[14211] | 1 | package org.greenstone.gsdl3.core;
|
---|
| 2 |
|
---|
| 3 | import org.greenstone.gsdl3.util.*;
|
---|
| 4 | import org.greenstone.gsdl3.action.*;
|
---|
| 5 | // XML classes
|
---|
| 6 | import org.w3c.dom.Node;
|
---|
| 7 | import org.w3c.dom.NodeList;
|
---|
| 8 | import org.w3c.dom.Document;
|
---|
| 9 | import org.w3c.dom.Element;
|
---|
| 10 |
|
---|
| 11 | // other java classes
|
---|
| 12 | import java.io.File;
|
---|
| 13 | import java.util.*;
|
---|
| 14 |
|
---|
| 15 | import org.apache.log4j.*;
|
---|
| 16 |
|
---|
| 17 | /** a Receptionist, used for oai metadata response xml generation.
|
---|
| 18 | * This receptionist talks to the message router directly,
|
---|
| 19 | * instead of via any action, hence no action map is needed.
|
---|
| 20 | * @see the basic Receptionist
|
---|
| 21 | */
|
---|
| 22 | public class OAIReceptionist implements ModuleInterface {
|
---|
| 23 |
|
---|
| 24 | static Logger logger = Logger.getLogger(org.greenstone.gsdl3.core.OAIReceptionist.class.getName());
|
---|
| 25 |
|
---|
| 26 | /** Instead of a config_params object, only a site_name is needed by oai receptionist. */
|
---|
| 27 | protected String site_name = null;
|
---|
| 28 | /** container Document to create XML Nodes for requests sent to message router
|
---|
| 29 | * Not used for response
|
---|
| 30 | */
|
---|
| 31 | protected Document doc=null;
|
---|
| 32 |
|
---|
| 33 | /** a converter class to parse XML and create Docs */
|
---|
| 34 | protected XMLConverter converter=null;
|
---|
| 35 |
|
---|
| 36 | /** the configure file of this receptionist passed from the oai servlet. */
|
---|
| 37 | protected Element oai_config = null;
|
---|
| 38 |
|
---|
| 39 | /** contained in the OAIConfig.xml deciding whether the resumptionToken should be in use */
|
---|
| 40 | protected int resume_after = -1 ;
|
---|
| 41 |
|
---|
| 42 | /** the message router that the Receptionist and Actions will talk to */
|
---|
| 43 | protected ModuleInterface mr = null;
|
---|
| 44 |
|
---|
| 45 | public OAIReceptionist() {
|
---|
| 46 | this.converter = new XMLConverter();
|
---|
| 47 | this.doc = this.converter.newDOM();
|
---|
| 48 |
|
---|
| 49 | }
|
---|
| 50 |
|
---|
| 51 | public void cleanUp() {}
|
---|
| 52 |
|
---|
| 53 | public void setSiteName(String site_name) {
|
---|
| 54 | this.site_name = site_name;
|
---|
| 55 | }
|
---|
| 56 | /** sets the message router - it should already be created and
|
---|
| 57 | * configured in the init() of a servlet (OAIServer, for example) before being passed to the receptionist*/
|
---|
| 58 | public void setMessageRouter(ModuleInterface mr) {
|
---|
| 59 | this.mr = mr;
|
---|
| 60 | }
|
---|
| 61 |
|
---|
| 62 | /** configures the receptionist */
|
---|
| 63 | public boolean configure(Element config) {
|
---|
| 64 |
|
---|
| 65 | if (this.mr==null) {
|
---|
| 66 | logger.error(" message routers must be set before calling oai configure");
|
---|
| 67 | return false;
|
---|
| 68 | }
|
---|
| 69 | if (config == null) {
|
---|
| 70 | logger.error(" oai configure file is null");
|
---|
| 71 | return false;
|
---|
| 72 | }
|
---|
| 73 | oai_config = config;
|
---|
| 74 | resume_after = getResumeAfter();
|
---|
| 75 |
|
---|
| 76 | //clear out expired resumption tokens stored in OAIResumptionToken.xml
|
---|
| 77 | OAIXML.init();
|
---|
| 78 | OAIXML.clearExpiredTokens();
|
---|
| 79 |
|
---|
| 80 | return true;
|
---|
| 81 | }
|
---|
| 82 | /** process using strings - just calls process using Elements */
|
---|
| 83 | public String process(String xml_in) {
|
---|
| 84 |
|
---|
[16688] | 85 | Node message_node = this.converter.getDOM(xml_in);
|
---|
| 86 | Node page = process(message_node);
|
---|
[14211] | 87 | return this.converter.getString(page);
|
---|
| 88 | }
|
---|
[16688] | 89 |
|
---|
[14211] | 90 | //Compose a message element used to send back to the OAIServer servlet.
|
---|
| 91 | //This method is only used within OAIReceptionist
|
---|
| 92 | private Element getMessage(Element e) {
|
---|
| 93 | Element msg = OAIXML.createElement(OAIXML.MESSAGE);
|
---|
| 94 | msg.appendChild(OAIXML.getResponse(e));
|
---|
| 95 | return msg;
|
---|
| 96 | }
|
---|
| 97 | /** process - produce xml data in response to a request
|
---|
| 98 | * if something goes wrong, it returns null -
|
---|
| 99 | */
|
---|
[16688] | 100 | public Node process(Node message_node) {
|
---|
[14211] | 101 | logger.error("OAIReceptionist received request");
|
---|
[16688] | 102 |
|
---|
| 103 | Element message = this.converter.nodeToElement(message_node);
|
---|
[14211] | 104 | logger.error(this.converter.getString(message));
|
---|
[16688] | 105 |
|
---|
[14211] | 106 | // check that its a correct message tag
|
---|
| 107 | if (!message.getTagName().equals(GSXML.MESSAGE_ELEM)) {
|
---|
| 108 | logger.error(" Invalid message. GSDL message should start with <"+GSXML.MESSAGE_ELEM+">, instead it starts with:"+message.getTagName()+".");
|
---|
| 109 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 110 | }
|
---|
| 111 |
|
---|
| 112 | // get the request out of the message - assume that there is only one
|
---|
| 113 | Element request = (Element)GSXML.getChildByTagName(message, GSXML.REQUEST_ELEM);
|
---|
| 114 | if (request == null) {
|
---|
| 115 | logger.error(" message had no request!");
|
---|
| 116 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 117 | }
|
---|
| 118 | //At this stage, the value of 'to' attribute of the request must be the 'verb'
|
---|
| 119 | //The only thing that the oai receptionist can be sure is that these verbs are valid, nothing else.
