1
|
package eu.dnetlib.data.mapreduce.hbase.statsExport.utils;
|
2
|
|
3
|
import com.google.common.collect.Multimap;
|
4
|
|
5
|
|
6
|
import eu.dnetlib.data.mapreduce.util.LicenseComparator;
|
7
|
import eu.dnetlib.data.proto.DatasourceProtos.Datasource;
|
8
|
import eu.dnetlib.data.proto.DatasourceProtos.Datasource.Metadata;
|
9
|
import eu.dnetlib.data.proto.FieldTypeProtos;
|
10
|
import eu.dnetlib.data.proto.FieldTypeProtos.Qualifier;
|
11
|
import eu.dnetlib.data.proto.FieldTypeProtos.StringField;
|
12
|
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty;
|
13
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
14
|
import eu.dnetlib.data.proto.OafProtos.OafEntity;
|
15
|
import eu.dnetlib.data.proto.OafProtos.OafRel;
|
16
|
import eu.dnetlib.data.proto.OrganizationProtos.Organization;
|
17
|
//import eu.dnetlib.data.proto.PersonProtos;
|
18
|
import eu.dnetlib.data.proto.ProjectProtos.Project;
|
19
|
import eu.dnetlib.data.proto.RelTypeProtos.RelType;
|
20
|
import eu.dnetlib.data.proto.ResultProtos.Result;
|
21
|
import eu.dnetlib.data.proto.ResultProtos.Result.Instance;
|
22
|
import org.apache.log4j.Logger;
|
23
|
import org.jsoup.Jsoup;
|
24
|
|
25
|
import java.text.SimpleDateFormat;
|
26
|
import java.util.ArrayList;
|
27
|
import java.util.Date;
|
28
|
import java.util.HashMap;
|
29
|
import java.util.List;
|
30
|
|
31
|
import org.w3c.dom.Element;
|
32
|
import org.w3c.dom.NodeList;
|
33
|
import org.xml.sax.InputSource;
|
34
|
import com.sun.org.apache.xerces.internal.parsers.DOMParser;
|
35
|
import org.w3c.dom.Document;
|
36
|
|
37
|
/**
|
38
|
* @author eri Simple serializer that parses input Oaf Protos and prepares them
|
39
|
* for sqoop
|
40
|
*/
|
41
|
public class Serializer {
|
42
|
|
43
|
public static String serialize(Oaf oaf, String DELIM, String ENCLOSING) {
|
44
|
|
45
|
switch (oaf.getKind()) {
|
46
|
case entity:
|
47
|
OafEntity valueEntity = oaf.getEntity();
|
48
|
|
49
|
switch (valueEntity.getType()) {
|
50
|
case datasource:
|
51
|
|
52
|
return buildDatasource(oaf, DELIM, ENCLOSING);
|
53
|
|
54
|
case organization:
|
55
|
|
56
|
return buildOrganization(oaf, DELIM, ENCLOSING);
|
57
|
|
58
|
case project:
|
59
|
|
60
|
return buildProject(oaf, DELIM, ENCLOSING);
|
61
|
case result:
|
62
|
|
63
|
return buildResult(oaf, DELIM, ENCLOSING);
|
64
|
/*
|
65
|
case person:
|
66
|
return buildPerson(oaf, DELIM, ENCLOSING);
|
67
|
*/
|
68
|
default:
|
69
|
break;
|
70
|
}
|
71
|
break;
|
72
|
case relation:
|
73
|
|
74
|
return buildRel(oaf.getRel(), DELIM, ENCLOSING);
|
75
|
|
76
|
}
|
77
|
return null;
|
78
|
}
|
79
|
|
80
|
public static String serialize(OafRel oaf, String DELIM, String ENCLOSING) {
|
81
|
|
82
|
switch (oaf.getRelType()) {
|
83
|
case resultProject:
|
84
|
return getResultProject(oaf, DELIM, ENCLOSING);
|
85
|
default:
|
86
|
return buildRel(oaf, DELIM, ENCLOSING);
|
87
|
}
|
88
|
}
|
89
|
|
90
|
private static String buildRel(OafRel Rel, String DELIM, String ENCLOSING) {
|
91
|
|
92
|
return cleanId(Rel.getTarget(), DELIM, ENCLOSING) + DELIM;
|
93
|
|
94
|
}
|
95
|
|
96
|
public static void extractRelations(Oaf oaf, String DELIM, String ENCLOSING, Multimap<String, String> relations) {
|
97
|
OafEntity valueEntity = oaf.getEntity();
|
98
|
getOriginalId(valueEntity, relations, DELIM, ENCLOSING);
|
99
|
|
100
|
switch (valueEntity.getType()) {
|
101
|
case datasource:
|
102
|
getDatasourceLanguages(valueEntity, relations, DELIM, ENCLOSING);
|
103
|
case result:
|
104
|
getResultTopics(valueEntity, relations, DELIM, ENCLOSING);
|
105
|
getResultLanguages(valueEntity, relations, DELIM, ENCLOSING);
|
106
|
getResultClassifications(valueEntity, relations, DELIM, ENCLOSING);
|
107
|
getResultDatasources(valueEntity, relations, DELIM, ENCLOSING);
|
108
|
getResultConcepts(valueEntity, relations, DELIM, ENCLOSING);
|
109
|
getResultDois(valueEntity, relations, DELIM, ENCLOSING);
|
110
|
getResultCitations(valueEntity, relations, DELIM, ENCLOSING);
|
111
|
getResultDescriptions(valueEntity, relations, DELIM, ENCLOSING);
|
112
|
|
113
|
case project:
|
114
|
getProjectKeywords(valueEntity, relations, DELIM, ENCLOSING);
|
115
|
getProjectSubjects(valueEntity, relations, DELIM, ENCLOSING);
|
116
|
|
117
|
default:
|
118
|
}
|
119
|
|
120
|
}
|
121
|
|
122
|
|
123
|
private static void getOriginalId(OafEntity oafEntity, Multimap<String, String> relations, String DELIM, String ENCLOSING) {
|
124
|
|
125
|
String relName = oafEntity.getType().toString().toLowerCase() + "Oid";
|
126
|
