1
|
package eu.dnetlib.data.mapreduce.hbase.lodExport.utils;
|
2
|
|
3
|
import eu.dnetlib.data.mapreduce.util.LicenseComparator;
|
4
|
import eu.dnetlib.data.proto.DatasourceProtos.Datasource;
|
5
|
import eu.dnetlib.data.proto.DatasourceProtos.Datasource.Metadata;
|
6
|
import eu.dnetlib.data.proto.FieldTypeProtos;
|
7
|
import eu.dnetlib.data.proto.FieldTypeProtos.StringField;
|
8
|
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty;
|
9
|
import eu.dnetlib.data.proto.OafProtos;
|
10
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
11
|
import eu.dnetlib.data.proto.OafProtos.OafEntity;
|
12
|
import eu.dnetlib.data.proto.OafProtos.OafRel;
|
13
|
import eu.dnetlib.data.proto.OrganizationProtos.Organization;
|
14
|
import eu.dnetlib.data.proto.PersonProtos;
|
15
|
import eu.dnetlib.data.proto.ProjectProtos.Project;
|
16
|
import eu.dnetlib.data.proto.RelTypeProtos.RelType;
|
17
|
import eu.dnetlib.data.proto.ResultProtos;
|
18
|
import eu.dnetlib.data.proto.ResultProtos.Result;
|
19
|
import eu.dnetlib.data.proto.ResultProtos.Result.Instance;
|
20
|
import org.apache.log4j.Logger;
|
21
|
|
22
|
import java.text.SimpleDateFormat;
|
23
|
import java.util.ArrayList;
|
24
|
import java.util.Date;
|
25
|
import java.util.List;
|
26
|
|
27
|
/**
|
28
|
* @author eri Simple serializer that parses input Oaf Protos and prepares them
|
29
|
* for sqoop
|
30
|
*/
|
31
|
public class Serializer {
|
32
|
|
33
|
private static String DELIM;
|
34
|
private Logger log = Logger.getLogger(this.getClass());
|
35
|
private String NULL_STRING = null;
|
36
|
private String NULL_NUM = "0";
|
37
|
|
38
|
private static String ENCLOSED;
|
39
|
private FundingParser fundingParser = new FundingParser();
|
40
|
|
41
|
|
42
|
public Serializer() {
|
43
|
}
|
44
|
|
45
|
public String serialize(Oaf oaf) {
|
46
|
|
47
|
switch (oaf.getKind()) {
|
48
|
case entity:
|
49
|
OafEntity valueEntity = oaf.getEntity();
|
50
|
|
51
|
|
52
|
switch (valueEntity.getType()) {
|
53
|
case datasource:
|
54
|
|
55
|
return buildDatasource(valueEntity);
|
56
|
|
57
|
case organization:
|
58
|
|
59
|
return buildOrganization(valueEntity);
|
60
|
|
61
|
case project:
|
62
|
|
63
|
return buildProject(valueEntity);
|
64
|
case result:
|
65
|
|
66
|
return buildResult(valueEntity);
|
67
|
case person:
|
68
|
return buildPerson(valueEntity);
|
69
|
default:
|
70
|
log.error("wrong type");
|
71
|
break;
|
72
|
}
|
73
|
break;
|
74
|
case relation:
|
75
|
OafRel valueRel = oaf.getRel();
|
76
|
|
77
|
return buildRel(valueRel);
|
78
|
|
79
|
}
|
80
|
|
81
|
return null;
|
82
|
|
83
|
}
|
84
|
|
85
|
public String serialize(OafRel oaf) {
|
86
|
|
87
|
switch (oaf.getRelType()) {
|
88
|
|
89
|
default:
|
90
|
return buildRel(oaf);
|
91
|
}
|
92
|
|
93
|
}
|
94
|
|
95
|
|
96
|
public ArrayList<String> extractRelations(Oaf oaf) {
|
97
|
OafEntity valueEntity = oaf.getEntity();
|
98
|
ArrayList<String> relations = new ArrayList<String>();
|
99
|
switch (valueEntity.getType()) {
|
100
|
case result:
|
101
|
relations.addAll(getResultDatasources(valueEntity));
|
102
|
relations.addAll(getDedups(valueEntity));
|
103
|
return relations;
|
104
|
case datasource:
|
105
|
relations.addAll(getDedups(valueEntity));
|
106
|
return relations;
|
107
|
case person:
|
108
|
relations.addAll(getDedups(valueEntity));
|
109
|
return relations;
|
110
|
default:
|
111
|
return null;
|
112
|
}
|
113
|
|
114
|
}
|
115
|
|
116
|
private String buildRel(OafRel Rel) {
|
117
|
|
118
|
return getStringField(Rel.getRelType().getDescriptorForType().getName()) + getStringField(Rel.getSource()) + getStringField(Rel.getTarget());
|
119
|
|
120
|
}
|
121
|
|
122
|
|
123
|
private String getResultResult(OafRel oaf) {
|
124
|
|
125
|
|
126
|
String buff = new String();
|
127
|
|
128
|
buff += getStringField(oaf.getTarget());
|
129
|
|
130
|
buff += getStringField(String.valueOf(oaf.getResultResult().getSimilarity().getSimilarity()));
|
131
|
/* • hasAmongTopNSimilarDocuments: r1 hasAmongTopNSimilarDocuments r2 means that Results r1 and r2 are similar, and that we also have r2 isAmongTopNSimilarDocuments of r1.
