1 | /*
|
---|
2 | * OAIReceptionist.java
|
---|
3 | * Copyright (C) 2012 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 | */
|
---|
19 |
|
---|
20 | package org.greenstone.gsdl3.core;
|
---|
21 |
|
---|
22 | import org.greenstone.gsdl3.util.*;
|
---|
23 | import org.greenstone.gsdl3.action.*;
|
---|
24 | // XML classes
|
---|
25 | import org.w3c.dom.Node;
|
---|
26 | import org.w3c.dom.NodeList;
|
---|
27 | import org.w3c.dom.Document;
|
---|
28 | import org.w3c.dom.Element;
|
---|
29 |
|
---|
30 | // other java classes
|
---|
31 | import java.io.File;
|
---|
32 | import java.util.*;
|
---|
33 |
|
---|
34 | import org.apache.log4j.*;
|
---|
35 |
|
---|
36 | /** a Receptionist, used for oai metadata response xml generation.
|
---|
37 | * This receptionist talks to the message router directly,
|
---|
38 | * instead of via any action, hence no action map is needed.
|
---|
39 | * @see the basic Receptionist
|
---|
40 | */
|
---|
41 | public class OAIReceptionist implements ModuleInterface {
|
---|
42 |
|
---|
43 | static Logger logger = Logger.getLogger(org.greenstone.gsdl3.core.OAIReceptionist.class.getName());
|
---|
44 |
|
---|
45 | /** Instead of a config_params object, only a site_name is needed by oai receptionist. */
|
---|
46 | protected String site_name = null;
|
---|
47 | /** The unique repository identifier */
|
---|
48 | protected String repository_id = null;
|
---|
49 |
|
---|
50 | /** the configure file of this receptionist passed from the oai servlet. */
|
---|
51 | protected Element oai_config = null;
|
---|
52 |
|
---|
53 | /** contained in the OAIConfig.xml deciding whether the resumptionToken should be in use */
|
---|
54 | protected int resume_after = -1 ;
|
---|
55 |
|
---|
56 | /** the message router that the Receptionist and Actions will talk to */
|
---|
57 | protected ModuleInterface mr = null;
|
---|
58 |
|
---|
59 | // Some of the data/responses will not change while the servlet is running, so
|
---|
60 | // we can cache them
|
---|
61 |
|
---|
62 | /** A list of all the collections available to this OAI server */
|
---|
63 | protected Element collection_list = null;
|
---|
64 | /** a vector of the names, for convenience */
|
---|
65 | protected Vector<String> collection_name_list = null;
|
---|
66 | /** a vector to be maintained in parallel to Vector collection_name_list in terms of adds/removes.
|
---|
67 | * This is a list of Integers that denote whether the associated collection in collection_name_list is currently ACTIVE or DEACTIVATED.
|
---|
68 | * The state for any collection can be changed when the building folder is moved to index during rebuilding, before and after which
|
---|
69 | * activateOrDeactivateCollection() here is called to deactivate a collection and then activate it again, respectively.
|
---|
70 | * Can't make collection_name_list a HashMap, since a map's keys are unique (i.e. a Set) whereas collection_name_list is a Vector.
|
---|
71 | * I've confirmed that collection_name_list is specifically a list of OAI collection names, not non-OAI collections.
|
---|
72 | */
|
---|
73 | protected Vector<Integer> collection_state_list = null;
|
---|
74 | /** Possible states for a collection.
|
---|
75 | * NOTE: Being done with Integer instead of int because they're put into Vectors and they can only contain Objects not primitive data types.
|
---|
76 | */
|
---|
77 | protected static final Integer DEACTIVATED = new Integer(0);
|
---|
78 | protected static final Integer ACTIVE = new Integer(1);
|
---|
79 |
|
---|
80 | /** If this is true, then there are no OAI enabled collections, so can always return noRecordsMatch (after validating the request params) */
|
---|
81 | protected boolean noRecordsMatch = false;
|
---|
82 |
|
---|
83 | /** A set of all known 'sets'. */
|
---|
84 | protected HashSet<String> set_oaiset = null;
|
---|
85 |
|
---|
86 | protected boolean has_super_colls = false;
|
---|
87 | /** a hash of super set-> collection list */
|
---|
88 | protected HashMap<String, Vector<String>> super_coll_map = null;
|
---|
89 | /** store the super coll elements for convenience */
|
---|
90 | HashMap<String, Element> super_coll_data = null;
|
---|
91 | /** store the metadata formats ??????*/
|
---|
92 | /** The identify response */
|
---|
93 | protected Element identify_response = null;
|
---|
94 | /** The list set response */
|
---|
95 | protected Element listsets_response = null;
|
---|
96 | /** the list metadata formats response */
|
---|
97 | protected Element listmetadataformats_response = null;
|
---|
98 |
|
---|
99 | public OAIReceptionist() {
|
---|
100 |
|
---|
101 | }
|
---|
102 |
|
---|
103 | public void cleanUp() {
|
---|
104 | if (this.mr != null) {
|
---|
105 |
|
---|
106 | this.mr.cleanUp();
|
---|
107 | }
|
---|
108 | OAIResumptionToken.saveTokensToFile();
|
---|
109 | }
|
---|
110 |
|
---|
111 | public void setSiteName(String site_name) {
|
---|
112 | this.site_name = site_name;
|
---|
113 | }
|
---|
114 | /** sets the message router - it should already be created and
|
---|
115 | * configured in the init() of a servlet (OAIServer, for example) before being passed to the receptionist*/
|
---|
116 | public void setMessageRouter(ModuleInterface mr) {
|
---|
117 | this.mr = mr;
|
---|
118 | }
|
---|
119 |
|
---|
120 | /** configures the receptionist */
|
---|
121 | public boolean configure(Element config) {
|
---|
122 |
|
---|
123 | if (this.mr==null) {
|
---|
124 | logger.error(" message routers must be set before calling oai configure");
|
---|
125 | return false;
|
---|
126 | }
|
---|
127 | if (config == null) {
|
---|
128 | logger.error(" oai configure file is null");
|
---|
129 | return false;
|
---|
130 | }
|
---|
131 | oai_config = config;
|
---|
132 | resume_after = getResumeAfter();
|
---|
133 |
|
---|
134 | repository_id = getRepositoryIdentifier();
|
---|
135 | configureSuperSetInfo();
|
---|
136 | if (!configureSetInfo()) {
|
---|
137 | // there are no sets
|
---|
138 | logger.error("No sets (collections) available for OAI");
|
---|
139 | return false;
|
---|
140 | }
|
---|
141 |
|
---|
142 | // load in tokens from OAIResumptionToken.xml, and then clear out any
|
---|
143 | // expired ones.
|
---|
144 | OAIResumptionToken.init();
|
---|
145 | OAIResumptionToken.clearExpiredTokens();
|
---|
146 |
|
---|
147 | return true;
|
---|
148 | }
|
---|
149 |
|
---|
150 | // assuming that sets are static. If collections change then the servlet
|
---|
151 | // should be restarted.
|
---|
152 | private boolean configureSuperSetInfo() {
|
---|
153 | // do we have any super colls listed in web/WEB-INF/classes/OAIConfig.xml?
|
---|
154 | // Will be like
|
---|
155 | // <oaiSuperSet>
|
---|
156 | // <SetSpec>xxx</SetSpec>
|
---|
157 | // <setName>xxx</SetName>
|
---|
158 | // <SetDescription>xxx</setDescription>
|
---|
159 | // </oaiSuperSet>
|
---|
160 | // The super set is listed in OAIConfig, and collections themselves state
|
---|
161 | // whether they are part of the super set or not.
