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