source: main/trunk/greenstone3/src/java/org/greenstone/gsdl3/util/OAIXML.java@ 28846

Last change on this file since 28846 was 28846, checked in by kjdon, 10 years ago

indented file in preparation for major changes
File size: 30.0 KB
Line 
1/*
2 * OAIXML.java
3 * Copyright (C) 2008 New Zealand Digital Library, http://www.nzdl.org
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
18 */
19package org.greenstone.gsdl3.util;
20
21import org.greenstone.util.GlobalProperties;
22
23import org.w3c.dom.*;
24
25import java.io.*;
26import java.net.*;
27import java.util.*;
28import java.text.DateFormat;
29import java.text.SimpleDateFormat;
30import org.apache.xerces.parsers.*;
31import org.apache.xml.serialize.*;
32
33// SAX
34import org.xml.sax.XMLReader;
35import org.xml.sax.SAXException;
36import org.xml.sax.SAXParseException;
37import org.xml.sax.helpers.DefaultHandler;
38import org.xml.sax.InputSource;
39
40// JAXP
41import javax.xml.parsers.FactoryConfigurationError;
42import javax.xml.parsers.ParserConfigurationException;
43import javax.xml.parsers.SAXParser;
44import javax.xml.parsers.SAXParserFactory;
45
46// import file Logger.java
47import org.apache.log4j.*;
48
49/** these constants are used for the OAI service */
50public class OAIXML {
51
52 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.util.GSXML.class.getName());
53
54 // the leading keyword of oai protocol
55 public static final String VERB = "verb";
56
57 // six valid oai verbs
58 public static final String GET_RECORD = "GetRecord";
59 public static final String LIST_RECORDS = "ListRecords";
60 public static final String LIST_IDENTIFIERS = "ListIdentifiers";
61 public static final String LIST_SETS = "ListSets";
62 public static final String LIST_METADATA_FORMATS = "ListMetadataFormats";
63 public static final String IDENTIFY = "Identify";
64
65 // other valid oai parameters
66 public static final String OAI_METADATAFORMAT = "OAIMetadataFormat";
67 public static final String METADATA_NAMESPACE = "metadataNamespace";
68 public static final String OAI_DC = "oai_dc";
69 public static final String DC = "dc";
70 public static final String METADATA_PREFIX = "metadataPrefix";
71 public static final String FROM = "from";
72 public static final String UNTIL = "until";
73 public static final String SET = "set";
74 public static final String RESUMPTION_TOKEN = "resumptionToken";
75 public static final String RESUMPTION_TOKEN_EXPIRATION = "resumptionTokenExpiration";
76 public static final String IDENTIFIER = "identifier";
77
78 public static final String USE_STYLESHEET = "useOAIStylesheet";
79 public static final String STYLESHEET = "OAIStylesheet";
80
81 // words used to compose oai responses and read in OAIConfig.xml
82 public static final String ADMIN_EMAIL = "adminEmail";
83 public static final String BAD_ARGUMENT = "badArgument";
84 public static final String BAD_RESUMPTION_TOKEN = "badResumptionToken";
85 public static final String BAD_VERB = "badVerb";
86 public static final String BASE_URL = "baseURL";
87 public static final String CANNOT_DISSEMINATE_FORMAT = "cannotDisseminateFormat";
88 public static final String CODE = "code";
89 public static final String COLLECTION = "collection";
90 public static final String COLLECTION_LIST = "collectionList";
91 public static final String COMPLETE_LIST_SIZE = "completeListSize";
92 public static final String COMPRESSION = "compression";
93 public static final String CURSOR = "cursor";
94 public static final String DATESTAMP = "datestamp";
95 public static final String DELETED_RECORD = "deletedRecord";
96 public static final String DESCRIPTION = "description";
97 public static final String EARLIEST_DATESTAMP = "earliestDatestamp";
98 public static final String ERROR = "error";
99 public static final String EXPIRATION_DATE = "expirationDate";
100 public static final String GRANULARITY = "granularity";
101 public static final String GS3OAI = "GS3OAI";
