Revision 40597
Added by Eri Katsari almost 9 years ago
modules/dnet-openaire-lodexport/trunk/src/main/java/eu/dnetlib/data/mapreduce/hbase/lodExport/utils/test | ||
---|---|---|
1 |
package eu.dnetlib.data.mapreduce.hbase.lodExport.utils; |
|
2 |
|
|
3 |
import eu.dnetlib.data.mapreduce.util.LicenseComparator; |
|
4 |
import eu.dnetlib.data.proto.DatasourceProtos.Datasource; |
|
5 |
import eu.dnetlib.data.proto.DatasourceProtos.Datasource.Metadata; |
|
6 |
import eu.dnetlib.data.proto.FieldTypeProtos; |
|
7 |
import eu.dnetlib.data.proto.FieldTypeProtos.StringField; |
|
8 |
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty; |
|
9 |
import eu.dnetlib.data.proto.OafProtos; |
|
10 |
import eu.dnetlib.data.proto.OafProtos.Oaf; |
|
11 |
import eu.dnetlib.data.proto.OafProtos.OafEntity; |
|
12 |
import eu.dnetlib.data.proto.OafProtos.OafRel; |
|
13 |
import eu.dnetlib.data.proto.OrganizationProtos.Organization; |
|
14 |
import eu.dnetlib.data.proto.PersonProtos; |
|
15 |
import eu.dnetlib.data.proto.ProjectProtos.Project; |
|
16 |
import eu.dnetlib.data.proto.RelTypeProtos.RelType; |
|
17 |
import eu.dnetlib.data.proto.ResultProtos; |
|
18 |
import eu.dnetlib.data.proto.ResultProtos.Result; |
|
19 |
import eu.dnetlib.data.proto.ResultProtos.Result.Instance; |
|
20 |
import org.apache.log4j.Logger; |
|
21 |
|
|
22 |
import java.text.SimpleDateFormat; |
|
23 |
import java.util.ArrayList; |
|
24 |
import java.util.Date; |
|
25 |
import java.util.HashMap; |
|
26 |
import java.util.List; |
|
27 |
|
|
28 |
/** |
|
29 |
* @author eri Simple serializer that parses input Oaf Protos and prepares them |
|
30 |
* for sqoop |
|
31 |
*/ |
|
32 |
public class Serializer { |
|
33 |
|
|
34 |
private static String DELIM; |
|
35 |
private Logger log = Logger.getLogger(this.getClass()); |
|
36 |
private String NULL_STRING = null; |
|
37 |
private String NULL_NUM = "0"; |
|
38 |
|
|
39 |
private static String ENCLOSED; |
|
40 |
private FundingParser fundingParser = new FundingParser(); |
|
41 |
|
|
42 |
|
|
43 |
public Serializer() { |
|
44 |
} |
|
45 |
|
|
46 |
public String serialize(Oaf oaf) { |
|
47 |
|
|
48 |
switch (oaf.getKind()) { |
|
49 |
case entity: |
|
50 |
OafEntity valueEntity = oaf.getEntity(); |
|
51 |
|
|
52 |
switch (valueEntity.getType()) { |
|
53 |
case datasource: |
|
54 |
|
|
55 |
return buildDatasource(valueEntity); |
|
56 |
|
|
57 |
case organization: |
|
58 |
|
|
59 |
return buildOrganization(valueEntity); |
|
60 |
|
|
61 |
case project: |
|
62 |
|
|
63 |
return buildProject(valueEntity); |
|
64 |
case result: |
|
65 |
|
|
66 |
return buildResult(valueEntity); |
|
67 |
case person: |
|
68 |
return buildPerson(valueEntity); |
|
69 |
default: |
|
70 |
log.error("wrong type"); |
|
71 |
break; |
|
72 |
} |
|
73 |
break; |
|
74 |
case relation: |
|
75 |
OafRel valueRel = oaf.getRel(); |
|
76 |
|
|
77 |
return buildRel(valueRel); |
|
78 |
|
|
79 |
} |
|
80 |
|
|
81 |
return null; |
|
82 |
|
|
83 |
} |
|
84 |
|
|
85 |
public String serialize(OafRel oaf) { |
|
86 |
|
|
87 |
// switch (oaf.getRelType()) { |
|
88 |
|
|
89 |
// default: |
|
90 |
return buildRel(oaf); |
|
91 |
// } |
|
92 |
} |
|
93 |
|
|
94 |
|
|
95 |
public HashMap<String, List<String>> extractRelations(Oaf oaf) { |
|
96 |
OafEntity valueEntity = oaf.getEntity(); |
|
97 |
switch (valueEntity.getType()) { |
|
98 |
case result: |
|
99 |
HashMap<String, List<String>> relations = new HashMap<String, List<String>>(); |
|
100 |
relations.putAll(getResultDatasources(valueEntity)); |
|
101 |
|
|
102 |
return relations; |
|
103 |
default: |
|
104 |
|
|
105 |
return null; |
|
106 |
} |
|
107 |
|
|
108 |
} |
|
109 |
|
|
110 |
private HashMap<String, List<String>> getResultDatasources(OafEntity valueEntity) { |
|
111 |
|
|
112 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
113 |
List<String> buffs = new ArrayList<String>(); |
|
114 |
Result result = valueEntity.getResult(); |
|
115 |
String relType="resultDatasource" ; |
|
116 |
|
|
117 |
|
|
118 |
|
|
119 |
//TODO hosted by |
|
120 |
for (Instance instance : (result.getInstanceList())) { |
|
121 |
|
|
122 |
|
|
123 |
String hostedBy = instance.getHostedby().getKey(); |
|
124 |
if (hostedBy != null && !hostedBy.isEmpty()) { |
|
125 |
buffs.add(cleanId(valueEntity.getId()) + DELIM + getStringField(hostedBy)); |
|
126 |
} |
|
127 |
} |
|
128 |
|
|
129 |
//TODO collected froms |
|
130 |
for (FieldTypeProtos.KeyValue collectedFromValue : (valueEntity.getCollectedfromList())) { |
|
131 |
String collectedFrom = collectedFromValue.getKey(); |
|
132 |
if (collectedFrom != null && !collectedFrom.isEmpty()) buffs.add((cleanId(valueEntity.getId()) + DELIM + getStringField(collectedFrom))); |
|
133 |
|
|
134 |
} |
|
135 |
|
|
136 |
rels.put(relType ,buffs); |
|
137 |
return rels; |
|
138 |
|
|
139 |
} |
|
140 |
|
|
141 |
|
|
142 |
private String getResultResult(OafRel oaf) { |
|
143 |
|
|
144 |
|
|
145 |
String buff = new String(); |
|
146 |
|
|
147 |
buff += getStringField(oaf.getTarget()); |
|
148 |
|
|
149 |
buff += getStringField(String.valueOf(oaf.getResultResult().getSimilarity().getSimilarity())); |
|
150 |
|
|
151 |
/* • hasAmongTopNSimilarDocuments: r1 hasAmongTopNSimilarDocuments r2 means that Results r1 and r2 are similar, and that we also have r2 isAmongTopNSimilarDocuments of r1. |
|
152 |
In OpenAIRE, N so far always equals 20. |
|
153 |
• isAmongTopNSimilarDocuments: r1 isAmongTopNSimilarDocuments r2 means that Results r1 and r2 are similar and that r2 hasAmongTopNSimilarDocuments of r1; |
|
154 |
• 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. |