|
132
|
In OpenAIRE, N so far always equals 20.
|
133
|
• isAmongTopNSimilarDocuments: r1 isAmongTopNSimilarDocuments r2 means that Results r1 and r2 are similar and that r2 hasAmongTopNSimilarDocuments of r1;
|
134
|
• isRelatedTo: two results are somehow related to each other. OpenAIRE may further refine the semantics of possible types of "relatedness" by adding new classes in the Qualifier.
|
135
|
Scheme "dnet:result_result_relations";
|
136
|
oaf. */
|
137
|
return buff;
|
138
|
}
|
139
|
|
140
|
private String getHeader(OafEntity data) {
|
141
|
String buff = new String();
|
142
|
|
143
|
// EntityType
|
144
|
buff += getStringField(data.getType().name());
|
145
|
// OpenaireID
|
146
|
buff += getStringField(getStringDateField(cleanId(data.getId())));
|
147
|
|
148
|
// dateOfTransformation
|
149
|
// TODO change to dateOfTransformation here when released
|
150
|
buff += getStringField(data.getDateofcollection());
|
151
|
// dateOfCollection
|
152
|
buff += getStringField(data.getDateofcollection());
|
153
|
|
154
|
String originalId = new String();
|
155
|
// originalId
|
156
|
|
157
|
for (String oid : data.getOriginalIdList()) {
|
158
|
originalId += oid + ";";
|
159
|
}
|
160
|
|
161
|
buff += getStringField(originalId);
|
162
|
|
163
|
return buff;
|
164
|
|
165
|
}
|
166
|
|
167
|
private String buildDatasource(OafEntity data) {
|
168
|
|
169
|
|
170
|
Datasource d = data.getDatasource();
|
171
|
|
172
|
Metadata metadata = d.getMetadata();
|
173
|
|
174
|
|
175
|
String buff = getHeader(data);
|
176
|
|
177
|
|
178
|
/*
|
179
|
|
180
|
OfficialName
|
181
|
Englishname
|
182
|
Websiteurl
|
183
|
Logourl
|
184
|
Contactemail
|
185
|
Namespaceprefix
|
186
|
Latitude
|
187
|
Longitude
|
188
|
Dateofvalidation
|
189
|
Description
|
190
|
SubjectsList
|
191
|
Odnumberofitems
|
192
|
Odnumberofitemsdate
|
193
|
Odpolicies
|
194
|
OdlanguagesList
|
195
|
OdcontenttypesList
|
196
|
Accessinfopackage
|
197
|
Releasestartdate
|
198
|
Releaseenddate
|
199
|
Missionstatementurl
|
200
|
Dataprovider
|
201
|
Serviceprovider
|
202
|
Databaseaccessrestriction
|
203
|
Datauploadtype
|
204
|
Datauploadrestriction
|
205
|
Versioning
|
206
|
Citationguidelineurl
|
207
|
Qualitymanagementkind
|
208
|
Pidsystems
|
209
|
Certificates
|
210
|
Policies
|
211
|
Trust
|
212
|
*/
|
213
|
|
214
|
//Datasourcetype
|
215
|
if (metadata.hasDatasourcetype()) {
|
216
|
buff += getStringField(metadata.getDatasourcetype().getClassname().replaceFirst(".*::", ""));
|
217
|
} else {
|
218
|
buff += getStringField(null);
|
219
|
}
|
220
|
|
221
|
//Openairecompatibility
|
222
|
buff += getStringField(metadata.getOpenairecompatibility().getClassname());
|
223
|
|
224
|
//Official Name
|
225
|
buff += getStringField(metadata.getOfficialname().getValue());
|
226
|
|
227
|
// English Name
|
228
|
buff += getStringField(metadata.getEnglishname().getValue());
|
229
|
|
230
|
//Website URL
|
231
|
buff += getStringField(metadata.getWebsiteurl().getValue());
|
232
|
|
233
|
//LogoURL
|
234
|
buff += getStringField(metadata.getLogourl().getValue());
|
235
|
|
236
|
//Email
|
237
|
buff += getStringField(metadata.getContactemail().getValue());
|
238
|
|
239
|
//Namespaceprefix
|
240
|
buff += getStringField(metadata.getNamespaceprefix().getValue());
|
241
|
|
242
|
// latitude
|
243
|
buff += getStringField(metadata.getLatitude().getValue());
|
244
|
|
245
|
// longtitude
|
246
|
buff += getStringField(metadata.getLongitude().getValue());
|
247
|
|
248
|
// dateofvalidation,
|
249
|
buff += getStringField(metadata.getDateofvalidation().getValue());
|
250
|
|
251
|
//Description
|
252
|
buff += getStringField(metadata.getDescription().getValue());
|
253
|
|
254
|
//subjects
|
255
|
String subj = new String();
|
256
|
for (StructuredProperty s : metadata.getSubjectsList()) {
|
257
|
subj += s.getValue() + ';';
|
258
|
|
259
|
}
|
260
|
|
261
|
//subjects
|
262
|
buff += getStringField(subj);
|
263
|
|
264
|
//Number of items
|
265
|
buff += getStringField(metadata.getOdnumberofitems().getValue());
|
266
|
|
267
|
//Date of number of items
|
268
|
buff += getStringField(metadata.getOdnumberofitemsdate().getValue());
|
269
|
|
270
|
// Policies
|
271
|
buff += getStringField(metadata.getOdpolicies().getValue());
|
272
|
|
273
|
//languages
|
274
|
String languages = new String();
