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.Metadata;
|
5
|
import eu.dnetlib.data.proto.FieldTypeProtos;
|
6
|
import eu.dnetlib.data.proto.FieldTypeProtos.StringField;
|
7
|
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty;
|
8
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
9
|
import eu.dnetlib.data.proto.OafProtos.OafEntity;
|
10
|
import eu.dnetlib.data.proto.OafProtos.OafRel;
|
11
|
import eu.dnetlib.data.proto.OrganizationProtos.Organization;
|
12
|
import eu.dnetlib.data.proto.PersonProtos;
|
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.jsoup.Jsoup;
|
17
|
|
18
|
import java.util.ArrayList;
|
19
|
import java.util.List;
|
20
|
import java.util.Set;
|
21
|
|
22
|
/**
|
23
|
* @author eri Simple serializer that parses input Oaf Protos and prepares them
|
24
|
* for sqoop
|
25
|
*/
|
26
|
public class NewSerializer {
|
27
|
|
28
|
public static String Serialize(Oaf oaf, String DELIM) {
|
29
|
|
30
|
switch (oaf.getKind()) {
|
31
|
case entity:
|
32
|
OafEntity valueEntity = oaf.getEntity();
|
33
|
switch (valueEntity.getType()) {
|
34
|
case datasource:
|
35
|
|
36
|
return buildDatasource(valueEntity, DELIM);
|
37
|
case organization:
|
38
|
|
39
|
return buildOrganization(valueEntity, DELIM);
|
40
|
|
41
|
case project:
|
42
|
|
43
|
return buildProject(valueEntity, DELIM);
|
44
|
|
45
|
case result:
|
46
|
|
47
|
return buildResult(valueEntity, DELIM);
|
48
|
case person:
|
49
|
return buildPerson(valueEntity, DELIM);
|
50
|
|
51
|
default:
|
52
|
|
53
|
break;
|
54
|
}
|
55
|
break;
|
56
|
case relation:
|
57
|
OafRel valueRel = oaf.getRel();
|
58
|
|
59
|
return Serialize(valueRel, DELIM);
|
60
|
|
61
|
}
|
62
|
|
63
|
return null;
|
64
|
|
65
|
}
|
66
|
|
67
|
|
68
|
public static void extractRelations(Oaf oaf, String DELIM, Set<String> relations) {
|
69
|
OafEntity valueEntity = oaf.getEntity();
|
70
|
switch (valueEntity.getType()) {
|
71
|
case result:
|
72
|
|
73
|
getResultDatasources(valueEntity, DELIM, relations);
|
74
|
getDedups(valueEntity, DELIM, relations);
|
75
|
|
76
|
case datasource:
|
77
|
getDedups(valueEntity, DELIM, relations);
|
78
|
|
79
|
case person:
|
80
|
getDedups(valueEntity, DELIM, relations);
|
81
|
case organization:
|
82
|
getDedups(valueEntity, DELIM, relations);
|
83
|
|
84
|
default:
|
85
|
|
86
|
}
|
87
|
|
88
|
}
|
89
|
|
90
|
public static String Serialize(OafRel Rel, String DELIM) {
|
91
|
StringBuilder buff ;
|
92
|
|
93
|
switch (Rel.getRelType()) {
|
94
|
case datasourceOrganization:
|
95
|
buff = new StringBuilder();
|
96
|
buff.append(Rel.getRelType().name()).append(DELIM).append("datasource").append(DELIM).append(cleanId(Rel.getSource())).append(DELIM)
|
97
|
.append("organization").append(DELIM).append(cleanId(Rel.getTarget())).append(DELIM).append("\n");
|
98
|
return buff.toString();
|
99
|
case resultResult:
|
100
|
buff = new StringBuilder();
|
101
|
buff.append(Rel.getRelType().name()).append(DELIM).append("result").append(DELIM).append(cleanId(Rel.getSource())).append(DELIM)
|
102
|
.append("result").append(DELIM).append(cleanId(Rel.getTarget())).append(DELIM);
|
103
|
return buff.toString();
|
104
|
case personPerson:
|
105
|
buff = new StringBuilder();
|
106
|
buff.append(Rel.getRelType().name()).append(DELIM).append("person").append(DELIM).append(cleanId(Rel.getSource())).append(DELIM)
|
107
|