|
---|
| 120 | String verb = request.getAttribute(GSXML.TO_ATT);
|
---|
| 121 | if (verb.equals(OAIXML.IDENTIFY)) {
|
---|
| 122 | return doIdentify();
|
---|
| 123 | }
|
---|
| 124 | if (verb.equals(OAIXML.LIST_METADATA_FORMATS)) {
|
---|
| 125 | return doListMetadataFormats(message);
|
---|
| 126 | }
|
---|
| 127 | if (verb.equals(OAIXML.LIST_SETS)) {
|
---|
| 128 | return doListSets(message);
|
---|
| 129 | }
|
---|
| 130 | if (verb.equals(OAIXML.GET_RECORD)) {
|
---|
| 131 | return doGetRecord(message);
|
---|
| 132 | }
|
---|
| 133 | if (verb.equals(OAIXML.LIST_IDENTIFIERS)) {
|
---|
| 134 | return doListIdentifiers(message);
|
---|
| 135 | }
|
---|
| 136 | if (verb.equals(OAIXML.LIST_RECORDS)) {
|
---|
| 137 | return doListRecords(message);
|
---|
| 138 | }
|
---|
| 139 | return getMessage(OAIXML.createErrorElement("Unexpected things happened", ""));
|
---|
| 140 |
|
---|
| 141 | }
|
---|
| 142 | /** send a request to the message router asking for a list of collections that support oai
|
---|
| 143 | * The type attribute must be changed from 'oaiService' to 'oaiSetList'
|
---|
| 144 | */
|
---|
| 145 | private NodeList getOAICollectionList() {
|
---|
| 146 | Element message = this.doc.createElement(OAIXML.MESSAGE);
|
---|
| 147 | Element request = this.doc.createElement(OAIXML.REQUEST);
|
---|
| 148 | message.appendChild(request);
|
---|
| 149 | request.setAttribute(OAIXML.TYPE, OAIXML.OAI_SET_LIST);
|
---|
| 150 | request.setAttribute(OAIXML.TO, "");
|
---|
[16688] | 151 | Node msg_node = mr.process(message);
|
---|
[14211] | 152 |
|
---|
[16688] | 153 | if (msg_node == null) {
|
---|
| 154 | logger.error("returned msg_node from mr is null");
|
---|
[14211] | 155 | return null;
|
---|
| 156 | }
|
---|
[16688] | 157 | Element resp = (Element)GSXML.getChildByTagName(msg_node, OAIXML.RESPONSE);
|
---|
[14211] | 158 | Element coll_list = (Element)GSXML.getChildByTagName(resp, OAIXML.COLLECTION_LIST);
|
---|
| 159 | if (coll_list == null) {
|
---|
| 160 | logger.error("coll_list is null");
|
---|
| 161 | return null;
|
---|
| 162 | }
|
---|
| 163 | //logger.info(GSXML.xmlNodeToString(coll_list));
|
---|
| 164 | NodeList list = coll_list.getElementsByTagName(OAIXML.COLLECTION);
|
---|
| 165 | int length = list.getLength();
|
---|
| 166 | if (length == 0) {
|
---|
| 167 | logger.error("length is 0");
|
---|
| 168 | return null;
|
---|
| 169 | }
|
---|
| 170 | return list;
|
---|
| 171 | }
|
---|
| 172 | /**Exclusively called by doListSets()*/
|
---|
| 173 | private void getSets(Element list_sets_elem, NodeList oai_coll, int start_point, int end_point) {
|
---|
| 174 | for (int i=start_point; i<end_point; i++) {
|
---|
| 175 | String coll_spec = ((Element)oai_coll.item(i)).getAttribute(OAIXML.NAME);
|
---|
| 176 | String coll_name = coll_spec.substring(coll_spec.indexOf(":") + 1);
|
---|
| 177 | Element set = OAIXML.createElement(OAIXML.SET);
|
---|
| 178 | Element set_spec = OAIXML.createElement(OAIXML.SET_SPEC);
|
---|
| 179 | GSXML.setNodeText(set_spec, coll_spec);
|
---|
| 180 | set.appendChild(set_spec);
|
---|
| 181 | Element set_name = OAIXML.createElement(OAIXML.SET_NAME);
|
---|
| 182 | GSXML.setNodeText(set_name, coll_name);
|
---|
| 183 | set.appendChild(set_name);
|
---|
| 184 | list_sets_elem.appendChild(set);
|
---|
| 185 | }
|
---|
| 186 | }
|
---|
| 187 | private int getResumeAfter() {
|
---|
| 188 | Element resume_after = (Element)GSXML.getChildByTagName(oai_config, OAIXML.RESUME_AFTER);
|
---|
| 189 | if(resume_after != null) return Integer.parseInt(GSXML.getNodeText(resume_after));
|
---|
| 190 | return -1;
|
---|
| 191 | }
|
---|
| 192 | /** method to compose a set element
|
---|
| 193 | */
|
---|
| 194 | private Element doListSets(Element msg){
|
---|
| 195 | logger.info("");
|
---|
| 196 | // option: resumptionToken
|
---|
| 197 | // exceptions: badArgument, badResumptionToken, noSetHierarchy
|
---|
| 198 | Element list_sets_elem = OAIXML.createElement(OAIXML.LIST_SETS);
|
---|
| 199 |
|
---|
| 200 | //ask the message router for a list of oai collections
|
---|
| 201 | NodeList oai_coll = getOAICollectionList();
|
---|
| 202 | int oai_coll_size = oai_coll.getLength();
|
---|
| 203 | if (oai_coll_size == 0) {
|
---|
| 204 | return getMessage(list_sets_elem);
|
---|
| 205 | }
|
---|
| 206 |
|
---|
| 207 | Element req = (Element)GSXML.getChildByTagName(msg, GSXML.REQUEST_ELEM);
|
---|
| 208 | if (req == null) {
|
---|
| 209 | logger.error("req is null");
|
---|
| 210 | return null;
|
---|
| 211 | }
|
---|
| 212 | //params list only contains the parameters other than the verb
|
---|
| 213 | NodeList params = GSXML.getChildrenByTagName(req, OAIXML.PARAM);
|
---|
| 214 | Element param = null;
|
---|
| 215 | int smaller = (oai_coll_size>resume_after)? resume_after : oai_coll_size;
|
---|
| 216 | if (params.getLength() > 1) {
|
---|
| 217 | //Bad argument. Can't be more than one parameters for ListMetadataFormats verb
|
---|
| 218 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 219 | }
|
---|
| 220 | if(params.getLength() == 0) {
|
---|
| 221 | //this is requesting a list of sets in the whole repository
|
---|
| 222 | /** there is no resumeptionToken in the request, we check whether we need
|
---|
| 223 | * to send out resumeptionToken by comparing the total number of sets in this
|
---|
| 224 | * repository and the specified value of resumeAfter
|
---|
| 225 | */
|
---|
| 226 | if(resume_after < 0 || oai_coll_size <= resume_after) {
|
---|
| 227 | //send the whole list of records
|
---|
| 228 | //all data are sent on the first request. Therefore there should be
|
---|
| 229 | //no resumeptionToken stored in OAIConfig.xml.
|
---|
| 230 | //As long as the verb is 'ListSets', we ignore the rest of the parameters
|
---|
| 231 | getSets(list_sets_elem, oai_coll, 0, oai_coll_size);
|
---|
| 232 | return getMessage(list_sets_elem);
|
---|
| 233 | }
|
---|
| 234 |
|
---|
| 235 | //append required sets to list_sets_elem (may be a complete or incomplete list)
|
---|
| 236 | getSets(list_sets_elem, oai_coll, 0, smaller);
|
---|
| 237 |
|
---|
| 238 | if(oai_coll_size > resume_after) {
|
---|
| 239 | //An incomplete list is sent; append a resumptionToken element
|
---|
| 240 | Element token = createResumptionTokenElement(oai_coll_size, 0, resume_after, true);
|
---|
| 241 | //store this token
|
---|
| 242 | OAIXML.addToken(token);
|
---|
| 243 |
|
---|
| 244 | list_sets_elem.appendChild(token);
|
---|
| 245 | }
|
---|
| 246 |
|
---|
| 247 | return getMessage(list_sets_elem);
|
---|
| 248 | }
|
---|
| 249 |
|
---|
| 250 | // The url should contain only one param called resumptionToken
|
---|
| 251 | // This is requesting a subsequent part of a list, following a previously sent incomplete list
|
---|
| 252 | param = (Element)params.item(0);
|
---|
| 253 | String param_name = param.getAttribute(OAIXML.NAME);
|
---|
| 254 | if (!param_name.equals(OAIXML.RESUMPTION_TOKEN)) {
|
---|
| 255 | //Bad argument
|
---|
| 256 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 257 | }
|
---|
| 258 | //get the token
|
---|
| 259 | String token = param.getAttribute(OAIXML.VALUE);
|
---|
| 260 | //validate the token string (the string has already been decoded in OAIServer, e.g.,
|
---|
| 261 | // replace %3A with ':')
|
---|
| 262 | if(OAIXML.containsToken(token) == false) {
|
---|
| 263 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_RESUMPTION_TOKEN, ""));
|
---|
| 264 | }
|
---|
| 265 | //take out the cursor value, which is the size of previously sent list
|
---|
| 266 | int index = token.indexOf(":");
|
---|
| 267 | int cursor = Integer.parseInt(token.substring(index + 1));
|
---|
| 268 | Element token_elem = null;
|
---|
| 269 |
|
---|
| 270 | // are we sending the final part of a complete list?
|
---|
| 271 | if(cursor + resume_after >= oai_coll_size) {
|
---|
| 272 | //Yes, we are.
|
---|
| 273 | //append required sets to list_sets_elem (list is complete)
|
---|
| 274 | getSets(list_sets_elem, oai_coll, cursor, oai_coll_size);
|
---|
| 275 | //An incomplete list is sent; append a resumptionToken element
|
---|
| 276 | token_elem = createResumptionTokenElement(oai_coll_size, cursor, -1, false);
|
---|
| 277 | list_sets_elem.appendChild(token_elem);
|
---|
| 278 | } else {
|
---|
| 279 | //No, we are not.
|
---|
| 280 | //append required sets to list_sets_elem (list is incomplete)
|
---|
| 281 | getSets(list_sets_elem, oai_coll, cursor, cursor + resume_after);
|
---|
| 282 | token_elem = createResumptionTokenElement(oai_coll_size, cursor, cursor + resume_after, true);
|
---|
| 283 | //store this token
|
---|
| 284 | OAIXML.addToken(token_elem);
|
---|
| 285 | list_sets_elem.appendChild(token_elem);
|
---|
| 286 | }
|
---|
| 287 | return getMessage(list_sets_elem);
|
---|
| 288 | }
|
---|
[24440] | 289 | private Element createResumptionTokenElement(int total_size, int cursor, int so_far_sent, boolean set_expiration, String metadata_prefix) {
|
---|
[14211] | 290 | Element token = OAIXML.createElement(OAIXML.RESUMPTION_TOKEN);
|
---|
| 291 | token.setAttribute(OAIXML.COMPLETE_LIST_SIZE, "" + total_size);
|
---|
| 292 | token.setAttribute(OAIXML.CURSOR, "" + cursor);
|
---|
| 293 |
|
---|
| 294 | if(set_expiration) {
|
---|
| 295 | /** read the resumptionTokenExpiration element in OAIConfig.xml and get the specified time value
|
---|
| 296 | * Use the time value plus the current system time to get the expiration date string.
