1
|
package eu.dnetlib.data.mapreduce.hbase.lodExport.utils;
|
2
|
|
3
|
import eu.dnetlib.data.mapreduce.util.LicenseComparator;
|
4
|
import eu.dnetlib.data.proto.*;
|
5
|
import eu.dnetlib.data.proto.DatasourceProtos.Datasource;
|
6
|
import eu.dnetlib.data.proto.DatasourceProtos.Datasource.Metadata;
|
7
|
import eu.dnetlib.data.proto.FieldTypeProtos.StringField;
|
8
|
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty;
|
9
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
10
|
import eu.dnetlib.data.proto.OafProtos.OafEntity;
|
11
|
import eu.dnetlib.data.proto.OafProtos.OafRel;
|
12
|
import eu.dnetlib.data.proto.OrganizationProtos.Organization;
|
13
|
import eu.dnetlib.data.proto.ProjectProtos.Project;
|
14
|
import eu.dnetlib.data.proto.ResultProtos.Result;
|
15
|
import eu.dnetlib.data.proto.ResultProtos.Result.Instance;
|
16
|
import org.apache.log4j.Logger;
|
17
|
|
18
|
import java.util.ArrayList;
|
19
|
import java.util.List;
|
20
|
|
21
|
/**
|
22
|
* @author eri Simple serializer that parses input Oaf Protos and prepares them
|
23
|
* for sqoop
|
24
|
*/
|
25
|
public class OldSerializer {
|
26
|
|
27
|
private static String DELIM;
|
28
|
private Logger log = Logger.getLogger(this.getClass());
|
29
|
private String NULL_STRING = "";
|
30
|
|
31
|
private static String SEPERATOR;
|
32
|
private FundingParser fundingParser;
|
33
|
|
34
|
|
35
|
public OldSerializer(String delim, String seperator) {
|
36
|
this.setDELIM(delim);
|
37
|
this.setSEPERATOR(seperator);
|
38
|
this.fundingParser = new FundingParser(delim);
|
39
|
}
|
40
|
|
41
|
|
42
|
public String serialize(Oaf oaf) {
|
43
|
|
44
|
switch (oaf.getKind()) {
|
45
|
case entity:
|
46
|
OafEntity valueEntity = oaf.getEntity();
|
47
|
|
48
|
|
49
|
switch (valueEntity.getType()) {
|
50
|
case datasource:
|
51
|
|
52
|
return buildDatasource(valueEntity);
|
53
|
case organization:
|
54
|
|
55
|
return buildOrganization(valueEntity);
|
56
|
|
57
|
case project:
|
58
|
|
59
|
return buildProject(valueEntity);
|
60
|
|
61
|
case result:
|
62
|
|
63
|
return buildResult(valueEntity);
|
64
|
case person:
|
65
|
return null;
|
66
|
// return buildPerson(valueEntity);
|
67
|
|
68
|
default:
|
69
|
log.error("wrong type");
|
70
|
break;
|
71
|
}
|
72
|
break;
|
73
|
case relation:
|
74
|
OafRel valueRel = oaf.getRel();
|
75
|
|
76
|
return buildRel(valueRel);
|
77
|
|
78
|
}
|
79
|
|
80
|
return null;
|
81
|
|
82
|
}
|
83
|
|
84
|
public String serialize(OafRel oaf) {
|
85
|
|
86
|
switch (oaf.getRelType()) {
|
87
|
|
88
|
default:
|
89
|
return buildRel(oaf);
|
90
|
|
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
|
case organization:
|
111
|
relations.addAll(getDedups(valueEntity));
|
112
|
return relations;
|
113
|
default:
|
114
|
return relations;
|
115
|
}
|
116
|
|
117
|
}
|
118
|
|
119
|
private String buildRel(OafRel Rel) {
|
120
|
switch (Rel.getRelType()) {
|
121
|
case datasourceOrganization:
|
122
|
return getStringField(Rel.getRelType().name()) + getStringField("datasource") + getStringField(Rel.getSource()) + getStringField("organization") + getStringField(Rel.getTarget());
|
123
|
case resultResult:
|
124
|
|
125
|