.append("person").append(DELIM).append(cleanId(Rel.getTarget())).append(DELIM);
|
108
|
return buff.toString();
|
109
|
case organizationOrganization:
|
110
|
buff = new StringBuilder();
|
111
|
buff.append(Rel.getRelType().name()).append(DELIM).append("organization").append(DELIM).append(cleanId(Rel.getSource())).append(DELIM)
|
112
|
.append("organization").append(DELIM).append(cleanId(Rel.getTarget())).append(DELIM);
|
113
|
return buff.toString();
|
114
|
case personResult:
|
115
|
buff = new StringBuilder();
|
116
|
buff.append(Rel.getRelType().name()).append(DELIM).append("person").append(DELIM).append(cleanId(Rel.getSource())).append(DELIM)
|
117
|
.append("result").append(DELIM).append(cleanId(Rel.getTarget())).append(DELIM);
|
118
|
return buff.toString();
|
119
|
case projectOrganization:
|
120
|
buff = new StringBuilder();
|
121
|
buff.append(Rel.getRelType().name()).append(DELIM).append("project").append(DELIM).append(cleanId(Rel.getSource())).append(DELIM)
|
122
|
.append("organization").append(DELIM).append(cleanId(Rel.getTarget())).append(DELIM);
|
123
|
return buff.toString();
|
124
|
case projectPerson:
|
125
|
buff = new StringBuilder();
|
126
|
buff.append(Rel.getRelType().name()).append(DELIM).append("project").append(DELIM).append(cleanId(Rel.getSource())).append(DELIM)
|
127
|
.append("person").append(DELIM).append(cleanId(Rel.getTarget())).append(DELIM);
|
128
|
return buff.toString();
|
129
|
case resultOrganization:
|
130
|
buff = new StringBuilder();
|
131
|
buff.append(Rel.getRelType().name()).append(DELIM).append("result").append(DELIM).append(cleanId(Rel.getSource())).append(DELIM)
|
132
|
.append("organization").append(DELIM).append(cleanId(Rel.getTarget())).append(DELIM);
|
133
|
return buff.toString();
|
134
|
//TOOD maybe switch them???
|
135
|
|
136
|
case resultProject:
|
137
|
buff = new StringBuilder();
|
138
|
buff.append(Rel.getRelType().name()).append(DELIM).append("result").append(DELIM).append(cleanId(Rel.getSource())).append(DELIM)
|
139
|
.append("project").append(DELIM).append(cleanId(Rel.getTarget())).append(DELIM);
|
140
|
return buff.toString();
|
141
|
default:
|
142
|
|
143
|
}
|
144
|
|
145
|
return "";
|
146
|
|
147
|
}
|
148
|
|
149
|
private static String getHeader(OafEntity data, String DELIM) {
|
150
|
String SEPERATOR = ";";
|
151
|
|
152
|
StringBuilder buff = new StringBuilder();
|
153
|
|
154
|
// EntityType
|
155
|
buff.append(data.getType().name() + DELIM);
|
156
|
|
157
|
// OpenaireID
|
158
|
buff.append(cleanId(data.getId()) + DELIM);
|
159
|
// dateOfTransformation
|
160
|
|
161
|
// TODO change to dateOfTransformation here when released
|
162
|
buff.append(cleanId(data.getDateoftransformation()) + DELIM);
|
163
|
|
164
|
// dateOfCollection
|
165
|
buff.append(cleanId(data.getDateofcollection()) + DELIM);
|
166
|
|
167
|
String dataStr = new String();
|
168
|
// originalId
|
169
|
|
170
|
for (String oid : data.getOriginalIdList()) {
|
171
|
dataStr += cleanId(oid) + SEPERATOR;
|
172
|
}
|
173
|
|
174
|
|
175
|
buff.append(dataStr + DELIM);
|
176
|
|
177
|
return buff.toString();
|
178
|
}
|
179
|
|
180
|
|
181
|
private static String buildDatasource(OafEntity data, String DELIM) {
|
182
|
String SEPERATOR = ";";
|
183
|
|
184
|
StringBuilder buff = new StringBuilder();
|
185
|
|
186
|
buff.append(getHeader(data, DELIM));
|
187
|
Metadata metadata = data.getDatasource().getMetadata();
|
188
|
|
189