102 public static final String GS_OAI_RESOURCE_URL = "gs.OAIResourceURL";
103 public static final String HAS_OAI = "hasOAI";
104 public static final String HEADER = "header";
105 public static final String ILLEGAL_OAI_VERB = "Illegal OAI verb";
106 public static final String INDEX_STEM = "indexStem";
107 public static final String INFO_METADATA = "Metadata"; // this has capital M
108 public static final String LASTMODIFIED = "lastmodified";
109 public static final String MAPPING = "mapping";
110 public static final String MAPPING_LIST = "mappingList";
111 public static final String MESSAGE = "message";
112 public static final String METADATA = "metadata";
113 public static final String METADATA_FORMAT = "metadataFormat";
114 public static final String NAME = "name";
115 public static final String NO_RECORDS_MATCH = "noRecordsMatch";
116 public static final String OAI = "OAI";
117 public static final String OAI_DASH_PMH = "OAI-PMH";
118 public static final String OAI_LASTMODIFIED = "oailastmodified";
119 public static final String OAIPMH = "OAIPMH";
120 public static final String OAI_RESUMPTION_TOKENS = "OAIResumptionTokens";
121 public static final String OAI_INFO = "oaiInfo";
122 public static final String OAI_SERVICE = "oaiService";
123 public static final String OAI_SET_LIST = "oaiSetList";
124 public static final String OAI_SERVICE_UNAVAILABLE = "OAI service unavailable";
125 public static final String OID = "OID";
126 public static final String PARAM = "param";
127 public static final String PARAM_LIST = "paramList";
128 public static final String PROTOCOL_VERSION = "protocolVersion";
129 public static final String RECORD = "record";
130 public static final String REQUEST = "request";
131 public static final String REPOSITORY_NAME = "repositoryName";
132 public static final String REPOSITORY_ID = "repositoryId";
133 public static final String RESPONSE = "response";
134 public static final String RESPONSE_DATE = "responseDate";
135 public static final String RESUME_AFTER = "resumeAfter";
136 public static final String SCHEMA = "schema";
137 public static final String SERVICE = "service";
138 public static final String SERVICE_UNAVAILABLE = "service unavailable";
139 public static final String SET_SPEC = "setSpec";
140 public static final String SET_NAME = "setName";
141 public static final String SET_DESCRIPTION = "setDescription";
142 public static final String SITE = "site";
143 public static final String TO = "to";
144 public static final String TYPE = "type";
145 public static final String VALUE = "value";
146
147 //Two error and exception conditions for the verb 'ListMetadataFormats'
148 public static final String ID_DOES_NOT_EXIST = "idDoesNotExist";
149 public static final String NO_METADATA_FORMATS = "noMetadataFormats";
150
151 // The node id in the collection database, which contains all the OIDs in the database
152 public static final String BROWSELIST = "browselist";
153
154 //system-dependent file separator, maybe '/' or '\'
155 public static final String FILE_SEPARATOR = File.separator;
156 public static final String OAI_VERSION1 = "1.0";
157 public static final String OAI_VERSION2 = "2.0";
158 /*************************above are final values****************************/
159
160 public static Element resumption_token_elem = null;
161 //used when saving the token file
162 public static File resumption_token_file = null;
163 //public static ArrayList token_list = new ArrayList();
164
165 //initialized in getOAIConfigXML()
166 public static Element oai_config_elem = null;
167
168 //stores the date format "yyyy-MM-ddTHH:mm:ssZ"
169 public static String granularity = "";
170
171 // http://www.openarchives.org/OAI/openarchivesprotocol.html#DatestampsRequests
172 // specifies that all repositories must support YYYY-MM-DD (yyyy-MM-dd in Java)
173 // this would be in addition to the other (optional) granularity of above that
174 // a repository may additionally choose to support.