|
275
|
for (StringField lang : metadata.getOdlanguagesList()) {
|
276
|
languages += lang.getValue() + ";";
|
277
|
}
|
278
|
buff += getStringField(languages);
|
279
|
|
280
|
|
281
|
// Content type
|
282
|
String contentType = new String();
|
283
|
for (StringField c : metadata.getOdcontenttypesList()) {
|
284
|
contentType += c.getValue() + ";";
|
285
|
}
|
286
|
buff += getStringField(contentType);
|
287
|
|
288
|
//Access info package
|
289
|
String accessInfo = new String();
|
290
|
for (StringField c : metadata.getAccessinfopackageList()) {
|
291
|
accessInfo += c.getValue() + ";";
|
292
|
}
|
293
|
buff += getStringField(accessInfo);
|
294
|
|
295
|
//Release start date
|
296
|
buff += getStringField(metadata.getReleasestartdate().getValue());
|
297
|
|
298
|
//Release end date
|
299
|
buff += getStringField(metadata.getReleaseenddate().getValue());
|
300
|
|
301
|
//Mission statement url
|
302
|
buff += getStringField(metadata.getMissionstatementurl().getValue());
|
303
|
|
304
|
//Data provider
|
305
|
buff += getStringField(String.valueOf(metadata.getDataprovider().getValue()));
|
306
|
|
307
|
//Service provider
|
308
|
buff += getStringField(String.valueOf(metadata.getServiceprovider().getValue()));
|
309
|
|
310
|
//Database access type
|
311
|
buff += getStringField(metadata.getDatabaseaccessrestriction().getValue());
|
312
|
|
313
|
//Data upload type
|
314
|
buff += getStringField(metadata.getDatauploadtype().getValue());
|
315
|
|
316
|
//Data upload restrictions
|
317
|
buff += getStringField(metadata.getDatauploadrestriction().getValue());
|
318
|
|
319
|
//Versioning
|
320
|
buff += getStringField(String.valueOf(metadata.getVersioning().getValue()));
|
321
|
|
322
|
//Citation guideline url
|
323
|
buff += getStringField(String.valueOf(metadata.getCitationguidelineurl().getValue()));
|
324
|
|
325
|
//Quality management kind
|
326
|
buff += getStringField(String.valueOf(metadata.getQualitymanagementkind().getValue()));
|
327
|
|
328
|
//PID systems
|
329
|
buff += getStringField(metadata.getPidsystems().getValue());
|
330
|
|
331
|
//Certificates
|
332
|
buff += getStringField(metadata.getCertificates().getValue());
|
333
|
|
334
|
//Policies
|
335
|
String policies = new String();
|
336
|
for (FieldTypeProtos.KeyValue property : metadata.getPoliciesList()) {
|
337
|
policies += property.getValue() + ";";
|
338
|
}
|
339
|
buff += getStringField(policies);
|
340
|
|
341
|
return buff;
|
342
|
}
|
343
|
|
344
|
|
345
|
private String buildOrganization(OafEntity data) {
|
346
|
|
347
|
|
348
|
Organization organization = data.getOrganization();
|
349
|
eu.dnetlib.data.proto.OrganizationProtos.Organization.Metadata metadata = organization.getMetadata();
|
350
|
|
351
|
String buff = getHeader(data);
|
352
|
|
353
|
//getLegalshortname
|
354
|
buff += getStringField(metadata.getLegalshortname().getValue());
|
355
|
// `name`,
|
356
|
buff += getStringField(metadata.getLegalname().getValue());
|
357
|
//website URL
|
358
|
buff += getStringField(metadata.getWebsiteurl().getValue());
|
359
|
//logourl
|
360
|
buff += getStringField(metadata.getLogourl().getValue());
|
361
|
// `country`,
|
362
|
buff += getStringField(metadata.getCountry().getClassname());
|
363
|
|
364
|
|
365
|
return buff;
|
366
|
|
367
|
}
|
368
|
|
369
|
private String buildResult(OafEntity data) {
|
370
|
|
371
|
Result result = data.getResult();
|
372
|
eu.dnetlib.data.proto.ResultProtos.Result.Metadata metadata = result.getMetadata();
|
373
|
|
374
|
|
375
|
String buff = getHeader(data);
|
376
|
// titleString
|
377
|
String titleString = new String();
|
378
|
String alternativeTitles = new String();
|
379
|
|
380
|
for (int i = 0; i < metadata.getTitleList().size(); i++) {
|
381
|
StructuredProperty title = metadata.getTitleList().get(i);
|
382
|
if (i == 0) {
|
383
|
titleString = title.getValue().replaceAll("\\s+", " ");
|
384
|
titleString = titleString.replaceAll("\n", " ");
|
385
|
} else {
|
386
|
alternativeTitles += title.getValue().replaceAll("\\s+", " ") + " ; ";
|
387
|
alternativeTitles = alternativeTitles.replaceAll("\n", " ");
|
388
|
}
|
389
|
break;
|
390
|
}
|
391
|
|
392
|
// pubtitle
|
393
|
buff += getStringField(titleString);
|
394
|
|
395
|
// alternative titles
|
396
|
//buff += getStringField(alternativeTitles); // null#!
