb_type);
235 if (!oaiinf_db.databaseOK()) {
236 logger.error("Couldn't create the oai-inf database of type "+infodb_type);
237 return false;
238 }
239 */
240
241 // Open databases for querying
242 String coll_db_file = GSFile.collectionDatabaseFile(this.site_home, this.cluster_name, index_stem, infodb_type);
243 if (!this.coll_db.openDatabase(coll_db_file, SimpleCollectionDatabase.READ)) {
244 logger.error("Could not open collection database!");
245 return false;
246 }
247 /*
248 // the oaiinf_db is called oai-inf.<infodb_type_extension>
249 String oaiinf_db_file = GSFile.OAIInfoDatabaseFile(this.site_home, this.cluster_name, "oai-inf", infodb_type);
250 File oaiinfFile = new File(oaiinf_db_file);
251
252 if(!oaiinfFile.exists()) {
253 logger.warn("oai-inf database for collection + " + this.cluster_name + " does not exist.");
254 oaiinf_db = null;
255 } else if (!this.oaiinf_db.openDatabase(oaiinf_db_file, SimpleCollectionDatabase.READ)) {
256 logger.warn("Could not open oai-inf database for collection + " + this.cluster_name + "!");
257 oaiinf_db = null;
258 }
259 */
260
261 return true;
262 }
263
264 /**
265 * @return the associated OAICollection's OAI_EARLIEST_TIMESTAMP_OID record's
266 * OAI_INF_TIMESTAMP field from the collection's oai-inf.db IN MILLISECONDS
267 */
268 /*
269 public long getEarliestTimestamp() {
270 long timestamp = -1;
271
272 DBInfo oai_info = null;
273 if(oaiinf_db != null) {
274 // get internal record containing the earliest timestamp of the collection
275 oai_info = this.oaiinf_db.getInfo(OAIXML.OAI_EARLIEST_TIMESTAMP_OID);
276 if (oai_info == null) {
277 logger.warn("Can't get collection " + this.cluster_name + "'s earliest timestamp from oai-inf db. No entry for 'OID' " + OAIXML.OAI_EARLIEST_TIMESTAMP_OID + " in the db.");
278 } else {
279 timestamp = Long.parseLong(oai_info.getInfo(OAIXML.OAI_INF_TIMESTAMP)) * 1000; // stored in seconds, so x1000 to convert to milliseconds
280 //logger.info("@@@ found earliest OAI timestamp for collection " + this.coll_name + ": " + timestamp + " (ms)");
281 }
282 }
283 return timestamp;
284 }
285 */
286
287 /*
288 protected Element findNamedMetadataFormat(Element list_meta_formats, String prefix) {
289 NodeList formats = list_meta_formats.getElementsByTagName(OAIXML.METADATA_FORMAT);
290 for (int i=0; i<formats.getLength(); i++) {
291 Element format = (Element)formats.item(i);
292 String meta_name = GSXML.getNodeText((Element)GSXML.getChildByTagName(format, OAIXML.METADATA_PREFIX));
293 if (prefix.equals(meta_name)) {
294 return format;
295 }
296 }
297 return null;
298 }
299 */
300
301 /** goes through the mappings from the collection one, and replaces existing ones in the main one */
302 /*
303 protected void addCollectionMappings(Element main_meta_format, Element coll_meta_format) {
304
305 Element element_list = (Element)GSXML.getChildByTagName(main_meta_format, OAIXML.ELEMENT+GSXML.LIST_MODIFIER);
306 Document doc = element_list.getOwnerDocument();
307 NodeList coll_elements = coll_meta_format.getElementsByTagName(OAIXML.ELEMENT);
308 if (coll_elements.getLength()==0) {
309 // no mappings to include
310 return;
311 }
312 for (int i=0; i<coll_elements.getLength(); i++) {
313 Element e = (Element)coll_elements.item(i);
314 String elem_name = e.getAttribute(GSXML.NAME_ATT);
315 Element main_elem = GSXML.getNamedElement(element_list, OAIXML.ELEMENT, GSXML.NAME_ATT, elem_name);
316 if (main_elem == null) {
317 logger.error(elem_name+" not found in meta format, not using it");
318 } else {
319 element_list.replaceChild(doc.importNode(e, true),main_elem );
320 }
321 }
322 }
323 */
324
325 /** goes through all the mappings and makes a set of all collection