175 public static final String default_granularity = "yyyy-MM-dd";
176
177 //this value is overriden in getOAIConfigXML()
178 public static long token_expiration = 7200;
179
180 /** which version of oai that this oaiserver supports; default is 2.0
181 * initialized in getOAIConfigXML()
182 */
183 public static String oai_version = "2.0";
184 public static String baseURL = "";
185
186 /**response owner document */
187 public static Document response_doc = new XMLConverter().newDOM();
188
189 public static String[] special_char = {"/", "?", "#", "=", "&", ":", ";", " ", "%", "+"};
190 public static String[] escape_sequence = {"%2F", "%3F", "%23", "%3D", "%26", "%3A", "%3B", "%20", "%25", "%2B"};
191 // /** key=special character; value=escaped sequence */
192 // public static HashMap encode_map = new HashMap();
193 // /** key=escaped sequence; value=special character */
194 // public static HashMap decode_map = new HashMap();
195
196 public static void init() {
197 resumption_token_elem = getOAIResumptionTokenXML();
198 }
199 public static String getOAIVersion() {
200 return oai_version;
201 }
202
203 public static String getBaseURL() {
204 return baseURL;
205 }
206
207 public static Element createElement(String tag_name) {
208 return response_doc.createElement(tag_name);
209 }
210 /**Compose a response element used when OAIPMH service sending responses thru
211 * ServiceCluster and MessageRouter, as they automatically wrap a message element
212 * on this response element
213 */
214 public static Element getResponse(Element core_msg) {
215 Element res = createElement(RESPONSE);
216 res.appendChild(response_doc.importNode(core_msg, true));
217 return res;
218 }
219 /** Read in OAIResumptionToken.xml (residing web/WEB-INF/classes/) */
220 public static Element getOAIResumptionTokenXML() {
221
222 // The system environment variable $GSDL3HOME(ends ../web) does not contain the file separator
223 resumption_token_file = new File(GlobalProperties.getGSDL3Home() + FILE_SEPARATOR +
224 "WEB-INF" + FILE_SEPARATOR + "classes" +FILE_SEPARATOR + "OAIResumptionToken.xml");
225 if (resumption_token_file.exists()) {
226 Document token_doc = parseXMLFile(resumption_token_file);
227 if (token_doc != null) {
228 resumption_token_elem = token_doc.getDocumentElement();
229 } else {
230 logger.error("Fail to parse resumption token file OAIReceptionToken.xml.");
231 return null;
232 }
233 //remove all expired tokens
234 clearExpiredTokens();
235 return resumption_token_elem;
236 }
237 //if resumption_token_file does not exist
238 logger.info("resumption token file: "+ resumption_token_file.getPath()+" not found! create an empty one.");
239 resumption_token_elem = createElement(OAI_RESUMPTION_TOKENS);
240 saveOAIResumptionTokenXML(resumption_token_elem);
241 return resumption_token_elem;
242 }
243 public static void saveOAIResumptionTokenXML(Element token_elem) {
244 if(writeXMLFile(resumption_token_file, token_elem.getOwnerDocument()) == false) {
245 logger.error("Fail to save the resumption token file");
246 }
247 }
248 public static void clearExpiredTokens() {
249 boolean token_deleted = false;
250 NodeList tokens = GSXML.getChildrenByTagName(resumption_token_elem, RESUMPTION_TOKEN);
251 for (int i=0; i<tokens.getLength(); i++) {
252 Element token_elem = (Element)tokens.item(i);
253 String expire_str = token_elem.getAttribute(EXPIRATION_DATE);
254 long datestamp = getTime(expire_str); // expire_str is in milliseconds
255 if(datestamp < System.currentTimeMillis()) {
256 resumption_token_elem.removeChild(token_elem);
257 token_elem = null;
258 token_deleted = true;
259 }
260 }
261
262 if(token_deleted) {
263 saveOAIResumptionTokenXML(resumption_token_elem);
264 }
265 }
266 public static boolean containsToken(String token) {
267 NodeList tokens = GSXML.getChildrenByTagName(resumption_token_elem, OAIXML.RESUMPTION_TOKEN);
268 for (int i=0; i<tokens.getLength(); i++) {
269 if(token.equals(GSXML.getNodeText((Element)tokens.item(i)).trim() ))
270 return true;
271 }
272 return false;
273 }
274 public static void addToken(Element token) {
275 Document doc = resumption_token_elem.getOwnerDocument();
276 resumption_token_elem.appendChild(duplicateElement(doc, token, true));
277 saveOAIResumptionTokenXML(resumption_token_elem);
278 }
279 public static void addToken(String token) {
280 Element te = resumption_token_elem.getOwnerDocument().createElement(OAIXML.RESUMPTION_TOKEN);
281 //add expiration att
282 resumption_token_elem.appendChild(te);
283 saveOAIResumptionTokenXML(resumption_token_elem);
284 }
285 public static boolean removeToken(String token) {
286 NodeList tokens = GSXML.getChildrenByTagName(resumption_token_elem, OAIXML.RESUMPTION_TOKEN);
287 int num_tokens = tokens.getLength();
288 for (int i=0; i<num_tokens; i++) {
289 Element e = (Element)(tokens.item(i));
290 if(token.equals(GSXML.getNodeText(e))) {
291 resumption_token_elem.removeChild(e);
292 saveOAIResumptionTokenXML(resumption_token_elem);
293 return true;
294 }
295 }
296 return false;
297 }
298 /** Read in OAIConfig.xml (residing web/WEB-INF/classes/) and use it to configure the receptionist etc.