|
397
|
// date of acceptance CHANGED THIS TO DATE FORMAT
|
398
|
buff += getStringDateField(metadata.getDateofacceptance().getValue());
|
399
|
|
400
|
// publisher
|
401
|
buff += getStringField(metadata.getPublisher().getValue());
|
402
|
|
403
|
|
404
|
//PID
|
405
|
String pids = new String();
|
406
|
for (StructuredProperty p : data.getPidList()) {
|
407
|
pids += p.getValue() + ";";
|
408
|
}
|
409
|
buff += getStringField(pids);
|
410
|
|
411
|
//language
|
412
|
buff += getStringField(metadata.getLanguage().getClassname());
|
413
|
|
414
|
// RelevantDate
|
415
|
String reldate = new String();
|
416
|
|
417
|
for (StructuredProperty p : metadata.getRelevantdateList()) {
|
418
|
reldate += p.getValue();
|
419
|
break;
|
420
|
}
|
421
|
buff += getStringField(reldate);
|
422
|
|
423
|
//Subject
|
424
|
String subjects = new String();
|
425
|
for (StructuredProperty subj : metadata.getSubjectList()) {
|
426
|
|
427
|
if (isValidTopic(subj.getValue())) {
|
428
|
if (!isNumeric(subj.getValue())) {
|
429
|
subjects += subj.getValue() + ";";
|
430
|
}
|
431
|
}
|
432
|
}
|
433
|
|
434
|
buff += subjects;
|
435
|
|
436
|
// TODO Instance
|
437
|
// buff += getStringField();
|
438
|
|
439
|
//TODO ExternalReference
|
440
|
|
441
|
|
442
|
//Source
|
443
|
String source = new String();
|
444
|
for (StringField s : metadata.getSourceList()) {
|
445
|
source += s.getValue() + ";";
|
446
|
}
|
447
|
|
448
|
buff += getStringField(source);
|
449
|
|
450
|
|
451
|
//TODO Format
|
452
|
buff += getStringField("");
|
453
|
//DOES NOT EXIST
|
454
|
/*String formatString = new String();
|
455
|
for (StringField format : metadata.getFormatList()) {
|
456
|
formatString = format.getValue();
|
457
|
break;}
|
458
|
buff += getStringField(formatString);
|
459
|
}*/
|
460
|
//Context
|
461
|
String context = new String();
|
462
|
for (Result.Context s : metadata.getContextList()) {
|
463
|
context += s.getId() + ";";
|
464
|
}
|
465
|
buff += getStringField(context);
|
466
|
|
467
|
//country TODO does not exist; throws error
|
468
|
String country = new String();
|
469
|
// for (Qualifier s : metadata.getCountryList()) {
|
470
|
// country += s.getClassname() + ";";
|
471
|
// }
|
472
|
|
473
|
buff += getStringField(country);
|
474
|
|
475
|
// access_mode,
|
476
|
buff += getStringField(getAccessMode(result));
|
477
|
//Best License
|
478
|
buff += getStringField(getBestLicense(result));
|
479
|
//Description
|
480
|
String description = new String();
|
481
|
|
482
|
for (StringField desc : metadata.getDescriptionList()) {
|
483
|
description += desc;
|
484
|
break;
|
485
|
}
|
486
|
buff += getStringField(description);
|
487
|
//Journal
|
488
|
buff += getStringField(metadata.getJournal().getName().replaceAll("\n", " ")); //#null#!
|
489
|
|
490
|
//journalName
|
491
|
buff += getStringField(metadata.getJournal().getName().replaceAll("\n", " ")); //#null#!
|
492
|
|
493
|
// TODO ERI SOS : HERE IN GET JOUTNAL. GET DATA INFO I CAN FIND PROVENANCE AND SIMILARITY
|
494
|
|
495
|
//ISSN
|
496
|
buff += getStringField(metadata.getJournal().getIssnLinking());
|
497
|
|
498
|
//embargoEndDate
|
499
|
buff += getStringField(metadata.getEmbargoenddate().getValue());
|
500
|
|
501
|
|
502
|
// `authors`,
|
503
|
int authors = 0;
|
504
|
String delayed = "no";
|
505
|
|
506
|
for (OafRel rel : data.getCachedRelList()) {
|
507
|
|
508
|
if (rel.getRelType().equals(RelType.personResult)) {
|
509
|
|
510
|
authors++;
|
511
|
} else if (rel.getRelType().equals(RelType.resultProject))
|
512
|
// TODO remember : in result Project, first id is project, second is result.