return getStringField(Rel.getRelType().name()) + getStringField("result") + getStringField(Rel.getSource()) + getStringField("result") + getStringField(Rel.getTarget());
|
126
|
case personPerson:
|
127
|
return getStringField(Rel.getRelType().name()) + getStringField(" person") + getStringField(Rel.getSource()) + getStringField(" person") + getStringField(Rel.getTarget());
|
128
|
|
129
|
case organizationOrganization:
|
130
|
return getStringField(Rel.getRelType().name()) + getStringField("organization") + getStringField(Rel.getSource()) + getStringField("organization") + getStringField(Rel.getTarget());
|
131
|
|
132
|
case personResult:
|
133
|
return getStringField(Rel.getRelType().name()) + getStringField("person") + getStringField(Rel.getSource()) + getStringField("result") + getStringField(Rel.getTarget());
|
134
|
|
135
|
case projectOrganization:
|
136
|
return getStringField(Rel.getRelType().name()) + getStringField("project") + getStringField(Rel.getSource()) + getStringField("organization") + getStringField(Rel.getTarget());
|
137
|
|
138
|
case projectPerson:
|
139
|
return getStringField(Rel.getRelType().name()) + getStringField("project") + getStringField(Rel.getSource()) + getStringField("person") + getStringField(Rel.getTarget());
|
140
|
|
141
|
case resultOrganization:
|
142
|
return getStringField(Rel.getRelType().name()) + getStringField("result") + getStringField(Rel.getSource()) + getStringField("organization") + getStringField(Rel.getTarget());
|
143
|
|
144
|
case resultProject:
|
145
|
return getStringField(Rel.getRelType().name()) + getStringField("result") + getStringField(Rel.getSource()) + getStringField("project") + getStringField(Rel.getTarget());
|
146
|
default:
|
147
|
return null;
|
148
|
}
|
149
|
|
150
|
}
|
151
|
|
152
|
|
153
|
private String getHeader(OafEntity data) {
|
154
|
String buff = new String();
|
155
|
|
156
|
// EntityType
|
157
|
buff += getStringField(data.getType().name());
|
158
|
|
159
|
// OpenaireID
|
160
|
buff += cleanId(data.getId()) + DELIM;
|
161
|
// dateOfTransformation
|
162
|
|
163
|
// TODO change to dateOfTransformation here when released
|
164
|
buff += getStringDateField(data.getDateofcollection());
|
165
|
|
166
|
// dateOfCollection
|
167
|
buff += getStringDateField(data.getDateofcollection());
|
168
|
|
169
|
String originalId = new String();
|
170
|
// originalId
|
171
|
|
172
|
for (String oid : data.getOriginalIdList()) {
|
173
|
originalId += oid + SEPERATOR;
|
174
|
}
|
175
|
|
176
|
buff += originalId + DELIM;
|
177
|
|
178
|
return buff;
|
179
|
|
180
|
}
|
181
|
|
182
|
|
183
|
private String buildDatasource(OafEntity data) {
|
184
|
|
185
|
|
186
|
Datasource d = data.getDatasource();
|
187
|
|
188
|
Metadata metadata = d.getMetadata();
|
189
|
|
190
|
|
191
|
String buff = getHeader(data);
|
192
|
|
193
|
//Datasourcetype
|
194
|
if (metadata.hasDatasourcetype()) {
|
195
|
buff += getStringField(metadata.getDatasourcetype().getClassname());
|
196
|
} else {
|
197
|
buff += getStringField(null);
|
198
|
}
|
199
|
|
200
|
//Openairecompatibility
|
201
|
buff += getStringField(metadata.getOpenairecompatibility().getClassname());
|
202
|
|
203
|
//OfficialName
|
204
|
buff += getStringField(metadata.getOfficialname().getValue());
|
205
|
|
206
|
// Englishname
|