for (String oid : oafEntity.getOriginalIdList()) {
|
127
|
relations.put(relName, cleanId(oid, DELIM, ENCLOSING));
|
128
|
}
|
129
|
|
130
|
}
|
131
|
|
132
|
private static void getProjectKeywords(OafEntity oafEntity, Multimap<String, String> relations, String DELIM, String ENCLOSING) {
|
133
|
relations.put("projectKeyword", getStringField(oafEntity.getProject().getMetadata().getKeywords().getValue(), DELIM, ENCLOSING));
|
134
|
|
135
|
}
|
136
|
|
137
|
private static void getProjectSubjects(OafEntity oafEntity, Multimap<String, String> relations, String DELIM, String ENCLOSING) {
|
138
|
for (StructuredProperty subj : oafEntity.getProject().getMetadata().getSubjectsList()) {
|
139
|
relations.put("projectSubject", getStringField(subj.getValue(), DELIM, ENCLOSING));
|
140
|
}
|
141
|
}
|
142
|
|
143
|
private static String getResultProject(OafRel oaf, String DELIM, String ENCLOSING) {
|
144
|
StringBuilder buff = new StringBuilder();
|
145
|
buff.append(cleanId(oaf.getTarget(), DELIM, ENCLOSING) + DELIM);
|
146
|
// TODO is declared as int!!!
|
147
|
long diff = DATEDIFF(oaf.getResultProject().getOutcome().getRelMetadata().getEnddate(), oaf.getResultProject().getOutcome().getRelMetadata().getStartdate());
|
148
|
if (diff < 0) {
|
149
|
diff = 0;
|
150
|
}
|
151
|
|
152
|
buff.append(getNumericField(String.valueOf(diff), DELIM, ENCLOSING));
|
153
|
return buff.toString();
|
154
|
}
|
155
|
|
156
|
|
157
|
private static void getDatasourceLanguages(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
158
|
Datasource d = valueEntity.getDatasource();
|
159
|
Metadata metadata = d.getMetadata();
|
160
|
|
161
|
for (StringField lang : metadata.getOdlanguagesList()) {
|
162
|
rels.put("datasourceLanguage", getStringField(lang.getValue(), DELIM, ENCLOSING));
|
163
|
}
|
164
|
}
|
165
|
|
166
|
private static void getResultLanguages(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
167
|
|
168
|
Result d = valueEntity.getResult();
|
169
|
Result.Metadata metadata = d.getMetadata();
|
170
|
if (metadata.getLanguage().getClassname() != null && !metadata.getLanguage().getClassname().isEmpty()) {
|
171
|
rels.put("resultLanguage", getStringField(metadata.getLanguage().getClassname(), DELIM, ENCLOSING));
|
172
|
}
|
173
|
|
174
|
}
|
175
|
|
176
|
private static void getResultDois(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
177
|
|
178
|
for (StructuredProperty pid : valueEntity.getPidList()) {
|
179
|
|
180
|
rels.put("resultPid",
|
181
|
getStringField(pid.getQualifier().getClassname(), DELIM, ENCLOSING) + getStringField(pid.getValue(), DELIM, ENCLOSING));
|
182
|
}
|
183
|
}
|
184
|
|
185
|
private static void getResultClassifications(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
186
|
|
187
|
Result result = valueEntity.getResult();
|
188
|
|
189
|
for (Instance instance : (result.getInstanceList())) {
|
190
|
String classification = instance.getInstancetype().getClassname();
|
191
|
if (classification != null && !classification.isEmpty()) {
|
192
|
rels.put("resultClassification", getStringField(instance.getInstancetype().getClassname(), DELIM, ENCLOSING));
|
193
|
// TODO HERE KEEP ONLY ONE CLASSIFICATIONS PER RESULT
|
194
|
break;
|
195
|
}
|
196
|
}
|
197
|
}
|
198
|
|
199
|
private static void getResultDescriptions(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
200
|
Result result = valueEntity.getResult();
|
201
|
//description
|
202
|
for (StringField s : result.getMetadata().getDescriptionList()) {
|
203
|
|
204
|
rels.put("resultDescription", getStringField(Jsoup.parse(s.getValue()).text(), DELIM, ENCLOSING));
|
205
|
}
|
206
|
}
|
207
|
|
208
|
private static void getResultConcepts(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
209
|
|
210
|
Result result = valueEntity.getResult();
|
211
|
|
212
|
|
213
|
for (Result.Context context : result.getMetadata().getContextList()) {
|
214
|
|
215
|
rels.put("resultConcept", cleanId(context.getId(), DELIM, ENCLOSING));
|
216
|
}
|
217
|
}
|
218
|
|
219
|
private static void getResultDatasources(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
220
|
Result result = valueEntity.getResult();
|
221
|
|
222
|
//TODO hosted by
|
223
|
for (Instance instance : (result.getInstanceList())) {
|
224
|
String hostedBy = instance.getHostedby().getKey();
|
225
|
if (hostedBy != null && !hostedBy.isEmpty()) {
|
226
|
rels.put("resultDatasource", cleanId(hostedBy, DELIM, ENCLOSING) + DELIM);
|
227
|
}
|
228
|
}
|
229
|
|
230
|
//TODO collected froms
|
231
|