|
|
190
|
//Datasourcetype
|
191
|
buff.append(clean(metadata.getDatasourcetype().getClassname()) + DELIM);
|
192
|
|
193
|
//Openairecompatibility
|
194
|
buff.append(clean(metadata.getOpenairecompatibility().getClassname()) + DELIM);
|
195
|
|
196
|
//OfficialName
|
197
|
buff.append(clean(metadata.getOfficialname().getValue()) + DELIM);
|
198
|
|
199
|
// Englishname
|
200
|
buff.append(clean(metadata.getEnglishname().getValue()) + DELIM);
|
201
|
|
202
|
//Websiteurl
|
203
|
buff.append(clean(metadata.getWebsiteurl().getValue()) + DELIM);
|
204
|
|
205
|
//LogoURL
|
206
|
buff.append(clean(metadata.getLogourl().getValue()) + DELIM);
|
207
|
|
208
|
//Contactemail
|
209
|
buff.append(clean(metadata.getContactemail().getValue()) + DELIM);
|
210
|
|
211
|
//Namespaceprefix
|
212
|
buff.append(clean(metadata.getNamespaceprefix().getValue()) + DELIM);
|
213
|
|
214
|
// latitude
|
215
|
buff.append(clean(metadata.getLatitude().getValue()) + DELIM);
|
216
|
|
217
|
// longtitude
|
218
|
buff.append(clean(metadata.getLongitude().getValue() + DELIM));
|
219
|
|
220
|
// dateofvalidation,
|
221
|
buff.append(clean(metadata.getDateofvalidation().getValue()) + DELIM);
|
222
|
|
223
|
//Description
|
224
|
buff.append(cleanText(metadata.getDescription().getValue(), DELIM) + DELIM);
|
225
|
|
226
|
//subjects
|
227
|
String subj = new String();
|
228
|
for (StructuredProperty s : metadata.getSubjectsList()) {
|
229
|
subj += clean(s.getValue()) + SEPERATOR;
|
230
|
|
231
|
}
|
232
|
|
233
|
//subjectList
|
234
|
buff.append(clean(subj) + DELIM);
|
235
|
|
236
|
//Number of items
|
237
|
buff.append(clean(metadata.getOdnumberofitems().getValue()) + DELIM);
|
238
|
|
239
|
//Date of number of items
|
240
|
buff.append(clean(metadata.getOdnumberofitemsdate().getValue()) + DELIM);
|
241
|
|
242
|
// Policies
|
243
|
buff.append(clean(metadata.getOdpolicies().getValue()) + DELIM);
|
244
|
|
245
|
//languages
|
246
|
String dataStr = new String();
|
247
|
|
248
|
for (StringField lang : metadata.getOdlanguagesList()) {
|
249
|
dataStr += clean(lang.getValue()) + SEPERATOR;
|
250
|
}
|
251
|
|
252
|
buff.append(dataStr + DELIM);
|
253
|
|
254
|
|
255
|
// Content type
|
256
|
dataStr = " ";
|
257
|
for (StringField c : metadata.getOdcontenttypesList()) {
|
258
|
dataStr += clean(c.getValue()) + SEPERATOR;
|
259
|
}
|
260
|
buff.append(dataStr + DELIM);
|
261
|
|
262
|
//Access info package
|
263
|
dataStr = " ";
|
264
|
|
265
|
for (StringField c : metadata.getAccessinfopackageList()) {
|
266
|
dataStr += clean(c.getValue()) + SEPERATOR;
|
267
|
}
|
268
|
|
269
|
buff.append(dataStr + DELIM);
|
270
|
|
271
|
//Release start date
|
272
|
buff.append(clean(metadata.getReleasestartdate().getValue()) + DELIM);
|
273
|
|
274
|
//Release end date
|
275
|
buff.append(clean(metadata.getReleaseenddate().getValue()) + DELIM);
|
276
|
|
277
|
//Mission statement url
|
278
|
buff.append(clean(metadata.getMissionstatementurl().getValue()) + DELIM);
|
279
|
|
280
|
//Data provider
|
281
|
buff.append(clean(String.valueOf(metadata.getDataprovider().getValue())) + DELIM);
|
282
|
|
283
|
//Service provider
|
284
|
buff.append(clean(String.valueOf(metadata.getServiceprovider().getValue())) + DELIM);
|
285
|
|
286
|
//Database access type
|
287
|
buff.append(clean(metadata.getDatabaseaccessrestriction().getValue()) + DELIM);
|
288
|
|
289
|
//Data upload type
|