299 * the oai_version and baseURL variables are also set in here.
300 * The init() method is also called in here. */
301 public static Element getOAIConfigXML() {
302 init();
303
304 // The system environment variable $GSDL3HOME(ends ../web) does not contain the file separator
305 File oai_config_file = new File(GlobalProperties.getGSDL3Home() + FILE_SEPARATOR +
306 "WEB-INF" + FILE_SEPARATOR + "classes" +FILE_SEPARATOR + "OAIConfig.xml");
307 if (!oai_config_file.exists()) {
308 logger.error(" oai config file: "+oai_config_file.getPath()+" not found!");
309 return null;
310 }
311 Document oai_config_doc = parseXMLFile(oai_config_file);
312 if (oai_config_doc != null) {
313 oai_config_elem = oai_config_doc.getDocumentElement();
314 } else {
315 logger.error("Fail to parse oai config file OAIConfig.xml.");
316 return null;
317 }
318
319 //initialize oai_version
320 Element protocol_version = (Element)GSXML.getChildByTagName(oai_config_elem, PROTOCOL_VERSION);
321 oai_version = GSXML.getNodeText(protocol_version).trim();
322
323 // initialize baseURL
324 Element base_url_elem = (Element)GSXML.getChildByTagName(oai_config_elem, BASE_URL);
325 baseURL = GSXML.getNodeText(base_url_elem);
326
327 //initialize token_expiration
328 Element expiration = (Element)GSXML.getChildByTagName(oai_config_elem, RESUMPTION_TOKEN_EXPIRATION);
329 String expire_str = GSXML.getNodeText(expiration).trim();
330 if (expiration != null && !expire_str.equals("")) {
331 token_expiration = Long.parseLong(expire_str);
332 }
333
334 // read granularity from the config file
335 Element granu_elem = (Element)GSXML.getChildByTagName(oai_config_elem, GRANULARITY);
336 //initialize the granu_str which might be used by other methods (eg, getDate())
337 granularity = GSXML.getNodeText(granu_elem).trim();
338
339 //change "yyyy-MM-ddTHH:mm:ssZ" to "yyyy-MM-dd'T'HH:mm:ss'Z'"
340 granularity = granularity.replaceAll("T", "'T'");
341 granularity = granularity.replaceAll("Z", "'Z'");
342 granularity = granularity.replaceAll("YYYY", "yyyy").replaceAll("DD", "dd").replaceAll("hh", "HH");
343 return oai_config_elem;
344 }
345
346 public static String[] getMetadataMapping(Element metadata_format) {
347
348 if (metadata_format == null) {
349 return null;
350 }
351 NodeList mappings = metadata_format.getElementsByTagName(MAPPING);
352 int size = mappings.getLength();
353 if (size == 0) {
354 logger.info("No metadata mappings are provided in OAIConfig.xml.");
355 return null;
356 }
357 String[] names = new String[size];
358 for (int i=0; i<size; i++) {
359 names[i] = GSXML.getNodeText((Element)mappings.item(i)).trim();
360 }
361 return names;
362
363 }
364
365 public static String[] getGlobalMetadataMapping(String prefix) {
366 Element list_meta_formats = (Element)GSXML.getChildByTagName(oai_config_elem, LIST_METADATA_FORMATS);
367 if(list_meta_formats == null) {
368 return null;
369 }
370 Element metadata_format = GSXML.getNamedElement(list_meta_formats, METADATA_FORMAT, METADATA_PREFIX, prefix);
371 if(metadata_format == null) {
372 return null;
373 }
374 return getMetadataMapping(metadata_format);
375 }
376
377
378 public static long getTokenExpiration() {
379 return token_expiration*1000; // in milliseconds
380 }
381
382 /** TODO: returns a basic response for appropriate oai version
383 *
384 */
