|
1
|
package eu.dnetlib.validator.service.impls.providers;
|
|
2
|
|
|
3
|
import java.io.IOException;
|
|
4
|
import java.io.StringReader;
|
|
5
|
import java.net.MalformedURLException;
|
|
6
|
import java.net.URL;
|
|
7
|
import java.net.URLEncoder;
|
|
8
|
import java.util.ArrayList;
|
|
9
|
import java.util.List;
|
|
10
|
|
|
11
|
import javax.xml.parsers.DocumentBuilder;
|
|
12
|
import javax.xml.parsers.DocumentBuilderFactory;
|
|
13
|
import javax.xml.parsers.ParserConfigurationException;
|
|
14
|
import javax.xml.xpath.XPath;
|
|
15
|
import javax.xml.xpath.XPathConstants;
|
|
16
|
import javax.xml.xpath.XPathExpression;
|
|
17
|
import javax.xml.xpath.XPathExpressionException;
|
|
18
|
import javax.xml.xpath.XPathFactory;
|
|
19
|
|
|
20
|
import org.w3c.dom.Document;
|
|
21
|
import org.w3c.dom.Element;
|
|
22
|
import org.w3c.dom.Node;
|
|
23
|
import org.w3c.dom.NodeList;
|
|
24
|
import org.w3c.dom.ls.DOMImplementationLS;
|
|
25
|
import org.w3c.dom.ls.LSSerializer;
|
|
26
|
import org.xml.sax.InputSource;
|
|
27
|
import org.xml.sax.SAXException;
|
|
28
|
|
|
29
|
import se.kb.oai.pmh.OaiPmhServer;
|
|
30
|
import se.kb.oai.pmh.ResumptionToken;
|
|
31
|
import se.kb.oai.pmh.Set;
|
|
32
|
import se.kb.oai.pmh.SetsList;
|
|
33
|
import eu.dnetlib.validator.engine.data.DataException;
|
|
34
|
import eu.dnetlib.validator.engine.data.Provider;
|
|
35
|
import eu.dnetlib.validator.engine.data.ResultSet;
|
|
36
|
import eu.dnetlib.validator.engine.execution.ValidationObject;
|
|
37
|
import eu.dnetlib.validator.service.impls.valobjs.XMLTextValidationObject;
|
|
38
|
|
|
39
|
/**
|
|
40
|
* A provider that retrieves records from an OAI-PMH repository. Resumption
|
|
41
|
* tokens are handled transparently.
|
|
42
|
*
|
|
43
|
* @author Manos Karvounis
|
|
44
|
*
|
|
45
|
*/
|
|
46
|
public class OAIPMHRecordProvider extends Provider {
|
|
47
|
|
|
48
|
private static final long serialVersionUID = 3386029339653670731L;
|
|
49
|
|
|
50
|
private int timeout;
|
|
51
|
private int delay;
|
|
52
|
private int retryDelay;
|
|
53
|
private int retryEfforts;
|
|
54
|
|
|
55
|
public static final String BASEURL = "BASEURL";
|
|
56
|
public static final String METADATA_PREFIX = "metadataPrefix";
|
|
57
|
/**
|
|
58
|
* optional
|
|
59
|
*/
|
|
60
|
public static final String FROM = "from";
|
|
61
|
/**
|
|
62
|
* optional
|
|
63
|
*/
|
|
64
|
public static final String UNTIL = "until";
|
|
65
|
/**
|
|
66
|
* optional
|
|
67
|
*/
|
|
68
|
public static final String SET = "set";
|
|
69
|
|
|
70
|
/**
|
|
71
|
* The maximum time to wait for a response from the repository (in millis)
|
|
72
|
*/
|
|
73
|
// public static final String TIMEOUT = "TIMEOUT";
|
|
74
|
/**
|
|
75
|
* How much time to wait between consecutive HTTP requests to the repository
|
|
76
|
* (in millis).
|
|
77
|
*/
|
|
78
|
// public static final String DELAY = "DELAY";
|
|
79
|
/**
|
|
80
|
* How much to wait if an HTTP request fails before trying again by
|
|
81
|
* resending the request.
|
|
82
|
*/
|
|
83
|
// public static final String RETRY_DELAY = "RETRY_DELAY";
|
|
84
|
/**
|
|
85
|
* If an HTTP requests fails, how many times to try to resend the request.