326 metadata names that could become an oai meta element - acts as
327 a reverse lookup for the mappings */
328 /*
329 protected HashSet<String> getAllCollectionElements(Element meta_format) {
330 HashSet<String> meta_name_set = new HashSet<String>();
331 NodeList elements = meta_format.getElementsByTagName(OAIXML.ELEMENT);
332 for (int i=0; i<elements.getLength(); i++) {
333 Element e = (Element)elements.item(i);
334 Element map = (Element)GSXML.getChildByTagName(e, OAIXML.MAPPING);
335 if (map == null) {
336 // there is no mapping, just use the element name
337 meta_name_set.add(e.getAttribute(GSXML.NAME_ATT));
338 } else {
339 String list_of_names = map.getAttribute(OAIXML.ELEMENTS);
340 String[] name_array = list_of_names.split(",");
341 for (int j=0; j<name_array.length; j++) {
342 meta_name_set.add(name_array[j]);
343 }
344 }
345 }
346 return meta_name_set;
347 }
348 */
349
350 /** returns a specific service description */
351 public Element getServiceDescription(Document doc, String service_id, String lang, String subset) {
352
353 /*
354 if (service_id.equals(OAIXML.LIST_RECORDS)) {
355 Element list_records = doc.createElement(GSXML.SERVICE_ELEM);
356 list_records.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_RECORDS);
357 list_records.setAttribute(GSXML.TYPE_ATT, "oai");
358 return list_records;
359 }
360
361 if (service_id.equals(OAIXML.LIST_IDENTIFIERS)) {
362 Element list_identifiers = doc.createElement(GSXML.SERVICE_ELEM);
363 list_identifiers.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_IDENTIFIERS);
364 list_identifiers.setAttribute(GSXML.TYPE_ATT, "oai");
365 return list_identifiers;
366 }
367 if (service_id.equals(OAIXML.LIST_SETS)) {
368 Element list_sets = doc.createElement(GSXML.SERVICE_ELEM);
369 list_sets.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_SETS);
370 list_sets.setAttribute(GSXML.TYPE_ATT, "oai");
371 return list_sets;
372 }
373 if (service_id.equals(OAIXML.LIST_METADATA_FORMATS)) {
374 Element list_metadata_formats = doc.createElement(GSXML.SERVICE_ELEM);
375 list_metadata_formats.setAttribute(GSXML.NAME_ATT, OAIXML.LIST_METADATA_FORMATS);
376 list_metadata_formats.setAttribute(GSXML.TYPE_ATT, "oai");
377 return list_metadata_formats;
378 }
379 */
380
381 if (service_id.equals(IIIFXML.GET_RECORD)) {
382 Element get_record = doc.createElement(GSXML.SERVICE_ELEM);
383 get_record.setAttribute(GSXML.NAME_ATT, IIIFXML.GET_RECORD);
384 get_record.setAttribute(GSXML.TYPE_ATT, "oai");
385 return get_record;
386 }
387
388 return null;
389 }
390
391 /** 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. */
392 /*
393 protected Element processListSets(Element req) {
394 return list_sets_response;
395 }
396 */
397
398 /** returns the actual record element used in the OAI GetRecord response */
399 protected Element processGetRecord(Element req) {
400 /** arguments:
401 identifier: required
402 metadataPrefix: required
403 * Exceptions: badArgument; cannotDisseminateFormat; idDoesNotExist
404 */
405 NodeList params = GSXML.getChildrenByTagName(req, GSXML.PARAM_ELEM);
406 HashMap<String, String> param_map = GSXML.getParamMap(params);
407
408 String prefix = param_map.get(OAIXML.METADATA_PREFIX);
409 if (prefix == null || prefix.equals("")) {
410 //Just a double-check
411 logger.error("the value of metadataPrefix att is not present in the request.");
412 return IIIFXML.createErrorResponse(OAIXML.CANNOT_DISSEMINATE_FORMAT, "");
413 }
414
415 // check that we support this format
416 if (!format_response_map.containsKey(prefix)) {
417 logger.error("metadata prefix is not supported for collection "+this.coll_name);
418 return IIIFXML.createErrorResponse(OAIXML.CANNOT_DISSEMINATE_FORMAT, "");