|
155 |
Scheme "dnet:result_result_relations"; |
|
156 |
oaf. */ |
|
157 |
|
|
158 |
return buff; |
|
159 |
} |
|
160 |
|
|
161 |
private String buildRel(OafRel Rel) { |
|
162 |
return getStringField(Rel.getSource()) + getStringField(Rel.getTarget()); |
|
163 |
} |
|
164 |
|
|
165 |
|
|
166 |
/* private String getResultProject(OafRel oaf) { |
|
167 |
String buff = new String();+ |
|
168 |
String result = oaf.getTarget(); |
|
169 |
|
|
170 |
buff += getStringField(result); |
|
171 |
// is declared as int!!! |
|
172 |
long diff = DATEDIFF(oaf.getResultProject().getOutcome().getRelMetadata().getEnddate(), oaf.getResultProject().getOutcome().getRelMetadata().getStartdate()); |
|
173 |
if (diff < 0) { |
|
174 |
diff = 0; |
|
175 |
} |
|
176 |
buff += getNumericField(String.valueOf(diff)); |
|
177 |
|
|
178 |
return buff; |
|
179 |
}*/ |
|
180 |
|
|
181 |
|
|
182 |
private String buildDatasource(OafEntity data) { |
|
183 |
|
|
184 |
String buff = new String(); |
|
185 |
|
|
186 |
Datasource d = data.getDatasource(); |
|
187 |
|
|
188 |
Metadata metadata = d.getMetadata(); |
|
189 |
|
|
190 |
//`original Id` |
|
191 |
String originalId = new String(); |
|
192 |
|
|
193 |
for (String oid : data.getOriginalIdList()) { |
|
194 |
originalId += oid + ";"; |
|
195 |
} |
|
196 |
|
|
197 |
buff += getStringField(originalId); |
|
198 |
|
|
199 |
//dateOfCollection |
|
200 |
buff += getStringDateField(data.getDateofcollection()); |
|
201 |
|
|
202 |
//Provider Type |
|
203 |
if (metadata.hasDatasourcetype()) { |
|
204 |
buff += getStringField(metadata.getDatasourcetype().getClassname().replaceFirst(".*::", "")); |
|
205 |
} else { |
|
206 |
buff += getStringField(null); |
|
207 |
} |
|
208 |
|
|
209 |
//Compatibility |
|
210 |
buff += getStringField(metadata.getOpenairecompatibility().getClassname()); |
|
211 |
|
|
212 |
//Official Name |
|
213 |
buff += getStringField(metadata.getOfficialname().getValue()); |
|
214 |
|
|
215 |
// English Name |
|
216 |
buff += getStringField(metadata.getEnglishname().getValue()); |
|
217 |
|
|
218 |
//Website URL |
|
219 |
buff += getStringField(metadata.getWebsiteurl().getValue()); |
|
220 |
|
|
221 |
//LogoURL |
|
222 |
buff += getStringField(metadata.getLogourl().getValue()); |
|
223 |
|
|
224 |
|
|
225 |
buff += getStringField(metadata.getContactemail().getValue()); |
|
226 |
|
|
227 |
//Namespaceprefix |
|
228 |
buff += getStringField(metadata.getNamespaceprefix().getValue()); |
|
229 |
|
|
230 |
// latitude |
|
231 |
buff += getStringField(metadata.getLatitude().getValue()); |
|
232 |
|
|
233 |
// longtitude |
|
234 |
buff += getStringField(metadata.getLongitude().getValue()); |
|
235 |
|
|
236 |
// dateofvalidation, |
|
237 |
buff += getStringField(metadata.getDateofvalidation().getValue()); |
|
238 |
|
|
239 |
//Description |
|
240 |
buff += getStringField(metadata.getDescription().getValue()); |
|
241 |
|
|
242 |
//subjects |
|
243 |
String subj = new String(); |
|
244 |
for (StructuredProperty s : metadata.getSubjectsList()) { |
|
245 |
subj += s.getValue() + ';'; |
|
246 |
|
|
247 |
} |
|
248 |
|
|
249 |
//subjects |
|
250 |
buff += getStringField(subj); |
|
251 |
|
|
252 |
//Number of items |
|
253 |
buff += getStringField(metadata.getOdnumberofitems().getValue()); |
|
254 |
|
|
255 |
//Date of number of items |
|
256 |
buff += getStringField(metadata.getOdnumberofitemsdate().getValue()); |
|
257 |
|
|
258 |
// Policies |
|
259 |
buff += getStringField(metadata.getOdpolicies().getValue()); |
|
260 |
|
|
261 |
//languages |
|
262 |
String languages = new String(); |
|
263 |
for (StringField lang : metadata.getOdlanguagesList()) { |
|
264 |
languages += lang.getValue() + ";"; |
|
265 |
} |
|
266 |
buff += getStringField(languages); |
|
267 |
|
|
268 |
|
|
269 |
// Content type |
|
270 |
String contentType = new String(); |
|
271 |
for (StringField c : metadata.getOdcontenttypesList()) { |
|
272 |
contentType += c.getValue() + ";"; |
|
273 |
} |
|
274 |
buff += getStringField(contentType); |
|
275 |
|
|
276 |
//Access info package |
|
277 |
String accessInfo = new String(); |
|
278 |
for (StringField c : metadata.getAccessinfopackageList()) { |
|
279 |
accessInfo += c.getValue() + ";"; |
|
280 |
} |
|
281 |
buff += getStringField(accessInfo); |
|
282 |
|
|
283 |
//Release start date |
|
284 |
buff += getStringField(metadata.getReleasestartdate().getValue()); |
|
285 |
|
|
286 |
//Release end date |
|
287 |
buff += getStringField(metadata.getReleaseenddate().getValue()); |
|
288 |
|
|
289 |
//Mission statement url |
|
290 |
buff += getStringField(metadata.getMissionstatementurl().getValue()); |
|
291 |
|
|
292 |
//Data provider |
|
293 |
buff += getStringField(String.valueOf(metadata.getDataprovider().getValue())); |
|
294 |
|
|
295 |
//Service provider |
|
296 |
buff += getStringField(String.valueOf(metadata.getServiceprovider().getValue())); |
|
297 |
|
|
298 |
//Database access type |
|
299 |
buff += getStringField(metadata.getDatabaseaccessrestriction().getValue()); |
|
300 |
|
|
301 |
//Data upload type |
|
302 |
buff += getStringField(metadata.getDatauploadtype().getValue()); |
|
303 |
|
|
304 |
//Data upload restrictions |
|
305 |
buff += getStringField(metadata.getDatauploadrestriction().getValue()); |
|
306 |
|
|
307 |
//Versioning |
|
308 |
buff += getStringField(String.valueOf(metadata.getVersioning().getValue())); |
|
309 |
|
|
310 |
//Citation guideline url |
|
311 |
buff += getStringField(String.valueOf(metadata.getCitationguidelineurl().getValue())); |
|
312 |
|
|
313 |