|
|
86
|
*/
|
|
87
|
// public static final String RETRY_EFFORTS = "RETRY_EFFORTS";
|
|
88
|
/**
|
|
89
|
* How many records to test.
|
|
90
|
*/
|
|
91
|
public static final String RECORDS = "records";
|
|
92
|
|
|
93
|
public OAIPMHRecordProvider() {
|
|
94
|
super(1);
|
|
95
|
}
|
|
96
|
|
|
97
|
@Override
|
|
98
|
public ResultSet<ValidationObject> getValidationObjects() throws ProviderException {
|
|
99
|
return new OAIPMHRecordResultSet(this);
|
|
100
|
}
|
|
101
|
|
|
102
|
@Override
|
|
103
|
public ResultSet<String> getValidationObjectIds() throws ProviderException {
|
|
104
|
return new OAIPMHRecordIdentifierResultSet(this);
|
|
105
|
}
|
|
106
|
|
|
107
|
@Override
|
|
108
|
public ValidationObject getValidationObject(String valObjId) throws ProviderException {
|
|
109
|
OAIPMHResultSet oai = new OAIPMHResultSet(this);
|
|
110
|
try {
|
|
111
|
return new XMLTextValidationObject(oai.getRecord(valObjId));
|
|
112
|
} catch (DataException e) {
|
|
113
|
log.error("", e);
|
|
114
|
throw new ProviderException();
|
|
115
|
}
|
|
116
|
}
|
|
117
|
|
|
118
|
private class OAIPMHResultSet {
|
|
119
|
protected DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
|
|
120
|
protected DocumentBuilder builder;
|
|
121
|
protected XPathFactory xfactory = XPathFactory.newInstance();
|
|
122
|
protected NodeList recordIds = null;
|
|
123
|
protected NodeList records = null;
|
|
124
|
protected int index = -1;
|
|
125
|
protected String resumptionToken = null;
|
|
126
|
protected String error = null;
|
|
127
|
private URLStreamer streamer = new URLStreamer();
|
|
128
|
|
|
129
|
|
|
130
|
public OAIPMHResultSet(OAIPMHRecordProvider prv) {
|
|
131
|
super();
|
|
132
|
try {
|
|
133
|
builder = factory.newDocumentBuilder();
|
|
134
|
} catch (ParserConfigurationException e) {
|
|
135
|
log.error("", e);
|
|
136
|
}
|
|
137
|
}
|
|
138
|
|
|
139
|
protected NodeList getIds() throws DataException {
|
|
140
|
NodeList recordIds = null;
|
|
141
|
String surl = pros.getProperty(BASEURL) + "?verb=ListIdentifiers";
|
|
142
|
if (resumptionToken == null || resumptionToken.trim().length() == 0) {
|
|
143
|
surl += "&metadataPrefix=" + pros.getProperty(METADATA_PREFIX);
|
|
144
|
if (pros.getProperty(FROM) != null)
|
|
145
|
surl += "&from=" + pros.getProperty(FROM);
|
|
146
|
if (pros.getProperty(UNTIL) != null)
|
|
147
|
surl += "&until=" + pros.getProperty(UNTIL);
|
|
148
|
if (!pros.getProperty(SET).equals("none"))
|
|
149
|
surl += "&set=" + pros.getProperty(SET);
|
|
150
|
} else {
|
|
151
|
surl += "&resumptionToken=" + resumptionToken;
|
|
152
|
}
|
|
153
|
|
|
154
|
log.debug("Issuing request "+surl);
|
|
155
|
|
|
156
|
String response = null;
|
|
157
|
try {
|
|
158
|
response = streamer.getResponse(new URL(surl), timeout, delay, retryDelay, retryEfforts);
|
|
159
|
} catch (NumberFormatException e) {
|
|
160
|
log.error("", e);
|
|
161
|
throw new DataException();
|
|
162
|
} catch (MalformedURLException e) {
|
|
163
|
log.error("", e);
|
|
164
|
throw new DataException();
|
|
165
|
} catch (IOException e) {
|
|
166
|
log.error("", e);
|
|
167
|
throw new DataException();
|
|
168
|
}
|
|
169
|
try {
|
|
170
|
InputSource is = new InputSource(new StringReader(response));