for (FieldTypeProtos.KeyValue collectedFromValue : (valueEntity.getCollectedfromList())) {
|
232
|
|
233
|
String collectedFrom = collectedFromValue.getKey();
|
234
|
if (collectedFrom != null && !collectedFrom.isEmpty()) {
|
235
|
rels.put("resultDatasource", cleanId(collectedFrom, DELIM, ENCLOSING) + DELIM);
|
236
|
}
|
237
|
}
|
238
|
}
|
239
|
|
240
|
private static void getResultTopics(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
241
|
|
242
|
Result d = valueEntity.getResult();
|
243
|
Result.Metadata metadata = d.getMetadata();
|
244
|
|
245
|
List<StructuredProperty> Topics = metadata.getSubjectList();
|
246
|
|
247
|
for (StructuredProperty topic : Topics) {
|
248
|
// TODO result topics
|
249
|
rels.put("resultTopic", getStringField(topic.getValue(), DELIM, ENCLOSING));
|
250
|
}
|
251
|
}
|
252
|
|
253
|
|
254
|
private static void getResultCitations(OafEntity oafEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
255
|
for (FieldTypeProtos.ExtraInfo extraInfo : oafEntity.getExtraInfoList()) {
|
256
|
if (extraInfo.getName().equals("result citations")) {
|
257
|
DOMParser parser = new DOMParser();
|
258
|
try {
|
259
|
parser.parse(new InputSource(new java.io.StringReader(extraInfo.getValue())));
|
260
|
Document doc = parser.getDocument();
|
261
|
doc.getDocumentElement().normalize();
|
262
|
|
263
|
NodeList citations = doc.getElementsByTagName("citation");
|
264
|
for (int temp = 0; temp < citations.getLength(); temp++) {
|
265
|
Element citation = (Element) citations.item(temp);
|
266
|
NodeList ids = citation.getElementsByTagName("id");
|
267
|
for(int temp1 = 0; temp1 < ids.getLength(); temp1++){
|
268
|
Element id = (Element) ids.item(temp1);
|
269
|
if(id.getAttribute("type").equals("openaire")){
|
270
|
//System.out.println(id.getAttribute("value"));
|
271
|
rels.put("resultCitation", id.getAttribute("value"));
|
272
|
}
|
273
|
}
|
274
|
}
|
275
|
} catch (Exception e) {
|
276
|
|
277
|
}
|
278
|
|
279
|
/*
|
280
|
rels.put("resultCitation", getStringField(extraInfo.getTrust(), DELIM, ENCLOSING) +
|
281
|
getStringField(extraInfo.getProvenance(), DELIM, ENCLOSING) + getStringField(extraInfo.getValue(), DELIM, ENCLOSING));
|
282
|
*/
|
283
|
}
|
284
|
|
285
|
}
|
286
|
}
|
287
|
|
288
|
private static String buildDatasource(Oaf oaf, String DELIM, String ENCLOSING) {
|
289
|
Metadata metadata = oaf.getEntity().getDatasource().getMetadata();
|
290
|
StringBuilder buff = new StringBuilder();
|
291
|
|
292
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
293
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
294
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
295
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
296
|
|
297
|
// name
|
298
|
if (metadata.getOfficialname().getValue().equalsIgnoreCase("unknown")) {
|
299
|
buff.append(getStringField("Unknown Repository", DELIM, ENCLOSING));
|
300
|
} else {
|
301
|
buff.append(getStringField(metadata.getOfficialname().getValue(), DELIM, ENCLOSING));
|
302
|
}
|
303
|
// type
|
304
|
|
305
|
if (metadata.hasDatasourcetype()) {
|
306
|
buff.append(getStringField(metadata.getDatasourcetype().getClassname().replaceFirst(".*::", ""), DELIM, ENCLOSING));
|
307
|
}
|
308
|
|
309
|
// compatibility,
|
310
|
buff.append(getStringField(metadata.getOpenairecompatibility().getClassname(), DELIM, ENCLOSING));
|
311
|
|
312
|
// latitude
|
313
|
buff.append(getLatLongField(metadata.getLatitude().getValue(), DELIM, ENCLOSING));
|
314
|
|
315
|
// longtitude
|
316
|
buff.append(getLatLongField(metadata.getLongitude().getValue(), DELIM, ENCLOSING));
|
317
|
|
318
|
// dateofvalidation,
|
319
|
buff.append(getStringDateField(metadata.getDateofvalidation().getValue(), DELIM, ENCLOSING));
|
320
|
|
321
|
// yearofvalidation,
|
322
|
buff.append(getYearInt(metadata.getDateofvalidation().getValue(), DELIM, ENCLOSING));
|
323
|
|
324
|
//website
|
325
|
buff.append(getStringField(metadata.getWebsiteurl().getValue(), DELIM, ENCLOSING));
|
326
|
|
327
|
//piwik_id
|
328
|
String piwik_id = "";
|
329
|
for (String oid : oaf.getEntity().getOriginalIdList()) {
|
330
|
if (oid.contains("piwik")) {
|
331
|
piwik_id = oid.split(":")[1];
|
332
|
break;
|
333
|
}
|
334
|
}
|
335
|
buff.append(getStringField(cleanNumber(piwik_id), DELIM, ENCLOSING));
|
336
|
|
337
|
//harvested
|
338
|
buff.append(getStringField("false", DELIM, ENCLOSING));
|
339
|
|
340
|
// deletedByInference
|
341
|
buff.append(getStringField(String.valueOf(oaf.getDataInfo().getDeletedbyinference()), DELIM, ENCLOSING));
|
342
|
|
343
|
// number??