|
---|
162 | NodeList super_coll_list = this.oai_config.getElementsByTagName(OAIXML.OAI_SUPER_SET);
|
---|
163 | this.super_coll_data = new HashMap<String, Element>();
|
---|
164 | if (super_coll_list.getLength() > 0) {
|
---|
165 | this.has_super_colls = true;
|
---|
166 | for (int i=0; i<super_coll_list.getLength(); i++) {
|
---|
167 | Element super_coll = (Element)super_coll_list.item(i);
|
---|
168 | Element set_spec = (Element)GSXML.getChildByTagName(super_coll, OAIXML.SET_SPEC);
|
---|
169 | if (set_spec != null) {
|
---|
170 | String name = GSXML.getNodeText(set_spec);
|
---|
171 | if (!name.equals("")) {
|
---|
172 | this.super_coll_data.put(name, super_coll);
|
---|
173 | logger.info("adding in super coll "+name);
|
---|
174 | }
|
---|
175 | }
|
---|
176 | }
|
---|
177 |
|
---|
178 | if (this.super_coll_data.size()==0) {
|
---|
179 | this.has_super_colls = false;
|
---|
180 | }
|
---|
181 | }
|
---|
182 | if (this.has_super_colls == true) {
|
---|
183 | this.super_coll_map = new HashMap<String, Vector<String>>();
|
---|
184 | }
|
---|
185 | return true;
|
---|
186 |
|
---|
187 | }
|
---|
188 | private boolean configureSetInfo() {
|
---|
189 | this.set_oaiset = new HashSet<String>();
|
---|
190 |
|
---|
191 | // First, we get a list of all the OAI enabled collections
|
---|
192 | // We get this by sending a listSets request to the MR
|
---|
193 | Document doc = XMLConverter.newDOM();
|
---|
194 | Element message = doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
195 |
|
---|
196 | Element request = GSXML.createBasicRequest(doc, OAIXML.OAI_SET_LIST, "", null);
|
---|
197 | message.appendChild(request);
|
---|
198 | Node msg_node = mr.process(message);
|
---|
199 |
|
---|
200 | if (msg_node == null) {
|
---|
201 | logger.error("returned msg_node from mr is null");
|
---|
202 | return false;
|
---|
203 | }
|
---|
204 | Element resp = (Element)GSXML.getChildByTagName(msg_node, GSXML.RESPONSE_ELEM);
|
---|
205 | Element coll_list = (Element)GSXML.getChildByTagName(resp, GSXML.COLLECTION_ELEM + GSXML.LIST_MODIFIER);
|
---|
206 | if (coll_list == null) {
|
---|
207 | logger.error("coll_list is null");
|
---|
208 | return false;
|
---|
209 | }
|
---|
210 |
|
---|
211 | this.collection_list = (Element)doc.importNode(coll_list, true);
|
---|
212 |
|
---|
213 | // go through and store a list of collection names for convenience
|
---|
214 | // also create a 'to' attribute for the next request to the MR, which
|
---|
215 | // is a ListSets request to each collection
|
---|
216 | Node child = this.collection_list.getFirstChild();
|
---|
217 | if (child == null) {
|
---|
218 | logger.error("collection list has no children");
|
---|
219 | noRecordsMatch = true;
|
---|
220 | return false;
|
---|
221 | }
|
---|
222 |
|
---|
223 | this.collection_name_list = new Vector<String>();
|
---|
224 | this.collection_state_list = new Vector<Integer>();
|
---|
225 | StringBuffer to = new StringBuffer();
|
---|
226 | boolean first = true;
|
---|
227 | while (child != null) {
|
---|
228 | if (child.getNodeName().equals(GSXML.COLLECTION_ELEM)) {
|
---|
229 | String coll_id =((Element) child).getAttribute(GSXML.NAME_ATT);
|
---|
230 | this.collection_name_list.add(coll_id);
|
---|
231 | this.collection_state_list.add(ACTIVE); // collections start out active (they get activated during configuring) until deactivation takes place
|
---|
232 | if (!first) {
|
---|
233 | to.append(',');
|
---|
234 | }
|
---|
235 | first = false;
|
---|
236 | to.append(coll_id+"/"+OAIXML.LIST_SETS);
|
---|
237 | }
|
---|
238 | child = child.getNextSibling();
|
---|
239 | }
|
---|
240 | if (first) {
|
---|
241 | // we haven't found any collections
|
---|
242 | logger.error("found no collection elements in collectionList");
|
---|
243 | noRecordsMatch = true;
|
---|
244 | return false;
|
---|
245 | }
|
---|
246 | Document listsets_doc = XMLConverter.newDOM();
|
---|
247 | Element listsets_element = listsets_doc.createElement(OAIXML.LIST_SETS);
|
---|
248 | this.listsets_response = getMessage(listsets_doc, listsets_element);
|
---|
249 |
|
---|
250 | // Now, for each collection, get a list of all its sets
|
---|
251 | // might include subsets (classifiers) or super colls
|
---|
252 | // We'll reuse the first message, changing its type and to atts
|
---|
253 | request.setAttribute(GSXML.TYPE_ATT, "");
|
---|
254 | request.setAttribute(GSXML.TO_ATT, to.toString());
|
---|
255 | // send to MR
|
---|
256 | msg_node = mr.process(message);
|
---|
257 | //logger.info("*** " + XMLConverter.getPrettyString(msg_node));
|
---|
258 | NodeList response_list = ((Element)msg_node).getElementsByTagName(GSXML.RESPONSE_ELEM);
|
---|
259 | for (int c=0; c<response_list.getLength(); c++) {
|
---|
260 | // for each collection's response
|
---|
261 | Element response = (Element)response_list.item(c);
|
---|
262 | String coll_name = GSPath.getFirstLink(response.getAttribute(GSXML.FROM_ATT));
|
---|
263 | logger.info("*** coll from response "+coll_name);
|
---|
264 | NodeList set_list = response.getElementsByTagName(OAIXML.SET);
|
---|
265 | for (int j=0; j<set_list.getLength(); j++) {
|
---|
266 | // now check if it a super collection
|
---|
267 | Element set = (Element)set_list.item(j);
|
---|
268 | String set_spec = GSXML.getNodeText((Element)GSXML.getChildByTagName(set, OAIXML.SET_SPEC));
|
---|
269 | logger.info("*** set spec = "+set_spec);
|
---|
270 | // this may change if we add site name back in
|
---|
271 | // setSpecs will be collname or collname:subset or supercollname
|
---|
272 | if (set_spec.indexOf(":")==-1 && ! set_spec.equals(coll_name)) {
|
---|
273 | // it must be a super coll spec
|
---|
274 | logger.info("*** found super coll, "+set_spec);
|
---|
275 | // check that it is a valid one from config
|
---|
276 | if (this.has_super_colls == true && this.super_coll_data.containsKey(set_spec)) {
|
---|
277 | Vector <String> subcolls = this.super_coll_map.get(set_spec);
|
---|
278 | if (subcolls == null) {
|
---|
279 | logger.info("*** it's new!!");
|
---|
280 | // not in there yet
|
---|
281 | subcolls = new Vector<String>();
|
---|
282 | this.set_oaiset.add(set_spec);
|
---|
283 | this.super_coll_map.put(set_spec, subcolls);
|
---|
284 | // the first time a supercoll is mentioned, add into the set list
|
---|
285 | logger.info("*** finding the set info "+XMLConverter.getPrettyString(this.super_coll_data.get(set_spec)));
|
---|
286 | listsets_element.appendChild(GSXML.duplicateWithNewName(listsets_doc, this.super_coll_data.get(set_spec), OAIXML.SET, true));
|
---|
287 | }
|
---|
288 | // add this collection to the list for the super coll
|
---|
289 | subcolls.add(coll_name);
|
---|
290 | }
|
---|
291 | } else { // its either the coll itself or a subcoll
|
---|
292 | // add in the set
|
---|
293 | listsets_element.appendChild(listsets_doc.importNode(set, true));
|
---|
294 | this.set_oaiset.add(set_spec);
|
---|
295 | }
|
---|
296 | } // for each set in the collection
|
---|
297 | } // for each OAI enabled collection
|
---|
298 |
|
---|
299 | return true;
|
---|
300 | }
|
---|
301 |
|
---|
302 | protected void resetMessageRouter() {
|
---|
303 | // we just need to send a configure request to MR
|
---|
304 | Document doc = XMLConverter.newDOM();
|
---|
305 | Element mr_request_message = doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
306 | Element mr_request = GSXML.createBasicRequest(doc, GSXML.REQUEST_TYPE_SYSTEM, "", null);
|
---|
307 | mr_request_message.appendChild(mr_request);
|
---|
308 |
|
---|
309 | Element system = doc.createElement(GSXML.SYSTEM_ELEM);
|
---|
310 | mr_request.appendChild(system);
|
---|
311 | system.setAttribute(GSXML.TYPE_ATT, GSXML.SYSTEM_TYPE_CONFIGURE);
|
---|
312 |
|
---|
313 | Element response = (Element) this.mr.process(mr_request_message);
|
---|
314 | logger.info("*** configure response = "+XMLConverter.getPrettyString(response));
|
---|
315 | }
|
---|
316 |
|
---|
317 | protected boolean activateOrDeactivateCollection(String collName, int activationState) {
|
---|
318 | // Send a request like: a=s&sa=<a|d>&st=collection&sn=<collName>
|
---|
319 | Document doc = XMLConverter.newDOM();
|
---|
320 | Element mr_request_message = doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
321 | Element mr_request = GSXML.createBasicRequest(doc, GSXML.REQUEST_TYPE_SYSTEM, "", null);
|
---|
322 | mr_request_message.appendChild(mr_request);
|
---|
323 |
|
---|
324 | Element system = doc.createElement(GSXML.SYSTEM_ELEM);
|
---|
325 | mr_request.appendChild(system);
|
---|
326 | if(activationState == OAIXML.ACTIVATION) {
|
---|
327 | system.setAttribute(GSXML.TYPE_ATT, GSXML.SYSTEM_TYPE_ACTIVATE);
|
---|
328 | } else {
|
---|
329 | system.setAttribute(GSXML.TYPE_ATT, GSXML.SYSTEM_TYPE_DEACTIVATE);
|
---|
330 | }
|
---|
331 | system.setAttribute(GSXML.SYSTEM_MODULE_TYPE_ATT, GSXML.COLLECTION_ELEM);
|
---|
332 | system.setAttribute(GSXML.SYSTEM_MODULE_NAME_ATT, collName);
|
---|
333 |
|
---|
334 | Element response = (Element) this.mr.process(mr_request_message);
|
---|
335 | logger.info("*** (de)activate response = "+XMLConverter.getPrettyString(response));
|
---|
336 |
|
---|
337 | boolean success = false;
|
---|
338 | NodeList elements = response.getElementsByTagName(GSXML.STATUS_ELEM);
|
---|
339 | if(elements.getLength() <= 0) {
|
---|
340 | logger.error("***** No result status");
|
---|
341 | return false;
|
---|
342 | }
|
---|
343 |
|
---|
344 | String result = GSXML.getNodeText((Element)elements.item(0));
|
---|
345 | if(result.contains("could not be")) { // could not be (de)activated
|
---|
346 | return false;
|
---|
347 | } else {
|
---|
348 | // the collection's state has successfully been set to the requested activationState (be it ACTIVE or DEACTIVATED)
|
---|
349 | Integer changedState = (activationState == OAIXML.ACTIVATION) ? ACTIVE : DEACTIVATED;
|
---|
350 | int index = this.collection_name_list.indexOf(collName); // TODO: Check that whatever collname passed in by servercontrol.pm is fully qualified, to account for super/sub collections.