|
|
171
|
Document doc = builder.parse(is);
|
|
172
|
XPath xpath = xfactory.newXPath();
|
|
173
|
XPathExpression expr = xpath.compile("OAI-PMH/ListIdentifiers/header/identifier/text()");
|
|
174
|
recordIds = (NodeList) expr.evaluate(doc, XPathConstants.NODESET);
|
|
175
|
XPath xpath2 = xfactory.newXPath();
|
|
176
|
XPathExpression expr2 = xpath2.compile("OAI-PMH/ListIdentifiers/resumptionToken/text()");
|
|
177
|
NodeList rtl = (NodeList) expr2.evaluate(doc, XPathConstants.NODESET);
|
|
178
|
if (rtl == null || rtl.getLength() == 0) {
|
|
179
|
log.debug("There seems to be no resumption token present");
|
|
180
|
resumptionToken = null;
|
|
181
|
}
|
|
182
|
else {
|
|
183
|
resumptionToken = rtl.item(0).getNodeValue();
|
|
184
|
resumptionToken = URLEncoder.encode(resumptionToken, "UTF-8");
|
|
185
|
log.debug("Found resumption token: "+resumptionToken);
|
|
186
|
}
|
|
187
|
} catch (Exception e) {
|
|
188
|
log.error("", e);
|
|
189
|
throw new DataException();
|
|
190
|
}
|
|
191
|
return recordIds;
|
|
192
|
}
|
|
193
|
|
|
194
|
protected NodeList getRecords() throws DataException {
|
|
195
|
NodeList records = null;
|
|
196
|
String surl = pros.getProperty(BASEURL) + "?verb=ListRecords";
|
|
197
|
if (resumptionToken == null || resumptionToken.trim().length() == 0) {
|
|
198
|
surl += "&metadataPrefix=" + pros.getProperty(METADATA_PREFIX);
|
|
199
|
if (pros.getProperty(FROM) != null)
|
|
200
|
surl += "&from=" + pros.getProperty(FROM);
|
|
201
|
if (pros.getProperty(UNTIL) != null)
|
|
202
|
surl += "&until=" + pros.getProperty(UNTIL);
|
|
203
|
if (!pros.getProperty(SET).equals("none"))
|
|
204
|
surl += "&set=" + pros.getProperty(SET);
|
|
205
|
} else {
|
|
206
|
surl += "&resumptionToken=" + resumptionToken;
|
|
207
|
}
|
|
208
|
String setError=null;
|
|
209
|
if (!pros.getProperty(SET).equals("none")) {
|
|
210
|
OaiPmhServer harvester = new OaiPmhServer(pros.getProperty(BASEURL));
|
|
211
|
try {
|
|
212
|
SetsList setList = harvester.listSets();
|
|
213
|
ResumptionToken token = setList.getResumptionToken();
|
|
214
|
List<Set> sets = new ArrayList<Set>();
|
|
215
|
sets.addAll(setList.asList());
|
|
216
|
while (token != null) {
|
|
217
|
setList = harvester.listSets(token);
|
|
218
|
token = setList.getResumptionToken();
|
|
219
|
sets.addAll(setList.asList());
|
|
220
|
}
|
|
221
|
List<String> ret = new ArrayList<String>();
|
|
222
|
for (Set set : sets) {
|
|
223
|
ret.add(set.getSpec().trim());
|
|
224
|
}
|
|
225
|
if (!ret.contains(pros.getProperty(SET))){
|
|
226
|
error = "Set: <b>'" + pros.getProperty(SET) + "'</b> is not exposed by the repository. \n Please make sure that 'ListSets' verb is configured correctly on your server as well as that the exposed sets list includes this set.";
|
|
227
|
setError = error;
|
|
228
|
}
|
|
229
|
} catch (Exception e) {
|
|
230
|
log.error("error getting sets from url: " + pros.getProperty(BASEURL), e);
|
|
231
|
}
|
|
232
|
}
|
|
233
|
log.debug("Issuing request "+surl);
|
|
234
|
|
|
235
|
String response = null;
|
|
236
|
try {
|
|
237
|
response = streamer.getResponse(new URL(surl), timeout, delay, retryDelay, retryEfforts);
|
|
238
|
} catch (NumberFormatException e) {
|
