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