419 }
420
421 Document doc = XMLConverter.newDOM();
422
423 String oid = param_map.get(IIIFXML.OID); // TODO should this be identifier???
424
425 /*
426 boolean OID_is_deleted = false;
427 long millis = -1;
428
429 DBInfo oai_info = null;
430 if(oaiinf_db != null) {
431 oai_info = this.oaiinf_db.getInfo(oid);
432 if (oai_info == null) {
433 logger.warn("OID: " + oid + " is not present in the collection's oai-inf database.");
434 } else {
435
436 // indexdb doesn't have info on deleted docs, only oaiinf db does.
437 // So only oaiinfdb has timestamps for deleted docs
438 // For non-deleted doc ids: also obtain timestamp from oaiinf db,
439 // but if the oaiinf db doesn't exist, resort to oailastmodified fields of indexdb.
440 String timestamp = oai_info.getInfo(OAIXML.OAI_INF_TIMESTAMP); // stored in seconds, like oailastmodified in the collection index db
441 millis = Long.parseLong(timestamp)*1000; // in milliseconds
442
443 String oaiinf_status = oai_info.getInfo(OAIXML.OAI_INF_STATUS);
444 if(oaiinf_status != null && oaiinf_status.equals(OAIXML.OAI_INF_DELETED)) {
445 OID_is_deleted = true;
446 }
447 }
448 }
449 */
450
451 //get a DBInfo object of the identifier; if this identifier is not present in the database,
452 // null is returned.
453 DBInfo info = this.coll_db.getInfo(oid);
454 if (info == null) {
455 //if(!OID_is_deleted) { // we don't expect to find entries for deleted docs in index db. // ****
456 logger.error("OID: " + oid + " is not present in the collection index database.");
457 return IIIFXML.createErrorResponse(IIIFXML.ID_DOES_NOT_EXIST, "");
458 //} // if doc deleted, id missing in indexdb is not an error: doc id would exist only in oai-inf db, marked as deleted 'D' // ****
459 }
460
461 /*
462 else if (millis == -1) { // so couldn't get doc lastmod from oaiinf db, get oailastmodified from collection's index db
463 millis = getDateStampMillis(info);
464 }
465 String oailastmodified = (millis == -1) ? "" : OAIXML.getTime(millis);
466 */
467
468 Element get_record_response = doc.createElement(GSXML.RESPONSE_ELEM);
469 Element get_record = doc.createElement(IIIFXML.GET_RECORD);
470 get_record_response.appendChild(get_record);
471 Element record = doc.createElement(IIIFXML.RECORD);
472 //compose the header element
473 record.appendChild(createHeaderElement(doc, oid));
474 //if(!OID_is_deleted) { // ****
475 //compose the metadata element
476 record.appendChild(createMetadataElement(doc, prefix, info));
477 // } // ****
478 get_record.appendChild(record);
479 return get_record_response;
480 }
481
482 /** return a list of records in specified set, containing metadata from specified prefix*/
483 /*
484 protected Element processListRecords(Element req) {
485 return processListIdentifiersOrRecords(req, OAIXML.LIST_RECORDS, true);
486 }
487 */
488 /** return a list of identifiers in specified set that contain metadata belonging to specified prefix. */
489 /*
490 protected Element processListIdentifiers(Element req) {
491 return processListIdentifiersOrRecords(req, OAIXML.LIST_IDENTIFIERS, false);
492 }
493 */
494
495 // Get a list of records/identifiers that match the parameters.