207
|
buff += getStringField(metadata.getEnglishname().getValue());
|
208
|
|
209
|
//Websiteurl
|
210
|
buff += getStringField(metadata.getWebsiteurl().getValue());
|
211
|
|
212
|
//LogoURL
|
213
|
buff += getStringField(metadata.getLogourl().getValue());
|
214
|
|
215
|
//Contactemail
|
216
|
buff += getStringField(metadata.getContactemail().getValue());
|
217
|
//Namespaceprefix
|
218
|
buff += getStringField(metadata.getNamespaceprefix().getValue());
|
219
|
|
220
|
// latitude
|
221
|
buff += getStringField(metadata.getLatitude().getValue());
|
222
|
|
223
|
// longtitude
|
224
|
buff += getStringField(metadata.getLongitude().getValue());
|
225
|
|
226
|
// dateofvalidation,
|
227
|
buff += getStringDateField(metadata.getDateofvalidation().getValue());
|
228
|
|
229
|
//Description
|
230
|
buff += getStringField(metadata.getDescription().getValue());
|
231
|
|
232
|
//subjects
|
233
|
String subj = new String();
|
234
|
for (StructuredProperty s : metadata.getSubjectsList()) {
|
235
|
subj += clean(s.getValue()) + SEPERATOR;
|
236
|
|
237
|
}
|
238
|
|
239
|
//subjectList
|
240
|
buff += getMultipleStringField(subj);
|
241
|
|
242
|
//Number of items
|
243
|
buff += getStringField(metadata.getOdnumberofitems().getValue());
|
244
|
|
245
|
//Date of number of items
|
246
|
buff += getStringDateField(metadata.getOdnumberofitemsdate().getValue());
|
247
|
|
248
|
// Policies
|
249
|
buff += getStringField(metadata.getOdpolicies().getValue());
|
250
|
|
251
|
//languages
|
252
|
String languages = new String();
|
253
|
for (StringField lang : metadata.getOdlanguagesList()) {
|
254
|
languages += clean(lang.getValue()) + SEPERATOR;
|
255
|
}
|
256
|
buff += getMultipleStringField(languages);
|
257
|
|
258
|
|
259
|
// Content type
|
260
|
String contentType = new String();
|
261
|
for (StringField c : metadata.getOdcontenttypesList()) {
|
262
|
contentType += clean(c.getValue()) + SEPERATOR;
|
263
|
}
|
264
|
buff += getMultipleStringField(contentType);
|
265
|
|
266
|
//Access info package
|
267
|
String accessInfo = new String();
|
268
|
for (StringField c : metadata.getAccessinfopackageList()) {
|
269
|
accessInfo += clean(c.getValue()) + SEPERATOR;
|
270
|
}
|
271
|
buff += getMultipleStringField(accessInfo);
|
272
|
|
273
|
//Release start date
|
274
|
buff += getStringDateField(metadata.getReleasestartdate().getValue());
|
275
|
|
276
|
|
277
|
//Release end date
|
278
|
buff += getStringDateField(metadata.getReleaseenddate().getValue());
|
279
|
|
280
|
//Mission statement url
|
281
|
buff += getStringField(metadata.getMissionstatementurl().getValue());
|
282
|
|
283
|
//Data provider
|
284
|
buff += getStringField(String.valueOf(metadata.getDataprovider().getValue()));
|
285
|
|
286
|
//Service provider
|
287
|
buff += getStringField(String.valueOf(metadata.getServiceprovider().getValue()));
|
288
|
|
289
|
//Database access type
|
290
|
buff += getStringField(metadata.getDatabaseaccessrestriction().getValue());
|
291
|
|
292
|
//Data upload type
|
293
|
buff += getStringField(metadata.getDatauploadtype().getValue());
|
294
|
|
295
|
//Data upload restrictions
|
296
|
buff += getStringField(metadata.getDatauploadrestriction().getValue());
|
297
|
|
298
|
|
299
|
//Versioning