|
344
|
buff.append(getStringField("1", DELIM, ENCLOSING));
|
345
|
|
346
|
return buff.toString();
|
347
|
|
348
|
}
|
349
|
|
350
|
private static String buildOrganization(Oaf oaf, String DELIM, String ENCLOSING) {
|
351
|
|
352
|
StringBuilder buff = new StringBuilder();
|
353
|
Organization.Metadata metadata = oaf.getEntity().getOrganization().getMetadata();
|
354
|
|
355
|
// `organization_datasources`,
|
356
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
357
|
// organization_projects
|
358
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
359
|
// `name`,
|
360
|
buff.append(getStringField(metadata.getLegalname().getValue(), DELIM, ENCLOSING));
|
361
|
|
362
|
// `country`,
|
363
|
buff.append(getStringField(metadata.getCountry().getClassid(), DELIM, ENCLOSING));
|
364
|
|
365
|
//website
|
366
|
buff.append(getStringField(metadata.getWebsiteurl().getValue(), DELIM, ENCLOSING));
|
367
|
|
368
|
// deletedByInference
|
369
|
buff.append(getStringField(String.valueOf(oaf.getDataInfo().getDeletedbyinference()), DELIM, ENCLOSING));
|
370
|
|
371
|
// number
|
372
|
buff.append(getStringField("1", DELIM, ENCLOSING));
|
373
|
|
374
|
return buff.toString();
|
375
|
}
|
376
|
|
377
|
private static String buildResult(Oaf oaf, String DELIM, String ENCLOSING) {
|
378
|
StringBuilder buff = new StringBuilder();
|
379
|
|
380
|
Result.Metadata metadata = oaf.getEntity().getResult().getMetadata();
|
381
|
|
382
|
// result_topics/
|
383
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
384
|
// result_languages
|
385
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
386
|
// `result_projects`,
|
387
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
388
|
// `result_datasources`,
|
389
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
390
|
// `result_classifications`,
|
391
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
392
|
/// `result_infrastructures`,
|
393
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
394
|
// `result_claims`,
|
395
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
396
|
// `result_results`,
|
397
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
398
|
// pubtitle
|
399
|
|
400
|
String titleString = new String();
|
401
|
|
402
|
for (int i = 0; i < metadata.getTitleList().size(); i++) {
|
403
|
StructuredProperty title = metadata.getTitleList().get(i);
|
404
|
|
405
|
if (i == 0) {
|
406
|
titleString = title.getValue().replaceAll("\\s+", " ");
|
407
|
titleString = titleString.replaceAll("\n", " ");
|
408
|
}
|
409
|
break;
|
410
|
}
|
411
|
|
412
|
// pubtitle
|
413
|
buff.append(getStringField(titleString, DELIM, ENCLOSING));
|
414
|
|
415
|
// format
|
416
|
String formatString = new String();
|
417
|
|
418
|
for (StringField format : metadata.getFormatList()) {
|
419
|
formatString += format.getValue() + ";";
|
420
|
|
421
|
}
|
422
|
|
423
|
buff.append(getStringField(formatString, DELIM, ENCLOSING));
|
424
|
// publisher
|
425
|
|
426
|
buff.append(getStringField(metadata.getPublisher().getValue(), DELIM, ENCLOSING));
|
427
|
|
428
|
// journal
|
429
|
|
430
|
buff.append(getStringField(metadata.getJournal().getName(), DELIM, ENCLOSING)); //#null#!