//Quality management kind |
|
314 |
buff += getStringField(String.valueOf(metadata.getQualitymanagementkind().getValue())); |
|
315 |
|
|
316 |
//PID systems |
|
317 |
buff += getStringField(metadata.getPidsystems().getValue()); |
|
318 |
|
|
319 |
//Certificates |
|
320 |
buff += getStringField(metadata.getCertificates().getValue()); |
|
321 |
|
|
322 |
//Policies |
|
323 |
String policies = new String(); |
|
324 |
for (FieldTypeProtos.KeyValue property : metadata.getPoliciesList()) { |
|
325 |
policies += property.getValue() + ";"; |
|
326 |
} |
|
327 |
buff += getStringField(policies); |
|
328 |
|
|
329 |
return buff; |
|
330 |
} |
|
331 |
|
|
332 |
|
|
333 |
private String buildOrganization(OafEntity data) { |
|
334 |
|
|
335 |
String buff = new String(); |
|
336 |
|
|
337 |
Organization organization = data.getOrganization(); |
|
338 |
eu.dnetlib.data.proto.OrganizationProtos.Organization.Metadata metadata = organization.getMetadata(); |
|
339 |
|
|
340 |
//`original Id` |
|
341 |
String originalId = new String(); |
|
342 |
for (String oid : data.getOriginalIdList()) { |
|
343 |
originalId += oid + ";"; |
|
344 |
} |
|
345 |
|
|
346 |
|
|
347 |
buff += getStringField(originalId); |
|
348 |
//dateOfCollection |
|
349 |
buff += getStringDateField(data.getDateofcollection()); |
|
350 |
|
|
351 |
|
|
352 |
//getLegalshortname |
|
353 |
buff += getStringField(metadata.getLegalshortname().getValue()); |
|
354 |
// `name`, |
|
355 |
buff += getStringField(metadata.getLegalname().getValue()); |
|
356 |
//website URL |
|
357 |
buff += getStringField(metadata.getWebsiteurl().getValue()); |
|
358 |
//logourl |
|
359 |
buff += getStringField(metadata.getLogourl().getValue()); |
|
360 |
// `country`, |
|
361 |
buff += getStringField(metadata.getCountry().getClassname()); |
|
362 |
|
|
363 |
|
|
364 |
return buff; |
|
365 |
|
|
366 |
} |
|
367 |
|
|
368 |
private String buildResult(OafEntity data) { |
|
369 |
String buff = new String(); |
|
370 |
|
|
371 |
Result result = data.getResult(); |
|
372 |
eu.dnetlib.data.proto.ResultProtos.Result.Metadata metadata = result.getMetadata(); |
|
373 |
|
|
374 |
//`original Id` |
|
375 |
String originalId = new String(); |
|
376 |
|
|
377 |
for (String oid : data.getOriginalIdList()) { |
|
378 |
originalId += oid + ";"; |
|
379 |
} |
|
380 |
|
|
381 |
buff += getStringField(originalId); |
|
382 |
|
|
383 |
//dateOfCollection |
|
384 |
buff += getStringDateField(data.getDateofcollection()); |
|
385 |
// titleString |
|
386 |
String titleString = new String(); |
|
387 |
String alternativeTitles = new String(); |
|
388 |
|
|
389 |
for (int i = 0; i < metadata.getTitleList().size(); i++) { |
|
390 |
StructuredProperty title = metadata.getTitleList().get(i); |
|
391 |
if (i == 0) { |
|
392 |
titleString = title.getValue().replaceAll("\\s+", " "); |
|
393 |
titleString = titleString.replaceAll("\n", " "); |
|
394 |
} else { |
|
395 |
alternativeTitles += title.getValue().replaceAll("\\s+", " ") + " ; "; |
|
396 |
alternativeTitles = alternativeTitles.replaceAll("\n", " "); |
|
397 |
} |
|
398 |
break; |
|
399 |
} |
|
400 |
|
|
401 |
// pubtitle |
|
402 |
buff += getStringField(titleString); |
|
403 |
|
|
404 |
// alternative titles |
|
405 |
//buff += getStringField(alternativeTitles); // null#! |
|
406 |
// date of acceptance CHANGED THIS TO DATE FORMAT |
|
407 |
buff += getStringDateField(metadata.getDateofacceptance().getValue()); |
|
408 |
|
|
409 |
// publisher |
|
410 |
buff += getStringField(metadata.getPublisher().getValue()); |
|
411 |
|
|
412 |
|
|
413 |
//PID |
|
414 |
String pids = new String(); |
|
415 |
for (StructuredProperty p : data.getPidList()) { |
|
416 |
pids += p.getValue() + ";"; |
|
417 |
} |
|
418 |
buff += getStringField(pids); |
|
419 |
|
|
420 |
//language |
|
421 |
buff += getStringField(metadata.getLanguage().getClassname()); |
|
422 |
|
|
423 |
// RelevantDate |
|
424 |
String reldate = new String(); |
|
425 |
|
|
426 |
for (StructuredProperty p : metadata.getRelevantdateList()) { |
|
427 |
reldate += p.getValue(); |
|
428 |
break; |
|
429 |
} |
|
430 |
buff += getStringField(reldate); |
|
431 |
|
|
432 |
//Subject |
|
433 |
String subjects = new String(); |
|
434 |
for (StructuredProperty subj : metadata.getSubjectList()) { |
|
435 |
|
|
436 |
if (isValidTopic(subj.getValue())) { |
|
437 |
if (!isNumeric(subj.getValue())) { |
|
438 |
subjects += subj.getValue() + ";"; |
|
439 |
} |
|
440 |
} |
|
441 |
} |
|
442 |
|
|
443 |
buff += subjects; |
|
444 |
|
|
445 |
// TODO Instance |
|
446 |
// buff += getStringField(); |
|
447 |
|
|
448 |
//TODO ExternalReference |
|
449 |
|
|
450 |
|
|
451 |
//Source |
|
452 |
String source = new String(); |
|
453 |
for (StringField s : metadata.getSourceList()) { |
|
454 |
source += s.getValue() + ";"; |
|
455 |
} |
|
456 |
|
|
457 |
buff += getStringField(source); |
|
458 |
|
|
459 |
|
|
460 |
//TODO Format |
|
461 |
buff += getStringField(""); |
|
462 |
//DOES NOT EXIST |
|
463 |
/*String formatString = new String(); |
|
464 |
for (StringField format : metadata.getFormatList()) { |
|
465 |
formatString = format.getValue(); |
|
466 |
break;} |
|
467 |
buff += getStringField(formatString); |
|
468 |
}*/ |
|
469 |
//Context |
|
470 |
String context = new String(); |
|
471 |
for (Result.Context s : metadata.getContextList()) { |
|
472 |
context += s.getId() + ";"; |
|
473 |
} |
|
474 |
buff += getStringField(context); |
|
475 |
|
|
476 |
//country TODO does not exist; throws error |
|
477 |
String country = new String(); |
|
478 |