|
513
|
|
514
|
{
|
515
|
String daysfromend = getYearDifferenceInteger(rel.getResultProject().getOutcome().getRelMetadata().getEnddate(), rel.getResultProject().getOutcome().getRelMetadata().getStartdate());
|
516
|
if (Integer.parseInt(daysfromend) > 0) {
|
517
|
delayed = "yes";
|
518
|
}
|
519
|
}
|
520
|
}
|
521
|
|
522
|
buff += getNumericField(String.valueOf(authors));
|
523
|
|
524
|
|
525
|
// TODO isRelatedTo
|
526
|
|
527
|
// resource type
|
528
|
buff += getStringField(metadata.getResourcetype().getClassname());
|
529
|
// device
|
530
|
buff += getStringField(metadata.getDevice().getValue());
|
531
|
// size
|
532
|
buff += getStringField(metadata.getSize().getValue());
|
533
|
// version
|
534
|
buff += getStringField(metadata.getVersion().getValue());
|
535
|
// metadata update
|
536
|
buff += getStringField(metadata.getLastmetadataupdate().getValue());
|
537
|
// metadata version
|
538
|
buff += getStringField(metadata.getMetadataversionnumber().getValue());
|
539
|
|
540
|
// `delayed`,
|
541
|
buff += getStringField(delayed);
|
542
|
|
543
|
// year
|
544
|
buff += getYearInt(metadata.getDateofacceptance().getValue());
|
545
|
|
546
|
// type
|
547
|
buff += getStringField(metadata.getResulttype().getClassname());
|
548
|
//classification
|
549
|
String classification = new String();
|
550
|
for (Instance instance : (result.getInstanceList())) {
|
551
|
|
552
|
if (instance.getInstancetype().getClassname() != null && !instance.getInstancetype().getClassname().isEmpty()) {
|
553
|
classification += instance.getInstancetype().getClassname() + ';';
|
554
|
}
|
555
|
}
|
556
|
|
557
|
buff += getStringField(classification);
|
558
|
|
559
|
// hosted by
|
560
|
String hostedBy = new String();
|
561
|
for (Instance instance : (result.getInstanceList())) {
|
562
|
String host = instance.getHostedby().getKey();
|
563
|
if (host != null && !host.isEmpty()) {
|
564
|
hostedBy += host + ";";
|
565
|
}
|
566
|
}
|
567
|
|
568
|
buff += getStringField(hostedBy);
|
569
|
|
570
|
// collectedfrom
|
571
|
String collectedFrom = new String();
|
572
|
for (FieldTypeProtos.KeyValue collectedFromValue : (data.getCollectedfromList())) {
|
573
|
|
574
|
String host = collectedFromValue.getKey();
|
575
|
if (host != null && !host.isEmpty()) {
|
576
|
collectedFrom += host + ';';
|
577
|
|
578
|
}
|
579
|
}
|
580
|
buff += getStringField(collectedFrom);
|
581
|
return buff;
|
582
|
}
|
583
|
|
584
|
|
585
|
private String buildProject(OafEntity data) {
|
586
|
|
587
|
|
588
|
Project project = data.getProject();
|
589
|
eu.dnetlib.data.proto.ProjectProtos.Project.Metadata metadata = project.getMetadata();
|
590
|
String buff = getHeader(data);
|
591
|
//Code
|
592
|
buff += getStringField(metadata.getCode().getValue());
|
593
|
// `url`,
|
594
|
buff += getStringField(metadata.getWebsiteurl().getValue());
|
595
|
|
596
|
// `acronym`,
|
597
|
String acronym = metadata.getAcronym().getValue();
|
598
|
if (acronym.equalsIgnoreCase("UNKNOWN")) {
|
599
|
acronym = metadata.getTitle().getValue();
|
600
|
}
|
601
|
|
602
|
buff += getStringField(acronym);
|
603
|
|
604
|
//title!