|
431
|
|
432
|
// year
|
433
|
buff.append(getYearInt(metadata.getDateofacceptance().getValue(), DELIM, ENCLOSING));
|
434
|
|
435
|
// date CHANGED THIS TO DATE FORMAT
|
436
|
buff.append(getStringDateField(metadata.getDateofacceptance().getValue(), DELIM, ENCLOSING));
|
437
|
|
438
|
// access_mode,
|
439
|
buff.append(getStringField(getAccessMode(oaf.getEntity().getResult()), DELIM, ENCLOSING));
|
440
|
|
441
|
// bestlicense
|
442
|
buff.append(getStringField(getBestLicense(oaf.getEntity().getResult()), DELIM, ENCLOSING));
|
443
|
|
444
|
// type
|
445
|
buff.append(getStringField(metadata.getResulttype().getClassname(), DELIM, ENCLOSING));
|
446
|
|
447
|
// embargo_end_date
|
448
|
buff.append(getStringDateField(metadata.getEmbargoenddate().getValue(), DELIM, ENCLOSING));
|
449
|
|
450
|
// `authors`,
|
451
|
int authors = 0;
|
452
|
|
453
|
String delayed = "no";
|
454
|
|
455
|
for (OafRel rel : oaf.getEntity().getCachedRelList()) {
|
456
|
|
457
|
/*
|
458
|
if (rel.getRelType().equals(RelType.personResult)) {
|
459
|
|
460
|
authors++;
|
461
|
} else
|
462
|
*/
|
463
|
if (rel.getRelType().equals(RelType.resultProject))
|
464
|
// TODO remember : in result Project, first id is project, second is
|
465
|
// result.
|
466
|
{
|
467
|
String daysfromend = getYearDifferenceInteger(rel.getResultProject().getOutcome().getRelMetadata().getEnddate(),
|
468
|
rel.getResultProject().getOutcome().getRelMetadata().getStartdate(), DELIM, ENCLOSING);
|
469
|
if (Integer.parseInt(daysfromend) > 0) {
|
470
|
delayed = "yes";
|
471
|
}
|
472
|
}
|
473
|
}
|
474
|
|
475
|
// `delayed`,
|
476
|
buff.append(getStringField(delayed, DELIM, ENCLOSING));
|
477
|
//authors
|
478
|
buff.append(getNumericField(String.valueOf(authors), DELIM, ENCLOSING));
|
479
|
|
480
|
String sources = new String();
|
481
|
|
482
|
|
483
|
for (Instance instance : (oaf.getEntity().getResult().getInstanceList())) {
|
484
|
List<String> urls = instance.getUrlList();
|
485
|
for (String url : urls) {
|
486
|
sources += cleanUrl(url, DELIM, ENCLOSING) + " ;";
|
487
|
}
|
488
|
}
|
489
|
|
490
|
//sources
|
491
|
sources = ENCLOSING + sources + ENCLOSING + DELIM;
|
492
|
|
493
|
buff.append(sources);
|
494
|
|
495
|
// deletedByInference
|
496
|
buff.append(getStringField(String.valueOf(oaf.getDataInfo().getDeletedbyinference()), DELIM, ENCLOSING));
|
497
|
|
498
|
|
499
|
// number??
|
500
|
buff.append(getStringField("1", DELIM, ENCLOSING));
|
501
|
return buff.toString();
|
502
|
|
503
|
}
|
504
|
|
505
|
|
506
|
/*
|
507
|
private static String getBestLicense(Result result) {
|
508
|
Qualifier bestLicense = null;
|
509
|
LicenseComparator lc = new LicenseComparator();
|
510
|
for (Instance instance : (result.getInstanceList())) {
|
511
|
if (lc.compare(bestLicense, instance.getLicence()) > 0) {
|
512
|
bestLicense = instance.getLicence();
|
513
|
}
|
514
|
}
|
515
|
if (bestLicense != null) {
|
516
|
return bestLicense.getClassname();
|
517
|
} else {
|
518
|
return null;
|
519
|
}
|
520
|
}
|
521
|
|
522
|
// TODO here iterate over all values
|
523
|
private static String getAccessMode(Result result) {
|
524
|
String accessMode = " ";
|
525
|
for (Instance instance : (result.getInstanceList())) {
|
526
|
if (instance.getLicence().getClassname() != null && !instance.getLicence().getClassname().isEmpty()) {
|
527
|
accessMode = instance.getLicence().getClassname();
|
528
|
break;
|
529
|
}
|
530
|
|
531
|
}
|
532
|
|
533
|
return accessMode;
|
534
|
}
|
535
|
*/
|
536
|
|
537
|
private static String getBestLicense(Result result) {
|
538
|
Qualifier bestLicense = null;
|
539
|
LicenseComparator lc = new LicenseComparator();
|
540
|
for (Instance instance : (result.getInstanceList())) {
|
541
|
if (lc.compare(bestLicense, instance.getAccessright()) > 0) {
|
542
|
bestLicense = instance.getAccessright();
|
543
|
}
|
544
|
}
|
545
|
if (bestLicense != null) {
|
546
|
return bestLicense.getClassname();
|
547
|
} else {
|
548
|
return null;
|
549
|
}
|
550
|
}
|
551
|
|
552
|
// TODO here iterate over all values
|
553
|
private static String getAccessMode(Result result) {
|
554
|
String accessMode = " ";
|
555
|
for (Instance instance : (result.getInstanceList())) {
|
556
|
if (instance.getAccessright().getClassname() != null && !instance.getAccessright().getClassname().isEmpty()) {
|
557
|
accessMode = instance.getAccessright().getClassname();
|
558
|
break;
|
559
|
}
|
560
|
|
561
|
}
|
562
|
|
563
|
return accessMode;
|
564
|
}
|
565
|
|
566
|
private static String buildProject(Oaf oaf, String DELIM, String ENCLOSING) {
|
567
|
|
568
|
StringBuilder buff = new StringBuilder();
|
569
|
Project.Metadata metadata = oaf.getEntity().getProject().getMetadata();
|
570
|
|
571
|
// project_organizations
|
572
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
573
|
|
574
|
// project_results
|
575
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
576
|
|
577
|
|
578
|
// `acronym`,
|
579
|
String acronym = metadata.getAcronym().getValue();
|
580
|
if (acronym.equalsIgnoreCase("UNKNOWN")) {
|
581
|
acronym = metadata.getTitle().getValue();
|
582
|
}
|
583
|
|
584
|
buff.append(getStringField(acronym, DELIM, ENCLOSING));
|
585
|
|
586
|
//title!