290
|
buff.append(clean(metadata.getDatauploadtype().getValue()) + DELIM);
|
291
|
|
292
|
//Data upload restrictions
|
293
|
buff.append(clean(metadata.getDatauploadrestriction().getValue()) + DELIM);
|
294
|
|
295
|
//Versioning
|
296
|
buff.append(clean(String.valueOf(metadata.getVersioning().getValue())) + DELIM);
|
297
|
|
298
|
//Citation guideline url
|
299
|
buff.append(clean(metadata.getCitationguidelineurl().getValue()) + DELIM);
|
300
|
|
301
|
//Quality management kind
|
302
|
buff.append(clean(metadata.getQualitymanagementkind().getValue()) + DELIM);
|
303
|
|
304
|
//PID systems
|
305
|
buff.append(clean(metadata.getPidsystems().getValue()) + DELIM);
|
306
|
|
307
|
//Certificates
|
308
|
buff.append(clean(metadata.getCertificates().getValue()) + DELIM);
|
309
|
|
310
|
//Policies
|
311
|
dataStr = " ";
|
312
|
for (FieldTypeProtos.KeyValue property : metadata.getPoliciesList()) {
|
313
|
dataStr += clean(property.getValue()) + SEPERATOR;
|
314
|
}
|
315
|
|
316
|
buff.append(dataStr + DELIM);
|
317
|
|
318
|
buff.append(getTrust(data) + DELIM);
|
319
|
return buff.toString();
|
320
|
}
|
321
|
|
322
|
|
323
|
private static String buildOrganization(OafEntity data, String DELIM) {
|
324
|
String SEPERATOR = ";";
|
325
|
|
326
|
StringBuilder buff = new StringBuilder();
|
327
|
buff.append(getHeader(data, DELIM));
|
328
|
|
329
|
Organization organization = data.getOrganization();
|
330
|
Organization.Metadata metadata = organization.getMetadata();
|
331
|
|
332
|
//getLegalshortname
|
333
|
buff.append(clean(metadata.getLegalshortname().getValue()) + DELIM);
|
334
|
// `name`,
|
335
|
buff.append(clean(metadata.getLegalname().getValue()) + DELIM);
|
336
|
//website URL
|
337
|
String[] split = metadata.getWebsiteurl().getValue().split(",");
|
338
|
String dataStr = new String();
|
339
|
|
340
|
for (String s : split) {
|
341
|
dataStr += s.replace(DELIM, " ") + SEPERATOR;
|
342
|
}
|
343
|
|
344
|
buff.append(dataStr + DELIM);
|
345
|
//logourl
|
346
|
buff.append(cleanUrl(metadata.getLogourl().getValue(), DELIM) + DELIM);
|
347
|
// `country`,
|
348
|
buff.append(clean(metadata.getCountry().getClassid()) + DELIM);
|
349
|
buff.append(getTrust(data) + DELIM);
|
350
|
|
351
|
return buff.toString();
|
352
|
|
353
|
}
|
354
|
|
355
|
static String getTrust(OafEntity data) {
|
356
|
for (FieldTypeProtos.ExtraInfo info : data.getExtraInfoList()) {
|
357
|
return (info.getTrust());
|
358
|
|
359
|
}
|
360
|
return " ";
|
361
|
}
|
362
|
|
363
|
|
364
|
private static String buildResult(OafEntity data, String DELIM) {
|
365
|
String SEPERATOR = ";";
|
366
|
|
367
|
Result.Metadata metadata = data.getResult().getMetadata();
|
368
|
|
369
|
|
370
|
StringBuilder buff = new StringBuilder();
|
371
|
|
372
|
buff.append(getHeader(data, DELIM));
|
373
|
|
374
|
// titleString
|
375
|
String dataStr = new String();
|
376
|
|
377
|
for (int i = 0; i < metadata.getTitleList().size(); i++) {
|
378
|
StructuredProperty title = metadata.getTitleList().get(i);
|
379
|
dataStr = clean(title.getValue());
|
380
|
break;
|
381
|
}
|
382
|
|
383
|
// pubtitle
|
384
|
buff.append(clean(dataStr) + DELIM);
|
385
|
|
386
|
// date of acceptance CHANGED THIS TO DATE FORMAT
|
387
|
buff.append(clean(metadata.getDateofacceptance().getValue()) + DELIM);
|
388
|
|
389
|
// publisher
|
390
|
buff.append(clean(metadata.getPublisher().getValue()) + DELIM);