// for (Qualifier s : metadata.getCountryList()) { |
|
479 |
// country += s.getClassname() + ";"; |
|
480 |
// } |
|
481 |
|
|
482 |
buff += getStringField(country); |
|
483 |
|
|
484 |
// access_mode, |
|
485 |
buff += getStringField(getAccessMode(result)); |
|
486 |
//Best License |
|
487 |
buff += getStringField(getBestLicense(result)); |
|
488 |
//Description |
|
489 |
String description = new String(); |
|
490 |
|
|
491 |
for (StringField desc : metadata.getDescriptionList()) { |
|
492 |
description += desc; |
|
493 |
break; |
|
494 |
} |
|
495 |
buff += getStringField(description); |
|
496 |
//Journal |
|
497 |
buff += getStringField(metadata.getJournal().getName().replaceAll("\n", " ")); //#null#! |
|
498 |
|
|
499 |
//journalName |
|
500 |
buff += getStringField(metadata.getJournal().getName().replaceAll("\n", " ")); //#null#! |
|
501 |
|
|
502 |
// TODO ERI SOS : HERE IN GET JOUTNAL. GET DATA INFO I CAN FIND PROVENANCE AND SIMILARITY |
|
503 |
|
|
504 |
//ISSN |
|
505 |
buff += getStringField(metadata.getJournal().getIssnLinking()); |
|
506 |
|
|
507 |
//embargoEndDate |
|
508 |
buff += getStringField(metadata.getEmbargoenddate().getValue()); |
|
509 |
|
|
510 |
|
|
511 |
// `authors`, |
|
512 |
int authors = 0; |
|
513 |
String delayed = "no"; |
|
514 |
|
|
515 |
for (OafRel rel : data.getCachedRelList()) { |
|
516 |
|
|
517 |
if (rel.getRelType().equals(RelType.personResult)) { |
|
518 |
|
|
519 |
authors++; |
|
520 |
} else if (rel.getRelType().equals(RelType.resultProject)) |
|
521 |
// TODO remember : in result Project, first id is project, second is result. |
|
522 |
|
|
523 |
{ |
|
524 |
String daysfromend = getYearDifferenceInteger(rel.getResultProject().getOutcome().getRelMetadata().getEnddate(), rel.getResultProject().getOutcome().getRelMetadata().getStartdate()); |
|
525 |
if (Integer.parseInt(daysfromend) > 0) { |
|
526 |
delayed = "yes"; |
|
527 |
} |
|
528 |
} |
|
529 |
} |
|
530 |
|
|
531 |
buff += getNumericField(String.valueOf(authors)); |
|
532 |
|
|
533 |
|
|
534 |
// TODO isRelatedTo |
|
535 |
|
|
536 |
// resource type |
|
537 |
buff += getStringField(metadata.getResourcetype().getClassname()); |
|
538 |
// device |
|
539 |
buff += getStringField(metadata.getDevice().getValue()); |
|
540 |
// size |
|
541 |
buff += getStringField(metadata.getSize().getValue()); |
|
542 |
// version |
|
543 |
buff += getStringField(metadata.getVersion().getValue()); |
|
544 |
// metadata update |
|
545 |
buff += getStringField(metadata.getLastmetadataupdate().getValue()); |
|
546 |
// metadata version |
|
547 |
buff += getStringField(metadata.getMetadataversionnumber().getValue()); |
|
548 |
|
|
549 |
// `delayed`, |
|
550 |
buff += getStringField(delayed); |
|
551 |
|
|
552 |
// year |
|
553 |
buff += getYearInt(metadata.getDateofacceptance().getValue()); |
|
554 |
|
|
555 |
// type |
|
556 |
buff += getStringField(metadata.getResulttype().getClassname()); |
|
557 |
//classification |
|
558 |
String classification = new String(); |
|
559 |
for (Instance instance : (result.getInstanceList())) { |
|
560 |
|
|
561 |
if (instance.getInstancetype().getClassname() != null && !instance.getInstancetype().getClassname().isEmpty()) { |
|
562 |
classification += instance.getInstancetype().getClassname() + ';'; |
|
563 |
} |
|
564 |
} |
|
565 |
|
|
566 |
buff += getStringField(classification); |
|
567 |
|
|
568 |
// hosted by |
|
569 |
String hostedBy = new String(); |
|
570 |
for (Instance instance : (result.getInstanceList())) { |
|
571 |
String host = instance.getHostedby().getKey(); |
|
572 |
if (host != null && !host.isEmpty()) { |
|
573 |
hostedBy += host + ";"; |
|
574 |
} |
|
575 |
} |
|
576 |
|
|
577 |
buff += getStringField(hostedBy); |
|
578 |
|
|
579 |
// collectedfrom |
|
580 |
String collectedFrom = new String(); |
|
581 |
for (FieldTypeProtos.KeyValue collectedFromValue : (data.getCollectedfromList())) { |
|
582 |
|
|
583 |
String host = collectedFromValue.getKey(); |
|
584 |
if (host != null && !host.isEmpty()) { |
|
585 |
collectedFrom += host + ';'; |
|
586 |
|
|
587 |
} |
|
588 |
} |
|
589 |
buff += getStringField(collectedFrom); |
|
590 |
return buff; |
|
591 |
} |
|
592 |
|
|
593 |
|
|
594 |
private String buildProject(OafEntity data) { |
|
595 |
|
|
596 |
String buff = new String(); |
|
597 |
|
|
598 |
Project project = data.getProject(); |
|
599 |
eu.dnetlib.data.proto.ProjectProtos.Project.Metadata metadata = project.getMetadata(); |
|
600 |
|
|
601 |
//`original Id` |
|
602 |
String originalId = new String(); |
|
603 |
|
|
604 |
for (String oid : data.getOriginalIdList()) { |
|
605 |
originalId += oid + ";"; |
|
606 |
} |
|
607 |
|
|
608 |
buff += getStringField(originalId); |
|
609 |
|
|
610 |
//dateOfCollection |
|
611 |
buff += getStringDateField(data.getDateofcollection()); |
|
612 |
|
|
613 |
//Code |
|
614 |
buff += getStringField(metadata.getCode().getValue()); |
|
615 |
// `url`, |
|
616 |
buff += getStringField(metadata.getWebsiteurl().getValue()); |
|
617 |
|
|
618 |
// `acronym`, |
|
619 |
String acronym = metadata.getAcronym().getValue(); |
|
620 |
if (acronym.equalsIgnoreCase("UNKNOWN")) { |
|
621 |
acronym = metadata.getTitle().getValue(); |
|
622 |
} |
|
623 |
|
|
624 |
buff += getStringField(acronym); |
|
625 |
|
|
626 |
//title! |
|
627 |
String title = getStringField(metadata.getTitle().getValue()); |
|
628 |
buff += getStringField(title); |
|
629 |
|
|
630 |
// startdate |
|
631 |
buff += getNumericField(metadata.getStartdate().getValue()); |
|
632 |
|
|
633 |
// enddate |
|
634 |
buff += getNumericField(metadata.getEnddate().getValue()); |