|
587
|
buff.append(getStringField(metadata.getTitle().getValue(), DELIM, ENCLOSING));
|
588
|
|
589
|
List<StringField> fundList = metadata.getFundingtreeList();
|
590
|
|
591
|
if (!fundList.isEmpty()) // `funding_lvl0`,
|
592
|
{
|
593
|
//TODO funder + 3 funding levels
|
594
|
/* funder text,
|
595
|
funding_lvl0 text,
|
596
|
funding_lvl1 text,
|
597
|
funding_lvl2 text,
|
598
|
funding_lvl3 text,*/
|
599
|
buff.append(FundingParser.getFundingInfo(fundList.get(0).getValue(), DELIM, ENCLOSING));
|
600
|
|
601
|
} else {
|
602
|
buff.append(FundingParser.getFundingInfo("", DELIM, ENCLOSING));
|
603
|
|
604
|
}
|
605
|
|
606
|
String sc39 = metadata.getEcsc39().getValue().toString();
|
607
|
if (sc39.equalsIgnoreCase("true") || sc39.equalsIgnoreCase("t") || sc39.contains("yes")) {
|
608
|
sc39 = "yes";
|
609
|
} else if (sc39.equalsIgnoreCase("false") || sc39.equalsIgnoreCase("f") || sc39.contains("no")) {
|
610
|
sc39 = "no";
|
611
|
}
|
612
|
|
613
|
buff.append(getStringField(sc39, DELIM, ENCLOSING));
|
614
|
|
615
|
|
616
|
//project_type
|
617
|
buff.append(getStringField(metadata.getContracttype().getClassid(),DELIM, ENCLOSING));
|
618
|
|
619
|
// `url`,
|
620
|
buff.append(getStringField(metadata.getWebsiteurl().getValue(), DELIM, ENCLOSING));
|
621
|
|
622
|
// start_year
|
623
|
|
624
|
buff.append(getYearInt(metadata.getStartdate().getValue(), DELIM, ENCLOSING));
|
625
|
|
626
|
// end_year
|
627
|
buff.append(getYearInt(metadata.getEnddate().getValue(), DELIM, ENCLOSING));
|
628
|
|
629
|
// duration enddate-startdate
|
630
|
|
631
|
buff.append(getYearDifferenceInteger(metadata.getEnddate().getValue(), metadata.getStartdate().getValue(), DELIM, ENCLOSING));
|
632
|
|
633
|
// haspubs
|
634
|
buff.append(getStringField("no", DELIM, ENCLOSING));
|
635
|
|
636
|
// numpubs
|
637
|
buff.append(getNumericField("0", DELIM, ENCLOSING));
|
638
|
|
639
|
// enddate
|
640
|
buff.append(getNumericField(metadata.getEnddate().getValue(), DELIM, ENCLOSING));
|
641
|
|
642
|
// startdate
|
643
|
buff.append(getNumericField(metadata.getStartdate().getValue(), DELIM, ENCLOSING));
|
644
|
|
645
|
// `daysforlastpub`,
|
646
|
buff.append(getNumericField("", DELIM, ENCLOSING));
|
647
|
|
648
|
// `delayedpubs`,
|
649
|
buff.append(getNumericField("", DELIM, ENCLOSING));
|
650
|
|
651
|
//call identifier
|
652
|
buff.append(getStringField(metadata.getCallidentifier().getValue(), DELIM, ENCLOSING));
|
653
|
//code
|
654
|
buff.append(getStringField(metadata.getCode().getValue(), DELIM, ENCLOSING));
|
655
|
|
656
|
//esc39
|
657
|
buff.append(getStringField(metadata.getEcsc39().getValue(), DELIM, ENCLOSING));
|
658
|
|
659
|
//getUrl
|
660
|
String sources = new String();
|
661
|
|
662
|
for (Instance instance : (oaf.getEntity().getResult().getInstanceList())) {
|
663
|
List<String> urls = instance.getUrlList();
|
664
|
for (String u : urls) {
|
665
|
sources += u + ";";
|
666
|
}
|
667
|
}
|
668
|
|
669
|
sources = cleanUrl(sources, DELIM, ENCLOSING);
|
670
|
sources = ENCLOSING + sources + ENCLOSING + DELIM;
|
671
|
|
672
|
buff.append(sources);
|
673
|
|
674
|
// deletedByInference
|
675
|
buff.append(getStringField(String.valueOf(oaf.getDataInfo().getDeletedbyinference()), DELIM, ENCLOSING));
|
676
|
|
677
|
// `number`
|
678
|
buff.append(getStringField("1", DELIM, ENCLOSING));
|
679
|
return buff.toString();
|
680
|
|
681
|
}
|
682
|
|
683
|
/*
|
684
|
private static String buildPerson(Oaf oaf, String DELIM, String ENCLOSING) {
|
685
|
StringBuilder buff = new StringBuilder();
|
686
|
|
687
|
PersonProtos.Person.Metadata metadata = oaf.getEntity().getPerson().getMetadata();
|
688
|
|
689
|
// `person_id`,
|
690
|
//buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
691
|
|