385 public static Element createBasicResponse(String verb, String[] pairs) {
386
387 Element response = createResponseHeader(verb);
388
389 //set the responseDate and request elements accordingly
390 Element request_elem = (Element)GSXML.getChildByTagName(response, REQUEST);
391 if (verb.equals("")) {
392 request_elem.setAttribute(VERB, verb);
393 }
394 int num_pairs = (pairs==null)? 0 : pairs.length;
395 for (int i=num_pairs - 1; i>=0; i--) {
396 int index = pairs[i].indexOf("=");
397 if (index != -1) {
398 String[] strs = pairs[i].split("=");
399 if(strs != null && strs.length == 2) {
400 request_elem.setAttribute(strs[0], oaiDecode(strs[1]));
401 }
402 }
403 }//end of for()
404
405 GSXML.setNodeText(request_elem, baseURL);
406
407 Node resp_date = GSXML.getChildByTagName(response, RESPONSE_DATE);
408 if (resp_date != null) {
409 GSXML.setNodeText((Element)resp_date, getCurrentUTCTime());
410 }
411
412 return response;
413 }
414 /** @param error_code the value of the code attribute
415 * @param error_text the node text of the error element
416 * @return an oai error element
417 * Used by receptionist
418 */
419 public static Element createErrorElement(String error_code, String error_text) {
420 Element error = createElement(ERROR);
421 error.setAttribute(CODE, error_code);
422 GSXML.setNodeText(error, error_text);
423 return error;
424 }
425
426 /** convert the escaped sequences (eg, '%3A') of those special characters back to their
427 * original form (eg, ':').
428 */
429 public static String oaiDecode(String escaped_str) {
430 logger.info("oaiDecode() " +escaped_str);
431 for (int i=0; i<special_char.length; i++) {
432 if (escaped_str.indexOf(escape_sequence[i]) != -1) {
433 escaped_str = escaped_str.replaceAll(escape_sequence[i], special_char[i]);
434 }
435 }
436 //escaped_str = escaped_str.replaceAll("%3A", ":");
437 return escaped_str;
438 }
439 /** convert those special characters (eg, ':') to their
440 * escaped sequences (eg, '%3A').
441 */
442 public static String oaiEncode(String original_str) {
443 logger.info("oaiEncode() " + original_str);
444 for (int i=0; i<special_char.length; i++) {
445 if (original_str.indexOf(special_char[i]) != -1) {
446 original_str = original_str.replaceAll(special_char[i], escape_sequence[i]);
447 }
448 }
449 //original_str = original_str.replaceAll(":", "%3A");
450 return original_str;
451 }
452 /** convert YYYY-MM_DDThh:mm:ssZ to yyyy-MM-ddTHH:mm:ssZ
453 */
454 public static String convertToJava(String oai_format) {
455 oai_format = oai_format.replaceAll("YYYY", "yyyy").replaceAll("DD", "dd").replaceAll("hh", "HH");
456 return oai_format;
457 }
458 /** convert yyyy-MM-ddTHH:mm:ssZ to YYYY-MM_DDThh:mm:ssZ
459 */
460 public static String convertToOAI(String java_format) {
461 java_format = java_format.replaceAll("yyyy", "YYYY").replaceAll("dd", "DD").replaceAll("HH", "hh");
462 return java_format;
463 }
464 public static String getCurrentUTCTime() {
465 Date current_utc = new Date(System.currentTimeMillis());
466 //granularity is in the form: yyyy-MM-dd'T'HH:mm:ss'Z '
467 DateFormat formatter = new SimpleDateFormat(granularity);
468 return formatter.format(current_utc);
469 }
470 /** get a Date object from a Date format pattern string
471 *
472 * @param pattern - in the form: 2007-06-14T16:48:25Z, for example.
