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