692
|
|
693
|
// person_result
|
694
|
buff.append(cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING) + DELIM);
|
695
|
|
696
|
|
697
|
//firstName
|
698
|
buff.append(getStringField(metadata.getFirstname().getValue(), DELIM, ENCLOSING));
|
699
|
|
700
|
//LastName
|
701
|
|
702
|
String secondName = new String();
|
703
|
for (StringField s : metadata.getSecondnamesList()) {
|
704
|
secondName += s.getValue().replace("\n", " ").replace(",", " ") + " ";
|
705
|
}
|
706
|
|
707
|
buff.append(getStringField(secondName, DELIM, ENCLOSING));
|
708
|
|
709
|
// `fullname`,
|
710
|
buff.append(getStringField(metadata.getFullname().getValue(), DELIM, ENCLOSING));
|
711
|
|
712
|
// `Nationality`,
|
713
|
buff.append(getStringField(metadata.getNationality().getClassid(), DELIM, ENCLOSING));
|
714
|
|
715
|
// `Email`,
|
716
|
buff.append(getStringField(metadata.getEmail().getValue(), DELIM, ENCLOSING));
|
717
|
|
718
|
// `Phone`,
|
719
|
buff.append(getStringField(metadata.getPhone().getValue(), DELIM, ENCLOSING));
|
720
|
|
721
|
//deletedByInference
|
722
|
|
723
|
buff.append(getStringField(String.valueOf(oaf.getDataInfo().getDeletedbyinference()), DELIM, ENCLOSING));
|
724
|
|
725
|
// `number`,
|
726
|
buff.append(getStringField("1", DELIM, ENCLOSING));
|
727
|
|
728
|
return buff.toString();
|
729
|
|
730
|
}
|
731
|
*/
|
732
|
|
733
|
|
734
|
private static String getYearDifferenceInteger(String enddate, String startdate, String DELIM, String ENCLOSING) {
|
735
|
|
736
|
if (enddate != null && !enddate.isEmpty() && startdate != null && !startdate.isEmpty()) {
|
737
|
|
738
|
String[] split = startdate.split("-");
|
739
|
|
740
|
if (split == null || split.length == 0) {
|
741
|
return ENCLOSING + "0" + ENCLOSING + DELIM;
|
742
|
}
|
743
|
|
744
|
int Startdate = Integer.parseInt(split[0]);
|
745
|
|
746
|
split = enddate.split("-");
|
747
|
|
748
|
if (split == null || split.length == 0) {
|
749
|
return ENCLOSING + "0" + ENCLOSING + DELIM;
|
750
|
}
|
751
|
|
752
|
int Enddate = Integer.parseInt(split[0]);
|
753
|
|
754
|
int diff = Enddate - Startdate;
|
755
|
|
756
|
return ENCLOSING + diff + ENCLOSING + DELIM;
|
757
|
|
758
|
}
|
759
|
|
760
|
return ENCLOSING + "0" + ENCLOSING + DELIM;
|
761
|
}
|
762
|
|
763
|
private static String getYearInt(String data, String DELIM, String ENCLOSING) {
|
764
|
if (data == null || data.isEmpty() || data.equals("-1")) {
|
765
|
return ENCLOSING + "0" + ENCLOSING + DELIM;
|
766
|
}
|
767
|
|
768
|
String[] split = data.split("-");
|
769
|
|
770
|
if (split == null || split.length == 0) {
|
771
|
return ENCLOSING + "0" + ENCLOSING + DELIM;
|
772
|
}
|
773
|
|
774
|
String year = split[0];
|
775
|
|
776
|
year = cleanNumber(year);
|
777
|
|
778
|
if (year == null || year.isEmpty()) year = "0";
|
779
|
|
780
|
return ENCLOSING + year + ENCLOSING + DELIM;
|
781
|
|
782
|
}
|
783
|
|
784
|
private static String cleanNumber(String number) {
|
785
|
number = number.replaceAll("[^A-Za-z0-9:,____]", "");
|
786
|
return number;
|
787
|
}
|
788
|
|
789
|
private static String getLatLongField(String data, String DELIM, String ENCLOSING) {
|
790
|
|
791
|
if (data == null || data.isEmpty()) return ENCLOSING + "null" + ENCLOSING + DELIM;
|
792
|
|
793
|
return ENCLOSING + data.replaceAll("[^-0-9.]+", "") + ENCLOSING + DELIM;
|
794
|
|
795
|
}
|
796
|
|
797
|
private static String getStringField(String data, String DELIM, String ENCLOSING) {
|
798
|
|
799
|
if (data == null || data.isEmpty()) return ENCLOSING + "null" + ENCLOSING + DELIM;
|
800
|
|
801
|
return ENCLOSING + clean(data, DELIM, ENCLOSING) + ENCLOSING + DELIM;
|
802
|
|
803
|
}
|
804
|
|
805
|
private static String getStringDateField(String data, String DELIM, String ENCLOSING) {
|