|
---|
| 297 | */
|
---|
[23938] | 298 | String expiration_date = OAIXML.getTime(System.currentTimeMillis() + OAIXML.getTokenExpiration()); // in milliseconds
|
---|
[14211] | 299 | token.setAttribute(OAIXML.EXPIRATION_DATE, expiration_date);
|
---|
| 300 | }
|
---|
| 301 |
|
---|
| 302 | if(so_far_sent > 0) {
|
---|
| 303 | //the format of resumptionToken is not defined by the OAI-PMH and should be
|
---|
| 304 | //considered opaque by the harvester (in other words, strictly follow what the
|
---|
| 305 | //data provider has to offer
|
---|
| 306 | //Here, we make use of the uniqueness of the system time
|
---|
[24440] | 307 | String tokenValue = OAIXML.GS3OAI + System.currentTimeMillis() + ":" + so_far_sent;
|
---|
| 308 | if(!metadata_prefix.equals("")) {
|
---|
| 309 | tokenValue = tokenValue + ":" + metadata_prefix;
|
---|
| 310 | }
|
---|
| 311 | GSXML.setNodeText(token, tokenValue);
|
---|
[14211] | 312 | }
|
---|
| 313 | return token;
|
---|
| 314 | }
|
---|
[24440] | 315 |
|
---|
| 316 | private Element createResumptionTokenElement(int total_size, int cursor, int so_far_sent, boolean set_expiration) {
|
---|
| 317 | return createResumptionTokenElement(total_size, cursor, so_far_sent, set_expiration, ""); // empty metadata_prefix
|
---|
| 318 | }
|
---|
| 319 |
|
---|
[14211] | 320 | /** if the param_map contains strings other than those in valid_strs, return false;
|
---|
| 321 | * otherwise true.
|
---|
| 322 | */
|
---|
| 323 | private boolean isValidParam(HashMap param_map, HashSet valid_strs) {
|
---|
| 324 | ArrayList param_list = new ArrayList(param_map.keySet());
|
---|
| 325 | for(int i=0; i<param_list.size(); i++) {
|
---|
| 326 | if (valid_strs.contains((String)param_list.get(i)) == false) {
|
---|
| 327 | return false;
|
---|
| 328 | }
|
---|
| 329 | }
|
---|
| 330 | return true;
|
---|
| 331 | }
|
---|
| 332 | private Element doListIdentifiers(Element msg) {
|
---|
| 333 | // option: from, until, set, metadataPrefix, resumptionToken
|
---|
| 334 | // exceptions: badArgument, badResumptionToken, cannotDisseminateFormat, noRecordMatch, and noSetHierarchy
|
---|
| 335 | HashSet valid_strs = new HashSet();
|
---|
| 336 | valid_strs.add(OAIXML.FROM);
|
---|
| 337 | valid_strs.add(OAIXML.UNTIL);
|
---|
| 338 | valid_strs.add(OAIXML.SET);
|
---|
| 339 | valid_strs.add(OAIXML.METADATA_PREFIX);
|
---|
| 340 | valid_strs.add(OAIXML.RESUMPTION_TOKEN);
|
---|
| 341 |
|
---|
| 342 | Element list_identifiers = OAIXML.createElement(OAIXML.LIST_IDENTIFIERS);
|
---|
| 343 | Element req = (Element)GSXML.getChildByTagName(msg, GSXML.REQUEST_ELEM);
|
---|
| 344 | if (req == null) { logger.error("req is null"); return null; }
|
---|
| 345 | NodeList params = GSXML.getChildrenByTagName(req, OAIXML.PARAM);
|
---|
| 346 | String coll_name = "";
|
---|
| 347 | String token = "";
|
---|
| 348 |
|
---|
| 349 | HashMap param_map = OAIXML.getParamMap(params);
|
---|
[23913] | 350 | if (!isValidParam(param_map, valid_strs)) {
|
---|
| 351 | logger.error("One of the params is invalid");
|
---|
| 352 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
[14211] | 353 | }
|
---|
[23913] | 354 | // param keys are valid, but if there are any date params, check they're of the right format
|
---|
| 355 | String from = (String)param_map.get(OAIXML.FROM);
|
---|
| 356 | if(from != null) {
|
---|
| 357 | Date from_date = OAIXML.getDate(from);
|
---|
| 358 | if(from_date == null) {
|
---|
| 359 | logger.error("invalid date: " + from);
|
---|
| 360 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 361 | }
|
---|
| 362 | }
|
---|
| 363 | String until = (String)param_map.get(OAIXML.UNTIL);
|
---|
| 364 | if(until != null) {
|
---|
| 365 | Date until_date = OAIXML.getDate(until);
|
---|
| 366 | if(until_date == null) {
|
---|
| 367 | logger.error("invalid date: " + until);
|
---|
| 368 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 369 | }
|
---|
| 370 | }
|
---|
| 371 | if(from != null && until != null) { // check they are of the same date-time format (granularity)
|
---|
| 372 | if(from.length() != until.length()) {
|
---|
| 373 | logger.error("The request has different granularities (date-time formats) for the From and Until date parameters.");
|
---|
| 374 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 375 | }
|
---|
| 376 | }
|
---|
| 377 |
|
---|
[14211] | 378 | //ask the message router for a list of oai collections
|
---|
| 379 | NodeList oai_coll = getOAICollectionList();
|
---|
| 380 | int oai_coll_size = oai_coll.getLength();
|
---|
| 381 | if (oai_coll_size == 0) {
|
---|
| 382 | logger.info("returned oai collection list is empty");
|
---|
| 383 | return getMessage(OAIXML.createErrorElement(OAIXML.NO_RECORDS_MATCH, ""));
|
---|
| 384 | }
|
---|
| 385 |
|
---|
| 386 | //Now we check if the optional argument 'set' has been specified in the params; if so,
|
---|
| 387 | //whether the specified setSpec is supported by this repository
|
---|
| 388 | boolean request_set = param_map.containsKey(OAIXML.SET);
|
---|
| 389 | if(request_set == true) {
|
---|
| 390 | boolean set_supported = false;
|
---|
| 391 | String set_spec_str = (String)param_map.get(OAIXML.SET);
|
---|
| 392 | // get the collection name
|
---|
| 393 | //if setSpec is supported by this repository, it must be in the form: site_name:coll_name
|
---|
| 394 | String[] strs = splitSetSpec(set_spec_str);
|
---|
| 395 | coll_name = strs[1];
|
---|
| 396 |
|
---|
| 397 | for(int i=0; i<oai_coll_size; i++) {
|
---|
| 398 | if(set_spec_str.equals(((Element)oai_coll.item(i)).getAttribute(OAIXML.NAME))) {
|
---|
| 399 | set_supported = true;
|
---|
| 400 | }
|
---|
| 401 | }
|
---|
| 402 | if(set_supported == false) {
|
---|
| 403 | logger.error("requested set is not found in this repository");
|
---|
| 404 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 405 | }
|
---|
| 406 | }
|
---|
| 407 |
|
---|
| 408 | //Is there a resumptionToken included which is requesting an incomplete list?