|
300
|
buff += getStringField(String.valueOf(metadata.getVersioning().getValue()));
|
301
|
|
302
|
//Citation guideline url
|
303
|
buff += getStringField(String.valueOf(metadata.getCitationguidelineurl().getValue()));
|
304
|
|
305
|
//Quality management kind
|
306
|
buff += getStringField(String.valueOf(metadata.getQualitymanagementkind().getValue()));
|
307
|
|
308
|
//PID systems
|
309
|
buff += getStringField(metadata.getPidsystems().getValue());
|
310
|
|
311
|
//Certificates
|
312
|
buff += getStringField(metadata.getCertificates().getValue());
|
313
|
|
314
|
//Policies
|
315
|
String policies = new String();
|
316
|
for (FieldTypeProtos.KeyValue property : metadata.getPoliciesList()) {
|
317
|
policies += clean(property.getValue()) + SEPERATOR;
|
318
|
}
|
319
|
buff += getMultipleStringField(policies);
|
320
|
|
321
|
buff += getTrust(data);
|
322
|
|
323
|
return buff;
|
324
|
|
325
|
}
|
326
|
|
327
|
|
328
|
private String buildOrganization(OafEntity data) {
|
329
|
|
330
|
|
331
|
Organization organization = data.getOrganization();
|
332
|
eu.dnetlib.data.proto.OrganizationProtos.Organization.Metadata metadata = organization.getMetadata();
|
333
|
|
334
|
String buff = getHeader(data);
|
335
|
|
336
|
//getLegalshortname
|
337
|
buff += getStringField(metadata.getLegalshortname().getValue());
|
338
|
// `name`,
|
339
|
buff += getStringField(metadata.getLegalname().getValue());
|
340
|
//website URL
|
341
|
String[] split = metadata.getWebsiteurl().getValue().split(",");
|
342
|
String urls = new String();
|
343
|
|
344
|
for (String s : split) {
|
345
|
urls += s + SEPERATOR;
|
346
|
}
|
347
|
buff += getMultipleStringField(urls);
|
348
|
|
349
|
//logourl
|
350
|
buff += getStringField(metadata.getLogourl().getValue());
|
351
|
// `country`,
|
352
|
buff += getStringField(metadata.getCountry().getClassid());
|
353
|
|
354
|
buff += getTrust(data);
|
355
|
return buff;
|
356
|
|
357
|
}
|
358
|
|
359
|
String getTrust(OafEntity data) {
|
360
|
for (FieldTypeProtos.ExtraInfo info : data.getExtraInfoList()) {
|
361
|
return getStringField(info.getTrust());
|
362
|
|
363
|
}
|
364
|
return NULL_STRING + DELIM;
|
365
|
}
|
366
|
|
367
|
|
368
|
private String buildResult(OafEntity data) {
|
369
|
|
370
|
Result result = data.getResult();
|
371
|
eu.dnetlib.data.proto.ResultProtos.Result.Metadata metadata = result.getMetadata();
|
372
|
|
373
|
|
374
|
String buff = getHeader(data);
|
375
|
|
376
|
|
377
|
// titleString
|
378
|
String titleString = new String();
|
379
|
String alternativeTitles = new String();
|
380
|
|
381
|
for (int i = 0; i < metadata.getTitleList().size(); i++) {
|
382
|
StructuredProperty title = metadata.getTitleList().get(i);
|
383
|
|
384
|
titleString = title.getValue();
|
385
|
|
386
|
break;
|
387
|
}
|
388
|
|
389
|
// pubtitle
|
390
|
buff += getStringField(titleString);
|
391
|
// date of acceptance CHANGED THIS TO DATE FORMAT
|
392
|
buff += getStringDateField(metadata.getDateofacceptance().getValue());
|
393
|
|
394
|
// publisher
|
395
|
buff += getStringField(metadata.getPublisher().getValue());
|
396
|
|
397
|
|
398
|
//PID
|
399
|
String pids = new String();
|
400
|
for (StructuredProperty p : data.getPidList()) {
|
401