496
497 //protected Element processListIdentifiersOrRecords(Element req, String response_name, boolean include_metadata) {
498 /** arguments:
499 metadataPrefix: required
500 * from: optional
501 * until: optional
502 * set: optional
503 * resumptionToken: exclusive and optional (ignored as it has been handled by OAIReceptionist)
504 * Exceptions: badArgument; cannotDisseminateFormat; idDoesNotExist
505 */
506 /*
507 NodeList params = GSXML.getChildrenByTagName(req, GSXML.PARAM_ELEM);
508
509 if(params.getLength() == 0) {
510 logger.error("must at least have the metadataPrefix parameter, can't be none");
511 return OAIXML.createErrorResponse(OAIXML.BAD_ARGUMENT, "");
512 }
513
514 HashMap<String, String> param_map = GSXML.getParamMap(params);
515
516 String prefix = "";
517 Date from_date = null;
518 Date until_date = null;
519
520 if(param_map.containsKey(OAIXML.METADATA_PREFIX) == false) {
521 //Just a double-check
522 logger.error("A param element containing the metadataPrefix is not present.");
523 return OAIXML.createErrorResponse(OAIXML.CANNOT_DISSEMINATE_FORMAT, "");
524 }
525 prefix = param_map.get(OAIXML.METADATA_PREFIX);
526 if (prefix == null || prefix.equals("")) {
527 //Just a double-check
528 logger.error("the value of metadataPrefix att is not present in the request.");
529 return OAIXML.createErrorResponse(OAIXML.CANNOT_DISSEMINATE_FORMAT, "");
530 }
531
532 if(param_map.containsKey(OAIXML.FROM)) {
533 String from = param_map.get(OAIXML.FROM);
534 from_date = OAIXML.getDate(from);
535 }
536 if(param_map.containsKey(OAIXML.UNTIL)) {
537 String until = param_map.get(OAIXML.UNTIL);
538 until_date = OAIXML.getDate(until);
539 }
540
541 if (!format_response_map.containsKey(prefix)) {
542 logger.error(prefix + " metadata prefix is not supported for collection "+this.coll_name);
543 return OAIXML.createErrorResponse(OAIXML.CANNOT_DISSEMINATE_FORMAT, "");
544 }
545
546 // get list of oids
547 ArrayList<String> oid_list = null;
548 if(oaiinf_db != null) { // try getting the OIDs from the oaiinf_db
549 oid_list = new ArrayList<String>(oaiinf_db.getAllKeys());
550
551 if(oid_list == null) { // try getting the OIDs from the oai entries in the index db
552 logger.warn("@@@@@@@@@@@@@ NO OIDs in oai-inf db for " + this.cluster_name);
553 oid_list = getChildrenIds(OAIXML.BROWSELIST);
554 }
555 }
556
557 if (oid_list == null) {
558 logger.error("No matched records found in collection: oai-inf and index db's browselist are empty");
559 return OAIXML.createErrorResponse(OAIXML.NO_RECORDS_MATCH, "");
560 }
561 // all validation is done
562
563 // get the list of elements that are in this metadata prefix
564 HashSet<String> set_of_elems = format_elements_map.get(prefix);
565
566 Document doc = XMLConverter.newDOM();
567 Element list_items_response = doc.createElement(GSXML.RESPONSE_ELEM);
568 Element list_items = doc.createElement(response_name);
569 list_items_response.appendChild(list_items);
570
571 for(int i=0; i<oid_list.size(); i++) {
572 String oid = oid_list.get(i);
573
574 if(oid.equals(OAIXML.OAI_EARLIEST_TIMESTAMP_OID)) { // internal id not doc id, so skip
575 continue;
576 }
577
578 boolean OID_is_deleted = false;
579 long millis = -1;
580
581 DBInfo oai_info = null;
582 if(oaiinf_db != null) {
583 oai_info = this.oaiinf_db.getInfo(oid);
584 if (oai_info == null) {
585 logger.warn("OID: " + oid + " is not present in the collection's oai-inf database.");
586 } else {
587
588 // indexdb doesn't have info on deleted docs, only oaiinf db does.
589 // So only oaiinfdb has timestamps for deleted docs
590 // For non-deleted doc ids: also obtain timestamp from oaiinf db,
591 // but if the oaiinf db doesn't exist, resort to oailastmodified fields of indexdb.