|
605
|
String title = getStringField(metadata.getTitle().getValue());
|
606
|
buff += getStringField(title);
|
607
|
|
608
|
// startdate
|
609
|
buff += getNumericField(metadata.getStartdate().getValue());
|
610
|
|
611
|
// enddate
|
612
|
buff += getNumericField(metadata.getEnddate().getValue());
|
613
|
|
614
|
//`Call identifer`
|
615
|
buff += getStringField(metadata.getCallidentifier().getValue());
|
616
|
|
617
|
//`KeyWords`
|
618
|
buff += getStringField(metadata.getKeywords().getValue());
|
619
|
|
620
|
//`Duration`
|
621
|
buff += getStringField(metadata.getDuration().getValue());
|
622
|
|
623
|
//esc39
|
624
|
buff += getStringField(metadata.getEcsc39().getValue().toString());
|
625
|
|
626
|
//`Contracttype`
|
627
|
buff += getStringField(metadata.getContracttype().getClassname());
|
628
|
|
629
|
//`OA mandate pubs` TODO DOES NOT EXIST
|
630
|
buff += getStringField(metadata.getOamandatepublications().getValue());
|
631
|
|
632
|
//`Subjects` TODO DOES NOT EXIST EITHER
|
633
|
String subjects = new String();
|
634
|
for (StructuredProperty s : metadata.getSubjectsList()) {
|
635
|
|
636
|
subjects += s.getValue() + ';';
|
637
|
}
|
638
|
buff += getStringField(subjects);
|
639
|
|
640
|
|
641
|
//`EC293`
|
642
|
buff += getStringField(metadata.getEcarticle293().getValue());
|
643
|
|
644
|
|
645
|
List<StringField> fundList = metadata.getFundingtreeList();
|
646
|
|
647
|
|
648
|
if (!fundList.isEmpty()) // `funding_lvl0`,
|
649
|
{
|
650
|
//TODO funder + 3 funding levels
|
651
|
/* funder text,
|
652
|
funding_lvl0 text,
|
653
|
funding_lvl1 text,
|
654
|
funding_lvl2 text,
|
655
|
funding_lvl3 text,*/
|
656
|
buff += this.fundingParser.getFundingInfo(fundList.get(0).getValue());
|
657
|
|
658
|
} else {
|
659
|
|
660
|
buff += this.fundingParser.getFundingInfo("");
|
661
|
}
|
662
|
|
663
|
return buff;
|
664
|
|
665
|
}
|
666
|
|
667
|
|
668
|
private String buildPerson(OafEntity data) {
|
669
|
|
670
|
PersonProtos.Person person = data.getPerson();
|
671
|
eu.dnetlib.data.proto.PersonProtos.Person.Metadata metadata = person.getMetadata();
|
672
|
|
673
|
String buff = getHeader(data);
|
674
|
|
675
|
// `firstname`,
|
676
|
buff += metadata.getFirstname();
|
677
|
|
678
|
// `secondNames`,
|
679
|
|
680
|
String secondNames = new String();
|
681
|
for (StringField s : metadata.getSecondnamesList()) {
|
682
|
|
683
|
secondNames += s.getValue() + ' ';
|
684
|
}
|
685
|
|
686
|
buff += getStringField(secondNames);
|
687
|
|
688
|
// `fullname`,
|
689
|
buff += getStringField(metadata.getFullname().getValue());
|
690
|
// `Fax`,
|
691
|
buff += getStringField(metadata.getFax().getValue());
|
692
|
// `Email`,
|
693
|
buff += getStringField(metadata.getEmail().getValue());
|
694
|
// `Phone`,
|
695
|
buff += getStringField(metadata.getPhone().getValue());
|
696
|
|
697
|
// `Nationality`,
|
698
|
buff += getStringField(metadata.getNationality().getClassname());
|
699
|
|
700
|
// `PIDS`,
|
701
|
String pids = new String();
|
702
|
for (StructuredProperty s : data.getPidList()) {
|
703
|
|
704
|
pids += s.getValue() + ";";
|
705
|
|
706
|
|
707
|
}
|
708
|
buff += getStringField(pids);
|
709
|
|
710
|
// `collected from`,
|
711
|
String collectedFrom = new String();
|
712
|
for (FieldTypeProtos.KeyValue s : data.getCollectedfromList()) {
|
713
|
collectedFrom += s.getValue() + ";";
|
714
|
}
|
715
|
buff += getStringField(collectedFrom);
|
716
|
return buff;
|
717
|
|
718
|
}
|
719
|
|
720
|
|
721
|
private ArrayList<String> getResultDatasources(OafEntity valueEntity) {
|
722
|
|
723
|
ArrayList<String> buffs = new ArrayList<String>();
|
724
|
Result result = valueEntity.getResult();
|
725
|
|
726
|
//TODO hosted by
|
727
|
for (Instance instance : (result.getInstanceList())) {
|
728
|
String hostedBy = instance.getHostedby().getKey();
|
729
|
|
730
|
if (hostedBy != null && !hostedBy.isEmpty()) {
|
731
|
buffs.add(getStringField("resultDatasource") + getStringField(hostedBy));
|
732
|
}
|
733
|
}
|
734
|
|
735
|
//TODO collected froms
|
736
|
for (FieldTypeProtos.KeyValue collectedFromValue : (valueEntity.getCollectedfromList())) {
|
737
|
String collectedFrom = collectedFromValue.getKey();
|
738
|
if (collectedFrom != null && !collectedFrom.isEmpty())
|
739
|
buffs.add((getStringField("resultDatasource") + getStringField(collectedFrom)));
|
740
|
|
741
|
}
|
742
|
return buffs;
|
743
|
|
744
|
}
|
745
|
|
746
|
|
747
|
public String cleanId(String value) {
|
748
|
if (value != null) {
|
749
|
// DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____::
|
750
|
|
751
|
|
752
|
// to datacite____:: )
|
753
|
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA
|
754
|
value = value.replaceFirst(".*\\|", "");
|
755
|
value = value.replaceAll("\n", "");
|
756
|
value = value.replaceAll(DELIM, "");
|
757
|
value = value.replaceAll(ENCLOSED, "");
|
758
|
value = value.trim();
|
759
|
|
760
|
}
|
761
|
if (value == null) {
|
762
|
return null;
|
763
|
}
|
764
|
return ENCLOSED + value + ENCLOSED;
|
765
|
|
766
|
}
|
767
|
|
768
|
|
769
|
private ArrayList<String> getDedups(OafEntity valueEntity) {
|
770
|
|
771
|
ArrayList<String> buffs = new ArrayList<String>();
|
772
|
String buff = getStringField("Dedup") + getStringField(cleanId(valueEntity.getId()));
|
773
|
|
774
|
//TODO hosted by
|
775
|
for (OafEntity child : valueEntity.getChildrenList()) {
|
776
|
{
|
777
|
if (child.getType() == valueEntity.getType()) { // if it is result, then its the deduplication
|
778
|
buff += getStringField(cleanId(child.getId()));
|
779
|
}
|
780
|
}
|
781
|
}
|
782
|
|
783
|
buffs.add(buff);
|
784
|
return buffs;
|
785
|
|
786
|
}
|
787
|
|
788
|
|
789
|
private String getNumericField(String data) {
|
790
|
if (data == null || data.isEmpty() || data.equals("")) {
|
791
|
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM;
|
792
|
} else {
|
793
|
|
794
|
return ENCLOSED + data + ENCLOSED + DELIM;
|
795
|
}
|
796
|
}
|
797
|
|
798
|
|
799
|
private String getYearDifferenceInteger(String enddate, String startdate) {
|
800
|
|
801
|
if (enddate != null && !enddate.isEmpty() && startdate != null && !startdate.isEmpty()) {
|
802
|
|
803
|
String[] split = startdate.split("-");
|
804
|
|
805
|
if (split == null || split.length == 0) {
|
806
|
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM;
|
807
|
}
|
808
|
|
809
|
int Startdate = Integer.parseInt(split[0]);
|
810
|
|
811
|
split = enddate.split("-");
|
812
|
|
813
|
if (split == null || split.length == 0) {
|
814
|
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM;
|
815
|
}
|
816
|
|
817
|
int Enddate = Integer.parseInt(split[0]);
|
818
|
|
819
|
int diff = Enddate - Startdate;
|
820
|
|
821
|
return ENCLOSED + diff + ENCLOSED + DELIM;
|
822
|
|
823
|
}
|
824
|
|
825
|
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM;
|
826
|
}
|
827
|
|
828
|
private String getYearInt(String data) {
|
829
|
if (data == null || data.isEmpty() || data.equals("-1")) {
|
830
|
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM;
|
831
|
}
|
832
|
|
833
|
String[] split = data.split("-");
|
834
|
|
835
|
if (split == null || split.length == 0) {
|
836
|
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM;
|
837
|
}
|
838
|
|
839
|
String year = split[0];
|
840
|
|
841
|
year = cleanNumber(year);
|
842
|
|
843
|
return ENCLOSED + year + ENCLOSED + DELIM;
|
844
|
|
845
|
|
846
|
}
|
847
|
|
848
|
private static String cleanNumber(String number) {
|
849
|
number = number.replaceAll("[^A-Za-z0-9:,____]", "");
|
850
|
|
851
|
return number;
|
852
|
}
|
853
|
|
854
|
private String getStringField(String data) {
|
855
|
|
856
|
if (data == null || data.isEmpty() || data.equals("")) {
|
857
|
|
858
|
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM;
|
859
|
} else {
|
860
|
|
861
|
String field = clean(data);
|
862
|
if (field == null) {
|
863
|
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM;
|
864
|
} else {
|
865
|
return field + DELIM;
|
866
|
}
|
867
|
}
|
868
|
}
|
869
|
|
870
|
private String getStringDateField(String data) {
|
871
|
|
872
|
if (data == null || data.isEmpty() || data.equals("") || data.equals("-1")) {
|
873
|
|
874
|
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM;
|
875
|
} else {
|
876
|
|
877
|
String field = clean(data);
|
878
|
if (field == null) {
|
879
|
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM;
|
880
|
} else {
|
881
|
return field + DELIM;
|
882
|
}
|
883
|
}
|
884
|
}
|
885
|
|
886
|
|
887
|
public String getId(OafProtos.Oaf oaf) {
|
888
|
switch (oaf.getKind()) {
|
889
|
case entity:
|
890
|
|
891
|
return cleanId(oaf.getEntity().getId());
|
892
|
case relation:
|
893
|
|
894
|
return cleanId(oaf.getRel().getSource());
|
895
|
|
896
|
}
|
897
|
return null;
|
898
|
|
899
|
}
|
900
|
|
901
|
private boolean isNumeric(String str) {
|
902
|
|
903
|
str = str.replaceAll("[^A-Za-z0-9 ]", "");
|
904
|
str = str.replaceAll(" ", "");
|
905
|
return str.matches("-?\\d+(\\.\\d+)?"); // match a number with optional
|
906
|
// '-' and decimal.