|
---|
351 | if(index != -1) { // shouldn't be? since startup never calls this method. This method is only called when building an existing collection?
|
---|
352 | this.collection_state_list.set(index, changedState);
|
---|
353 | } else {
|
---|
354 | logger.error("@@@@ index == -1, could not find collection " + collName + " in collection_name_list");
|
---|
355 | }
|
---|
356 | return true;
|
---|
357 | }
|
---|
358 | }
|
---|
359 |
|
---|
360 | /** process using strings - just calls process using Elements */
|
---|
361 | public String process(String xml_in) {
|
---|
362 |
|
---|
363 | Node message_node = XMLConverter.getDOM(xml_in);
|
---|
364 | Node page = process(message_node);
|
---|
365 | return XMLConverter.getString(page);
|
---|
366 | }
|
---|
367 |
|
---|
368 | //Compose a message/response element used to send back to the OAIServer servlet.
|
---|
369 | //This method is only used within OAIReceptionist
|
---|
370 | private Element getMessage(Document doc, Element e) {
|
---|
371 | Element msg = doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
372 | Element response = doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
373 | msg.appendChild(response);
|
---|
374 | response.appendChild(e);
|
---|
375 | return msg;
|
---|
376 | }
|
---|
377 |
|
---|
378 | /** process - produce xml data in response to a request
|
---|
379 | * if something goes wrong, it returns null -
|
---|
380 | */
|
---|
381 | public Node process(Node message_node) {
|
---|
382 | logger.info("*** OAIReceptionist received request");
|
---|
383 |
|
---|
384 | Element message = GSXML.nodeToElement(message_node);
|
---|
385 | logger.info("*** " + XMLConverter.getString(message));
|
---|
386 |
|
---|
387 | // check that its a correct message tag
|
---|
388 | if (!message.getTagName().equals(GSXML.MESSAGE_ELEM)) {
|
---|
389 | logger.error(" Invalid message. GSDL message should start with <"+GSXML.MESSAGE_ELEM+">, instead it starts with:"+message.getTagName()+".");
|
---|
390 | return OAIXML.createErrorMessage(OAIXML.BAD_ARGUMENT, "Internal messaging error");
|
---|
391 | }
|
---|
392 |
|
---|
393 | // get the request out of the message - assume that there is only one
|
---|
394 | Element request = (Element)GSXML.getChildByTagName(message, GSXML.REQUEST_ELEM);
|
---|
395 | if (request == null) {
|
---|
396 | logger.error(" message had no request!");
|
---|
397 | return OAIXML.createErrorMessage(OAIXML.BAD_ARGUMENT, "Internal messaging error");
|
---|
398 | }
|
---|
399 |
|
---|
400 | // Special cases: certain non-OAI commands/non-verbs are recognised
|
---|
401 | // special case, reset=true for reloading the MR and recept data
|
---|
402 | String reset = request.getAttribute("reset");
|
---|
403 | if (!reset.equals("")) {
|
---|
404 | resetMessageRouter();
|
---|
405 | configureSetInfo();
|
---|
406 | return OAIXML.createResetResponse(true);
|
---|
407 | }
|
---|
408 |
|
---|
409 | // special case 2: activate=<collname> or deactivate=<collname> can be passed to oaiserver servlet too
|
---|
410 | if (request.hasAttribute(GSXML.SYSTEM_TYPE_ACTIVATE)) {
|
---|
411 | String collname = request.getAttribute(GSXML.SYSTEM_TYPE_ACTIVATE);
|
---|
412 | // don't bother activating if it's not an OAI collection
|
---|
413 | if (!this.collection_name_list.contains(collname)) {
|
---|
414 | return OAIXML.createDeActivationOfNonOAICollResponse(OAIXML.ACTIVATION, collname);
|
---|
415 | }
|
---|
416 | boolean success = activateOrDeactivateCollection(collname, OAIXML.ACTIVATION);
|
---|
417 | return OAIXML.createActivationStateResponse(success, OAIXML.ACTIVATION, collname);
|
---|
418 | } else if (request.hasAttribute(GSXML.SYSTEM_TYPE_DEACTIVATE)) {
|
---|
419 | String collname = request.getAttribute(GSXML.SYSTEM_TYPE_DEACTIVATE);
|
---|
420 | // don't bother deactivating if it's not an OAI collection
|
---|
421 | if (!this.collection_name_list.contains(collname)) {
|
---|
422 | return OAIXML.createDeActivationOfNonOAICollResponse(OAIXML.DEACTIVATION, collname);
|
---|
423 | }
|
---|
424 | boolean success = activateOrDeactivateCollection(collname, OAIXML.DEACTIVATION);
|
---|
425 | return OAIXML.createActivationStateResponse(success, OAIXML.DEACTIVATION, collname);
|
---|
426 | }
|
---|
427 |
|
---|
428 | //At this stage, the value of 'to' attribute of the request must be the 'verb'
|
---|
429 | //The only thing that the oai receptionist can be sure is that these verbs are valid, nothing else.
|
---|
430 | String verb = request.getAttribute(GSXML.TO_ATT);
|
---|
431 | if (verb.equals(OAIXML.IDENTIFY)) {
|
---|
432 | return doIdentify();
|
---|
433 | }
|
---|
434 | if (verb.equals(OAIXML.LIST_METADATA_FORMATS)) {
|
---|
435 | return doListMetadataFormats(request);
|
---|
436 | }
|
---|
437 | if (verb.equals(OAIXML.LIST_SETS)) {
|
---|
438 | // we have composed the list sets response on init
|
---|
439 | // Note this means that list sets never uses resumption tokens
|
---|
440 | return this.listsets_response;
|
---|
441 | }
|
---|
442 | if (verb.equals(OAIXML.GET_RECORD)) {
|
---|
443 | return doGetRecord(request);
|
---|
444 | }
|
---|
445 | if (verb.equals(OAIXML.LIST_IDENTIFIERS)) {
|
---|
446 | return doListIdentifiersOrRecords(request,OAIXML.LIST_IDENTIFIERS , OAIXML.HEADER);
|
---|
447 | }
|
---|
448 | if (verb.equals(OAIXML.LIST_RECORDS)) {
|
---|
449 | return doListIdentifiersOrRecords(request, OAIXML.LIST_RECORDS, OAIXML.RECORD);
|
---|
450 | }
|
---|
451 | // should never get here as verbs were checked in OAIServer
|
---|
452 | return OAIXML.createErrorMessage(OAIXML.BAD_VERB, "Unexpected things happened");
|
---|
453 |
|
---|
454 | }
|
---|
455 |
|
---|
456 |
|
---|
457 | private int getResumeAfter() {
|
---|
458 | Element resume_after = (Element)GSXML.getChildByTagName(oai_config, OAIXML.RESUME_AFTER);
|
---|
459 | if(resume_after != null) return Integer.parseInt(GSXML.getNodeText(resume_after));
|
---|
460 | return -1;
|
---|
461 | }
|
---|
462 | private String getRepositoryIdentifier() {
|
---|
463 | Element ri = (Element)GSXML.getChildByTagName(oai_config, OAIXML.REPOSITORY_IDENTIFIER);
|
---|
464 | if (ri != null) {
|
---|
465 | return GSXML.getNodeText(ri);
|
---|
466 | }
|
---|
467 | return "";
|
---|
468 | }
|
---|
469 |
|
---|
470 |
|
---|
471 | /** if the param_map contains strings other than those in valid_strs, return false;
|
---|
472 | * otherwise true.