|
635 |
|
|
636 |
//`Call identifer` |
|
637 |
buff += getStringField(metadata.getCallidentifier().getValue()); |
|
638 |
|
|
639 |
//`KeyWords` |
|
640 |
buff += getStringField(metadata.getKeywords().getValue()); |
|
641 |
|
|
642 |
//`Duration` |
|
643 |
buff += getStringField(metadata.getDuration().getValue()); |
|
644 |
|
|
645 |
//esc39 |
|
646 |
buff += getStringField(metadata.getEcsc39().getValue().toString()); |
|
647 |
|
|
648 |
//`Contracttype` |
|
649 |
buff += getStringField(metadata.getContracttype().getClassname()); |
|
650 |
|
|
651 |
//`OA mandate pubs` TODO DOES NOT EXIST |
|
652 |
buff += getStringField(metadata.getOamandatepublications().getValue()); |
|
653 |
|
|
654 |
//`Subjects` TODO DOES NOT EXIST EITHER |
|
655 |
String subjects = new String(); |
|
656 |
for (StructuredProperty s : metadata.getSubjectsList()) { |
|
657 |
|
|
658 |
subjects += s.getValue() + ';'; |
|
659 |
} |
|
660 |
buff += getStringField(subjects); |
|
661 |
|
|
662 |
|
|
663 |
//`EC293` |
|
664 |
buff += getStringField(metadata.getEcarticle293().getValue()); |
|
665 |
|
|
666 |
|
|
667 |
List<StringField> fundList = metadata.getFundingtreeList(); |
|
668 |
|
|
669 |
|
|
670 |
if (!fundList.isEmpty()) // `funding_lvl0`, |
|
671 |
{ |
|
672 |
//TODO funder + 3 funding levels |
|
673 |
/* funder text, |
|
674 |
funding_lvl0 text, |
|
675 |
funding_lvl1 text, |
|
676 |
funding_lvl2 text, |
|
677 |
funding_lvl3 text,*/ |
|
678 |
buff += this.fundingParser.getFundingInfo(fundList.get(0).getValue()); |
|
679 |
|
|
680 |
} else { |
|
681 |
|
|
682 |
buff += this.fundingParser.getFundingInfo(""); |
|
683 |
} |
|
684 |
|
|
685 |
return buff; |
|
686 |
|
|
687 |
} |
|
688 |
|
|
689 |
|
|
690 |
private String buildPerson(OafEntity data) { |
|
691 |
|
|
692 |
String buff = new String(); |
|
693 |
|
|
694 |
PersonProtos.Person person = data.getPerson(); |
|
695 |
eu.dnetlib.data.proto.PersonProtos.Person.Metadata metadata = person.getMetadata(); |
|
696 |
|
|
697 |
|
|
698 |
//`original Id` |
|
699 |
String originalId = new String(); |
|
700 |
|
|
701 |
for (String oid : data.getOriginalIdList()) { |
|
702 |
originalId += oid + ";"; |
|
703 |
} |
|
704 |
|
|
705 |
buff += getStringField(originalId); |
|
706 |
//dateOfCollection |
|
707 |
buff += getStringDateField(data.getDateofcollection()); |
|
708 |
|
|
709 |
// `firstname`, |
|
710 |
buff += metadata.getFirstname(); |
|
711 |
|
|
712 |
// `secondNames`, |
|
713 |
|
|
714 |
String secondNames = new String(); |
|
715 |
for (StringField s : metadata.getSecondnamesList()) { |
|
716 |
|
|
717 |
secondNames += s.getValue() + ' '; |
|
718 |
} |
|
719 |
|
|
720 |
buff += getStringField(secondNames); |
|
721 |
|
|
722 |
// `fullname`, |
|
723 |
buff += getStringField(metadata.getFullname().getValue()); |
|
724 |
// `Fax`, |
|
725 |
buff += getStringField(metadata.getFax().getValue()); |
|
726 |
// `Email`, |
|
727 |
buff += getStringField(metadata.getEmail().getValue()); |
|
728 |
// `Phone`, |
|
729 |
buff += getStringField(metadata.getPhone().getValue()); |
|
730 |
|
|
731 |
// `Nationality`, |
|
732 |
buff += getStringField(metadata.getNationality().getClassname()); |
|
733 |
|
|
734 |
// `PIDS`, |
|
735 |
String pids = new String(); |
|
736 |
for (StructuredProperty s : data.getPidList()) { |
|
737 |
|
|
738 |
pids += s.getValue() + ";"; |
|
739 |
|
|
740 |
|
|
741 |
} |
|
742 |
buff += getStringField(pids); |
|
743 |
|
|
744 |
// `collected from`, |
|
745 |
String collectedFrom = new String(); |
|
746 |
for (FieldTypeProtos.KeyValue s : data.getCollectedfromList()) { |
|
747 |
collectedFrom += s.getValue() + ";"; |
|
748 |
} |
|
749 |
buff += getStringField(collectedFrom); |
|
750 |
return buff; |
|
751 |
|
|
752 |
} |
|
753 |
|
|
754 |
public String cleanId(String value) { |
|
755 |
if (value != null) { |
|
756 |
// DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____:: |
|
757 |
|
|
758 |
|
|
759 |
// to datacite____:: ) |
|
760 |
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA |
|
761 |
value = value.replaceFirst(".*\\|", ""); |
|
762 |
value = value.replaceAll("\n", ""); |
|
763 |
value = value.replaceAll(DELIM, ""); |
|
764 |
value = value.replaceAll(ENCLOSED, ""); |
|
765 |
value = value.trim(); |
|
766 |
|
|
767 |
} |
|
768 |
if (value == null) { |
|
769 |
return null; |
|
770 |
} |
|
771 |
return ENCLOSED + value + ENCLOSED; |
|
772 |
|
|
773 |
} |
|
774 |
|
|
775 |
|
|
776 |
private String getNumericField(String data) { |
|
777 |
if (data == null || data.isEmpty() || data.equals("")) { |
|
778 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
779 |
} else { |
|
780 |
|
|
781 |
return ENCLOSED + data + ENCLOSED + DELIM; |
|
782 |
} |
|
783 |
} |
|
784 |
|
|
785 |
|
|
786 |
private String getYearDifferenceInteger(String enddate, String startdate) { |
|
787 |
|
|
788 |
if (enddate != null && !enddate.isEmpty() && startdate != null && !startdate.isEmpty()) { |
|
789 |
|
|
790 |
String[] split = startdate.split("-"); |
|
791 |
|
|
792 |
if (split == null || split.length == 0) { |
|
793 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
794 |
} |
|
795 |
|
|
796 |
int Startdate = Integer.parseInt(split[0]); |
|
797 |
|
|
798 |
split = enddate.split("-"); |
|
799 |
|
|
800 |
if (split == null || split.length == 0) { |
|
801 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
802 |
} |
|
803 |
|
|
804 |
int Enddate = Integer.parseInt(split[0]); |
|
805 |
|
|
806 |
int diff = Enddate - Startdate; |
|
807 |
|
|
808 |
return ENCLOSED + diff + ENCLOSED + DELIM; |