592 String timestamp = oai_info.getInfo(OAIXML.OAI_INF_TIMESTAMP); // stored in seconds like oailastmodified in the collection index db
593 millis = Long.parseLong(timestamp)*1000; // in milliseconds
594
595 String oaiinf_status = oai_info.getInfo(OAIXML.OAI_INF_STATUS);
596 if(oaiinf_status != null && oaiinf_status.equals(OAIXML.OAI_INF_DELETED)) {
597 OID_is_deleted = true;
598 }
599 }
600 }
601 DBInfo info = this.coll_db.getInfo(oid);
602 if (info == null) { // can happen if oid was deleted, in which case only oai_info keeps a record of the oid
603 if(!OID_is_deleted) { // we don't expect to find entries for deleted docs in index db.
604 logger.error("Collection database does not contain information about oid: " +oid);
605 }
606 }
607 else if (millis == -1) { // so couldn't get doc lastmod from oaiinf db, get oailastmodified from collection's index db
608
609 millis = getDateStampMillis(info);
610 }
611
612 Date this_date = null;
613 String oailastmodified = (millis == -1) ? "" : OAIXML.getTime(millis);
614
615 if (millis == -1) {
616 if (from_date != null || until_date !=null) {
617 continue; // if this doc doesn't have a date for some reason, and
618 // we are doing a date range, then don't include it.
619 }
620 } else {
621 this_date = new Date(millis);
622 if (from_date != null) {
623 if(this_date.before(from_date)) {
624 continue;
625 }
626 }
627 if (until_date != null) {
628 if (this_date.after(until_date)) {
629 continue;
630 }
631 }
632 }
633
634 //compose the header element, which we'll be appending no matter what
635 Element header = createHeaderElement(doc, oid, oailastmodified, OID_is_deleted);
636
637 if (include_metadata) { // doing ListRecords
638 // compose a record for adding header and metadata
639 Element record = doc.createElement(OAIXML.RECORD);
640 list_items.appendChild(record);
641 //insert the header element
642 record.appendChild(header);
643 //Now check that this id has metadata for the required prefix.
644 if (info != null && documentContainsMetadata(info, set_of_elems)) {
645 // YES, it does have some metadata for this prefix
646 //compose the metadata element
647 record.appendChild(createMetadataElement(doc, prefix, info));
648 } // otherwise the oid was 'deleted' and only in the oai-inf db and not in the info (collection index) db
649 } else { // doing ListIdentifiers
650 //append the header element
651 list_items.appendChild(header);
652 }
653
654 }//end of for(int i=0; i<oid_list.size(); i++) of doing thru each record
655
656 return list_items_response;
657
658 }
659*/
660
661 // have implemented setDescription as an element, instead of a container containing metadata
662/*
663 private boolean configureSetInfo() {
664
665 Document doc = XMLConverter.newDOM();
666 this.list_sets_response = doc.createElement(GSXML.RESPONSE_ELEM);
667 Element list_sets_elem = doc.createElement(OAIXML.LIST_SETS);
668 this.list_sets_response.appendChild(list_sets_elem);
669 String set_name = this.coll_name;
670 String set_description = null;
671 Element name_elem = (Element)GSXML.getChildByTagName(this.config_info, OAIXML.SET_NAME);
672 if (name_elem!=null) {
673 set_name = GSXML.getNodeText(name_elem);
674 if (set_name.equals("")) {
675 set_name = this.coll_name; // default to coll name if can't find one
676 }
677 }
678 Element description_elem = (Element)GSXML.getChildByTagName(this.config_info, OAIXML.SET_DESCRIPTION);
679 if (description_elem!=null) {
680 set_description = GSXML.getNodeText(description_elem);
681 if (set_description.equals("")) {
682 set_description = null;
683 }
684 }
685 Element coll_set = OAIXML.createSet(doc, this.coll_name, set_name, set_description);
686 list_sets_elem.appendChild(coll_set);
687
688 // are we part of any super sets?