473 * @return a Date object - null if the pattern is not in the specified form
474 */
475
476 public static Date getDate(String pattern) {
477 if (pattern == null || pattern.equals("")) {
478 return null;
479 }
480 Date date = null;
481 // String str = pattern.replaceAll("T", " ");
482 // str = str.replaceAll("Z", "");
483 SimpleDateFormat sdf = null;
484 try {
485 sdf = new SimpleDateFormat(granularity);
486 date = sdf.parse(pattern);
487 } catch(Exception e) {
488 if(!default_granularity.equals(granularity)) { // try validating against default granularity
489 try {
490 date = null;
491 sdf = null;
492 sdf = new SimpleDateFormat(default_granularity);
493 date = sdf.parse(pattern);
494 } catch(Exception ex) {
495 logger.error("invalid date format: " + pattern);
496 return null;
497 }
498 } else {
499 logger.error("invalid date format: " + pattern);
500 return null;
501 }
502 }
503 return date;
504 }
505 /** get the million second value from a string representing time in a pattern
506 * (eg, 2007-06-14T16:48:25Z)
507 */
508 public static long getTime(String pattern) {
509 if (pattern == null || pattern.equals("")) {
510 return -1;
511 }
512 Date date = null;
513 SimpleDateFormat sdf = null;
514 try {
515 //granularity is a global variable in the form: yyyy-MM-ddTHH:mm:ssZ
516 sdf = new SimpleDateFormat(granularity);
517 date = sdf.parse(pattern);
518 } catch(Exception e) {
519 if(!default_granularity.equals(granularity)) { // try validating against default granularity
520 try {
521 date = null;
522 sdf = null;
523 sdf = new SimpleDateFormat(default_granularity);
524 date = sdf.parse(pattern);
525 } catch(Exception ex) {
526 logger.error("invalid date format: " + pattern);
527 return -1;
528 }
529 } else {
530 logger.error("invalid date format: " + pattern);
531 return -1;
532 }
533 }
534 return date.getTime();
535 }
536 /** get the string representation of a time from a long value(long type)
537 */
538 public static String getTime(long milliseconds) {
539 Date date = new Date(milliseconds);
540 SimpleDateFormat sdf = new SimpleDateFormat(granularity);
541 return sdf.format(date);
542 }
543 public static Element createResponseHeader(String verb) {
544 String tag_name = (oai_version.equals(OAI_VERSION2))? OAI_DASH_PMH : verb;
545 Element oai = response_doc.createElement(tag_name);
546 Element resp_date = response_doc.createElement(RESPONSE_DATE);
547 Element req = response_doc.createElement(REQUEST);
548 oai.appendChild(resp_date);
549 oai.appendChild(req);
550
551 if(oai_version.equals(OAI_VERSION2)) {
552 oai.setAttribute("xmlns", "http://www.openarchives.org/OAI/2.0/");
553 oai.setAttribute("xmlns:xsi", "http://www.w3.org/2001/XMLSchema-instance");
554 oai.setAttribute("xsi:schemaLocation", "http://www.openarchives.org/OAI/2.0/ \n http://www.openarchives.org/OAI/2.0/OAI-PMH.xsd");
555 } else {
556 oai.setAttribute("xmlns", "http://www.openarchives.com/OAI/1.1/OAI_" + verb);
557 oai.setAttribute("xmlns:xsi", "http://www.w3.org/2001/XMLSchema-instance");
558 oai.setAttribute("xsi:schemaLocation", "http://www.openarchives.org/OAI/1.1/OAI_" + verb + "\n http://www.openarchives.org/OAI/1.1/OAI_" + verb + ".xsd");
559 }
560 return oai;
561 }
562 public static Element getMetadataPrefixElement(String tag_name, String version) {
563 //examples of tag_name: dc, oai_dc:dc, etc.