|
391
|
|
392
|
|
393
|
//PID
|
394
|
dataStr = " ";
|
395
|
for (StructuredProperty p : data.getPidList()) {
|
396
|
dataStr += clean(p.getValue()) + SEPERATOR;
|
397
|
}
|
398
|
|
399
|
buff.append(dataStr + DELIM);
|
400
|
|
401
|
//language
|
402
|
buff.append(clean(metadata.getLanguage().getClassid()) + DELIM);
|
403
|
|
404
|
// RelevantDate
|
405
|
dataStr = " ";
|
406
|
|
407
|
for (StructuredProperty p : metadata.getRelevantdateList()) {
|
408
|
dataStr += clean(p.getValue());
|
409
|
break;
|
410
|
}
|
411
|
|
412
|
buff.append(dataStr + DELIM);
|
413
|
|
414
|
//Subject
|
415
|
dataStr = " ";
|
416
|
for (StructuredProperty subj : metadata.getSubjectList()) {
|
417
|
|
418
|
if (subj.getValue() != null && !subj.getValue().isEmpty()) {
|
419
|
dataStr += clean(subj.getValue()) + SEPERATOR;
|
420
|
}
|
421
|
}
|
422
|
|
423
|
buff.append(dataStr + DELIM);
|
424
|
|
425
|
//TODO ExternalReference
|
426
|
|
427
|
buff.append(" " + DELIM);
|
428
|
|
429
|
//Source
|
430
|
dataStr = " ";
|
431
|
for (StringField s : metadata.getSourceList()) {
|
432
|
dataStr += clean(s.getValue()) + SEPERATOR;
|
433
|
}
|
434
|
|
435
|
buff.append(dataStr + DELIM);
|
436
|
|
437
|
//TODO Format
|
438
|
buff.append(" " + DELIM);
|
439
|
|
440
|
//Context
|
441
|
dataStr = " ";
|
442
|
for (Result.Context s : metadata.getContextList()) {
|
443
|
dataStr += clean(s.getId()) + SEPERATOR;
|
444
|
}
|
445
|
buff.append(dataStr + DELIM);
|
446
|
|
447
|
//country
|
448
|
|
449
|
String country = " ";
|
450
|
|
451
|
for (FieldTypeProtos.Qualifier c : metadata.getCountryList()) {
|
452
|
country += clean(c.getClassid()) + SEPERATOR;
|
453
|
}
|
454
|
|
455
|
buff.append(country + DELIM);
|
456
|
|
457
|
//Best License
|
458
|
buff.append(getBestLicense(data.getResult()) + DELIM);
|
459
|
|
460
|
//Description
|
461
|
dataStr = " ";
|
462
|
|
463
|
for (StringField desc : metadata.getDescriptionList()) {
|
464
|
dataStr += clean(cleanText(desc.getValue(), DELIM));
|
465
|
break;
|
466
|
}
|
467
|
|
468
|
buff.append(dataStr + DELIM);
|
469
|
|
470
|
//Journal
|
471
|
buff.append(clean(metadata.getJournal().getName()) + DELIM); //#null#!
|
472
|
|
473
|
|
474
|
// TODO ERI SOS : HERE IN GET JOUTNAL. GET DATA INFO I CAN FIND PROVENANCE AND SIMILARITY
|
475
|
|
476
|
// TODO isRelatedTo
|
477
|
|
478
|
// resource type
|
479
|
buff.append(clean(metadata.getResourcetype().getClassname()) + DELIM);
|
480
|
// device
|
481
|
buff.append(clean(metadata.getDevice().getValue()) + DELIM);
|
482
|
// size
|
483
|
buff.append(clean(metadata.getSize().getValue()) + DELIM);
|
484
|
// version
|
485
|
buff.append(clean(metadata.getVersion().getValue()) + DELIM);
|
486
|
|
487
|
// metadata update
|
488
|
buff.append(clean(metadata.getLastmetadataupdate().getValue()) + DELIM);
|
489
|
// metadata version
|
490
|
buff.append(clean(metadata.getMetadataversionnumber().getValue()) + DELIM);
|
491
|
|
492
|
|
493
|
// year
|
494
|
buff.append(clean(getYearInt(metadata.getDateofacceptance().getValue())) + DELIM);
|
495
|
|
496
|
// type
|
497
|
buff.append(clean(metadata.getResulttype().getClassname()) + DELIM);
|
498
|
|
499
|
buff.append(getTrust(data) + DELIM);
|
500
|
|
501
|
|
502
|
return buff.toString();
|
503
|
}
|
504
|
|
505
|
|
506
|
private static String buildProject(OafEntity data, String DELIM) {
|
507
|
String SEPERATOR = ";";
|
508
|
|
509
|