689 NodeList super_set_list = GSXML.getChildrenByTagName(this.config_info, OAIXML.OAI_SUPER_SET);
690 for (int i=0; i<super_set_list.getLength(); i++) {
691 String super_name = ((Element)super_set_list.item(i)).getAttribute(GSXML.NAME_ATT);
692 if (super_name != null && !super_name.equals("")) {
693 list_sets_elem.appendChild(OAIXML.createSet(doc, super_name, super_name, null));
694 }
695 }
696 return true;
697 }
698*/
699
700 /** create the metadata element used when processing ListRecords/GetRecord requests
701 */
702
703 protected Element createMetadataElement(Document doc, String prefix, DBInfo info) {
704 // the <metadata> element
705 Element metadata = doc.createElement(OAIXML.METADATA);
706 // the <oai:dc namespace...> element
707 Element prfx_str_elem = (Element)doc.importNode(this.format_meta_elem_map.get(prefix), true);
708 metadata.appendChild(prfx_str_elem);
709
710 Element meta_format_element = GSXML.getNamedElement(this.meta_formats_definition, OAIXML.METADATA_FORMAT, GSXML.NAME_ATT, prefix);
711 NodeList elements = meta_format_element.getElementsByTagName(OAIXML.ELEMENT);
712 // for each element in the definition
713 for (int i=0; i<elements.getLength(); i++) {
714 Element e = (Element)elements.item(i);
715 Element map = (Element)GSXML.getChildByTagName(e, OAIXML.MAPPING);
716 if (map == null) {
717 // look up the element name
718 addMetadata(prfx_str_elem, e.getAttribute(GSXML.NAME_ATT), info);
719 } else {
720 // we go though the list of names in the mapping
721 addMetadata(prfx_str_elem, e.getAttribute(GSXML.NAME_ATT), map.getAttribute(OAIXML.SELECT), map.getAttribute(OAIXML.ELEMENTS), info);
722 }
723 }
724 // output any metadata that is not just a simple mapping
725 // addCustomMetadata(prfx_str_elem, prefix, info); // ****
726 return metadata;
727 }
728
729 /** a simple addMetadata where we look for meta_name metadata, and add as that name*/
730
731 protected void addMetadata(Element meta_list_elem, String meta_name, DBInfo info) {
732 Vector<String> values = info.getMultiInfo(meta_name);
733 if (values != null && values.size()!=0) {
734 for (int i=0; i<values.size(); i++) {
735 addMetadataElement(meta_list_elem, meta_name, values.get(i));
736 }
737 }
738 }
739
740 /** more complicated addMetadata - can add multiple items. */
741
742 protected void addMetadata(Element meta_list_elem, String new_meta_name, String select_type, String name_list, DBInfo info) {
743 String[] names = name_list.split(",");
744 for (int i=0; i<names.length; i++) {
745 Vector<String> values;
746 // some special words
747 if (names[i].startsWith(OAIXML.GSF_LINK_PREFIX)) {
748 values = new Vector<String>();
749 String base_url = OAIXML.getBaseURL(); // e.g. e.g. http://host:port/greenstone3/oaiserver
750 String link_url = null;
751 if (names[i].equals(OAIXML.GSF_LINK_PREFIX+OAIXML.LINK_TYPE_DOCUMENT)) {
752 link_url = base_url.replace("oaiserver", "library") + "/collection/" + this.coll_name + "/document/" + info.getInfo("Identifier");
753 } else if (names[i].equals(OAIXML.GSF_LINK_PREFIX+OAIXML.LINK_TYPE_SOURCE)) {
754 String srcfile = info.getInfo("srclinkFile");
755 if (!srcfile.equals("")) {
756 link_url = base_url.replace("oaiserver", "") + "sites/"
757 + this.site_name
758 + "/collect/" + this.coll_name + "/index/assoc/"
759 + info.getInfo("assocfilepath") + "/" + srcfile;
760 }
761 }
762 if (link_url !=null) {
763 values.add(link_url);
764 }
765 } else {
766 values = info.getMultiInfo(names[i]);
767 }
768 if (values == null || values.size()==0) {
769 continue;
770 }
771 for (int j=0; j<values.size(); j++) {
772 addMetadataElement(meta_list_elem, new_meta_name, values.get(j));
773 if (select_type.equals(OAIXML.SELECT_SINGLE_VALUE)) {
774 return; // only want to add one value
775 }
776 }
777 if (select_type.equals(OAIXML.SELECT_FIRST_VALID_META)) {
778 return; // we have added all values of this meta elem
779 }
780 // otherwise, we will keep going through the list and add them all.