|
239
|
log.error("", e);
|
|
240
|
throw new DataException();
|
|
241
|
} catch (MalformedURLException e) {
|
|
242
|
log.error("", e);
|
|
243
|
throw new DataException();
|
|
244
|
} catch (IOException e) {
|
|
245
|
log.error("", e);
|
|
246
|
throw new DataException();
|
|
247
|
}
|
|
248
|
try {
|
|
249
|
InputSource is = new InputSource(new StringReader(response));
|
|
250
|
Document doc = builder.parse(is);
|
|
251
|
XPath xpath = xfactory.newXPath();
|
|
252
|
// xpath.setNamespaceContext(new NamespaceResolver(doc));
|
|
253
|
XPathExpression expr = xpath.compile("OAI-PMH/ListRecords/record");
|
|
254
|
records = (NodeList) expr.evaluate(doc, XPathConstants.NODESET);
|
|
255
|
XPath xpath2 = xfactory.newXPath();
|
|
256
|
XPathExpression expr2 = xpath2.compile("OAI-PMH/ListRecords/resumptionToken/text()");
|
|
257
|
NodeList rtl = (NodeList) expr2.evaluate(doc, XPathConstants.NODESET);
|
|
258
|
log.debug("Check number of records: "+ records.getLength());
|
|
259
|
if (records.getLength() == 0) {
|
|
260
|
log.debug("There are no records: "+ records.getLength());
|
|
261
|
XPath xpath3 = xfactory.newXPath();
|
|
262
|
XPathExpression expr3 = xpath3.compile("OAI-PMH/error/text()");
|
|
263
|
error = "The response on request: <a href=\""+ surl + "\">"+surl+"</a> was the following: ";
|
|
264
|
error += "<br><b>" + (String) expr3.evaluate(doc, XPathConstants.STRING) + "</b>";
|
|
265
|
if (setError != null) {
|
|
266
|
error += "<br>" + setError;
|
|
267
|
}
|
|
268
|
log.debug("Error: "+ error);
|
|
269
|
}
|
|
270
|
|
|
271
|
if (rtl == null || rtl.getLength() == 0) {
|
|
272
|
log.debug("There seems to be no resumption token present");
|
|
273
|
resumptionToken = null;
|
|
274
|
}
|
|
275
|
else {
|
|
276
|
resumptionToken = rtl.item(0).getNodeValue();
|
|
277
|
resumptionToken = URLEncoder.encode(resumptionToken, "UTF-8");
|
|
278
|
log.debug("Found resumption token: "+resumptionToken);
|
|
279
|
}
|
|
280
|
} catch (Exception e) {
|
|
281
|
log.error("", e);
|
|
282
|
throw new DataException();
|
|
283
|
}
|
|
284
|
return records;
|
|
285
|
}
|
|
286
|
|
|
287
|
protected Document getRecord(String id) throws DataException {
|
|
288
|
String surl = pros.getProperty(BASEURL) + "?verb=GetRecord";
|
|
289
|
surl += "&metadataPrefix=" + pros.getProperty(METADATA_PREFIX);
|
|
290
|
surl += "&identifier=" + id;
|
|
291
|
|
|
292
|
log.debug("Issuing request: "+surl);
|
|
293
|
|
|
294
|
String response = null;
|
|
295
|
try {
|
|
296
|
response = streamer.getResponse(new URL(surl), timeout, delay, retryDelay, retryEfforts);
|
|
297
|
} catch (NumberFormatException e) {
|
|
298
|
log.error("", e);
|
|
299
|
throw new DataException();
|
|
300
|
} catch (MalformedURLException e) {
|
|
301
|
log.error("", e);
|
|
302
|
throw new DataException();
|
|
303
|
} catch (IOException e) {
|
|
304
|
log.error("", e);
|
|
305
|
throw new DataException();
|
|
306
|
}
|
|
307
|
|
|
308
|
InputSource is = new InputSource(new StringReader(response));
|
|
309
|
Document doc;
|
|
310
|
try {
|
|
311
|
doc = builder.parse(is);
|
|
312
|
} catch (SAXException e) {
|
|
313
|
log.error("", e);
|
|
314
|
throw new DataException();
|
|
315
|
} catch (IOException e) {
|
|
316
|