|
pids += clean(p.getValue()) + SEPERATOR;
|
402
|
}
|
403
|
buff += getMultipleStringField(pids);
|
404
|
|
405
|
//language
|
406
|
buff += getStringField(metadata.getLanguage().getClassid());
|
407
|
|
408
|
// RelevantDate
|
409
|
String reldate = new String();
|
410
|
|
411
|
for (StructuredProperty p : metadata.getRelevantdateList()) {
|
412
|
reldate += p.getValue();
|
413
|
break;
|
414
|
}
|
415
|
|
416
|
buff += getStringDateField(reldate);
|
417
|
|
418
|
//Subject
|
419
|
String subjects = new String();
|
420
|
for (StructuredProperty subj : metadata.getSubjectList()) {
|
421
|
|
422
|
if (subj.getValue() != null && !subj.getValue().isEmpty()) {
|
423
|
subjects += clean(subj.getValue()) + SEPERATOR;
|
424
|
|
425
|
}
|
426
|
}
|
427
|
|
428
|
buff += getMultipleStringField(subjects);
|
429
|
|
430
|
//TODO ExternalReference
|
431
|
|
432
|
buff += getStringField(" ");
|
433
|
|
434
|
//Source
|
435
|
String source = new String();
|
436
|
for (StringField s : metadata.getSourceList()) {
|
437
|
source += clean(s.getValue()) + SEPERATOR;
|
438
|
}
|
439
|
|
440
|
buff += getMultipleStringField(source);
|
441
|
|
442
|
|
443
|
//TODO Format
|
444
|
buff += getStringField("");
|
445
|
|
446
|
//Context
|
447
|
String context = new String();
|
448
|
for (Result.Context s : metadata.getContextList()) {
|
449
|
context += s.getId() + SEPERATOR;
|
450
|
}
|
451
|
buff += getMultipleStringField(context);
|
452
|
|
453
|
|
454
|
//country TODO does not exist; throws error
|
455
|
String country = new String();
|
456
|
|
457
|
|
458
|
buff += getStringField(country);
|
459
|
|
460
|
|
461
|
//Best License
|
462
|
buff += getStringField(getBestLicense(result));
|
463
|
//Description
|
464
|
String description = new String();
|
465
|
|
466
|
for (StringField desc : metadata.getDescriptionList()) {
|
467
|
description += desc;
|
468
|
break;
|
469
|
}
|
470
|
|
471
|
buff += getStringField(description);
|
472
|
//Journal
|
473
|
buff += getStringField(metadata.getJournal().getName()); //#null#!
|
474
|
|
475
|
|
476
|
// TODO ERI SOS : HERE IN GET JOUTNAL. GET DATA INFO I CAN FIND PROVENANCE AND SIMILARITY
|
477
|
|
478
|
|
479
|
// TODO isRelatedTo
|
480
|
|
481
|
// resource type
|
482
|
buff += getStringField(metadata.getResourcetype().getClassname());
|
483
|
// device
|
484
|
buff += getStringField(metadata.getDevice().getValue());
|
485
|
// size
|
486
|
buff += getStringField(metadata.getSize().getValue());
|
487
|
// version
|
488
|
buff += getStringField(metadata.getVersion().getValue());
|
489
|
|
490
|
// metadata update
|
491
|
buff += getStringDateField(metadata.getLastmetadataupdate().getValue());
|
492
|
// metadata version
|
493
|
buff += getStringField(metadata.getMetadataversionnumber().getValue());
|
494
|
|
495
|
|
496
|
// year
|
497
|
buff += getYearInt(metadata.getDateofacceptance().getValue());
|
498
|
|
499
|
// type
|
500
|
buff += getStringField(metadata.getResulttype().getClassname());
|
501
|
|
502
|
|
503
|
buff += getTrust(data);
|
504
|
|
505
|
|
506
|
return buff;
|
507
|
}
|
508
|
|
509
|
|
510
|
private String buildProject(OafEntity data) {
|
511
|
|
512
|
|
513
|
Project project = data.getProject();
|
514
|