781 }
782 }
783
784 // specific metadata formats might need to do some custom metadata that is not
785 //just a standard mapping.
786 protected void addCustomMetadata(Element meta_list_elem, String prefix, DBInfo info) {
787
788
789 }
790
791 /** create the actual metadata element for the list */
792
793 protected void addMetadataElement(Element meta_list_elem, String name, String value) {
794
795 Element meta = GSXML.createTextElement(meta_list_elem.getOwnerDocument(), name, value);
796 meta_list_elem.appendChild(meta);
797 }
798
799
800 /** create a header element used when processing requests like GetRecord
801 */
802 protected Element createHeaderElement(Document doc, String oid) {
803
804 Element header = doc.createElement(OAIXML.HEADER);
805
806 // if deleted, get the date and change oailastmodified to timestamp in oaiinfo
807 /*
808 if(deleted) {
809 header.setAttribute(OAIXML.OAI_INF_STATUS, OAIXML.HEADER_STATUS_ATTR_DELETED); // set the header status to deleted
810 // then the timestamp for deletion will be from oai-inf database
811 }
812 */
813 Element identifier = doc.createElement(IIIFXML.IDENTIFIER);
814 GSXML.setNodeText(identifier, coll_name + ":" + oid);
815 header.appendChild(identifier);
816 /*
817 Element set_spec = doc.createElement(OAIXML.SET_SPEC);
818 GSXML.setNodeText(set_spec, coll_name);
819 header.appendChild(set_spec);
820 Element datestamp = doc.createElement(OAIXML.DATESTAMP);
821 GSXML.setNodeText(datestamp, oailastmodified);
822 header.appendChild(datestamp);
823 */
824 return header;
825 }
826
827
828 /** return the metadata information */
829/*
830 protected Element processListMetadataFormats(Element req) {
831 // the request sent here must contain an OID. see doListMetadataFormats() in OAIReceptionist
832 Element param = GSXML.getNamedElement(req, GSXML.PARAM_ELEM, GSXML.NAME_ATT, OAIXML.OID);
833 if (param == null) {
834 logger.error("An element containing the OID attribute not is present.");
835 return OAIXML.createErrorResponse(OAIXML.ID_DOES_NOT_EXIST, "");
836 }
837 String oid = param.getAttribute(GSXML.VALUE_ATT);
838 if (oid == null || oid.equals("")) {
839 logger.error("No OID is present in the request.");
840 return OAIXML.createErrorResponse(OAIXML.ID_DOES_NOT_EXIST, "");
841 }
842*/
843 /*
844 ArrayList<String> oid_list = null;
845 if(oaiinf_db != null) { // try getting the OIDs from the oaiinf_db
846 oid_list = new ArrayList<String>(oaiinf_db.getAllKeys());
847
848 if(oid_list == null) { // try getting the OIDs from the oai entries in the index db
849 oid_list = getChildrenIds(OAIXML.BROWSELIST);
850 }
851 }
852 */
853/*
854 // assume meta formats are only for OIDs that have not been deleted
855 // so don't need to check oai-inf db, and can just check collection's index db for list of OIDs
856 ArrayList<String> oid_list = getChildrenIds(OAIXML.BROWSELIST);
857 if (oid_list == null || oid_list.contains(oid) == false) {
858 logger.error("OID: " + oid + " is not present in the database.");
859 Element e= OAIXML.createErrorResponse(OAIXML.ID_DOES_NOT_EXIST, "");
860// logger.error((new XMLConverter()).getPrettyString (e));
861 return e;
862 }
863
864 DBInfo info = null;
865 info = this.coll_db.getInfo(oid);
866 if (info == null) { //just double check
867 return OAIXML.createErrorResponse(OAIXML.OAI_SERVICE_UNAVAILABLE, "");
868 }
869
870 Document doc = XMLConverter.newDOM();