log.error("", e);
|
|
317
|
throw new DataException();
|
|
318
|
}
|
|
319
|
|
|
320
|
return doc;
|
|
321
|
}
|
|
322
|
}
|
|
323
|
|
|
324
|
private class OAIPMHRecordResultSet extends OAIPMHResultSet implements ResultSet<ValidationObject> {
|
|
325
|
|
|
326
|
public OAIPMHRecordResultSet(OAIPMHRecordProvider prv) {
|
|
327
|
super(prv);
|
|
328
|
// TODO Auto-generated constructor stub
|
|
329
|
}
|
|
330
|
|
|
331
|
@Override
|
|
332
|
public String getError() {
|
|
333
|
if (error != null)
|
|
334
|
log.debug("An error occured "+ this.error);
|
|
335
|
else
|
|
336
|
log.debug("No errors on request");
|
|
337
|
return this.error;
|
|
338
|
}
|
|
339
|
|
|
340
|
@Override
|
|
341
|
public boolean next() throws DataException {
|
|
342
|
index++;
|
|
343
|
|
|
344
|
log.debug("Moving cursor to result "+index);
|
|
345
|
if (records == null || index >= records.getLength()) {
|
|
346
|
// if we have previously received some results and there no more to take
|
|
347
|
if (records != null && (resumptionToken == null || resumptionToken.trim().length() == 0))
|
|
348
|
return false;
|
|
349
|
index = -1;
|
|
350
|
records = getRecords();
|
|
351
|
return next();
|
|
352
|
}
|
|
353
|
return true;
|
|
354
|
}
|
|
355
|
|
|
356
|
@Override
|
|
357
|
public ValidationObject get() throws DataException {
|
|
358
|
XMLTextValidationObject ret = null;
|
|
359
|
|
|
360
|
Document newXmlDocument;
|
|
361
|
try {
|
|
362
|
newXmlDocument = DocumentBuilderFactory.newInstance().newDocumentBuilder().newDocument();
|
|
363
|
|
|
364
|
Element root = newXmlDocument.createElement("root");
|
|
365
|
newXmlDocument.appendChild(root);
|
|
366
|
Node node = records.item(index);
|
|
367
|
Node copyNode = newXmlDocument.importNode(node, true);
|
|
368
|
root.appendChild(copyNode);
|
|
369
|
// printXmlDocument(newXmlDocument);
|
|
370
|
ret = new XMLTextValidationObject(newXmlDocument);
|
|
371
|
XPathFactory factory = XPathFactory.newInstance();
|
|
372
|
XPath xPath = factory.newXPath();
|
|
373
|
ret.setId(xPath.evaluate("header/identifier", records.item(index)));
|
|
374
|
ret.setStatus(xPath.evaluate("header/@status", records.item(index)));
|
|
375
|
|
|
376
|
} catch (ParserConfigurationException e) {
|
|
377
|
log.error("error getting object"+ e);
|
|
378
|
} catch (XPathExpressionException e) {
|
|
379
|
log.error("error getting object"+ e);
|
|
380
|
}
|
|
381
|
return ret;
|
|
382
|
}
|
|
383
|
|
|
384
|
}
|
|
385
|
|
|
386
|
public static void printXmlDocument(Document document) {
|
|
387
|
DOMImplementationLS domImplementationLS =
|
|
388
|
(DOMImplementationLS) document.getImplementation();
|
|
389
|
LSSerializer lsSerializer =
|
|
390
|
domImplementationLS.createLSSerializer();
|
|
391
|
String string = lsSerializer.writeToString(document);
|
|
392
|
System.out.println(string);
|
|
393
|
}
|
|
394
|
|
|
395
|
private class OAIPMHRecordIdentifierResultSet extends OAIPMHResultSet implements ResultSet<String> {
|
|
396
|
|
|
397
|
public OAIPMHRecordIdentifierResultSet(OAIPMHRecordProvider prv) {
|
|
398
|
super(prv);
|
|
399
|
// TODO Auto-generated constructor stub
|
|
400
|
}
|
|
401
|
|
|
402
|
@Override
|
|
403
|
public boolean next() throws DataException {
|
|
404