StringBuilder buff = new StringBuilder();
|
510
|
|
511
|
buff.append(getHeader(data, DELIM));
|
512
|
Project.Metadata metadata = data.getProject().getMetadata();
|
513
|
|
514
|
|
515
|
//Code
|
516
|
buff.append(metadata.getCode().getValue() + DELIM);
|
517
|
// `Websiteurl`,
|
518
|
buff.append(clean(metadata.getWebsiteurl().getValue()) + DELIM);
|
519
|
//TODO here
|
520
|
|
521
|
// `Acronym`,
|
522
|
buff.append(clean(metadata.getAcronym().getValue()) + DELIM);
|
523
|
|
524
|
//Title
|
525
|
buff.append(clean(metadata.getTitle().getValue()) + DELIM);
|
526
|
|
527
|
// Startdate
|
528
|
buff.append(clean(metadata.getStartdate().getValue()) + DELIM);
|
529
|
|
530
|
// Enddate
|
531
|
buff.append(clean(metadata.getEnddate().getValue()) + DELIM);
|
532
|
|
533
|
//`Call identifer`
|
534
|
buff.append(clean(metadata.getCallidentifier().getValue()) + DELIM);
|
535
|
|
536
|
//`KeyWords`
|
537
|
buff.append(clean(metadata.getKeywords().getValue()) + DELIM);
|
538
|
|
539
|
//`Duration`
|
540
|
buff.append(clean(metadata.getDuration().getValue()) + DELIM);
|
541
|
|
542
|
//esc39
|
543
|
buff.append(clean(metadata.getEcsc39().getValue()) + DELIM);
|
544
|
|
545
|
//`Contracttype`
|
546
|
buff.append(clean(metadata.getContracttype().getClassname()) + DELIM);
|
547
|
|
548
|
//`OA mandate pubs` TODO DOES NOT EXIST
|
549
|
buff.append(clean(metadata.getOamandatepublications().getValue()) + DELIM);
|
550
|
//`Subjects`
|
551
|
String dataStr = new String();
|
552
|
for (StructuredProperty s : metadata.getSubjectsList()) {
|
553
|
|
554
|
dataStr += clean(s.getValue()) + SEPERATOR;
|
555
|
}
|
556
|
buff.append(dataStr + DELIM);
|
557
|
|
558
|
//`EC293`
|
559
|
buff.append(clean(metadata.getEcarticle293().getValue()) + DELIM);
|
560
|
|
561
|
List<StringField> fundList = metadata.getFundingtreeList();
|
562
|
|
563
|
if (!fundList.isEmpty()) // `funding_lvl0`,
|
564
|
{//TODO funder + 3 funding levels
|
565
|
/* funder text,
|
566
|
funding_lvl0 text,
|
567
|
funding_lvl1 text,
|
568
|
funding_lvl2 text,
|
569
|
funding_lvl3 text,*/
|
570
|
buff.append(FundingParser.getFundingInfo(fundList.get(0).getValue(), DELIM));
|
571
|
} else {
|
572
|
|
573
|
buff.append(FundingParser.getFundingInfo("", DELIM));
|
574
|
}
|
575
|
|
576
|
buff.append(getTrust(data) + DELIM);
|
577
|
|
578
|
return buff.toString();
|
579
|
|
580
|
}
|
581
|
|
582
|
|
583
|
private static String buildPerson(OafEntity data, String DELIM) {
|
584
|
String SEPERATOR = ";";
|
585
|
|
586
|
PersonProtos.Person person = data.getPerson();
|
587
|
PersonProtos.Person.Metadata metadata = person.getMetadata();
|
588
|
|
589
|
StringBuilder buff = new StringBuilder();
|
590
|
|
591
|
buff.append(getHeader(data, DELIM));
|
592
|
|
593
|
// `firstname`,
|
594
|
buff.append(clean(metadata.getFirstname().getValue()) + DELIM);
|
595
|
|
596
|
// `secondNames`,
|
597
|
String dataStr = new String();
|
598
|
|
599
|
for (StringField s : metadata.getSecondnamesList()) {
|
600
|
dataStr += clean(s.getValue()) + ' ';
|
601
|
}
|
602
|
|
603
|
buff.append(dataStr + DELIM);
|
604
|
|
605
|
// `fullname`,
|
606
|
buff.append(clean(metadata.getFullname().getValue()) + DELIM);
|
607
|
|
608
|
// `Fax`,
|
609
|
buff.append(clean(metadata.getFax().getValue()) + DELIM);
|
610
|
|
611
|
// `Email`,
|
612
|
buff.append(clean(metadata.getEmail().getValue()) + DELIM);
|
613
|
|
614
|
// `Phone`,
|