|
---|
| 409 | if(param_map.containsKey(OAIXML.RESUMPTION_TOKEN)) {
|
---|
| 410 | // validate resumptionToken
|
---|
| 411 | token = (String)param_map.get(OAIXML.RESUMPTION_TOKEN);
|
---|
| 412 | logger.info("has resumptionToken" + token);
|
---|
| 413 | if(OAIXML.containsToken(token) == false) {
|
---|
| 414 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_RESUMPTION_TOKEN, ""));
|
---|
| 415 | }
|
---|
| 416 | }
|
---|
| 417 |
|
---|
[23913] | 418 | // Custom test that expects a metadataPrefix comes here at end so that the official params can
|
---|
| 419 | // be tested first for errors and their error responses sent off. Required for OAI validation
|
---|
| 420 | if (!param_map.containsKey(OAIXML.METADATA_PREFIX)) {
|
---|
| 421 | logger.error("contains invalid params or no metadataPrefix");
|
---|
| 422 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 423 | }
|
---|
| 424 |
|
---|
| 425 | //Now that we got a prefix, check and see if it's supported by this repository
|
---|
| 426 | String prefix_value = (String)param_map.get(OAIXML.METADATA_PREFIX);
|
---|
| 427 | if (containsMetadataPrefix(prefix_value) == false) {
|
---|
| 428 | logger.error("requested prefix is not found in OAIConfig.xml");
|
---|
| 429 | return getMessage(OAIXML.createErrorElement(OAIXML.CANNOT_DISSEMINATE_FORMAT, ""));
|
---|
| 430 | }
|
---|
| 431 |
|
---|
[14211] | 432 | //Now that all validation has been done, I hope, we can send request to the message router
|
---|
| 433 | Element result = null;
|
---|
| 434 | String verb = req.getAttribute(OAIXML.TO);
|
---|
| 435 | NodeList param_list = req.getElementsByTagName(OAIXML.PARAM);
|
---|
| 436 | ArrayList retain_param_list = new ArrayList();
|
---|
| 437 | for (int j=0; j<param_list.getLength(); j++) {
|
---|
| 438 | Element e = OAIXML.duplicateElement(msg.getOwnerDocument(), (Element)param_list.item(j), true);
|
---|
| 439 | retain_param_list.add(e);
|
---|
| 440 | }
|
---|
| 441 |
|
---|
| 442 | //re-organize the request element
|
---|
| 443 | // reset the 'to' attribute
|
---|
| 444 | if (request_set == false) {
|
---|
| 445 | logger.info("requesting identifiers of all collections");
|
---|
| 446 | for(int i=0; i<oai_coll_size; i++) {
|
---|
| 447 | if(req == null) {
|
---|
| 448 | req = msg.getOwnerDocument().createElement(GSXML.REQUEST_ELEM);
|
---|
| 449 | msg.appendChild(req);
|
---|
| 450 | for (int j=0; j<retain_param_list.size(); j++) {
|
---|
| 451 | req.appendChild((Element)retain_param_list.get(j));
|
---|
| 452 | }
|
---|
| 453 | }
|
---|
| 454 | String full_name = ((Element)oai_coll.item(i)).getAttribute(OAIXML.NAME);
|
---|
| 455 | coll_name = full_name.substring(full_name.indexOf(":") + 1);
|
---|
| 456 | req.setAttribute(OAIXML.TO, coll_name + "/" + verb);
|
---|
[16688] | 457 | Node n = mr.process(msg);
|
---|
| 458 | Element e = converter.nodeToElement(n);
|
---|
[14211] | 459 | result = collectAll(result, e, verb, OAIXML.HEADER);
|
---|
| 460 |
|
---|
| 461 | //clear the content of the old request element
|
---|
| 462 | msg.removeChild(req);
|
---|
| 463 | req = null;
|
---|
| 464 | }
|
---|
| 465 | } else {
|
---|
| 466 | req.setAttribute(OAIXML.TO, coll_name + "/" + verb);
|
---|
[16688] | 467 | Node result_node = mr.process(msg);
|
---|
| 468 | result = converter.nodeToElement(result_node);
|
---|
[14211] | 469 | }
|
---|
| 470 |
|
---|
| 471 | if (result == null) {
|
---|
| 472 | logger.info("message router returns null");
|
---|
| 473 | return getMessage(OAIXML.createErrorElement("Internal service returns null", ""));
|
---|
| 474 | }
|
---|
| 475 | Element res = (Element)GSXML.getChildByTagName(result, OAIXML.RESPONSE);
|
---|
| 476 | if(res == null) {
|
---|
| 477 | logger.info("response element in xml_result is null");
|
---|
| 478 | return getMessage(OAIXML.createErrorElement("Internal service returns null", ""));
|
---|
| 479 | }
|
---|
| 480 | NodeList header_list = res.getElementsByTagName(OAIXML.HEADER);
|
---|
| 481 | int num_headers = header_list.getLength();
|
---|
| 482 | if(num_headers == 0) {
|
---|
| 483 | logger.info("message router returns 0 headers.");
|
---|
| 484 | return getMessage(OAIXML.createErrorElement(OAIXML.NO_RECORDS_MATCH, ""));
|
---|
| 485 | }
|
---|
| 486 |
|
---|
| 487 | //The request coming in does not contain a token, but we have to check the resume_after value and see if we need to issue a resumption token and
|
---|
| 488 | // save the token as well.
|
---|
| 489 | if (token.equals("") == true) {
|
---|
| 490 | if(resume_after < 0 || num_headers <= resume_after) {
|
---|
| 491 | //send the whole list of records
|
---|
| 492 | return result;
|
---|
| 493 | }
|
---|
| 494 |
|
---|
| 495 | //append required number of records (may be a complete or incomplete list)
|
---|
| 496 | getRecords(list_identifiers, header_list, 0, resume_after);
|
---|
| 497 | //An incomplete list is sent; append a resumptionToken element
|
---|
| 498 | Element token_elem = createResumptionTokenElement(num_headers, 0, resume_after, true);
|
---|
| 499 | //store this token
|
---|
| 500 | OAIXML.addToken(token_elem);
|
---|
| 501 |
|
---|
| 502 | list_identifiers.appendChild(token_elem);
|
---|
| 503 | return getMessage(list_identifiers);
|
---|
| 504 | }
|
---|
| 505 |
|
---|
| 506 | if (token.equals("") == false) {
|
---|
| 507 | //get an appropriate number of records (partial list) according to the token
|
---|
| 508 | //take out the cursor value, which is the size of previously sent list
|
---|
| 509 | int index = token.indexOf(":");
|
---|
| 510 | int cursor = Integer.parseInt(token.substring(index + 1));
|
---|
| 511 | Element token_elem = null;
|
---|
| 512 |
|
---|
| 513 | // are we sending the final part of a complete list?
|
---|
| 514 | if(cursor + resume_after >= num_headers) {
|
---|
| 515 | //Yes, we are.
|
---|
| 516 | //append required records to list_records (list is complete)
|
---|
| 517 | getRecords(list_identifiers, header_list, cursor, num_headers);
|
---|
| 518 | //An incomplete list is sent; append a resumptionToken element
|
---|
| 519 | token_elem = createResumptionTokenElement(num_headers, cursor, -1, false);
|
---|
| 520 | list_identifiers.appendChild(token_elem);
|
---|
| 521 | } else {
|
---|
| 522 | //No, we are not.