|
809 |
|
|
810 |
} |
|
811 |
|
|
812 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
813 |
} |
|
814 |
|
|
815 |
private String getYearInt(String data) { |
|
816 |
if (data == null || data.isEmpty() || data.equals("-1")) { |
|
817 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
818 |
} |
|
819 |
|
|
820 |
String[] split = data.split("-"); |
|
821 |
|
|
822 |
if (split == null || split.length == 0) { |
|
823 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
824 |
} |
|
825 |
|
|
826 |
String year = split[0]; |
|
827 |
|
|
828 |
year = cleanNumber(year); |
|
829 |
|
|
830 |
return ENCLOSED + year + ENCLOSED + DELIM; |
|
831 |
|
|
832 |
|
|
833 |
} |
|
834 |
|
|
835 |
private static String cleanNumber(String number) { |
|
836 |
number = number.replaceAll("[^A-Za-z0-9:,____]", ""); |
|
837 |
|
|
838 |
return number; |
|
839 |
} |
|
840 |
|
|
841 |
private String getStringField(String data) { |
|
842 |
|
|
843 |
if (data == null || data.isEmpty() || data.equals("")) { |
|
844 |
|
|
845 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
846 |
} else { |
|
847 |
|
|
848 |
String field = clean(data); |
|
849 |
if (field == null) { |
|
850 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
851 |
} else { |
|
852 |
return field + DELIM; |
|
853 |
} |
|
854 |
} |
|
855 |
} |
|
856 |
|
|
857 |
private String getStringDateField(String data) { |
|
858 |
|
|
859 |
if (data == null || data.isEmpty() || data.equals("") || data.equals("-1")) { |
|
860 |
|
|
861 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
862 |
} else { |
|
863 |
|
|
864 |
String field = clean(data); |
|
865 |
if (field == null) { |
|
866 |
return ENCLOSED + NULL_NUM + ENCLOSED + DELIM; |
|
867 |
} else { |
|
868 |
return field + DELIM; |
|
869 |
} |
|
870 |
} |
|
871 |
} |
|
872 |
|
|
873 |
|
|
874 |
public String getId(OafProtos.Oaf oaf) { |
|
875 |
switch (oaf.getKind()) { |
|
876 |
case entity: |
|
877 |
|
|
878 |
return cleanId(oaf.getEntity().getId()); |
|
879 |
case relation: |
|
880 |
|
|
881 |
return cleanId(oaf.getRel().getSource()); |
|
882 |
|
|
883 |
} |
|
884 |
return null; |
|
885 |
|
|
886 |
} |
|
887 |
|
|
888 |
private boolean isNumeric(String str) { |
|
889 |
|
|
890 |
str = str.replaceAll("[^A-Za-z0-9 ]", ""); |
|
891 |
str = str.replaceAll(" ", ""); |
|
892 |
return str.matches("-?\\d+(\\.\\d+)?"); // match a number with optional |
|
893 |
// '-' and decimal. |
|
894 |
} |
|
895 |
|
|
896 |
// there are topics with "null" as value -> replace them |
|
897 |
private boolean isValidTopic(String t) { |
|
898 |
|
|
899 |
if (t == null || t.isEmpty()) { |
|
900 |
return false; |
|
901 |
} |
|
902 |
|
|
903 |
if (t.equals("") || t.equals(" ")) { |
|
904 |
return false; |
|
905 |
} |
|
906 |
if (t.equals("null") || t.equals("Null") || t.equals("NULL")) { |
|
907 |
return false; |
|
908 |
} |
|
909 |
|
|
910 |
if (t.equals(ENCLOSED + ENCLOSED + DELIM) || t.equals(ENCLOSED + NULL_STRING + ENCLOSED + DELIM)) { |
|
911 |
return false; |
|
912 |
} |
|
913 |
// skip dedups |
|
914 |
if (t.contains("ddc:")) { |
|
915 |
|
|
916 |
return false; |
|
917 |
} |
|
918 |
return true; |
|
919 |
} |
|
920 |
|
|
921 |
|
|
922 |
private String getBestLicense(ResultProtos.Result result) { |
|
923 |
FieldTypeProtos.Qualifier bestLicense = null; |
|
924 |
LicenseComparator lc = new LicenseComparator(); |
|
925 |
for (ResultProtos.Result.Instance instance : (result.getInstanceList())) { |
|
926 |
if (lc.compare(bestLicense, instance.getLicence()) > 0) { |
|
927 |
bestLicense = instance.getLicence(); |
|
928 |
} |
|
929 |
} |
|
930 |
if (bestLicense != null) { |
|
931 |
return bestLicense.getClassname(); |
|
932 |
} else { |
|
933 |
return null; |
|
934 |
} |
|
935 |
} |
|
936 |
|
|
937 |
// here iterate over all values |
|
938 |
private String getAccessMode(ResultProtos.Result result) { |
|
939 |
String accessMode = null; |
|
940 |
for (ResultProtos.Result.Instance instance : (result.getInstanceList())) { |
|
941 |
if (instance.getLicence().getClassname() != null && !instance.getLicence().getClassname().isEmpty()) { |
|
942 |
accessMode = instance.getLicence().getClassname(); |
|
943 |
break; |
|
944 |
} |
|
945 |
|
|
946 |
} |
|
947 |
|
|
948 |
|
|
949 |
return accessMode; |
|
950 |
} |
|
951 |
|
|
952 |
|
|
953 |
public String getId(OafProtos.OafRel relOaf) { |
|
954 |
return cleanId(relOaf.getSource()); |
|
955 |
} |
|
956 |
|
|
957 |
private String clean(String value) { |
|
958 |
if (value != null) { |
|
959 |
// TODO DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____:: |
|
960 |
// to datacite____:: ) |
|
961 |
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA |
|
962 |
value = value.replaceFirst(".*\\|", ""); |
|
963 |
value = value.replaceAll(DELIM, ""); |
|
964 |
value = value.replaceAll(",", ""); |
|
965 |
value = value.replaceAll("\"", ""); |
|
966 |
value = value.replaceAll("'", ""); |
|
967 |
value = value.replaceAll(ENCLOSED, ""); |
|
968 |
value = value.replaceAll("\\r\\n|\\r|\\n", " "); |
|
969 |
value = value.replaceAll("\\s+", " "); |
|
970 |
value = value.replaceAll("(\\r|\\n)", " "); |
|
971 |
value = value.replaceAll("\\t", " "); |
|
972 |
|
|
973 |
// value = value.replaceAll("[^A-Za-z0-9:,____-;:]", " "); |
|
974 |
value = value.trim(); |
|
975 |
|
|
976 |
} |
|
977 |
if (value == null) { |
|
978 |
return null; |
|
979 |
} |
|
980 |
return ENCLOSED + value + ENCLOSED; |
|
981 |
|
|
982 |
} |
|
983 |
|
|
984 |
|
|
985 |
public long DATEDIFF(String startDate, String endDate) { |
|
986 |
long MILLISECS_PER_DAY = 24 * 60 * 60 * 1000; |