|
907
|
}
|
908
|
|
909
|
// there are topics with "null" as value -> replace them
|
910
|
private boolean isValidTopic(String t) {
|
911
|
|
912
|
if (t == null || t.isEmpty()) {
|
913
|
return false;
|
914
|
}
|
915
|
|
916
|
if (t.equals("") || t.equals(" ")) {
|
917
|
return false;
|
918
|
}
|
919
|
if (t.equals("null") || t.equals("Null") || t.equals("NULL")) {
|
920
|
return false;
|
921
|
}
|
922
|
|
923
|
if (t.equals(ENCLOSED + ENCLOSED + DELIM) || t.equals(ENCLOSED + NULL_STRING + ENCLOSED + DELIM)) {
|
924
|
return false;
|
925
|
}
|
926
|
// skip dedups
|
927
|
if (t.contains("ddc:")) {
|
928
|
|
929
|
return false;
|
930
|
}
|
931
|
return true;
|
932
|
}
|
933
|
|
934
|
|
935
|
private String getBestLicense(ResultProtos.Result result) {
|
936
|
FieldTypeProtos.Qualifier bestLicense = null;
|
937
|
LicenseComparator lc = new LicenseComparator();
|
938
|
for (ResultProtos.Result.Instance instance : (result.getInstanceList())) {
|
939
|
if (lc.compare(bestLicense, instance.getLicence()) > 0) {
|
940
|
bestLicense = instance.getLicence();
|
941
|
}
|
942
|
}
|
943
|
if (bestLicense != null) {
|
944
|
return bestLicense.getClassname();
|
945
|
} else {
|
946
|
return null;
|
947
|
}
|
948
|
}
|
949
|
|
950
|
// here iterate over all values
|
951
|
private String getAccessMode(ResultProtos.Result result) {
|
952
|
String accessMode = null;
|
953
|
for (ResultProtos.Result.Instance instance : (result.getInstanceList())) {
|
954
|
if (instance.getLicence().getClassname() != null && !instance.getLicence().getClassname().isEmpty()) {
|
955
|
accessMode = instance.getLicence().getClassname();
|
956
|
break;
|
957
|
}
|
958
|
|
959
|
}
|
960
|
|
961
|
|
962
|
return accessMode;
|
963
|
}
|
964
|
|
965
|
|
966
|
private String clean(String value) {
|
967
|
if (value != null) {
|
968
|
// TODO DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____::
|
969
|
// to datacite____:: )
|
970
|
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA
|
971
|
value = value.replaceFirst(".*\\|", "");
|
972
|
value = value.replaceAll(DELIM, "");
|
973
|
value = value.replaceAll(",", "");
|
974
|
value = value.replaceAll("\"", "");
|
975
|
value = value.replaceAll("'", "");
|
976
|
value = value.replaceAll(ENCLOSED, "");
|
977
|
value = value.replaceAll("\\r\\n|\\r|\\n", " ");
|
978
|
value = value.replaceAll("\\s+", " ");
|
979
|
value = value.replaceAll("(\\r|\\n)", " ");
|
980
|
value = value.replaceAll("\\t", " ");
|
981
|
|
982
|
// value = value.replaceAll("[^A-Za-z0-9:,____-;:]", " ");
|
983
|
value = value.trim();
|
984
|
|
985
|
}
|
986
|
if (value == null) {
|
987
|
return null;
|
988
|
}
|
989
|
return ENCLOSED + value + ENCLOSED;
|
990
|
|
991
|
}
|
992
|
|
993
|
|
994
|
public long DATEDIFF(String startDate, String endDate) {
|
995
|
long MILLISECS_PER_DAY = 24 * 60 * 60 * 1000;
|
996
|
long days = 0l;
|
997
|
SimpleDateFormat format = new SimpleDateFormat("yyyy-MM-dd"); // "dd/MM/yyyy HH:mm:ss");
|
998
|
// <startdate>2011-09-01</startdate>
|
999
|
// <enddate>2015-08-31</enddate>
|
1000
|
Date dateIni = null;
|
1001
|
Date dateFin = null;
|
1002
|
|
1003
|
if (startDate == null || startDate.isEmpty() || endDate == null || endDate.isEmpty()) {
|
1004
|
return 0;
|
1005
|
}
|
1006
|
try {
|
1007
|
dateIni = (Date) format.parse(startDate);
|
1008
|
dateFin = (Date) format.parse(endDate);
|
1009
|
days = (dateFin.getTime() - dateIni.getTime()) / MILLISECS_PER_DAY;
|
1010
|
} catch (Exception e) {
|
1011
|
log.error(e.toString());
|
1012
|
return 0;
|
1013
|
}
|
1014
|
|
1015
|
return days;
|
1016
|
}
|
1017
|
|
1018
|
public String getDELIM() {
|
1019
|
return DELIM;
|
1020
|
}
|
1021
|
|
1022
|
public void setDELIM(String dELIM) {
|
1023
|
DELIM = dELIM;
|
1024
|
}
|
1025
|
|
1026
|
public String getNULL_STRING() {
|
1027
|
return NULL_STRING;
|
1028
|
}
|
1029
|
|
1030
|
public void setNULL_STRING(String nULL_STRING) {
|
1031
|
NULL_STRING = nULL_STRING;
|
1032
|
}
|
1033
|
|
1034
|
public String getNULL_NUM() {
|
1035
|
return NULL_NUM;
|
1036
|
}
|
1037
|
|
1038
|
public void setNULL_NUM(String nULL_NUM) {
|
1039
|
NULL_NUM = nULL_NUM;
|
1040
|
}
|
1041
|
|
1042
|
public String getENCLOSED() {
|
1043
|
return ENCLOSED;
|
1044
|
}
|
1045
|
|
1046
|
public void setENCLOSED(String eNCLOSED) {
|
1047
|
ENCLOSED = eNCLOSED;
|
1048
|
}
|
1049
|
|
1050
|
|
1051
|
}
|