|
---|
| 523 | //append required records to list_records (list is incomplete)
|
---|
| 524 | getRecords(list_identifiers, header_list, cursor, cursor + resume_after);
|
---|
| 525 | token_elem = createResumptionTokenElement(num_headers, cursor, cursor + resume_after, true);
|
---|
| 526 | //store this token
|
---|
| 527 | OAIXML.addToken(token_elem);
|
---|
| 528 | list_identifiers.appendChild(token_elem);
|
---|
| 529 | }
|
---|
| 530 |
|
---|
| 531 | return getMessage(list_identifiers);
|
---|
| 532 | }//end of if(!token.equals(""))
|
---|
| 533 |
|
---|
| 534 | return result;
|
---|
| 535 | }
|
---|
| 536 | private Element doListRecords(Element msg){
|
---|
| 537 | logger.info("");
|
---|
| 538 | // option: from, until, set, metadataPrefix, and resumptionToken
|
---|
| 539 | // exceptions: badArgument, badResumptionToken, cannotDisseminateFormat, noRecordMatch, and noSetHierarchy
|
---|
| 540 | HashSet valid_strs = new HashSet();
|
---|
| 541 | valid_strs.add(OAIXML.FROM);
|
---|
| 542 | valid_strs.add(OAIXML.UNTIL);
|
---|
| 543 | valid_strs.add(OAIXML.SET);
|
---|
| 544 | valid_strs.add(OAIXML.METADATA_PREFIX);
|
---|
| 545 | valid_strs.add(OAIXML.RESUMPTION_TOKEN);
|
---|
| 546 |
|
---|
| 547 | Element list_records = OAIXML.createElement(OAIXML.LIST_RECORDS);
|
---|
| 548 | Element req = (Element)GSXML.getChildByTagName(msg, GSXML.REQUEST_ELEM);
|
---|
| 549 | if (req == null) { logger.error("req is null"); return null; }
|
---|
| 550 | NodeList params = GSXML.getChildrenByTagName(req, OAIXML.PARAM);
|
---|
| 551 |
|
---|
| 552 | String coll_name = "";
|
---|
| 553 | String token = "";
|
---|
| 554 |
|
---|
| 555 | if(params.getLength() == 0) {
|
---|
| 556 | logger.error("must at least have the metadataPrefix parameter, can't be none");
|
---|
| 557 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 558 | }
|
---|
| 559 |
|
---|
| 560 | HashMap param_map = OAIXML.getParamMap(params);
|
---|
[23913] | 561 | if (!isValidParam(param_map, valid_strs)) {
|
---|
| 562 | logger.error("One of the params is invalid");
|
---|
| 563 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 564 | }
|
---|
| 565 | // param keys are valid, but if there are any date params, check they're of the right format
|
---|
| 566 | String from = (String)param_map.get(OAIXML.FROM);
|
---|
| 567 | if(from != null) {
|
---|
| 568 | Date from_date = OAIXML.getDate(from);
|
---|
| 569 | if(from_date == null) {
|
---|
| 570 | logger.error("invalid date: " + from);
|
---|
| 571 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 572 | }
|
---|
| 573 | }
|
---|
| 574 | String until = (String)param_map.get(OAIXML.UNTIL);
|
---|
| 575 | Date until_date = null;
|
---|
| 576 | if(until != null) {
|
---|
| 577 | until_date = OAIXML.getDate(until);
|
---|
| 578 | if(until_date == null) {
|
---|
| 579 | logger.error("invalid date: " + until);
|
---|
| 580 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 581 | }
|
---|
| 582 | }
|
---|
| 583 | if(from != null && until != null) { // check they are of the same date-time format (granularity)
|
---|
| 584 | if(from.length() != until.length()) {
|
---|
| 585 | logger.error("The request has different granularities (date-time formats) for the From and Until date parameters.");
|
---|
| 586 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 587 | }
|
---|
| 588 | }
|
---|
[14211] | 589 |
|
---|
| 590 | //ask the message router for a list of oai collections
|
---|
| 591 | NodeList oai_coll = getOAICollectionList();
|
---|
| 592 | int oai_coll_size = oai_coll.getLength();
|
---|
| 593 | if (oai_coll_size == 0) {
|
---|
| 594 | logger.info("returned oai collection list is empty");
|
---|
| 595 | return getMessage(OAIXML.createErrorElement(OAIXML.NO_RECORDS_MATCH, ""));
|
---|
| 596 | }
|
---|
| 597 |
|
---|
| 598 | //Now we check if the optional argument 'set' has been specified in the params; if so,
|
---|
| 599 | //whether the specified setSpec is supported by this repository
|
---|
| 600 | boolean request_set = param_map.containsKey(OAIXML.SET);
|
---|
| 601 | if(request_set == true) {
|
---|
| 602 | boolean set_supported = false;
|
---|
| 603 | String set_spec_str = (String)param_map.get(OAIXML.SET);
|
---|
| 604 | // get the collection name
|
---|
| 605 | //if setSpec is supported by this repository, it must be in the form: site_name:coll_name
|
---|
| 606 | String[] strs = splitSetSpec(set_spec_str);
|
---|
| 607 | // name_of_site = strs[0];
|
---|
| 608 | coll_name = strs[1];
|
---|
| 609 | //logger.info("param contains set: "+coll_name);
|
---|
| 610 |
|
---|
| 611 | for(int i=0; i<oai_coll_size; i++) {
|
---|
| 612 | if(set_spec_str.equals(((Element)oai_coll.item(i)).getAttribute(OAIXML.NAME))) {
|
---|
| 613 | set_supported = true;
|
---|
| 614 | }
|
---|
| 615 | }
|
---|
| 616 | if(set_supported == false) {
|
---|
| 617 | logger.error("requested set is not found in this repository");
|
---|
| 618 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 619 | }
|
---|
| 620 | }
|
---|
| 621 |
|
---|
| 622 | //Is there a resumptionToken included which is requesting an incomplete list?
|
---|
| 623 | if(param_map.containsKey(OAIXML.RESUMPTION_TOKEN)) {
|
---|
| 624 | // validate resumptionToken
|
---|
| 625 | //if (the token value is not found in the token xml file) {
|
---|
| 626 | // return getMessage(OAIXML.createErrorElement(OAIXML.BAD_RESUMPTION_TOKEN, ""));
|
---|
| 627 | //} else {
|
---|
| 628 | // use the request to get a complete list of records from the message router
|
---|
| 629 | // and issue the subsequent part of that complete list according to the token.
|
---|
| 630 | // store a new token if necessary.
|
---|
| 631 | //}
|
---|
| 632 | token = (String)param_map.get(OAIXML.RESUMPTION_TOKEN);
|
---|
[24440] | 633 | logger.info("has resumptionToken: " + token);
|
---|
[14211] | 634 | if(OAIXML.containsToken(token) == false) {
|
---|
| 635 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_RESUMPTION_TOKEN, ""));
|
---|
| 636 | }
|
---|
| 637 | }
|
---|
[23913] | 638 |
|
---|
| 639 | // Moved the additional custom test that mandates the metadataPrefix here, since official
|
---|
| 640 | // errors should be caught first, so that their error responses can be sent off first
|
---|
| 641 | // such that GS2's oaiserver will validate properly.
|
---|
| 642 | if (!param_map.containsKey(OAIXML.METADATA_PREFIX)) {
|
---|
[24440] | 643 | if(!token.equals("")) { // resumptiontoken
|
---|
| 644 | int lastIndex = token.lastIndexOf(":");
|
---|
| 645 | if(lastIndex != token.indexOf(":")) { // if a meta_prefix is suffixed to the usual token,
|
---|
| 646 | // put that in the map and remove it from the end of the stored token
|
---|
| 647 | String meta_prefix = token.substring(lastIndex+1);
|
---|
| 648 | param_map.put(OAIXML.METADATA_PREFIX, meta_prefix);
|
---|
| 649 | token = token.substring(0, lastIndex);
|
---|
| 650 | param_map.put(OAIXML.RESUMPTION_TOKEN, token);
|
---|
| 651 |
|
---|
| 652 | // Add to request <param name="metadataPrefix" value="oai_dc"/>
|
---|
| 653 | // need to add metaprefix as param to request, else a request
|
---|
| 654 | // for subsequent records when working with resumption tokens will fail
|
---|
| 655 | Element paramEl = req.getOwnerDocument().createElement(OAIXML.PARAM);
|
---|
| 656 | paramEl.setAttribute(OAIXML.NAME, OAIXML.METADATA_PREFIX);
|
---|
| 657 | paramEl.setAttribute(OAIXML.VALUE, meta_prefix);
|
---|
| 658 | req.appendChild(paramEl);
|
---|
| 659 | }
|
---|
| 660 | } else { // no metadata_prefix
|
---|
| 661 |
|
---|
| 662 | // it must have a metadataPrefix
|
---|
| 663 | /** Here I disagree with the OAI specification: even if a resumptionToken is
|
---|
| 664 | * included in the request, the metadataPrefix is a must argument. Otherwise
|
---|
| 665 | * how would we know what metadataPrefix the harvester requested in his last request?