eu.dnetlib.data.proto.ProjectProtos.Project.Metadata metadata = project.getMetadata();
|
515
|
|
516
|
|
517
|
String buff = getHeader(data);
|
518
|
|
519
|
|
520
|
//Code
|
521
|
buff += getStringField(metadata.getCode().getValue());
|
522
|
// `Websiteurl`,
|
523
|
buff += getStringField(metadata.getWebsiteurl().getValue());
|
524
|
|
525
|
|
526
|
// `Acronym`,
|
527
|
buff += getStringField(metadata.getAcronym().getValue());
|
528
|
|
529
|
//Titlem
|
530
|
buff += getStringField(metadata.getTitle().getValue());
|
531
|
|
532
|
// Startdate
|
533
|
buff += getNumericField(metadata.getStartdate().getValue());
|
534
|
|
535
|
// Enddate
|
536
|
buff += getNumericField(metadata.getEnddate().getValue());
|
537
|
|
538
|
//`Call identifer`
|
539
|
buff += getStringField(metadata.getCallidentifier().getValue());
|
540
|
|
541
|
//`KeyWords`
|
542
|
buff += getStringField(metadata.getKeywords().getValue());
|
543
|
|
544
|
//`Duration`
|
545
|
buff += getStringField(metadata.getDuration().getValue());
|
546
|
|
547
|
//esc39
|
548
|
buff += getStringField(metadata.getEcsc39().getValue().toString());
|
549
|
|
550
|
//`Contracttype`
|
551
|
buff += getStringField(metadata.getContracttype().getClassname());
|
552
|
|
553
|
//`OA mandate pubs` TODO DOES NOT EXIST
|
554
|
buff += getStringField(metadata.getOamandatepublications().getValue());
|
555
|
// buff += getStringField("");
|
556
|
|
557
|
//`Subjects` TODO DOES NOT EXIST EITHER
|
558
|
String subjects = new String();
|
559
|
for (StructuredProperty s : metadata.getSubjectsList()) {
|
560
|
|
561
|
subjects += clean(s.getValue()) + SEPERATOR;
|
562
|
}
|
563
|
buff += getMultipleStringField(subjects);
|
564
|
|
565
|
|
566
|
//`EC293`
|
567
|
buff += getStringField(metadata.getEcarticle293().getValue());
|
568
|
|
569
|
|
570
|
List<StringField> fundList = metadata.getFundingtreeList();
|
571
|
|
572
|
|
573
|
if (!fundList.isEmpty()) // `funding_lvl0`,
|
574
|
{
|
575
|
//TODO funder + 3 funding levels
|
576
|
/* funder text,
|
577
|
funding_lvl0 text,
|
578
|
funding_lvl1 text,
|
579
|
funding_lvl2 text,
|
580
|
funding_lvl3 text,*/
|
581
|
buff += this.fundingParser.getFundingInfo(fundList.get(0).getValue());
|
582
|
|
583
|
} else {
|
584
|
|
585
|
buff += this.fundingParser.getFundingInfo("");
|
586
|
}
|
587
|
|
588
|
|
589
|
buff += getTrust(data);
|
590
|
|
591
|
return buff;
|
592
|
|
593
|
}
|
594
|
|
595
|
|
596
|
private String buildPerson(OafEntity data) {
|
597
|
|
598
|
PersonProtos.Person person = data.getPerson();
|
599
|
eu.dnetlib.data.proto.PersonProtos.Person.Metadata metadata = person.getMetadata();
|
600
|
|
601
|
String buff = getHeader(data);
|
602
|
|
603
|
|
604
|
// `firstname`,
|
605
|
buff += getStringField(metadata.getFirstname().getValue());
|
606
|
|
607
|
// `secondNames`,
|
608
|
|
609
|
String secondNames = new String();
|
610
|
for (StringField s : metadata.getSecondnamesList()) {
|
611
|
|
612
|
secondNames += s.getValue() + ' ';
|
613
|
}
|
614
|
|
615
|
buff += getStringField(secondNames);
|
616
|
|
617
|
// `fullname`,
|
618
|
buff += getStringField(metadata.getFullname().getValue());
|
619
|
// `Fax`,
|
620
|
buff += getStringField(metadata.getFax().getValue());
|
621
|
// `Email`,
|
622
|
buff += getStringField(metadata.getEmail().getValue());