|
987 |
long days = 0l; |
|
988 |
SimpleDateFormat format = new SimpleDateFormat("yyyy-MM-dd"); // "dd/MM/yyyy HH:mm:ss"); |
|
989 |
// <startdate>2011-09-01</startdate> |
|
990 |
// <enddate>2015-08-31</enddate> |
|
991 |
Date dateIni = null; |
|
992 |
Date dateFin = null; |
|
993 |
|
|
994 |
if (startDate == null || startDate.isEmpty() || endDate == null || endDate.isEmpty()) { |
|
995 |
return 0; |
|
996 |
} |
|
997 |
try { |
|
998 |
dateIni = (Date) format.parse(startDate); |
|
999 |
dateFin = (Date) format.parse(endDate); |
|
1000 |
days = (dateFin.getTime() - dateIni.getTime()) / MILLISECS_PER_DAY; |
|
1001 |
} catch (Exception e) { |
|
1002 |
log.error(e.toString()); |
|
1003 |
return 0; |
|
1004 |
} |
|
1005 |
|
|
1006 |
return days; |
|
1007 |
} |
|
1008 |
|
|
1009 |
public String getDELIM() { |
|
1010 |
return DELIM; |
|
1011 |
} |
|
1012 |
|
|
1013 |
public void setDELIM(String dELIM) { |
|
1014 |
DELIM = dELIM; |
|
1015 |
} |
|
1016 |
|
|
1017 |
public String getNULL_STRING() { |
|
1018 |
return NULL_STRING; |
|
1019 |
} |
|
1020 |
|
|
1021 |
public void setNULL_STRING(String nULL_STRING) { |
|
1022 |
NULL_STRING = nULL_STRING; |
|
1023 |
} |
|
1024 |
|
|
1025 |
public String getNULL_NUM() { |
|
1026 |
return NULL_NUM; |
|
1027 |
} |
|
1028 |
|
|
1029 |
public void setNULL_NUM(String nULL_NUM) { |
|
1030 |
NULL_NUM = nULL_NUM; |
|
1031 |
} |
|
1032 |
|
|
1033 |
public String getENCLOSED() { |
|
1034 |
return ENCLOSED; |
|
1035 |
} |
|
1036 |
|
|
1037 |
public void setENCLOSED(String eNCLOSED) { |
|
1038 |
ENCLOSED = eNCLOSED; |
|
1039 |
} |
|
1040 |
|
|
1041 |
} |
modules/dnet-openaire-lodexport/trunk/src/main/java/eu/dnetlib/data/mapreduce/hbase/lodExport/utils/Serializer.java | ||
---|---|---|
22 | 22 |
import java.text.SimpleDateFormat; |
23 | 23 |
import java.util.ArrayList; |
24 | 24 |
import java.util.Date; |
25 |
import java.util.HashMap; |
|
26 | 25 |
import java.util.List; |
27 | 26 |
|
28 | 27 |
/** |
... | ... | |
31 | 30 |
*/ |
32 | 31 |
public class Serializer { |
33 | 32 |
|
34 |
private static String DELIM;
|
|
35 |
private Logger log = Logger.getLogger(this.getClass());
|
|
36 |
private String NULL_STRING = null;
|
|
37 |
private String NULL_NUM = "0";
|
|
33 |
private static String DELIM;
|
|
34 |
private Logger log = Logger.getLogger(this.getClass());
|
|
35 |
private String NULL_STRING = null;
|
|
36 |
private String NULL_NUM = "0";
|
|
38 | 37 |
|
39 |
private static String ENCLOSED;
|
|
40 |
private FundingParser fundingParser = new FundingParser();
|
|
38 |
private static String ENCLOSED;
|
|
39 |
private FundingParser fundingParser = new FundingParser();
|
|
41 | 40 |
|
42 | 41 |
|
43 |
public Serializer() {
|
|
44 |
}
|
|
42 |
public Serializer() {
|
|
43 |
}
|
|
45 | 44 |
|
46 |
public String serialize(Oaf oaf) {
|
|
45 |
public String serialize(Oaf oaf) {
|
|
47 | 46 |
|
48 |
switch (oaf.getKind()) {
|
|
49 |
case entity:
|
|
50 |
OafEntity valueEntity = oaf.getEntity();
|
|
47 |
switch (oaf.getKind()) {
|
|
48 |
case entity:
|
|
49 |
OafEntity valueEntity = oaf.getEntity();
|
|
51 | 50 |
|
52 |
switch (valueEntity.getType()) { |
|
53 |
case datasource: |
|
54 | 51 |
|
55 |
return buildDatasource(valueEntity); |
|
52 |
switch (valueEntity.getType()) { |
|
53 |
case datasource: |
|
56 | 54 |
|
57 |
case organization:
|
|
55 |
return buildDatasource(valueEntity);
|
|
58 | 56 |
|
59 |
return buildOrganization(valueEntity);
|
|
57 |
case organization:
|
|
60 | 58 |
|
61 |
case project:
|
|
59 |
return buildOrganization(valueEntity);
|
|
62 | 60 |
|
63 |
return buildProject(valueEntity); |
|
64 |
case result: |
|
61 |
case project: |
|
65 | 62 |
|
66 |
return buildResult(valueEntity); |
|
67 |
case person: |
|
68 |
return buildPerson(valueEntity); |
|
69 |
default: |
|
70 |
log.error("wrong type"); |
|
71 |
break; |
|
72 |
} |
|
73 |
break; |
|
74 |
case relation: |
|
75 |
OafRel valueRel = oaf.getRel(); |
|
63 |
return buildProject(valueEntity); |
|
64 |
case result: |
|
76 | 65 |
|
77 |
return buildRel(valueRel); |
|
66 |
return buildResult(valueEntity); |
|
67 |
case person: |
|
68 |
return buildPerson(valueEntity); |
|
69 |
default: |
|
70 |
log.error("wrong type"); |
|
71 |
break; |
|
72 |
} |
|
73 |
break; |
|
74 |
case relation: |
|
75 |
OafRel valueRel = oaf.getRel(); |
|
78 | 76 |
|
79 |
}
|
|
77 |
return buildRel(valueRel);
|
|
80 | 78 |
|
81 |
return null;
|
|
79 |
}
|
|
82 | 80 |
|
83 |
}
|
|
81 |
return null;
|
|
84 | 82 |
|
85 |
public String serialize(OafRel oaf) {
|
|
83 |
}
|
|
86 | 84 |
|
87 |
// switch (oaf.getRelType()) {
|
|
85 |
public String serialize(OafRel oaf) {
|
|
88 | 86 |
|
89 |
// default: |
|
90 |
return buildRel(oaf); |
|
91 |
// } |
|
92 |
} |
|
87 |
switch (oaf.getRelType()) { |
|
93 | 88 |
|
89 |
default: |
|
90 |
return buildRel(oaf); |
|
91 |
} |
|
94 | 92 |
|
95 |
public HashMap<String, List<String>> extractRelations(Oaf oaf) { |
|
96 |
OafEntity valueEntity = oaf.getEntity(); |
|
97 |
switch (valueEntity.getType()) { |
|
98 |
case result: |
|
99 |
HashMap<String, List<String>> relations = new HashMap<String, List<String>>(); |
|
100 |
relations.putAll(getResultDatasources(valueEntity)); |
|
93 |
} |
|
101 | 94 |
|
102 |
return relations; |
|
103 |
default: |
|
95 |
private String getResultResult(OafRel oaf) { |
|
104 | 96 |
|
105 |
return null; |
|
106 |
} |
|
107 | 97 |
|
108 |
}
|
|
98 |
String buff = new String();
|
|
109 | 99 |
|
110 |
private HashMap<String, List<String>> getResultDatasources(OafEntity valueEntity) {
|
|
100 |
buff += getStringField(oaf.getTarget());