|
---|
| 666 | */
|
---|
| 667 | logger.error("no metadataPrefix");
|
---|
| 668 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 669 | }
|
---|
[23913] | 670 | }
|
---|
[24440] | 671 |
|
---|
[23913] | 672 | //Now that we got a prefix, check and see if it's supported by this repository
|
---|
| 673 | String prefix_value = (String)param_map.get(OAIXML.METADATA_PREFIX);
|
---|
| 674 | if (containsMetadataPrefix(prefix_value) == false) {
|
---|
[24440] | 675 | logger.error("requested prefix is not found in OAIConfig.xml");
|
---|
| 676 | return getMessage(OAIXML.createErrorElement(OAIXML.CANNOT_DISSEMINATE_FORMAT, ""));
|
---|
[23913] | 677 | }
|
---|
[24440] | 678 |
|
---|
[23913] | 679 |
|
---|
[14211] | 680 | //Now that all validation has been done, I hope, we can send request to the message router
|
---|
| 681 | Element result = null;
|
---|
| 682 | String verb = req.getAttribute(OAIXML.TO);
|
---|
| 683 | NodeList param_list = req.getElementsByTagName(OAIXML.PARAM);
|
---|
| 684 | ArrayList retain_param_list = new ArrayList();
|
---|
| 685 | for (int j=0; j<param_list.getLength(); j++) {
|
---|
| 686 | Element e = OAIXML.duplicateElement(msg.getOwnerDocument(), (Element)param_list.item(j), true);
|
---|
| 687 | retain_param_list.add(e);
|
---|
| 688 | }
|
---|
| 689 |
|
---|
| 690 | //re-organize the request element
|
---|
| 691 | // reset the 'to' attribute
|
---|
| 692 | if (request_set == false) {
|
---|
| 693 | //coll_name could be "", which means it's requesting all records of all collections
|
---|
| 694 | //we send a request to each collection asking for its records
|
---|
| 695 | for(int i=0; i<oai_coll_size; i++) {
|
---|
| 696 | if(req == null) {
|
---|
| 697 | req = msg.getOwnerDocument().createElement(GSXML.REQUEST_ELEM);
|
---|
| 698 | msg.appendChild(req);
|
---|
| 699 | for (int j=0; j<retain_param_list.size(); j++) {
|
---|
| 700 | req.appendChild((Element)retain_param_list.get(j));
|
---|
| 701 | }
|
---|
| 702 | }
|
---|
| 703 | String full_name = ((Element)oai_coll.item(i)).getAttribute(OAIXML.NAME);
|
---|
| 704 | coll_name = full_name.substring(full_name.indexOf(":") + 1);
|
---|
| 705 | req.setAttribute(OAIXML.TO, coll_name + "/" + verb);
|
---|
| 706 | //logger.info(GSXML.xmlNodeToString(req));
|
---|
[16688] | 707 | Node n = mr.process(msg);
|
---|
| 708 | Element e = converter.nodeToElement(n);
|
---|
[14211] | 709 | result = collectAll(result, e, verb, OAIXML.RECORD);
|
---|
[24440] | 710 |
|
---|
[14211] | 711 | //clear the content of the old request element
|
---|
| 712 | msg.removeChild(req);
|
---|
| 713 | req = null;
|
---|
| 714 | }
|
---|
| 715 | } else {
|
---|
| 716 | req.setAttribute(OAIXML.TO, coll_name + "/" + verb);
|
---|
[16688] | 717 |
|
---|
| 718 | Node result_node = mr.process(msg);
|
---|
| 719 | result = converter.nodeToElement(result_node);
|
---|
[14211] | 720 | }
|
---|
| 721 |
|
---|
| 722 | if (result == null) {
|
---|
| 723 | logger.info("message router returns null");
|
---|
| 724 | return getMessage(OAIXML.createErrorElement("Internal service returns null", ""));
|
---|
| 725 | }
|
---|
| 726 | Element res = (Element)GSXML.getChildByTagName(result, OAIXML.RESPONSE);
|
---|
| 727 | if(res == null) {
|
---|
| 728 | logger.info("response element in xml_result is null");
|
---|
| 729 | return getMessage(OAIXML.createErrorElement("Internal service returns null", ""));
|
---|
| 730 | }
|
---|
| 731 | NodeList record_list = res.getElementsByTagName(OAIXML.RECORD);
|
---|
| 732 | int num_records = record_list.getLength();
|
---|
| 733 | if(num_records == 0) {
|
---|
| 734 | logger.info("message router returns 0 records.");
|
---|
| 735 | return getMessage(OAIXML.createErrorElement(OAIXML.NO_RECORDS_MATCH, ""));
|
---|
| 736 | }
|
---|
| 737 |
|
---|
| 738 | //The request coming in does not contain a token, but we have to check the resume_after value and see if we need to issue a resumption token and
|
---|
| 739 | // save the token as well.
|
---|
| 740 | if (token.equals("") == true) {
|
---|
| 741 | if(resume_after < 0 || num_records <= resume_after) {
|
---|
| 742 | //send the whole list of records
|
---|
| 743 | return result;
|
---|
| 744 | }
|
---|
| 745 |
|
---|
| 746 | //append required number of records (may be a complete or incomplete list)
|
---|
| 747 | getRecords(list_records, record_list, 0, resume_after);
|
---|
| 748 | //An incomplete list is sent; append a resumptionToken element
|
---|
[24440] | 749 | Element token_elem = createResumptionTokenElement(num_records, 0, resume_after, true, (String)param_map.get(OAIXML.METADATA_PREFIX));
|
---|
[14211] | 750 | //store this token
|
---|
| 751 | OAIXML.addToken(token_elem);
|
---|
| 752 |
|
---|
| 753 | list_records.appendChild(token_elem);
|
---|
| 754 | return getMessage(list_records);
|
---|
| 755 | }
|
---|
| 756 |
|
---|
| 757 | if (token.equals("") == false) {
|
---|
| 758 | //get an appropriate number of records (partial list) according to the token
|
---|
| 759 | //take out the cursor value, which is the size of previously sent list
|
---|
| 760 | int index = token.indexOf(":");
|
---|
| 761 | int cursor = Integer.parseInt(token.substring(index + 1));
|
---|
| 762 | Element token_elem = null;
|
---|
| 763 |
|
---|
| 764 | // are we sending the final part of a complete list?
|
---|
| 765 | if(cursor + resume_after >= num_records) {
|
---|
| 766 | //Yes, we are.
|
---|
| 767 | //append required records to list_records (list is complete)
|
---|
| 768 | getRecords(list_records, record_list, cursor, num_records);
|
---|
| 769 | //An incomplete list is sent; append a resumptionToken element
|
---|
[24440] | 770 | token_elem = createResumptionTokenElement(num_records, cursor, -1, false, (String)param_map.get(OAIXML.METADATA_PREFIX));
|
---|
| 771 | list_records.appendChild(token_elem);
|
---|
| 772 |
|
---|
[14211] | 773 | } else {
|
---|
| 774 | //No, we are not.
|
---|
| 775 | //append required records to list_records (list is incomplete)
|
---|
| 776 | getRecords(list_records, record_list, cursor, cursor + resume_after);
|
---|
[24440] | 777 | token_elem = createResumptionTokenElement(num_records, cursor, cursor + resume_after, true, (String)param_map.get(OAIXML.METADATA_PREFIX));
|
---|
[14211] | 778 | //store this token
|
---|
| 779 | OAIXML.addToken(token_elem);
|
---|
| 780 | list_records.appendChild(token_elem);
|
---|
| 781 | }
|
---|
| 782 |
|
---|
| 783 | return getMessage(list_records);
|
---|
| 784 | }//end of if(!token.equals(""))
|
---|
| 785 |
|
---|
| 786 | return result;//a backup return
|
---|
| 787 | }
|
---|
| 788 | // method exclusively used by doListRecords/doListIdentifiers
|
---|
| 789 | private void getRecords(Element verb_elem, NodeList list, int start_point, int end_point) {
|
---|
| 790 | for (int i=start_point; i<end_point; i++) {
|
---|
| 791 | verb_elem.appendChild(verb_elem.getOwnerDocument().importNode(list.item(i), true));
|
---|
| 792 | }
|
---|
| 793 | }
|
---|
| 794 | private Element collectAll(Element result, Element msg, String verb, String elem_name) {
|
---|
| 795 | if(result == null) {
|
---|
| 796 | //in the first round, result is null
|
---|
| 797 | return msg;
|
---|
| 798 | }
|
---|
| 799 | Element res_in_result = (Element)GSXML.getChildByTagName(result, OAIXML.RESPONSE);
|
---|
[23901] | 800 | if(res_in_result == null) { // return the results of all other collections accumulated so far
|
---|
| 801 | return msg;
|
---|
| 802 | }
|
---|
[14211] | 803 | Element verb_elem = (Element)GSXML.getChildByTagName(res_in_result, verb);
|
---|
| 804 | if(msg == null) {
|
---|
| 805 | return result;
|
---|
| 806 | }
|
---|
| 807 |
|
---|
| 808 | //e.g., get all <record> elements from the returned message. There may be none of
|
---|
| 809 | //such element, for example, the collection service returned an error message
|
---|
| 810 | NodeList elem_list = msg.getElementsByTagName(elem_name);
|
---|
| 811 |
|
---|
| 812 | for (int i=0; i<elem_list.getLength(); i++) {
|
---|
| 813 | verb_elem.appendChild(res_in_result.getOwnerDocument().importNode(elem_list.item(i), true));
|
---|
| 814 | }
|
---|
| 815 | return result;
|
---|
| 816 | }
|
---|
| 817 | /** there are three possible exception conditions: bad argument, idDoesNotExist, and noMetadataFormats.
|
---|
| 818 | * The first one is handled here, and the last two are processed by OAIPMH.
|
---|
| 819 | */
|
---|
| 820 | private Element doListMetadataFormats(Element msg) {
|
---|
| 821 | //if the verb is ListMetadataFormats, there could be only one parameter: identifier
|
---|
| 822 | //, or there is no parameter; otherwise it is an error
|
---|
| 823 | //logger.info("" + this.converter.getString(msg));
|
---|
| 824 |
|
---|
| 825 | Element list_metadata_formats = OAIXML.createElement(OAIXML.LIST_METADATA_FORMATS);
|
---|
| 826 |
|
---|
| 827 | Element req = (Element)GSXML.getChildByTagName(msg, GSXML.REQUEST_ELEM);
|
---|
| 828 | if (req == null) { logger.error(""); return null; }
|
---|
| 829 | NodeList params = GSXML.getChildrenByTagName(req, OAIXML.PARAM);
|
---|
| 830 | Element param = null;
|
---|
| 831 | if(params.getLength() == 0) {
|
---|
| 832 | //this is requesting metadata formats for the whole repository
|
---|
| 833 | //read the oaiConfig.xml file, return the metadata formats specified there.