|
index++;
|
|
405
|
log.debug("Moving cursor to result "+index);
|
|
406
|
if (recordIds == null || index >= recordIds.getLength()) {
|
|
407
|
// if we have previously received some results and there no more to take
|
|
408
|
if (recordIds != null && (resumptionToken == null || resumptionToken.trim().length() == 0))
|
|
409
|
return false;
|
|
410
|
index = -1;
|
|
411
|
recordIds = getIds();
|
|
412
|
return next();
|
|
413
|
}
|
|
414
|
return true;
|
|
415
|
}
|
|
416
|
|
|
417
|
@Override
|
|
418
|
public String get() throws DataException {
|
|
419
|
String id = recordIds.item(index).getNodeValue();
|
|
420
|
|
|
421
|
log.debug("Returing object with id "+id);
|
|
422
|
|
|
423
|
return id;
|
|
424
|
}
|
|
425
|
|
|
426
|
@Override
|
|
427
|
public String getError() {
|
|
428
|
// TODO Auto-generated method stub
|
|
429
|
return null;
|
|
430
|
}
|
|
431
|
|
|
432
|
}
|
|
433
|
|
|
434
|
@Override
|
|
435
|
public ResultSet<ValidationObject> getValidationObjects(String entity)
|
|
436
|
throws ProviderException {
|
|
437
|
// TODO Auto-generated method stub
|
|
438
|
return null;
|
|
439
|
}
|
|
440
|
|
|
441
|
public Integer getTimeout() {
|
|
442
|
return timeout;
|
|
443
|
}
|
|
444
|
|
|
445
|
public void setTimeout(Integer timeout) {
|
|
446
|
this.timeout = timeout;
|
|
447
|
}
|
|
448
|
|
|
449
|
public Integer getDelay() {
|
|
450
|
return delay;
|
|
451
|
}
|
|
452
|
|
|
453
|
public void setDelay(Integer delay) {
|
|
454
|
this.delay = delay;
|
|
455
|
}
|
|
456
|
|
|
457
|
public Integer getRetryDelay() {
|
|
458
|
return retryDelay;
|
|
459
|
}
|
|
460
|
|
|
461
|
public void setRetryDelay(Integer retryDelay) {
|
|
462
|
this.retryDelay = retryDelay;
|
|
463
|
}
|
|
464
|
|
|
465
|
public Integer getRetryEfforts() {
|
|
466
|
return retryEfforts;
|
|
467
|
}
|
|
468
|
|
|
469
|
public void setRetryEfforts(Integer retryEfforts) {
|
|
470
|
this.retryEfforts = retryEfforts;
|
|
471
|
}
|
|
472
|
|
|
473
|
/*
|
|
474
|
class NamespaceResolver implements NamespaceContext {
|
|
475
|
|
|
476
|
private static final String OAI_NS = "http://www.openarchives.org/OAI/2.0/";
|
|
477
|
private static final String OAI_DC_NS = "http://www.openarchives.org/OAI/2.0/oai_dc/";
|
|
478
|
private static final String DC_NS = "http://purl.org/dc/elements/1.1/";
|
|
479
|
|
|
480
|
private final Document document;
|
|
481
|
|
|
482
|
public NamespaceResolver(Document document) {
|
|
483
|
this.document = document;
|
|
484
|
}
|
|
485
|
|
|
486
|
public String getNamespaceURI(String prefix) {
|
|
487
|
log.debug("prefix: " + prefix);
|
|
488
|
if ("".equals(prefix) || "oai".equals(prefix)){
|
|
489
|
return OAI_NS;
|
|
490
|
}else if ("oai_dc".equals(prefix)){
|
|
491
|
return OAI_DC_NS;
|
|
492
|
}else if ("dc".equals(prefix)){
|
|
493
|
return DC_NS;
|
|
494
|
}
|
|
495
|
return "";
|
|
496
|
}
|
|
497
|
|
|
498
|
public String getPrefix(String namespaceURI) {
|
|
499
|
log.debug("prefix: " + namespaceURI);
|
|
500
|
return "";
|
|
501
|
}
|
|
502
|
|
|
503
|
@SuppressWarnings("rawtypes")
|
|
504
|
public Iterator getPrefixes(String namespaceURI) {
|
|
505
|
// not implemented
|
|
506
|
return null;
|
|
507
|
}
|
|
508
|
|
|
509
|
} */
|
|
510
|
|
|
511
|
}
|