|
623
|
// `Phone`,
|
624
|
buff += getStringField(metadata.getPhone().getValue());
|
625
|
|
626
|
// `Nationality`,
|
627
|
buff += getStringField(metadata.getNationality().getClassid());
|
628
|
|
629
|
// `PIDS`,
|
630
|
String pids = new String();
|
631
|
for (StructuredProperty s : data.getPidList()) {
|
632
|
|
633
|
pids += s.getValue() + ";";
|
634
|
|
635
|
|
636
|
}
|
637
|
buff += getStringField(pids);
|
638
|
|
639
|
|
640
|
buff += getTrust(data);
|
641
|
|
642
|
return buff;
|
643
|
|
644
|
}
|
645
|
|
646
|
|
647
|
private ArrayList<String> getResultDatasources(OafEntity valueEntity) {
|
648
|
|
649
|
ArrayList<String> buffs = new ArrayList<String>();
|
650
|
Result result = valueEntity.getResult();
|
651
|
|
652
|
//TODO hosted by
|
653
|
for (Instance instance : (result.getInstanceList())) {
|
654
|
String hostedBy = instance.getHostedby().getKey();
|
655
|
|
656
|
if (hostedBy != null && !hostedBy.isEmpty()) {
|
657
|
buffs.add(getStringField("resultDatasource") + getStringField("result") + getStringField(cleanId(valueEntity.getId())) + getStringField("datasource") + getStringField(hostedBy));
|
658
|
}
|
659
|
}
|
660
|
|
661
|
//TODO collected froms
|
662
|
for (FieldTypeProtos.KeyValue collectedFromValue : (valueEntity.getCollectedfromList())) {
|
663
|
String collectedFrom = collectedFromValue.getKey();
|
664
|
if (collectedFrom != null && !collectedFrom.isEmpty())
|
665
|
buffs.add((getStringField("resultDatasource") + getStringField("result") + getStringField(cleanId(valueEntity.getId())) + getStringField("datasource") + getStringField(collectedFrom)));
|
666
|
|
667
|
}
|
668
|
return buffs;
|
669
|
|
670
|
}
|
671
|
|
672
|
|
673
|
public String cleanId(String value) {
|
674
|
if (value != null) {
|
675
|
// DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____::
|
676
|
|
677
|
// to datacite____:: )
|
678
|
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA
|
679
|
value = value.replaceFirst(".*\\|", "");
|
680
|
value = value.replace("\n", "");
|
681
|
}
|
682
|
|
683
|
return value;
|
684
|
|
685
|
}
|
686
|
|
687
|
//TODO make them in pairs
|
688
|
private ArrayList<String> getDedups(OafEntity valueEntity) {
|
689
|
|
690
|
ArrayList<String> buffs = new ArrayList<String>();
|
691
|
|
692
|
if (!valueEntity.getChildrenList().isEmpty()) {
|
693
|
|
694
|
String header = getStringField("dedup") + getStringField(valueEntity.getType().name()) + getStringField(cleanId(valueEntity.getId())) + getStringField(valueEntity.getType().name());
|
695
|
|
696
|
for (OafEntity child : valueEntity.getChildrenList()) {
|
697
|
{
|
698
|
if (child.getType() == valueEntity.getType()) { // if it is result, then its the deduplication
|
699
|
|
700
|
buffs.add(header + getStringField(cleanId(child.getId())));
|
701
|
|
702
|
}
|
703
|
}
|
704
|
}
|
705
|
|
706
|
}
|
707
|
|
708
|
return buffs;
|
709
|
|
710
|
}
|
711
|
|
712
|
|
713
|
private String getNumericField(String data) {
|
714
|
if (data == null || data.isEmpty() || data.equals("")) {
|
715
|
return NULL_STRING + DELIM;
|
716
|
} else {
|
717
|
|
718
|
return data + DELIM;
|
719
|
}
|
720
|
}
|
721
|
|
722
|
|
723
|
private String getYearInt(String data) {
|
724
|