|
---|
| 834 | Element oai_config = OAIXML.getOAIConfigXML();
|
---|
| 835 | if (oai_config == null) {
|
---|
| 836 | return getMessage(OAIXML.createErrorElement(OAIXML.ERROR, OAIXML.SERVICE_UNAVAILABLE));
|
---|
| 837 | } else {
|
---|
| 838 | Element format_list = (Element)GSXML.getChildByTagName(oai_config, OAIXML.LIST_METADATA_FORMATS);
|
---|
| 839 | if(format_list == null) {
|
---|
| 840 | logger.error("OAIConfig.xml must contain the supported metadata formats");
|
---|
| 841 | return getMessage(list_metadata_formats);
|
---|
| 842 | }
|
---|
| 843 | NodeList formats = format_list.getElementsByTagName(OAIXML.METADATA_FORMAT);
|
---|
| 844 | for(int i=0; i<formats.getLength(); i++) {
|
---|
| 845 | Element meta_fmt = OAIXML.createElement(OAIXML.METADATA_FORMAT);
|
---|
| 846 | Element first_meta_format = (Element)formats.item(i);
|
---|
| 847 | //the element also contains mappings, but we don't want them
|
---|
| 848 | meta_fmt.appendChild(meta_fmt.getOwnerDocument().importNode(GSXML.getChildByTagName(first_meta_format, OAIXML.METADATA_PREFIX), true));
|
---|
| 849 | meta_fmt.appendChild(meta_fmt.getOwnerDocument().importNode(GSXML.getChildByTagName(first_meta_format, OAIXML.SCHEMA), true));
|
---|
| 850 | meta_fmt.appendChild(meta_fmt.getOwnerDocument().importNode(GSXML.getChildByTagName(first_meta_format, OAIXML.METADATA_NAMESPACE), true));
|
---|
| 851 | list_metadata_formats.appendChild(meta_fmt);
|
---|
| 852 | }
|
---|
| 853 | return getMessage(list_metadata_formats);
|
---|
| 854 | }
|
---|
| 855 |
|
---|
| 856 | } else if (params.getLength() > 1) {
|
---|
| 857 | //Bad argument. Can't be more than one parameters for ListMetadataFormats verb
|
---|
| 858 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 859 | } else {
|
---|
| 860 | // This is a request for the metadata of a particular item with an identifier
|
---|
| 861 | /**the request xml is in the form: <request>
|
---|
| 862 | * <param name=.../>
|
---|
| 863 | * </request>
|
---|
| 864 | *And there is a param element and one element only. (No paramList element in between).
|
---|
| 865 | */
|
---|
| 866 | param = (Element)params.item(0);
|
---|
| 867 | String param_name = param.getAttribute(OAIXML.NAME);
|
---|
| 868 | String identifier = "";
|
---|
| 869 | if (!param_name.equals(OAIXML.IDENTIFIER)) {
|
---|
| 870 | //Bad argument
|
---|
| 871 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 872 | } else {
|
---|
| 873 | identifier = param.getAttribute(OAIXML.VALUE);
|
---|
| 874 | // the identifier is in the form: <site_name>:<coll_name>:<OID>
|
---|
| 875 | // so it must contain at least two ':' characters
|
---|
| 876 | String[] strs = identifier.split(":");
|
---|
| 877 | if(strs == null || strs.length < 3) {
|
---|
| 878 | // the OID may also contain ':'
|
---|
| 879 | logger.error("identifier is not in the form site:coll:id" + identifier);
|
---|
| 880 | return getMessage(OAIXML.createErrorElement(OAIXML.ID_DOES_NOT_EXIST, ""));
|
---|
| 881 | }
|
---|
| 882 |
|
---|
| 883 | // send request to message router
|
---|
| 884 | // get the names
|
---|
| 885 | strs = splitNames(identifier);
|
---|
[23913] | 886 | if(strs == null || strs.length < 3) {
|
---|
| 887 | logger.error("identifier is not in the form site:coll:id" + identifier);
|
---|
| 888 | return getMessage(OAIXML.createErrorElement(OAIXML.ID_DOES_NOT_EXIST, ""));
|
---|
| 889 | }
|
---|
[14211] | 890 | String name_of_site = strs[0];
|
---|
| 891 | String coll_name = strs[1];
|
---|
| 892 | String oid = strs[2];
|
---|
| 893 |
|
---|
| 894 | //re-organize the request element
|
---|
| 895 | // reset the 'to' attribute
|
---|
| 896 | String verb = req.getAttribute(OAIXML.TO);
|
---|
| 897 | req.setAttribute(OAIXML.TO, coll_name + "/" + verb);
|
---|
| 898 | // reset the identifier element
|
---|
| 899 | param.setAttribute(OAIXML.NAME, OAIXML.OID);
|
---|
| 900 | param.setAttribute(OAIXML.VALUE, oid);
|
---|
| 901 |
|
---|
| 902 | //Now send the request to the message router to process
|
---|
[16688] | 903 | Node result_node = mr.process(msg);
|
---|
| 904 | return converter.nodeToElement(result_node);
|
---|
[14211] | 905 | }
|
---|
| 906 | }
|
---|
| 907 |
|
---|
| 908 | }
|
---|
| 909 | private void appendParam(Element req, String name, String value) {
|
---|
| 910 | Element param = req.getOwnerDocument().createElement(OAIXML.PARAM);
|
---|
| 911 | param.setAttribute(OAIXML.NAME, name);
|
---|
| 912 | param.setAttribute(OAIXML.VALUE, value);
|
---|
| 913 | req.appendChild(param);
|
---|
| 914 | }
|
---|
| 915 | private void copyElement(Element identify, String tag_name) {
|
---|
| 916 | Element from_repository_name = (Element)GSXML.getChildByTagName(oai_config, tag_name);
|
---|
| 917 | if(from_repository_name != null) {
|
---|
| 918 | Element this_repository_name = OAIXML.createElement(tag_name);
|
---|
| 919 | GSXML.setNodeText(this_repository_name, GSXML.getNodeText(from_repository_name));
|
---|
| 920 | identify.appendChild(this_repository_name);
|
---|
| 921 | }
|
---|
| 922 | }
|
---|
| 923 | private Element doIdentify() {
|
---|
| 924 | //The validation for this verb has been done in OAIServer.validate(). So no bother here.
|
---|
| 925 | logger.info("");
|
---|
| 926 |
|
---|
| 927 | Element identify = OAIXML.createElement(OAIXML.IDENTIFY);
|
---|
| 928 | //do the repository name
|
---|
| 929 | copyElement(identify, OAIXML.REPOSITORY_NAME);
|
---|
| 930 | //do the baseurl
|
---|
| 931 | copyElement(identify, OAIXML.BASE_URL);
|
---|
| 932 | //do the protocol version
|
---|
| 933 | copyElement(identify, OAIXML.PROTOCOL_VERSION);
|
---|
[23944] | 934 |
|
---|
[14211] | 935 | //There can be more than one admin email according to the OAI specification
|
---|
| 936 | NodeList admin_emails = GSXML.getChildrenByTagName(oai_config, OAIXML.ADMIN_EMAIL);
|
---|
| 937 | int num_admin = 0;
|
---|
| 938 | Element from_admin_email = null;
|
---|
| 939 | if (admin_emails != null) {
|
---|
| 940 | num_admin = admin_emails.getLength();
|
---|
| 941 | }
|
---|
| 942 | for (int i=0; i<num_admin; i++) {
|
---|
| 943 | copyElement(identify, OAIXML.ADMIN_EMAIL);
|
---|
| 944 | }
|
---|
| 945 |
|
---|
| 946 | //do the earliestDatestamp
|
---|
| 947 | //send request to mr to search through the earliest datestamp amongst all oai collections in the repository.