871 Element list_metadata_formats_response = doc.createElement(GSXML.RESPONSE_ELEM);
872
873 Element list_metadata_formats = doc.createElement(OAIXML.LIST_METADATA_FORMATS);
874 list_metadata_formats_response.appendChild(list_metadata_formats);
875 boolean has_meta_format = false;
876
877 // for each format in format_elements_map
878 Iterator<String> it = format_elements_map.keySet().iterator();
879 while (it.hasNext()) {
880 String format = it.next();
881 HashSet<String> set_of_elems = format_elements_map.get(format);
882 if (documentContainsMetadata(info, set_of_elems)) {
883 // add this format into the response
884 has_meta_format = true;
885 list_metadata_formats.appendChild(doc.importNode(format_response_map.get(format), true));
886 }
887 }
888
889 if (has_meta_format == false) {
890 logger.error("Specified metadata names are not contained in the database.");
891 return OAIXML.createErrorResponse(OAIXML.NO_METADATA_FORMATS, "");
892 } else {
893 return list_metadata_formats_response;
894 }
895 }
896*/
897/*
898 protected boolean documentContainsMetadata(DBInfo info, HashSet<String> set_of_elems) {
899 if (set_of_elems.size() == 0) {
900 return false;
901 }
902 Iterator<String> i = set_of_elems.iterator();
903 while (i.hasNext()) {
904 if (!info.getInfo(i.next()).equals("")) {
905 return true;
906 }
907 }
908 return false;
909 }
910*/
911
912 /** returns a list of the child ids in order, null if no children */
913/*
914 protected ArrayList<String> getChildrenIds(String node_id) {
915 DBInfo info = this.coll_db.getInfo(node_id);
916 if (info == null) {
917 return null;
918 }
919
920 String contains = info.getInfo("contains");
921 if (contains.equals("")) {
922 return null;
923 }
924 ArrayList<String> children = new ArrayList<String>();
925 StringTokenizer st = new StringTokenizer(contains, ";");
926 while (st.hasMoreTokens()) {
927 String child_id = st.nextToken().replaceAll("\"", node_id);
928 children.add(child_id);
929 }
930 return children;
931 }
932*/
933
934 /**method to check whether any of the 'metadata_names' is contained in the 'info'.
935 * The name may be in the form: <name>,<mapped name>, in which the mapped name is
936 * optional. The mapped name is looked up in the DBInfo; if not present, use the first
937 * name which is mandatory.
938 */
939/*
940 protected boolean containsMetadata(DBInfo info, String[] metadata_names) {
941 if (metadata_names == null) return false;
942 logger.info("checking metadata names in db.");
943 for(int i=0; i<metadata_names.length; i++) {
944 int index = metadata_names[i].indexOf(",");
945 String meta_name = (index == -1) ? metadata_names[i] :
946 metadata_names[i].substring(index + 1);
947
948 if(info.getInfo(meta_name).equals("") == false) {
949 return true;
950 }
951 }
952 return false;
953 }
954*/
955/*
956 protected long getDateStampMillis(DBInfo info) {
957 // gs.OAIDateStamp is in YYYY-MM-DD
958 String time_stamp = info.getInfo(OAIXML.GS_OAI_DATE_STAMP);
959 long millis = -1;
960 if (!time_stamp.equals("")) {
961 millis = OAIXML.getTime(time_stamp);
962 }
963 if (millis == -1) {
964 // oailastmodified is in seconds
965 time_stamp = info.getInfo(OAIXML.OAI_LASTMODIFIED);
966 if (!time_stamp.equals("")) {
967 millis = Long.parseLong(time_stamp)*1000;
968 }
969 }
970 return millis;
971
972
973 }
974*/
975}
976
977
Note: See TracBrowser for help on using the repository browser.