if (data == null || data.isEmpty() || data.equals("-1")) {
|
725
|
return NULL_STRING + DELIM;
|
726
|
}
|
727
|
String[] split = data.split("-");
|
728
|
|
729
|
if (split == null || split.length == 0) {
|
730
|
return NULL_STRING + DELIM;
|
731
|
}
|
732
|
|
733
|
String year = split[0];
|
734
|
|
735
|
year = cleanNumber(year);
|
736
|
|
737
|
return year + DELIM;
|
738
|
|
739
|
|
740
|
}
|
741
|
|
742
|
private static String cleanNumber(String number) {
|
743
|
number = number.replaceAll("[^A-Za-z0-9:,____]", "");
|
744
|
|
745
|
return number;
|
746
|
}
|
747
|
|
748
|
|
749
|
private String getStringField(String data) {
|
750
|
|
751
|
if (data == null || data.isEmpty() || data.equals("")) {
|
752
|
|
753
|
return NULL_STRING + DELIM;
|
754
|
} else {
|
755
|
|
756
|
String field = clean(data);
|
757
|
if (field == null) {
|
758
|
return NULL_STRING + DELIM;
|
759
|
} else {
|
760
|
return field + DELIM;
|
761
|
}
|
762
|
}
|
763
|
}
|
764
|
|
765
|
|
766
|
private String getMultipleStringField(String data) {
|
767
|
if (data == null || data.isEmpty() || data.equals("")) {
|
768
|
return NULL_STRING + DELIM;
|
769
|
}
|
770
|
|
771
|
return data + DELIM;
|
772
|
|
773
|
}
|
774
|
|
775
|
private String getStringDateField(String data) {
|
776
|
|
777
|
if (data == null || data.isEmpty() || data.equals("") || data.equals("-1")) {
|
778
|
|
779
|
return NULL_STRING + DELIM;
|
780
|
}
|
781
|
return data + DELIM;
|
782
|
|
783
|
}
|
784
|
|
785
|
|
786
|
public String getId(OafProtos.Oaf oaf) {
|
787
|
switch (oaf.getKind()) {
|
788
|
case entity:
|
789
|
|
790
|
return cleanId(oaf.getEntity().getId());
|
791
|
case relation:
|
792
|
|
793
|
return cleanId(oaf.getRel().getSource());
|
794
|
|
795
|
}
|
796
|
return null;
|
797
|
|
798
|
}
|
799
|
|
800
|
|
801
|
private String getBestLicense(ResultProtos.Result result) {
|
802
|
FieldTypeProtos.Qualifier bestLicense = null;
|
803
|
LicenseComparator lc = new LicenseComparator();
|
804
|
for (ResultProtos.Result.Instance instance : (result.getInstanceList())) {
|
805
|
if (lc.compare(bestLicense, instance.getLicence()) > 0) {
|
806
|
bestLicense = instance.getLicence();
|
807
|
}
|
808
|
}
|
809
|
if (bestLicense != null) {
|
810
|
return bestLicense.getClassname();
|
811
|
} else {
|
812
|
return null;
|
813
|
}
|
814
|
}
|
815
|
|
816
|
|
817
|
private String clean(String value) {
|
818
|
if (value != null) {
|
819
|
// TODO DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____::
|
820
|
// to datacite____:: )
|
821
|
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA
|
822
|
value = value.replaceFirst(".*\\|", "");
|
823
|
value = value.replaceAll("[\"\\r\\\\;]", "");
|
824
|
value=value.replace(SEPERATOR," ");
|
825
|
value = value.replaceAll("[^a-zA-Z0-9 .-_:/]+", "");
|
826
|
}
|
827
|
|
828
|
return value;
|
829
|
|
830
|
}
|
831
|
|
832
|
|
833
|
public String getDELIM() {
|
834
|
return DELIM;
|
835
|
}
|
836
|
|
837
|
public void setDELIM(String dELIM) {
|
838
|
DELIM = dELIM;
|
839
|
}
|
840
|
|
841
|
|
842
|
public static String getSEPERATOR() {
|
843
|
return SEPERATOR;
|
844
|
}
|
845
|
|
846
|
public static void setSEPERATOR(String SEPERATOR) {
|
847
|
OldSerializer.SEPERATOR = SEPERATOR;
|
848
|
}
|
849
|
}
|