615
|
buff.append(clean(metadata.getPhone().getValue()) + DELIM);
|
616
|
|
617
|
// `Nationality`,
|
618
|
buff.append(clean(metadata.getNationality().getClassid()) + DELIM);
|
619
|
|
620
|
// `PIDS`,
|
621
|
dataStr = " ";
|
622
|
for (StructuredProperty s : data.getPidList()) {
|
623
|
|
624
|
dataStr += cleanId(s.getValue()) + ";";
|
625
|
}
|
626
|
buff.append(dataStr + DELIM);
|
627
|
|
628
|
buff.append(getTrust(data) + DELIM);
|
629
|
|
630
|
return buff.toString();
|
631
|
|
632
|
}
|
633
|
|
634
|
|
635
|
private static void getResultDatasources(OafEntity valueEntity, String DELIM, Set<String> returnList) {
|
636
|
String SEPERATOR = ";";
|
637
|
|
638
|
Result result = valueEntity.getResult();
|
639
|
|
640
|
if (valueEntity.getId().contains("dedup")) return;
|
641
|
|
642
|
//TODO hosted by
|
643
|
for (Instance instance : (result.getInstanceList())) {
|
644
|
String hostedBy = instance.getHostedby().getKey();
|
645
|
|
646
|
if (hostedBy != null && !hostedBy.isEmpty()) {
|
647
|
returnList.add("resultDatasource" + DELIM + "result" + DELIM +
|
648
|
cleanId(valueEntity.getId()) + DELIM + "datasource" + DELIM + cleanId(hostedBy) + DELIM);
|
649
|
}
|
650
|
}
|
651
|
|
652
|
//TODO collected froms
|
653
|
for (FieldTypeProtos.KeyValue collectedFromValue : (valueEntity.getCollectedfromList())) {
|
654
|
String collectedFrom = collectedFromValue.getKey();
|
655
|
if (collectedFrom != null && !collectedFrom.isEmpty())
|
656
|
returnList.add(("resultDatasource" + DELIM
|
657
|
+ "result" + DELIM + cleanId(valueEntity.getId()) + DELIM
|
658
|
+ "datasource" + DELIM + cleanId(collectedFrom) + DELIM));
|
659
|
|
660
|
}
|
661
|
|
662
|
|
663
|
}
|
664
|
|
665
|
public static String cleanText(String value, String DELIM) {
|
666
|
value = value.replace(DELIM, " ");
|
667
|
|
668
|
value = Jsoup.parse(value).text();
|
669
|
//PROSOXI PROSOXI DEN TO PIANEI AUTO H REGEX GIA TO REPLACE!!!!
|
670
|
|
671
|
value = value.replace("\n", " ");
|
672
|
value = value.replace(">", " ");
|
673
|
value = value.replace("<", " ");
|
674
|
value = value.replace(",", " ");
|
675
|
value = value.replace("\"", " ");
|
676
|
value = value.replace("'", " ");
|
677
|
|
678
|
value = value.replace("«", " ");
|
679
|
value = value.replace("»", " ");
|
680
|
value = value.replace("!", " ");
|
681
|
value = value.replace("#", " ");
|
682
|
|
683
|
|
684
|
value = value.replaceAll("[^a-zA-Z0-9 .-_:/@+=]+", " ");
|
685
|
|
686
|
return value;
|
687
|
}
|
688
|
|
689
|
public static String cleanUrl(String value, String DELIM) {
|
690
|
value = value.replace(DELIM, " ");
|
691
|
value = value.replace(" ", "");
|
692
|
value = value.replace("\"", " ");
|
693
|
value = value.replace("\n", "");
|
694
|
|
695
|
|
696
|
value = value.replace(">", " ");
|
697
|
value = value.replace("<", " ");
|
698
|
value = value.replace(",", " ");
|
699
|
value = value.replace("\"", " ");
|
700
|
value = value.replace("'", " ");
|
701
|
value = value.replace("«", " ");
|
702
|
value = value.replace("»", " ");
|
703
|
|
704
|
value = value.replace("!", " ");
|
705
|
value = value.replace("#", " ");
|
706
|
|
707
|
|
708
|
return value;
|
709
|
}
|
710
|
|
711
|
|
712
|
public static String cleanId(String value) {
|
713
|
if (value != null) {
|
714
|
// DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____::
|
715
|
// to datacite____:: )
|
716
|