|
---|
473 | */
|
---|
474 | private boolean areAllParamsValid(HashMap<String, String> param_map, HashSet<String> valid_strs) {
|
---|
475 | ArrayList<String> param_list = new ArrayList<String>(param_map.keySet());
|
---|
476 | for(int i=0; i<param_list.size(); i++) {
|
---|
477 | logger.info("*** param, key = "+param_list.get(i)+", value = "+param_map.get(param_list.get(i)));
|
---|
478 | if (valid_strs.contains(param_list.get(i)) == false) {
|
---|
479 | return false;
|
---|
480 | }
|
---|
481 | }
|
---|
482 | return true;
|
---|
483 | }
|
---|
484 |
|
---|
485 | private Element doListIdentifiersOrRecords(Element req, String verb, String record_type) {
|
---|
486 | // options: from, until, set, metadataPrefix, resumptionToken
|
---|
487 | // exceptions: badArgument, badResumptionToken, cannotDisseminateFormat, noRecordMatch, and noSetHierarchy
|
---|
488 | HashSet<String> valid_strs = new HashSet<String>();
|
---|
489 | valid_strs.add(OAIXML.FROM);
|
---|
490 | valid_strs.add(OAIXML.UNTIL);
|
---|
491 | valid_strs.add(OAIXML.SET);
|
---|
492 | valid_strs.add(OAIXML.METADATA_PREFIX);
|
---|
493 | valid_strs.add(OAIXML.RESUMPTION_TOKEN);
|
---|
494 |
|
---|
495 | Document result_doc = XMLConverter.newDOM();
|
---|
496 | Element result_element = result_doc.createElement(verb);
|
---|
497 | boolean result_token_needed = false; // does this result need to include a
|
---|
498 | // resumption token
|
---|
499 |
|
---|
500 | NodeList params = GSXML.getChildrenByTagName(req, GSXML.PARAM_ELEM);
|
---|
501 |
|
---|
502 | HashMap<String, String> param_map = GSXML.getParamMap(params);
|
---|
503 |
|
---|
504 | // are all the params valid?
|
---|
505 | if (!areAllParamsValid(param_map, valid_strs)) {
|
---|
506 | logger.error("One of the params is invalid");
|
---|
507 | return OAIXML.createErrorMessage(OAIXML.BAD_ARGUMENT, "There was an invalid parameter");
|
---|
508 | // TODO, need to tell the user which one was invalid ??
|
---|
509 | }
|
---|
510 |
|
---|
511 | // Do we have a resumption token??
|
---|
512 | String token = null;
|
---|
513 | String from = null;
|
---|
514 | String until = null;
|
---|
515 | boolean set_requested = false;
|
---|
516 | String set_spec_str = null;
|
---|
517 | String prefix_value = null;
|
---|
518 | int cursor = 0;
|
---|
519 | int current_cursor = 0;
|
---|
520 | String current_set = null;
|
---|
521 | long initial_time = 0;
|
---|
522 |
|
---|
523 | int total_size = -1; // we are only going to set this in resumption
|
---|
524 | // token if it is easy to work out, i.e. not sending extra requests to
|
---|
525 | // MR just to calculate total size
|
---|
526 |
|
---|
527 | if(param_map.containsKey(OAIXML.RESUMPTION_TOKEN)) {
|
---|
528 | // Is it an error to have other arguments? Do we need to check to make sure that resumptionToken is the only arg??
|
---|
529 | // validate resumptionToken
|
---|
530 | token = param_map.get(OAIXML.RESUMPTION_TOKEN);
|
---|
531 | logger.info("has resumptionToken " + token);
|
---|
532 | if(OAIResumptionToken.isValidToken(token) == false) {
|
---|
533 | logger.error("token is not valid");
|
---|
534 | return OAIXML.createErrorMessage(OAIXML.BAD_RESUMPTION_TOKEN, "");
|
---|
535 | }
|
---|
536 | result_token_needed = true; // we always need to send a token back if we have started with one. It may be empty if we are returning the end of the list
|
---|
537 | // initialise the request params from the stored token data
|
---|
538 | HashMap<String, String> token_data = OAIResumptionToken.getTokenData(token);
|
---|
539 | from = token_data.get(OAIXML.FROM);
|
---|
540 | until = token_data.get(OAIXML.UNTIL);
|
---|
541 | set_spec_str = token_data.get(OAIXML.SET);
|
---|
542 | if (set_spec_str != null) {
|
---|
543 | set_requested = true;
|
---|
544 | }
|
---|
545 | prefix_value = token_data.get(OAIXML.METADATA_PREFIX);
|
---|
546 | current_set = token_data.get(OAIResumptionToken.CURRENT_SET);
|
---|
547 | try {
|
---|
548 | cursor = Integer.parseInt(token_data.get(OAIXML.CURSOR));
|
---|
549 | cursor = cursor + resume_after; // increment cursor
|
---|
550 | current_cursor = Integer.parseInt(token_data.get(OAIResumptionToken.CURRENT_CURSOR));
|
---|
551 | initial_time = Long.parseLong(token_data.get(OAIResumptionToken.INITIAL_TIME));
|
---|
552 | } catch (NumberFormatException e) {
|
---|
553 | logger.error("tried to parse int from cursor data and failed");
|
---|
554 | }
|
---|
555 |
|
---|
556 | // check that the collections/sets haven't changed since the token was issued
|
---|
557 | if (collectionsChangedSinceTime(set_spec_str, initial_time)) {
|
---|
558 | logger.error("one of the collections in set "+set_spec_str+" has changed since token issued. Expiring the token");
|
---|
559 | OAIResumptionToken.expireToken(token);
|
---|
560 | return OAIXML.createErrorMessage(OAIXML.BAD_RESUMPTION_TOKEN, "Repository data has changed since this token was issued. Resend original request");
|
---|
561 | }
|
---|
562 | }
|
---|
563 | else {
|
---|
564 | // no resumption token, lets check the other params
|
---|
565 | // there must be a metadataPrefix
|
---|
566 | if (!param_map.containsKey(OAIXML.METADATA_PREFIX)) {
|
---|
567 | logger.error("metadataPrefix param required");
|
---|
568 | return OAIXML.createErrorMessage(OAIXML.BAD_ARGUMENT, "metadataPrefix param required");
|
---|
569 | }
|
---|
570 |
|
---|
571 | //if there are any date params, check they're of the right format
|
---|
572 | Date from_date = null;
|
---|
573 | Date until_date = null;
|
---|
574 |
|
---|
575 | from = param_map.get(OAIXML.FROM);
|
---|
576 | if(from != null) {
|
---|
577 | from_date = OAIXML.getDate(from);
|
---|
578 | if(from_date == null) {
|
---|
579 | logger.error("invalid date: " + from);
|
---|
580 | return OAIXML.createErrorMessage(OAIXML.BAD_ARGUMENT, "invalid format for "+ OAIXML.FROM);
|
---|
581 | }
|
---|
582 | }
|
---|
583 | until = param_map.get(OAIXML.UNTIL);
|
---|
584 | if(until != null) {
|
---|
585 | until_date = OAIXML.getDate(until);
|
---|
586 | if(until_date == null) {
|
---|
587 | logger.error("invalid date: " + until);
|
---|
588 | return OAIXML.createErrorMessage(OAIXML.BAD_ARGUMENT, "invalid format for "+ OAIXML.UNTIL);
|
---|
589 | }
|
---|
590 | }
|
---|
591 |
|
---|
592 | if(from != null && until != null) { // check they are of the same date-time format (granularity)
|
---|
593 | if(from.length() != until.length()) {
|
---|
594 | logger.error("The request has different granularities (date-time formats) for the From and Until date parameters.");
|
---|
595 | return OAIXML.createErrorMessage(OAIXML.BAD_ARGUMENT, "The request has different granularities (date-time formats) for the From and Until date parameters.");
|
---|
596 | }
|
---|
597 |
|
---|
598 | if(from_date.compareTo(until_date) > 0) { // from date can't be later than until date
|
---|
599 | return OAIXML.createErrorMessage(OAIXML.NO_RECORDS_MATCH, "");
|
---|
600 | }
|
---|
601 | }
|
---|
602 |
|
---|
603 | if(until_date != null) {
|
---|
604 |
|
---|
605 | // Also call until_date.compareTo(earliestdatestamp) as the until date can't precede the earliest timestamp
|
---|
606 | // Unfortunately, this test has to be done after the granularity test
|
---|
607 | // compareTo() returns the value 0 if the argument Date is equal to this Date; a value less than 0 if this Date is before
|
---|
608 | // the Date argument; and a value greater than 0 if this Date is after the Date argument.