806
|
if (data == null || data.isEmpty() || data.equals("") || data.equals("-1")) {
|
807
|
return ENCLOSING + "0" + ENCLOSING + DELIM;
|
808
|
} else {
|
809
|
data = data.replace(DELIM, " ");
|
810
|
data = data.replace(ENCLOSING, " ");
|
811
|
data = data.replaceAll("\\r\\n|\\r|\\n", "");
|
812
|
return ENCLOSING + data + ENCLOSING + DELIM;
|
813
|
}
|
814
|
}
|
815
|
|
816
|
private static String getNumericField(String data, String DELIM, String ENCLOSING) {
|
817
|
if (data == null || data.isEmpty() || data.equals("")) {
|
818
|
return ENCLOSING + "0" + ENCLOSING + DELIM;
|
819
|
} else {
|
820
|
return ENCLOSING + data + ENCLOSING + DELIM;
|
821
|
}
|
822
|
}
|
823
|
|
824
|
public static String getId(Oaf oaf, String DELIM, String ENCLOSING) {
|
825
|
switch (oaf.getKind()) {
|
826
|
case entity:
|
827
|
return cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING);
|
828
|
case relation:
|
829
|
|
830
|
return cleanId(oaf.getRel().getSource(), DELIM, ENCLOSING);
|
831
|
|
832
|
}
|
833
|
return null;
|
834
|
|
835
|
}
|
836
|
|
837
|
public static String getId(OafRel relOaf, String DELIM, String ENCLOSING) {
|
838
|
return cleanId(relOaf.getSource(), DELIM, ENCLOSING);
|
839
|
}
|
840
|
|
841
|
public static String clean(String value, String DELIM, String ENCLOSING) {
|
842
|
if (value != null) {
|
843
|
|
844
|
value = value.replaceAll("[\"\\r\\\\;]", "");
|
845
|
value = value.replace(DELIM, " ");
|
846
|
value = value.replace(ENCLOSING, " ");
|
847
|
value = value.replaceAll("\\r\\n|\\r|\\n", " ");
|
848
|
//value = value.replace("\"", "");
|
849
|
//value = value.replace("'", "");
|
850
|
//value = value.replace("«", " ");
|
851
|
//value = value.replace("»", " ");
|
852
|
|
853
|
//value = value.replaceAll("[^a-zA-Z0-9 .-_:/!@+=]+", " ");
|
854
|
|
855
|
return value;
|
856
|
|
857
|
} else {
|
858
|
return "";
|
859
|
|
860
|
}
|
861
|
|
862
|
}
|
863
|
|
864
|
public static String cleanId(String value, String DELIM, String ENCLOSING) {
|
865
|
if (value != null) {
|
866
|
// TODO DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____::
|
867
|
|
868
|
// to datacite____:: )
|
869
|
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA
|
870
|
value = value.replaceFirst(".*\\|", "");
|
871
|
value = value.replace("\n", "");
|
872
|
value = value.replace(ENCLOSING, "");
|
873
|
value = value.replace(DELIM, "");
|
874
|
value = value.replace("\"", "");
|
875
|
value = value.replace("«", " ");
|
876
|
value = value.replace("»", " ");
|
877
|
}
|
878
|
|
879
|
return ENCLOSING + value + ENCLOSING;
|
880
|
|
881
|
}
|
882
|
|
883
|
|
884
|
public static String cleanUrl(String value, String DELIM, String ENCLOSING) {
|
885
|
value = value.replace(DELIM, " ");
|
886
|
value = value.replace(ENCLOSING, " ");
|
887
|
value = value.replace(" ", "");
|
888
|
value = value.replace("\n", "");
|
889
|
return value;
|
890
|
}
|
891
|
|
892
|
|
893
|
public static long DATEDIFF(String startDate, String endDate) {
|
894
|
long MILLISECS_PER_DAY = 24 * 60 * 60 * 1000;
|
895
|
long days = 0l;
|
896
|
SimpleDateFormat format = new SimpleDateFormat("yyyy-MM-dd"); // "dd/MM/yyyy HH:mm:ss");
|
897
|
// <startdate>2011-09-01</startdate>
|
898
|
// <enddate>2015-08-31</enddate>
|
899
|
Date dateIni = null;
|
900
|
Date dateFin = null;
|
901
|
|
902
|
if (startDate == null || startDate.isEmpty() || endDate == null || endDate.isEmpty()) {
|
903
|
return 0;
|
904
|
}
|
905
|
try {
|
906
|
dateIni = (Date) format.parse(startDate);
|
907
|
dateFin = (Date) format.parse(endDate);
|
908
|
days = (dateFin.getTime() - dateIni.getTime()) / MILLISECS_PER_DAY;
|
909
|
} catch (Exception e) {
|
910
|
|
911
|
return 0;
|
912
|
}
|
913
|
|
914
|
return days;
|
915
|
}
|
916
|
|
917
|
|
918
|
}
|