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