|
---|
609 | long earliestDatestamp = getEarliestDateStamp(collection_list);
|
---|
610 | String earliestDatestamp_str = OAIXML.getTime(earliestDatestamp);
|
---|
611 | Date earliestDatestamp_date = OAIXML.getDate(earliestDatestamp_str);
|
---|
612 |
|
---|
613 | if(until_date.compareTo(earliestDatestamp_date) < 0) {
|
---|
614 | return OAIXML.createErrorMessage(OAIXML.NO_RECORDS_MATCH, "");
|
---|
615 | }
|
---|
616 | }
|
---|
617 |
|
---|
618 |
|
---|
619 | // check the set arg is a set we know about
|
---|
620 | set_requested = param_map.containsKey(OAIXML.SET);
|
---|
621 | set_spec_str = null;
|
---|
622 | if(set_requested == true) {
|
---|
623 | set_spec_str = param_map.get(OAIXML.SET);
|
---|
624 | if (!this.set_oaiset.contains(set_spec_str)) {
|
---|
625 | // the set is not one we know about
|
---|
626 | logger.error("requested set is not found in this repository");
|
---|
627 | return OAIXML.createErrorMessage(OAIXML.BAD_ARGUMENT, "invalid set parameter");
|
---|
628 |
|
---|
629 | }
|
---|
630 | }
|
---|
631 | // Is the metadataPrefix arg one this repository supports?
|
---|
632 | prefix_value = param_map.get(OAIXML.METADATA_PREFIX);
|
---|
633 | if (repositorySupportsMetadataPrefix(prefix_value) == false) {
|
---|
634 | logger.error("requested metadataPrefix is not found in OAIConfig.xml");
|
---|
635 | return OAIXML.createErrorMessage(OAIXML.CANNOT_DISSEMINATE_FORMAT, "metadata format "+prefix_value+" not supported by this repository");
|
---|
636 | }
|
---|
637 |
|
---|
638 | } // else no resumption token, check other params
|
---|
639 |
|
---|
640 | // Whew. Now we have validated the params, we can work on doing the actual
|
---|
641 | // request
|
---|
642 |
|
---|
643 |
|
---|
644 | Document doc = XMLConverter.newDOM();
|
---|
645 | Element mr_msg = doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
646 | Element mr_req = doc.createElement(GSXML.REQUEST_ELEM);
|
---|
647 | // TODO does this need a type???
|
---|
648 | mr_msg.appendChild(mr_req);
|
---|
649 |
|
---|
650 | // copy in the from/until params if there
|
---|
651 | if (from != null) {
|
---|
652 | mr_req.appendChild(GSXML.createParameter(doc, OAIXML.FROM, from));
|
---|
653 | }
|
---|
654 | if (until != null) {
|
---|
655 | mr_req.appendChild(GSXML.createParameter(doc, OAIXML.UNTIL, until));
|
---|
656 | }
|
---|
657 | // add metadataPrefix
|
---|
658 | mr_req.appendChild(GSXML.createParameter(doc, OAIXML.METADATA_PREFIX, prefix_value));
|
---|
659 |
|
---|
660 | // do we have a set???
|
---|
661 | // if no set, we send to all collections in the collection list
|
---|
662 | // if super set, we send to all collections in super set list
|
---|
663 | // if a single collection, send to it
|
---|
664 | // if a subset, send to the collection
|
---|
665 | Vector<String> current_coll_list = getCollectionListForSet(set_spec_str); // return value is now safe to modify such as with .remove() operations
|
---|
666 | boolean single_collection = false;
|
---|
667 | if (current_coll_list.size() == 1) {
|
---|
668 | single_collection = true;
|
---|
669 |
|
---|
670 | // now handle any deactivated OAI collections. OAI collections can be deactivated briefly during build, when the building folder gets moved to index
|
---|
671 | String collName = current_coll_list.get(0); //set_spec_str;
|
---|
672 |
|
---|
673 | int index = collection_name_list.indexOf(collName);
|
---|
674 | if(collection_state_list.get(index).equals(DEACTIVATED)) {
|
---|
675 | // forced to send this as an OAI error message, since the XML output is of OAI schema
|
---|
676 | // and has to match http://www.openarchives.org/OAI/2.0/OAI-PMH.xsd, which has no info message
|
---|
677 | //return OAIXML.createCollectionDeactivatedMessage(collName);
|
---|
678 | String errorCode = (record_type.equals(OAIXML.RECORD)) ? OAIXML.NO_RECORDS_MATCH : OAIXML.ID_DOES_NOT_EXIST; // GetRecords vs GetIdentifiers request
|
---|
679 | return OAIXML.createErrorMessage(errorCode, "OAI collection " + collName + " is temporarily inactive. Likely it's in the final stages of being rebuilt. Check back shortly.");
|
---|
680 | }
|
---|
681 | } else {
|
---|
682 |
|
---|
683 | for(int i = 0; i < collection_name_list.size(); i++) {
|
---|
684 | if(collection_state_list.get(i).equals(DEACTIVATED)) {
|
---|
685 | String collName = collection_name_list.get(i);
|
---|
686 | // remove from the list of collections for which we're going to get identifiers/records
|
---|
687 | // so we won't get identifiers/records for any deactivated collections
|
---|
688 | current_coll_list.remove(collName);
|
---|
689 | }
|
---|
690 | }
|
---|
691 |
|
---|
692 | // Have we 0 active OAI collections?
|
---|
693 | if (current_coll_list.size() == 0) {
|
---|
694 | // 0 collections are active, but (the remaining one) would have been deactivated, not other reason
|
---|
695 | String errorCode = (record_type.equals(OAIXML.RECORD)) ? OAIXML.NO_RECORDS_MATCH : OAIXML.ID_DOES_NOT_EXIST; // GetRecords vs GetIdentifiers request
|
---|
696 | return OAIXML.createErrorMessage(errorCode, "OAI collections temporarily inactive. Likely because of (a) collection rebuilding. Check back shortly.");
|
---|
697 | }
|
---|
698 | }
|
---|
699 | if (set_spec_str != null && set_spec_str.indexOf(":") != -1) {
|
---|
700 | // we have a subset - add the set param back in
|
---|
701 | mr_req.appendChild(GSXML.createParameter(doc, OAIXML.SET, set_spec_str));
|
---|
702 | }
|
---|
703 |
|
---|
704 | int num_collected_records = 0;
|
---|
705 | int start_point = current_cursor; // may not be 0 if we are using a resumption token
|
---|
706 | String resumption_collection = "";
|
---|
707 | boolean empty_result_token = false; // if we are sending the last part of a list, then the token value will be empty
|
---|
708 |
|
---|
709 | // iterate through the list of collections and send the request to each
|
---|
710 |
|
---|
711 | int start_coll=0;
|
---|
712 | if (current_set != null) {
|
---|
713 | // we are resuming a previous request, need to locate the first collection
|
---|
714 | for (int i=0; i<current_coll_list.size(); i++) {
|
---|
715 | if (current_set.equals(current_coll_list.get(i))) {
|
---|
716 | start_coll = i;
|
---|
717 | break;
|
---|
718 | }
|
---|
719 | }
|
---|
720 | }
|
---|
721 |
|
---|
722 | for (int i=start_coll; i<current_coll_list.size(); i++) {
|
---|
723 | String current_coll = current_coll_list.get(i);
|
---|
724 | mr_req.setAttribute(GSXML.TO_ATT, current_coll+"/"+verb);
|
---|
725 |
|
---|
726 | Element result = (Element)mr.process(mr_msg);
|
---|
727 | //logger.info("*** " + verb+ " result for coll "+current_coll);
|
---|
728 | //logger.info("*** " + XMLConverter.getPrettyString(result));
|
---|
729 | if (result == null) {
|
---|
730 | logger.info("message router returns null");
|
---|
731 | // do what??? carry on? fail??
|
---|
732 | return OAIXML.createErrorMessage("Internal service returns null", "");
|
---|
733 | }
|
---|
734 | Element res = (Element)GSXML.getChildByTagName(result, GSXML.RESPONSE_ELEM);
|
---|
735 | if(res == null) {
|
---|
736 | logger.info("response element in xml_result is null");
|
---|
737 | return OAIXML.createErrorMessage("Internal service returns null", "");
|
---|
738 | }
|
---|
739 | NodeList record_list = res.getElementsByTagName(record_type);
|
---|
740 | int num_records = record_list.getLength();
|
---|
741 | if(num_records == 0) {
|
---|
742 | logger.info("message router returns 0 records for coll "+current_coll);
|
---|
743 | continue; // try the next collection
|
---|
744 | }
|
---|
745 | if (single_collection) {
|
---|
746 | total_size = num_records;
|
---|
747 | }
|
---|
748 | int records_to_add = (resume_after > 0 ? resume_after - num_collected_records : num_records);
|
---|
749 | if (records_to_add > (num_records-start_point)) {
|
---|
750 | records_to_add = num_records-start_point;
|
---|
751 | }
|
---|
752 | addRecordsToList(result_doc, result_element, record_list, start_point, records_to_add);
|
---|
753 | num_collected_records += records_to_add;
|
---|
754 |
|
---|
755 | // do we need to stop here, and do we need to issue a resumption token?
|
---|
756 | if (resume_after > 0 && num_collected_records == resume_after) {
|
---|
757 | // we have finished collecting records at the moment.
|
---|
758 | // but are we conincidentally at the end? or are there more to go?
|
---|
759 | if (records_to_add < (num_records - start_point)) {
|
---|
760 | // we have added less than this collection had
|
---|
761 | start_point += records_to_add;
|
---|
762 | resumption_collection = current_coll;
|
---|
763 | result_token_needed = true;
|
---|
764 | }
|
---|
765 | else {
|
---|
766 | // we added all this collection had to offer
|
---|
767 | // is there another collection in the list??