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA
|
717
|
value = value.replaceFirst(".*\\|", "");
|
718
|
value = value.replace("\n", " ");
|
719
|
value = value.replace("\"", "");
|
720
|
value = value.replace("'", "");
|
721
|
value = value.replace("#", "");
|
722
|
value = value.replace("!", "");
|
723
|
|
724
|
value = value.replace("«", " ");
|
725
|
value = value.replace("»", " ");
|
726
|
|
727
|
}
|
728
|
|
729
|
|
730
|
return value;
|
731
|
|
732
|
}
|
733
|
|
734
|
|
735
|
//TODO make them in pairs
|
736
|
private static void getDedups(OafEntity valueEntity, String DELIM, Set<String> returnList) {
|
737
|
if (!valueEntity.getChildrenList().isEmpty() && valueEntity.getId().contains("dedup")) {
|
738
|
ArrayList<String> entries = new ArrayList<String>();
|
739
|
|
740
|
for (OafEntity child : valueEntity.getChildrenList()) {
|
741
|
if (child.getType() == valueEntity.getType() && !child.getId().contains("dedup")) {
|
742
|
// if it is result, then its the deduplication
|
743
|
entries.add(cleanId(child.getId()));
|
744
|
}
|
745
|
}
|
746
|
|
747
|
for (int i = 0; i < entries.size() - 1; i++) {
|
748
|
for (int j = i + 1; j < entries.size(); j++) {
|
749
|
returnList.add("dedup" + DELIM +
|
750
|
valueEntity.getType().name() + DELIM +
|
751
|
entries.get(i) + DELIM +
|
752
|
valueEntity.getType().name() + DELIM +
|
753
|
entries.get(j) + DELIM);
|
754
|
}
|
755
|
}
|
756
|
|
757
|
}
|
758
|
|
759
|
}
|
760
|
|
761
|
|
762
|
private static String getYearInt(String data) {
|
763
|
if (data == null || data.isEmpty() || data.equals("-1")) {
|
764
|
return " ";
|
765
|
}
|
766
|
String[] split = data.split("-");
|
767
|
|
768
|
if (split == null || split.length == 0) {
|
769
|
return " ";
|
770
|
}
|
771
|
|
772
|
|
773
|
return split[0];
|
774
|
|
775
|
|
776
|
}
|
777
|
|
778
|
|
779
|
private static String getBestLicense(Result result) {
|
780
|
FieldTypeProtos.Qualifier bestLicense = null;
|
781
|
LicenseComparator lc = new LicenseComparator();
|
782
|
for (Instance instance : (result.getInstanceList())) {
|
783
|
if (lc.compare(bestLicense, instance.getLicence()) > 0) {
|
784
|
bestLicense = instance.getLicence();
|
785
|
}
|
786
|
}
|
787
|
if (bestLicense != null) {
|
788
|
return bestLicense.getClassname();
|
789
|
} else {
|
790
|
return null;
|
791
|
}
|
792
|
}
|
793
|
|
794
|
|
795
|
private static String clean(String value) {
|
796
|
String SEPERATOR = ";";
|
797
|
String enclosing = "\"";
|
798
|
|
799
|
|
800
|
if (value != null) {
|
801
|
// TODO DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____::
|
802
|
// to datacite____:: )
|
803
|
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA
|
804
|
// value = value.replaceFirst(".*\\|", "");
|
805
|
|
806
|
value = value.replaceAll("[\"\\r\\\\;]", "");
|
807
|
|
808
|
value = value.replace(SEPERATOR, " ");
|
809
|
value = value.replace(">", " ");
|
810
|
value = value.replace("<", " ");
|
811
|
value = value.replace(",", " ");
|
812
|
value = value.replace("\"", " ");
|
813
|
value = value.replace("'", " ");
|
814
|
value = value.replace(enclosing, " ");
|
815
|
value = value.replace("«", " ");
|
816
|
value = value.replace("»", " ");
|
817
|
value = value.replace("!", " ");
|
818
|
value = value.replace("#", " ");
|
819
|
|
820
|
value = value.replaceAll("[^a-zA-Z0-9 .-_:/@+=]+", " ");
|
821
|
}
|
822
|
|
823
|
|
824
|
return value;
|
825
|
|
826
|
}
|
827
|
|
828
|
}
|