564 Element oai = response_doc.createElement(tag_name);
565 if (version.equals(OAI_VERSION2)) {
566 oai.setAttribute("xmlns:oai_dc", "http://www.openarchives.org/OAI/2.0/oai_dc/");
567 oai.setAttribute("xmlns:dc", "http://purl.org/dc/elements/1.1/");
568 oai.setAttribute("xmlns:xsi", "http://www.w3.org/2001/XMLSchema-instance");
569 oai.setAttribute("xsi:schemaLocation", "http://www.openarchives.org/OAI/2.0/oai_dc/ \n http://www.openarchives.org/OAI/2.0/oai_dc.xsd");
570 } else {
571 oai.setAttribute("xmlns", "ttp://www.openarchives.com/OAI/1.1/");
572 oai.setAttribute("xmlns:xsi", "http://www.w3.org/2001/XMLSchema-instance");
573 oai.setAttribute("xsi:schemaLocation", "http://www.openarchives.org/OAI/1.1/" + tag_name + ".xsd");
574 }
575
576 return oai;
577 }
578 public static HashMap<String, Node> getChildrenMapByTagName(Node n, String tag_name) {
579
580 HashMap<String, Node> map= new HashMap<String, Node>();
581 Node child = n.getFirstChild();
582 while (child!=null) {
583 String name = child.getNodeName();
584 if(name.equals(tag_name)) {
585 map.put(name, child);
586 }
587 child = child.getNextSibling();
588 }
589 return map;
590 }
591
592 /** Duplicates an element */
593 public static Element duplicateElement (Document owner, Element element, boolean with_attributes) {
594 return duplicateElementNS (owner, element, null, with_attributes);
595 }
596
597 /** Duplicates an element */
598 public static Element duplicateElementNS (Document owner,
599 Element element,
600 String namespace_uri,
601 boolean with_attributes) {
602 Element duplicate;
603 if (namespace_uri == null) {
604 duplicate = owner.createElement (element.getTagName ());
605 } else {
606 duplicate = owner.createElementNS (namespace_uri, element.getTagName ());
607 }
608 // Copy element attributes
609 if (with_attributes) {
610 NamedNodeMap attributes = element.getAttributes ();
611 for (int i = 0; i < attributes.getLength (); i++) {
612 Node attribute = attributes.item (i);
613 duplicate.setAttribute (attribute.getNodeName (), attribute.getNodeValue ());
614 }
615 }
616
617 // Copy element children
618 NodeList children = element.getChildNodes ();
619 for (int i = 0; i < children.getLength (); i++) {
620 Node child = children.item (i);
621 duplicate.appendChild (owner.importNode (child, true));
622 }
623
624 return duplicate;
625 }
626
627 public static void copyElement(Element to, Element from, String elem_name) {
628
629 Document to_doc = to.getOwnerDocument();
630 Node child = from.getFirstChild();
631 while (child != null) {
632 if (child.getNodeName().equals(elem_name)) {
633 to.appendChild(to_doc.importNode(child, true));
634 return;
635 }
636 child = child.getNextSibling();
637 }
638 }
639
640 public static HashMap<String, String> getParamMap(NodeList params) {
641 HashMap<String, String> map = new HashMap<String, String>();
642 for(int i=0; i<params.getLength(); i++) {
643 Element param = (Element)params.item(i);
644 String param_name = param.getAttribute(OAIXML.NAME);
645 String param_value = param.getAttribute(OAIXML.VALUE);
646 map.put(param_name, param_value);
647 }
648 return map;
649 }
650
651
652 /** Parse an XML document from a given file */
653 static public Document parseXMLFile (File xml_file) {
654 // No file? No point trying!