|
---|
768 | if (i<current_coll_list.size()-1) {
|
---|
769 | result_token_needed = true;
|
---|
770 | start_point = 0;
|
---|
771 | resumption_collection = current_coll_list.get(i+1);
|
---|
772 | }
|
---|
773 | else {
|
---|
774 | // we have finished one collection and there are no more collection
|
---|
775 | // if we need to send a resumption token (in this case, only because we started with one, then it will be empty
|
---|
776 | logger.info("*** at end of list, need empty result token");
|
---|
777 | empty_result_token = true;
|
---|
778 | }
|
---|
779 | }
|
---|
780 | break;
|
---|
781 | }
|
---|
782 | start_point = 0; // only the first one will have start non-zero, if we
|
---|
783 | // have a resumption token
|
---|
784 |
|
---|
785 | } // for each collection
|
---|
786 |
|
---|
787 | if (num_collected_records ==0) {
|
---|
788 | // there were no matching results
|
---|
789 | return OAIXML.createErrorMessage(OAIXML.NO_RECORDS_MATCH, "");
|
---|
790 | }
|
---|
791 |
|
---|
792 | if (num_collected_records < resume_after) {
|
---|
793 | // we have been through all collections, and there are no more
|
---|
794 | // if we need a result token - only because we started with one, so we need to send an empty one, then make sure everyone knows we are just sending an empty one
|
---|
795 | if (result_token_needed) {
|
---|
796 | empty_result_token = true;
|
---|
797 | }
|
---|
798 | }
|
---|
799 |
|
---|
800 | if (result_token_needed) {
|
---|
801 | // we need a resumption token
|
---|
802 | if (empty_result_token) {
|
---|
803 | logger.info("*** have empty result token");
|
---|
804 | token = "";
|
---|
805 | } else {
|
---|
806 | if (token != null) {
|
---|
807 | // we had a token for this request, we can just update it
|
---|
808 | token = OAIResumptionToken.updateToken(token, ""+cursor, resumption_collection, ""+start_point);
|
---|
809 | } else {
|
---|
810 | // we are generating a new one
|
---|
811 | token = OAIResumptionToken.createAndStoreResumptionToken(set_spec_str, prefix_value, from, until, ""+cursor, resumption_collection, ""+start_point );
|
---|
812 | }
|
---|
813 | }
|
---|
814 |
|
---|
815 | // result token XML
|
---|
816 | long expiration_date = -1;
|
---|
817 | if (empty_result_token) {
|
---|
818 | // we know how many records in total as we have sent them all
|
---|
819 | total_size = cursor+num_collected_records;
|
---|
820 | } else {
|
---|
821 | // non-empty token, set the expiration date
|
---|
822 | expiration_date = OAIResumptionToken.getExpirationDate(token);
|
---|
823 | }
|
---|
824 | Element token_elem = OAIXML.createResumptionTokenElement(result_doc, token, total_size, cursor, expiration_date);
|
---|
825 | // OAIXML.addToken(token_elem); // store it
|
---|
826 | result_element.appendChild(token_elem); // add to the result
|
---|
827 | }
|
---|
828 |
|
---|
829 |
|
---|
830 | return getMessage(result_doc, result_element);
|
---|
831 | }
|
---|
832 |
|
---|
833 | private Vector<String> getCollectionListForSet(String set) {
|
---|
834 | if (set == null) {
|
---|
835 | // no set requested, need the complete collection list
|
---|
836 |
|
---|
837 | // Important to return a clone of the member variable, since collection names can be locally removed from the Vector by
|
---|
838 | // the caller for any collection that is deactivated. This is so the caller can determine which collections are active and
|
---|
839 | // thus have records/identifiers to display. We don't want to remove elements from the member variable collection_name_list itself.
|
---|
840 | return (Vector<String>)this.collection_name_list.clone();
|
---|
841 | }
|
---|
842 | if (has_super_colls && super_coll_map.containsKey(set)) {
|
---|
843 | Vector<String> supercoll_list = super_coll_map.get(set);
|
---|
844 | return (Vector<String>)supercoll_list.clone(); // see comment just above, don't want member data structure modified by caller's destructive operations
|
---|
845 | }
|
---|
846 |
|
---|
847 | Vector<String> coll_list = new Vector<String>();
|
---|
848 | if (set.indexOf(":") != -1) {
|
---|
849 | String col_name = set.substring(0, set.indexOf(":"));
|
---|
850 | coll_list.add(col_name);
|
---|
851 | }
|
---|
852 | else {
|
---|
853 | coll_list.add(set);
|
---|
854 | }
|
---|
855 | return coll_list;
|
---|
856 | }
|
---|
857 | private void addRecordsToList(Document doc, Element result_element, NodeList
|
---|
858 | record_list, int start_point, int num_records) {
|
---|
859 | int end_point = start_point + num_records;
|
---|
860 | for (int i=start_point; i<end_point; i++) {
|
---|
861 | result_element.appendChild(doc.importNode(record_list.item(i), true));
|
---|
862 | }
|
---|
863 | }
|
---|
864 |
|
---|
865 | private Element collectAll(Element result, Element msg, String verb, String elem_name) {
|
---|
866 | if(result == null) {
|
---|
867 | //in the first round, result is null
|
---|
868 | return msg;
|
---|
869 | }
|
---|
870 | Element res_in_result = (Element)GSXML.getChildByTagName(result, GSXML.RESPONSE_ELEM);
|
---|
871 | if(res_in_result == null) { // return the results of all other collections accumulated so far
|
---|
872 | return msg;
|
---|
873 | }
|
---|
874 | Element verb_elem = (Element)GSXML.getChildByTagName(res_in_result, verb);
|
---|
875 | if(msg == null) {
|
---|
876 | return result;
|
---|
877 | }
|
---|
878 |
|
---|
879 | //e.g., get all <record> elements from the returned message. There may be none of
|
---|
880 | //such element, for example, the collection service returned an error message
|
---|
881 | NodeList elem_list = msg.getElementsByTagName(elem_name);
|
---|
882 |
|
---|
883 | for (int i=0; i<elem_list.getLength(); i++) {
|
---|
884 | verb_elem.appendChild(res_in_result.getOwnerDocument().importNode(elem_list.item(i), true));
|
---|
885 | }
|
---|
886 | return result;
|
---|
887 | }
|
---|
888 |
|
---|
889 |
|
---|
890 | /** there are three possible exception conditions: bad argument, idDoesNotExist, and noMetadataFormat.
|
---|
891 | * The first one is handled here, and the last two are processed by OAIPMH.
|
---|
892 | */
|
---|
893 | private Element doListMetadataFormats(Element req) {
|
---|
894 | //if the verb is ListMetadataFormats, there could be only one parameter: identifier
|
---|
895 | //, or there is no parameter; otherwise it is an error
|
---|
896 | //logger.info("" + XMLConverter.getString(msg));
|
---|
897 |
|
---|
898 | NodeList params = GSXML.getChildrenByTagName(req, GSXML.PARAM_ELEM);
|
---|
899 | Element param = null;
|
---|
900 | Document lmf_doc = XMLConverter.newDOM();
|
---|
901 | if(params.getLength() == 0) {
|
---|
902 | //this is requesting metadata formats for the whole repository
|
---|
903 | //read the oaiConfig.xml file, return the metadata formats specified there.
|
---|
904 | if (this.listmetadataformats_response != null) {
|
---|
905 | // we have already created it
|
---|
906 | return this.listmetadataformats_response;
|
---|
907 | }
|
---|
908 |
|
---|
909 | Element list_metadata_formats = lmf_doc.createElement(OAIXML.LIST_METADATA_FORMATS);
|
---|
910 | // get all the formats out of oai_config
|
---|
911 | NodeList formats = oai_config.getElementsByTagName(OAIXML.METADATA_FORMAT);
|
---|
912 | if (formats.getLength() ==0) {
|
---|
913 | logger.error("OAIConfig.xml must contain the supported metadata formats");
|
---|
914 | // TODO this is internal error, what to do???
|
---|
915 | return getMessage(lmf_doc, list_metadata_formats);
|
---|
916 | }
|
---|
917 |
|
---|
918 | for(int i=0; i<formats.getLength(); i++) {
|
---|
919 | Element f = OAIXML.getMetadataFormatShort(lmf_doc, (Element)formats.item(i));
|
---|
920 | list_metadata_formats.appendChild(f);
|
---|
921 | }
|
---|
922 | this.listmetadataformats_response = getMessage(lmf_doc, list_metadata_formats);
|
---|
923 | return this.listmetadataformats_response;
|
---|
924 |
|
---|
925 | }
|
---|
926 |
|
---|
927 | if (params.getLength() > 1) {
|
---|
928 | //Bad argument. Can't be more than one parameters for ListMetadataFormats verb
|
---|
929 | return OAIXML.createErrorMessage(OAIXML.BAD_ARGUMENT, "");
|
---|
930 | }
|
---|
931 |
|
---|
932 | // This is a request for the metadata of a particular item with an identifier
|
---|
933 | /**the request xml is in the form: <request>
|
---|
934 | * <param name=.../>
|
---|
935 | * </request>
|
---|
936 | *And there is a param element and one element only. (No paramList element in between).