|
---|
| 948 | //ask the message router for a list of oai collections
|
---|
| 949 | NodeList oai_coll = getOAICollectionList();
|
---|
[23913] | 950 | long earliestDatestamp = getEarliestDateStamp(oai_coll);
|
---|
| 951 | String earliestDatestamp_str = OAIXML.getTime(earliestDatestamp);
|
---|
[14211] | 952 | Element earliestDatestamp_elem = OAIXML.createElement(OAIXML.EARLIEST_DATESTAMP);
|
---|
| 953 | GSXML.setNodeText(earliestDatestamp_elem, earliestDatestamp_str);
|
---|
| 954 | identify.appendChild(earliestDatestamp_elem);
|
---|
[23944] | 955 |
|
---|
| 956 | //do the deletedRecord
|
---|
| 957 | copyElement(identify, OAIXML.DELETED_RECORD);
|
---|
| 958 | //do the granularity
|
---|
| 959 | copyElement(identify, OAIXML.GRANULARITY);
|
---|
[14211] | 960 |
|
---|
| 961 | return getMessage(identify);
|
---|
| 962 | }
|
---|
| 963 | //split setSpec (site_name:coll_name) into an array of strings
|
---|
| 964 | //It has already been checked that the set_spec contains at least one ':'
|
---|
| 965 | private String[] splitSetSpec(String set_spec) {
|
---|
| 966 | logger.info(set_spec);
|
---|
| 967 | String[] strs = new String[2];
|
---|
| 968 | int colon_index = set_spec.indexOf(":");
|
---|
| 969 | strs[0] = set_spec.substring(0, colon_index);
|
---|
| 970 | strs[1] = set_spec.substring(colon_index + 1);
|
---|
| 971 | return strs;
|
---|
| 972 | }
|
---|
| 973 | /** split the identifier into <site + collection + OID> as an array
|
---|
| 974 | It has already been checked that the 'identifier' contains at least two ':'
|
---|
| 975 | */
|
---|
| 976 | private String[] splitNames(String identifier) {
|
---|
| 977 | logger.info(identifier);
|
---|
| 978 | String [] strs = new String[3];
|
---|
| 979 | int first_colon = identifier.indexOf(":");
|
---|
[23913] | 980 | if(first_colon == -1) {
|
---|
| 981 | return null;
|
---|
| 982 | }
|
---|
[14211] | 983 | strs[0] = identifier.substring(0, first_colon);
|
---|
| 984 |
|
---|
| 985 | String sr = identifier.substring(first_colon + 1);
|
---|
| 986 | int second_colon = sr.indexOf(":");
|
---|
| 987 | //logger.error(first_colon + " " + second_colon);
|
---|
| 988 | strs[1] = sr.substring(0, second_colon);
|
---|
| 989 |
|
---|
| 990 | strs[2] = sr.substring(second_colon + 1);
|
---|
| 991 | return strs;
|
---|
| 992 | }
|
---|
| 993 | /** validate if the specified metadata prefix value is supported by the repository
|
---|
| 994 | * by checking it in the OAIConfig.xml
|
---|
| 995 | */
|
---|
| 996 | private boolean containsMetadataPrefix(String prefix_value) {
|
---|
| 997 | NodeList prefix_list = oai_config.getElementsByTagName(OAIXML.METADATA_PREFIX);
|
---|
| 998 |
|
---|
| 999 | for(int i=0; i<prefix_list.getLength(); i++) {
|
---|
| 1000 | if(prefix_value.equals(GSXML.getNodeText((Element)prefix_list.item(i)).trim() )) {
|
---|
| 1001 | return true;
|
---|
| 1002 | }
|
---|
| 1003 | }
|
---|
| 1004 | return false;
|
---|
| 1005 | }
|
---|
| 1006 | private Element doGetRecord(Element msg){
|
---|
| 1007 | logger.info("");
|
---|
| 1008 | /** arguments:
|
---|
| 1009 | identifier: required
|
---|
| 1010 | metadataPrefix: required
|
---|
| 1011 | * Exceptions: badArgument; cannotDisseminateFormat; idDoesNotExist
|
---|
| 1012 | */
|
---|
| 1013 | Element get_record = OAIXML.createElement(OAIXML.GET_RECORD);
|
---|
| 1014 |
|
---|
| 1015 | HashSet valid_strs = new HashSet();
|
---|
| 1016 | valid_strs.add(OAIXML.IDENTIFIER);
|
---|
| 1017 | valid_strs.add(OAIXML.METADATA_PREFIX);
|
---|
| 1018 |
|
---|
| 1019 | Element req = (Element)GSXML.getChildByTagName(msg, GSXML.REQUEST_ELEM);
|
---|
| 1020 | NodeList params = GSXML.getChildrenByTagName(req, OAIXML.PARAM);
|
---|
| 1021 | HashMap param_map = OAIXML.getParamMap(params);
|
---|
| 1022 |
|
---|
| 1023 | if(!isValidParam(param_map, valid_strs) ||
|
---|
| 1024 | params.getLength() == 0 ||
|
---|
| 1025 | param_map.containsKey(OAIXML.IDENTIFIER) == false ||
|
---|
| 1026 | param_map.containsKey(OAIXML.METADATA_PREFIX) == false ) {
|
---|
| 1027 | logger.error("must have the metadataPrefix/identifier parameter.");
|
---|
| 1028 | return getMessage(OAIXML.createErrorElement(OAIXML.BAD_ARGUMENT, ""));
|
---|
| 1029 | }
|
---|
| 1030 |
|
---|
| 1031 | String prefix = (String)param_map.get(OAIXML.METADATA_PREFIX);
|
---|
| 1032 | String identifier = (String)param_map.get(OAIXML.IDENTIFIER);
|
---|
| 1033 |
|
---|
| 1034 | // verify the metadata prefix
|
---|
| 1035 | if (containsMetadataPrefix(prefix) == false) {
|
---|
| 1036 | logger.error("requested prefix is not found in OAIConfig.xml");
|
---|
| 1037 | return getMessage(OAIXML.createErrorElement(OAIXML.CANNOT_DISSEMINATE_FORMAT, ""));
|
---|
| 1038 | }
|
---|
| 1039 |
|
---|
| 1040 | // get the names
|
---|
| 1041 | String[] strs = splitNames(identifier);
|
---|
| 1042 | if(strs == null || strs.length < 3) {
|
---|
| 1043 | logger.error("identifier is not in the form site:coll:id" + identifier);
|
---|
| 1044 | return getMessage(OAIXML.createErrorElement(OAIXML.ID_DOES_NOT_EXIST, ""));
|
---|
| 1045 | }
|
---|
| 1046 | String name_of_site = strs[0];
|
---|
| 1047 | String coll_name = strs[1];
|
---|
| 1048 | String oid = strs[2];
|
---|
| 1049 |
|
---|
| 1050 | //re-organize the request element
|
---|
| 1051 | // reset the 'to' attribute
|
---|
| 1052 | String verb = req.getAttribute(OAIXML.TO);
|
---|
| 1053 | req.setAttribute(OAIXML.TO, coll_name + "/" + verb);
|
---|
| 1054 | // reset the identifier element
|
---|
| 1055 | Element param = GSXML.getNamedElement(req, OAIXML.PARAM, OAIXML.NAME, OAIXML.IDENTIFIER);
|
---|
| 1056 | if (param != null) {
|
---|
| 1057 | param.setAttribute(OAIXML.NAME, OAIXML.OID);
|
---|
| 1058 | param.setAttribute(OAIXML.VALUE, oid);
|
---|
| 1059 | }
|
---|
| 1060 |
|
---|
| 1061 | //Now send the request to the message router to process
|
---|
[16688] | 1062 | Node result_node = mr.process(msg);
|
---|
| 1063 | return converter.nodeToElement(result_node);
|
---|
[14211] | 1064 | }
|
---|
[23913] | 1065 |
|
---|
| 1066 | // See OAIConfig.xml
|
---|
| 1067 | // dynamically works out what the earliestDateStamp is, since it varies by collection
|
---|
[23938] | 1068 | // returns this time in *milliseconds*.
|
---|
[23913] | 1069 | protected long getEarliestDateStamp(NodeList oai_coll) {
|
---|
| 1070 | //do the earliestDatestamp
|
---|
[23938] | 1071 | long earliestDatestamp = System.currentTimeMillis();
|
---|
[23913] | 1072 | int oai_coll_size = oai_coll.getLength();
|
---|
| 1073 | if (oai_coll_size == 0) {
|
---|
| 1074 | logger.info("returned oai collection list is empty. Setting repository earliestDatestamp to be 1970-01-01.");
|
---|
[23938] | 1075 | earliestDatestamp = 0;
|
---|
[23913] | 1076 | }
|
---|
[23938] | 1077 | // the earliestDatestamp is now stored as a metadata element in the collection's buildConfig.xml file
|
---|
| 1078 | // we get the earliestDatestamp among the collections
|
---|
[23913] | 1079 | for(int i=0; i<oai_coll_size; i++) {
|
---|
[23938] | 1080 | long coll_earliestDatestamp = Long.parseLong(((Element)oai_coll.item(i)).getAttribute(OAIXML.EARLIEST_DATESTAMP));
|
---|
| 1081 | earliestDatestamp = (earliestDatestamp > coll_earliestDatestamp)? coll_earliestDatestamp : earliestDatestamp;
|
---|
[23913] | 1082 | }
|
---|
[23938] | 1083 |
|
---|
| 1084 | return earliestDatestamp*1000; // converting from seconds to milliseconds
|
---|
[23913] | 1085 | }
|
---|
[14211] | 1086 | }
|
---|