655 if (xml_file.exists () == false) {
656 return null;
657 }
658 Document doc = null;
659 try {
660 doc = parseXML (new FileInputStream (xml_file));
661 }
662 catch (Exception exception) {
663 logger.error(exception.toString());
664 return null;
665 }
666 return doc;
667 }
668
669
670 /** Parse an XML document from a given input stream */
671 static public Document parseXML (InputStream xml_input_stream) {
672 Document document = null;
673
674 try {
675 InputStreamReader isr = new InputStreamReader (xml_input_stream, "UTF-8");
676 Reader xml_reader = new BufferedReader (isr);
677 document = parseXML (xml_reader);
678 isr.close ();
679 xml_input_stream.close ();
680 }
681 catch (Exception exception) {
682 logger.error(exception.toString());
683 }
684
685 return document;
686 }
687
688 /** Parse an XML document from a given reader */
689 static public Document parseXML (Reader xml_reader) {
690 Document document = null;
691
692 try {
693 InputSource isc = new InputSource (xml_reader);
694 DOMParser parser = new DOMParser ();
695 parser.setFeature ("http://xml.org/sax/features/validation", false);
696 parser.setFeature ("http://apache.org/xml/features/nonvalidating/load-external-dtd", false);
697 // May or may not be ignored, the documentation for Xerces is contradictory. If it works then parsing -should- be faster.
698 parser.setFeature ("http://apache.org/xml/features/dom/defer-node-expansion", true);
699 parser.setFeature ("http://apache.org/xml/features/dom/include-ignorable-whitespace", false);
700 parser.parse (isc);
701 document = parser.getDocument ();
702 }
703 catch (SAXException exception) {
704 System.err.println ("SAX exception: " + exception.getMessage ());
705 logger.error(exception.toString());
706 }
707 catch (Exception exception) {
708 logger.error(exception.toString());
709 }
710
711 return document;
712 }
713 /** Write an XML document to a given file */
714 static public boolean writeXMLFile (File xml_file, Document document) {
715 try {
716 OutputStream os = new FileOutputStream (xml_file);
717 // Create an output format for our document.
718 OutputFormat f = new OutputFormat (document);
719 f.setEncoding ("UTF-8");
720 f.setIndenting (true);
721 f.setLineWidth (0); // Why isn't this working!
722 f.setPreserveSpace (false);
723 // Create the necessary writer stream for serialization.
724 OutputStreamWriter osw = new OutputStreamWriter (os, "UTF-8");
725 Writer w = new BufferedWriter (osw);
726 // Generate a new serializer from the above.
727 XMLSerializer s = new XMLSerializer (w, f);
728 s.asDOMSerializer ();
729 // Finally serialize the document to file.
730 s.serialize (document);
731 // And close.
732 os.close ();
733 return true;
734 }
735 catch (Exception exception) {
736 logger.error(exception.toString());
737 return false;
738 }
739 }
740
741
742 public static Element createOAIIdentifierXML(String repository_id, String sample_collection, String sample_doc_id) {
743 String xml = "<oai-identifier xmlns=\"http://www.openarchives.org/OAI/2.0/oai-identifier\"\n xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\"\n xsi:schemaLocation=\"http://www.openarchives.org/OAI/2.0/oai-identifier\n http://www.openarchives.org/OAI/2.0/oai-identifier.xsd\">\n <scheme>oai</scheme>\n<repositoryIdentifier>" + repository_id + "</repositoryIdentifier>\n<delimiter>:</delimiter>\n<sampleIdentifier>oai:"+repository_id+":"+sample_collection+":"+sample_doc_id+"</sampleIdentifier>\n</oai-identifier>";
744
745 Document xml_doc = new XMLConverter().getDOM(xml);
746 return (Element)response_doc.importNode(xml_doc.getDocumentElement(), true);
747
748
749 }
750
751 public static Element createGSDLElement() {
752 String xml = "<gsdl xmlns=\"http://www.greenstone.org/namespace/gsdl_oaiinfo/1.0/gsdl_oaiinfo\"\n xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\"\n xsi:schemaLocation=\"http://www.greenstone.org/namespace/gsdl_oaiinfo/1.0/gsdl_oaiinfo\n http://www.greenstone.org/namespace/gsdl_oaiinfo/1.0/gsdl_oaiinfo.xsd\"></gsdl>";
753 Document xml_doc = new XMLConverter().getDOM(xml);
754 return (Element)response_doc.importNode(xml_doc.getDocumentElement(), true);
755
756
757 }
758}
759
760
761
762
763
764
Note: See TracBrowser for help on using the repository browser.