|
---|
937 | */
|
---|
938 | param = (Element)params.item(0);
|
---|
939 | String param_name = param.getAttribute(GSXML.NAME_ATT);
|
---|
940 | String identifier = "";
|
---|
941 | if (!param_name.equals(OAIXML.IDENTIFIER)) {
|
---|
942 | //Bad argument
|
---|
943 | return OAIXML.createErrorMessage(OAIXML.BAD_ARGUMENT, "");
|
---|
944 | }
|
---|
945 |
|
---|
946 | identifier = param.getAttribute(GSXML.VALUE_ATT);
|
---|
947 | // the identifier is in the form: <coll_name>:<OID>
|
---|
948 | // so it must contain at least one ':' characters
|
---|
949 | // (the oid itself may contain : chars)
|
---|
950 | String[] strs = identifier.split(":", 2);
|
---|
951 | if(strs.length != 2) {
|
---|
952 | logger.error("identifier is not in the form coll:id" + identifier);
|
---|
953 | return OAIXML.createErrorMessage(OAIXML.ID_DOES_NOT_EXIST, "");
|
---|
954 | }
|
---|
955 |
|
---|
956 | // send request to message router
|
---|
957 | // get the names
|
---|
958 | String coll_name = strs[0];
|
---|
959 | String oid = strs[1];
|
---|
960 |
|
---|
961 | Document msg_doc = XMLConverter.newDOM();
|
---|
962 | Element message = msg_doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
963 | String verb = req.getAttribute(GSXML.TO_ATT);
|
---|
964 | String new_to = coll_name + "/" + verb;
|
---|
965 | Element request = GSXML.createBasicRequest(msg_doc, "oai???", new_to, null);
|
---|
966 | message.appendChild(request);
|
---|
967 | // add the id param
|
---|
968 | GSXML.addParameterToList(request, OAIXML.OID, oid);
|
---|
969 |
|
---|
970 | //Now send the request to the message router to process
|
---|
971 | Node result_node = mr.process(message);
|
---|
972 | return GSXML.nodeToElement(result_node);
|
---|
973 | }
|
---|
974 |
|
---|
975 | private void copyNamedElementfromConfig(Element to_elem, String element_name) {
|
---|
976 | Element original_element = (Element)GSXML.getChildByTagName(oai_config, element_name);
|
---|
977 | if(original_element != null) {
|
---|
978 | GSXML.copyNode(to_elem, original_element);
|
---|
979 | }
|
---|
980 | }
|
---|
981 |
|
---|
982 |
|
---|
983 | private Element doIdentify() {
|
---|
984 | //The validation for this verb has been done in OAIServer.validate(). So no bother here.
|
---|
985 | logger.info("");
|
---|
986 | if (this.identify_response != null) {
|
---|
987 | // we have already created it
|
---|
988 | return getMessage(this.identify_response.getOwnerDocument(), this.identify_response);
|
---|
989 | }
|
---|
990 | Document doc = XMLConverter.newDOM();
|
---|
991 | Element identify = doc.createElement(OAIXML.IDENTIFY);
|
---|
992 | //do the repository name
|
---|
993 | copyNamedElementfromConfig(identify, OAIXML.REPOSITORY_NAME);
|
---|
994 | //do the baseurl
|
---|
995 | copyNamedElementfromConfig(identify, OAIXML.BASE_URL);
|
---|
996 | //do the protocol version
|
---|
997 | copyNamedElementfromConfig(identify, OAIXML.PROTOCOL_VERSION);
|
---|
998 |
|
---|
999 | //There can be more than one admin email according to the OAI specification
|
---|
1000 | NodeList admin_emails = GSXML.getChildrenByTagName(oai_config, OAIXML.ADMIN_EMAIL);
|
---|
1001 | int num_admin = 0;
|
---|
1002 | Element from_admin_email = null;
|
---|
1003 | if (admin_emails != null) {
|
---|
1004 | num_admin = admin_emails.getLength();
|
---|
1005 | }
|
---|
1006 | for (int i=0; i<num_admin; i++) {
|
---|
1007 | GSXML.copyNode(identify, admin_emails.item(i));
|
---|
1008 | }
|
---|
1009 |
|
---|
1010 | //do the earliestDatestamp
|
---|
1011 | //send request to mr to search through the earliest datestamp amongst all oai collections in the repository.
|
---|
1012 | //ask the message router for a list of oai collections
|
---|
1013 | //NodeList oai_coll = getOAICollectionList();
|
---|
1014 | long earliestDatestamp = getEarliestDateStamp(collection_list);
|
---|
1015 | String earliestDatestamp_str = OAIXML.getTime(earliestDatestamp);
|
---|
1016 | Element earliestDatestamp_elem = doc.createElement(OAIXML.EARLIEST_DATESTAMP);
|
---|
1017 | GSXML.setNodeText(earliestDatestamp_elem, earliestDatestamp_str);
|
---|
1018 | identify.appendChild(earliestDatestamp_elem);
|
---|
1019 |
|
---|
1020 | //do the deletedRecord
|
---|
1021 | copyNamedElementfromConfig(identify, OAIXML.DELETED_RECORD);
|
---|
1022 | //do the granularity
|
---|
1023 | copyNamedElementfromConfig(identify, OAIXML.GRANULARITY);
|
---|
1024 |
|
---|
1025 | // output the oai identifier
|
---|
1026 | Element description = doc.createElement(OAIXML.DESCRIPTION);
|
---|
1027 | identify.appendChild(description);
|
---|
1028 | // TODO, make this a valid id
|
---|
1029 | Element oaiIdentifier = OAIXML.createOAIIdentifierXML(doc, repository_id, "lucene-jdbm-demo", "ec159e");
|
---|
1030 | description.appendChild(oaiIdentifier);
|
---|
1031 |
|
---|
1032 | // if there are any oaiInfo metadata, add them in too.
|
---|
1033 | Element info = (Element)GSXML.getChildByTagName(oai_config, OAIXML.OAI_INFO);
|
---|
1034 | if (info != null) {
|
---|
1035 | NodeList meta = GSXML.getChildrenByTagName(info, OAIXML.METADATA);
|
---|
1036 | if (meta != null && meta.getLength() > 0) {
|
---|
1037 | Element gsdl = OAIXML.createGSDLElement(doc);
|
---|
1038 | description.appendChild(gsdl);
|
---|
1039 | for (int m = 0; m<meta.getLength(); m++) {
|
---|
1040 | GSXML.copyNode(gsdl, meta.item(m));
|
---|
1041 | }
|
---|
1042 |
|
---|
1043 | }
|
---|
1044 | }
|
---|
1045 | this.identify_response = identify;
|
---|
1046 | return getMessage(doc, identify);
|
---|
1047 | }
|
---|
1048 | /** split the identifier into <collection + OID> as an array
|
---|
1049 | It has already been checked that the 'identifier' contains at least one ':'
|
---|
1050 | */
|
---|
1051 |
|
---|
1052 | /** validate if the specified metadata prefix value is supported by the repository
|
---|
1053 | * by checking it in the OAIConfig.xml
|
---|
1054 | */
|
---|
1055 | private boolean repositorySupportsMetadataPrefix(String prefix_value) {
|
---|
1056 | NodeList prefix_list = oai_config.getElementsByTagName(OAIXML.METADATA_PREFIX);
|
---|
1057 |
|
---|
1058 | for(int i=0; i<prefix_list.getLength(); i++) {
|
---|
1059 | if(prefix_value.equals(GSXML.getNodeText((Element)prefix_list.item(i)).trim() )) {
|
---|
1060 | return true;
|
---|
1061 | }
|
---|
1062 | }
|
---|
1063 | return false;
|
---|
1064 | }
|
---|
1065 | private Element doGetRecord(Element req){
|
---|
1066 | logger.info("");
|
---|
1067 | /** arguments:
|
---|
1068 | identifier: required
|
---|
1069 | metadataPrefix: required
|
---|
1070 | * Exceptions: badArgument; cannotDisseminateFormat; idDoesNotExist
|
---|
1071 | */
|
---|
1072 | Document doc = XMLConverter.newDOM();
|
---|
1073 | Element get_record = doc.createElement(OAIXML.GET_RECORD);
|
---|
1074 |
|
---|
1075 | HashSet<String> valid_strs = new HashSet<String>();
|
---|
1076 | valid_strs.add(OAIXML.IDENTIFIER);
|
---|
1077 | valid_strs.add(OAIXML.METADATA_PREFIX);
|
---|
1078 |
|
---|
1079 | NodeList params = GSXML.getChildrenByTagName(req, GSXML.PARAM_ELEM);
|
---|
1080 | HashMap<String, String> param_map = GSXML.getParamMap(params);
|
---|
1081 |
|
---|
1082 | if(!areAllParamsValid(param_map, valid_strs) ||
|
---|
1083 | params.getLength() == 0 ||
|
---|
1084 | param_map.containsKey(OAIXML.IDENTIFIER) == false ||
|
---|
1085 | param_map.containsKey(OAIXML.METADATA_PREFIX) == false ) {
|
---|
1086 | logger.error("must have the metadataPrefix/identifier parameter.");
|
---|
1087 | return OAIXML.createErrorMessage(OAIXML.BAD_ARGUMENT, "");
|
---|
1088 | }
|
---|
1089 |
|
---|
1090 | String prefix = param_map.get(OAIXML.METADATA_PREFIX);
|
---|
1091 | String identifier = param_map.get(OAIXML.IDENTIFIER);
|
---|
1092 |
|
---|
1093 | // verify the metadata prefix
|
---|
1094 | if (repositorySupportsMetadataPrefix(prefix) == false) {
|
---|
1095 | logger.error("requested prefix is not found in OAIConfig.xml");
|
---|
1096 | return OAIXML.createErrorMessage(OAIXML.CANNOT_DISSEMINATE_FORMAT, "");
|
---|
1097 | }
|
---|
1098 |
|
---|
1099 | // get the names
|
---|
1100 | String[] strs = identifier.split(":", 2);
|
---|
1101 | if(strs == null || strs.length < 2) {
|
---|
1102 | logger.error("identifier is not in the form coll:id" + identifier);
|
---|
1103 | return OAIXML.createErrorMessage(OAIXML.ID_DOES_NOT_EXIST, "");
|
---|
1104 | }
|
---|
1105 | //String name_of_site = strs[0];
|
---|
1106 | String coll_name = strs[0];
|
---|
1107 | String oid = strs[1];
|
---|
1108 |
|
---|
1109 | //re-organize the request element
|
---|
1110 | // reset the 'to' attribute
|
---|
1111 | String verb = req.getAttribute(GSXML.TO_ATT);
|
---|
1112 | req.setAttribute(GSXML.TO_ATT, coll_name + "/" + verb);
|
---|
1113 | // reset the identifier element
|
---|
1114 | Element param = GSXML.getNamedElement(req, GSXML.PARAM_ELEM, GSXML.NAME_ATT, OAIXML.IDENTIFIER);
|
---|
1115 | if (param != null) {
|
---|
1116 | param.setAttribute(GSXML.NAME_ATT, OAIXML.OID);
|
---|
1117 | param.setAttribute(GSXML.VALUE_ATT, oid);
|
---|
1118 | }
|
---|
1119 |
|
---|
1120 | //Now send the request to the message router to process
|
---|
1121 | Element msg = doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
1122 | msg.appendChild(doc.importNode(req, true));
|
---|
1123 | Node result_node = mr.process(msg);
|
---|
1124 | return GSXML.nodeToElement(result_node);
|
---|
1125 | }
|
---|
1126 |
|
---|
1127 | // See OAIConfig.xml
|
---|
1128 | // dynamically works out what the earliestDateStamp is, since it varies by collection
|
---|
1129 | // returns this time in *milliseconds*.
|
---|
1130 | protected long getEarliestDateStamp(Element oai_coll_list) {
|
---|
1131 | // config earliest datstamp
|
---|
1132 | long config_datestamp = 0;
|
---|
1133 | Element config_datestamp_elem = (Element)GSXML.getChildByTagName(this.oai_config, OAIXML.EARLIEST_DATESTAMP);
|
---|
1134 | if (config_datestamp_elem != null) {
|
---|
1135 | String datest = GSXML.getNodeText(config_datestamp_elem);
|
---|
1136 | config_datestamp = OAIXML.getTime(datest);
|
---|
1137 | if (config_datestamp == -1) {
|
---|
1138 | config_datestamp = 0;
|
---|
1139 | }
|
---|
1140 | }
|
---|
1141 | //do the earliestDatestamp
|
---|
1142 | long current_time = System.currentTimeMillis();
|
---|
1143 | long earliestDatestamp = current_time;
|
---|
1144 | NodeList oai_coll = oai_coll_list.getElementsByTagName(GSXML.COLLECTION_ELEM);
|
---|
1145 | int oai_coll_size = oai_coll.getLength();
|
---|
1146 | if (oai_coll_size == 0) {
|
---|
1147 | logger.info("returned oai collection list is empty. Setting repository earliestDatestamp to be the earliest datestamp from OAIConfig.xml, or 1970-01-01 if not specified.");
|
---|
1148 | return config_datestamp;
|
---|
1149 | }
|
---|
1150 | // the earliestDatestamp is now stored as a metadata element in the collection's buildConfig.xml file
|
---|
1151 | // we get the earliestDatestamp among the collections
|
---|
1152 | for(int i=0; i<oai_coll_size; i++) {
|
---|
1153 | String collName = collection_name_list.get(i);
|
---|
1154 | long coll_earliestDatestamp = Long.parseLong(((Element)oai_coll.item(i)).getAttribute(OAIXML.EARLIEST_OAI_DATESTAMP)); // Taken from oai-inf db's OAI_EARLIEST_TIMESTAMP_OID entry, -1 if not found
|
---|
1155 |
|
---|
1156 | if (coll_earliestDatestamp > 0 && earliestDatestamp > coll_earliestDatestamp) {
|
---|
1157 | earliestDatestamp = coll_earliestDatestamp;
|
---|
1158 | //logger.info("@@@ Found earlier timestamp: " + earliestDatestamp + " ms");
|
---|
1159 | }
|
---|
1160 | }
|
---|
1161 |
|
---|
1162 | // we're no longer trying fallbacks for earliestDatestamp (other than the extreme fallback of
|
---|
1163 | // unix epoch time) because, going forward, all collections will have oai-inf db containing
|
---|
1164 | // an entry for earliesttimestamp. And all OAICollections will moreover have them stored and
|
---|
1165 | // will return them upon calling getEarliestOAIDatestamp().
|
---|
1166 | /*
|
---|
1167 | if(earliestDatestamp == current_time) {
|
---|
1168 | logger.info("Can't determine earliesttimestamp from oai-inf.db for any OAI collection. Trying timestamps in build config...");
|
---|
1169 | for(int i=0; i<oai_coll_size; i++) {
|
---|
1170 | String collName = collection_name_list.get(i);
|
---|
1171 | long coll_earliestDatestamp = Long.parseLong(((Element)oai_coll.item(i)).getAttribute(OAIXML.EARLIEST_DATESTAMP)); // Taken from the earliest datestamp field in buildcfg
|
---|
1172 | if (coll_earliestDatestamp == 0) {
|
---|
1173 | // try last modified
|
---|
1174 | coll_earliestDatestamp = Long.parseLong(((Element)oai_coll.item(i)).getAttribute(OAIXML.LAST_MODIFIED));
|
---|
1175 | //logger.info("@@@ Falling back to using collection " + collName + "'s lastmodified date as its earliest timestamp: " + coll_earliestDatestamp);
|
---|
1176 | }
|
---|
1177 | if (coll_earliestDatestamp > 0) {
|
---|
1178 | earliestDatestamp = (earliestDatestamp > coll_earliestDatestamp)? coll_earliestDatestamp : earliestDatestamp;
|
---|
1179 | }
|
---|
1180 | }
|
---|
1181 | }
|
---|
1182 | */
|
---|
1183 |
|
---|
1184 | if (earliestDatestamp == current_time) {
|
---|
1185 | logger.info("no collection had a real datestamp, using value from OAIConfig");
|
---|
1186 | return config_datestamp;
|
---|
1187 | }
|
---|
1188 | return earliestDatestamp;
|
---|
1189 | }
|
---|
1190 |
|
---|
1191 | private boolean collectionsChangedSinceTime(String set_spec_str, long initial_time) {
|
---|
1192 |
|
---|
1193 | // we need to look though all collections in the set to see if any have last modified dates > initial_time
|
---|
1194 | Vector<String> set_coll_list = getCollectionListForSet(set_spec_str);
|
---|
1195 |
|
---|
1196 | Node child = this.collection_list.getFirstChild();
|
---|
1197 | while (child != null) {
|
---|
1198 | if (child.getNodeName().equals(GSXML.COLLECTION_ELEM)) {
|
---|
1199 | String coll_id =((Element) child).getAttribute(GSXML.NAME_ATT);
|
---|
1200 | if (set_coll_list.contains(coll_id)) {
|
---|
1201 | long last_modified = Long.parseLong(((Element)child).getAttribute(OAIXML.LAST_MODIFIED));
|
---|
1202 | if (initial_time < last_modified) {
|
---|
1203 | return true;
|
---|
1204 | }
|
---|
1205 | }
|
---|
1206 | }
|
---|
1207 | child = child.getNextSibling();
|
---|
1208 | }
|
---|
1209 | return false;
|
---|
1210 |
|
---|
1211 | }
|
---|
1212 |
|
---|
1213 | }
|
---|
1214 |
|
---|
1215 |
|
---|