|
|
111 | 101 |
|
112 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
113 |
List<String> buffs = new ArrayList<String>(); |
|
114 |
Result result = valueEntity.getResult(); |
|
115 |
String relType="resultDatasource" ; |
|
102 |
buff += getStringField(String.valueOf(oaf.getResultResult().getSimilarity().getSimilarity())); |
|
103 |
/* • hasAmongTopNSimilarDocuments: r1 hasAmongTopNSimilarDocuments r2 means that Results r1 and r2 are similar, and that we also have r2 isAmongTopNSimilarDocuments of r1. |
|
104 |
In OpenAIRE, N so far always equals 20. |
|
105 |
• isAmongTopNSimilarDocuments: r1 isAmongTopNSimilarDocuments r2 means that Results r1 and r2 are similar and that r2 hasAmongTopNSimilarDocuments of r1; |
|
106 |
• 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. |
|
107 |
Scheme "dnet:result_result_relations"; |
|
108 |
oaf. */ |
|
109 |
return buff; |
|
110 |
} |
|
116 | 111 |
|
112 |
private String buildRel(OafRel Rel) { |
|
117 | 113 |
|
114 |
return getStringField(Rel.getRelType().getDescriptorForType().getName()) + getStringField(Rel.getSource()) + getStringField(Rel.getTarget()); |
|
118 | 115 |
|
119 |
//TODO hosted by |
|
120 |
for (Instance instance : (result.getInstanceList())) { |
|
116 |
} |
|
121 | 117 |
|
118 |
private String buildDatasource(OafEntity data) { |
|
122 | 119 |
|
123 |
String hostedBy = instance.getHostedby().getKey(); |
|
124 |
if (hostedBy != null && !hostedBy.isEmpty()) { |
|
125 |
buffs.add(cleanId(valueEntity.getId()) + DELIM + getStringField(hostedBy)); |
|
126 |
} |
|
127 |
} |
|
120 |
String buff = new String(); |
|
128 | 121 |
|
129 |
//TODO collected froms |
|
130 |
for (FieldTypeProtos.KeyValue collectedFromValue : (valueEntity.getCollectedfromList())) { |
|
131 |
String collectedFrom = collectedFromValue.getKey(); |
|
132 |
if (collectedFrom != null && !collectedFrom.isEmpty()) buffs.add((cleanId(valueEntity.getId()) + DELIM + getStringField(collectedFrom))); |
|
122 |
Datasource d = data.getDatasource(); |
|
133 | 123 |
|
134 |
}
|
|
124 |
Metadata metadata = d.getMetadata();
|
|
135 | 125 |
|
136 |
rels.put(relType ,buffs);
|
|
137 |
return rels;
|
|
126 |
//`original Id`
|
|
127 |
String originalId = new String();
|
|
138 | 128 |
|
139 |
} |
|
129 |
for (String oid : data.getOriginalIdList()) { |
|
130 |
originalId += oid + ";"; |
|
131 |
} |
|
140 | 132 |
|
133 |
buff += getStringField(originalId); |
|
141 | 134 |
|
142 |
private String getResultResult(OafRel oaf) { |
|
135 |
//dateOfCollection |
|
136 |
buff += getStringDateField(data.getDateofcollection()); |
|
143 | 137 |
|
138 |
//Provider Type |
|
139 |
if (metadata.hasDatasourcetype()) { |
|
140 |
buff += getStringField(metadata.getDatasourcetype().getClassname().replaceFirst(".*::", "")); |
|
141 |
} else { |
|
142 |
buff += getStringField(null); |
|
143 |
} |
|
144 | 144 |
|
145 |
String buff = new String(); |
|
145 |
//Compatibility |
|
146 |
buff += getStringField(metadata.getOpenairecompatibility().getClassname()); |
|
146 | 147 |
|
147 |
buff += getStringField(oaf.getTarget()); |
|
148 |
//Official Name |
|
149 |
buff += getStringField(metadata.getOfficialname().getValue()); |
|
148 | 150 |
|
149 |
buff += getStringField(String.valueOf(oaf.getResultResult().getSimilarity().getSimilarity())); |
|
151 |
// English Name |
|
152 |
buff += getStringField(metadata.getEnglishname().getValue()); |
|
150 | 153 |
|
151 |
/* • hasAmongTopNSimilarDocuments: r1 hasAmongTopNSimilarDocuments r2 means that Results r1 and r2 are similar, and that we also have r2 isAmongTopNSimilarDocuments of r1. |
|
152 |
In OpenAIRE, N so far always equals 20. |
|
153 |
• isAmongTopNSimilarDocuments: r1 isAmongTopNSimilarDocuments r2 means that Results r1 and r2 are similar and that r2 hasAmongTopNSimilarDocuments of r1; |
|
154 |
• 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. |
|
155 |
Scheme "dnet:result_result_relations"; |
|
156 |
oaf. */ |
|
154 |
//Website URL |
|
155 |
buff += getStringField(metadata.getWebsiteurl().getValue()); |
|
157 | 156 |
|
158 |
return buff;
|
|
159 |
}
|
|
157 |
//LogoURL
|
|
158 |
buff += getStringField(metadata.getLogourl().getValue());
|
|
160 | 159 |
|
161 |
private String buildRel(OafRel Rel) { |
|
162 |
return getStringField(Rel.getSource()) + getStringField(Rel.getTarget()); |
|
163 |
} |
|
160 |
|
|
161 |
buff += getStringField(metadata.getContactemail().getValue()); |
|
164 | 162 |
|
163 |
//Namespaceprefix |
|
164 |
buff += getStringField(metadata.getNamespaceprefix().getValue()); |
|
165 | 165 |
|
166 |
/* private String getResultProject(OafRel oaf) { |
|
167 |
String buff = new String();+ |
|
168 |
String result = oaf.getTarget(); |
|
166 |
// latitude |
|
167 |
buff += getStringField(metadata.getLatitude().getValue()); |
|
169 | 168 |
|
170 |
buff += getStringField(result); |
|
171 |
// is declared as int!!! |
|
172 |
long diff = DATEDIFF(oaf.getResultProject().getOutcome().getRelMetadata().getEnddate(), oaf.getResultProject().getOutcome().getRelMetadata().getStartdate()); |
|
173 |
if (diff < 0) { |
|
174 |
diff = 0; |
|
175 |
} |
|
176 |
buff += getNumericField(String.valueOf(diff)); |
|
169 |
// longtitude |
|
170 |
buff += getStringField(metadata.getLongitude().getValue()); |
|
177 | 171 |
|
178 |
return buff;
|
|
179 |
}*/
|
|
172 |
// dateofvalidation,
|
|
173 |
buff += getStringField(metadata.getDateofvalidation().getValue());
|
|
180 | 174 |
|
175 |
//Description |
|
176 |
buff += getStringField(metadata.getDescription().getValue()); |
|
181 | 177 |
|
182 |
private String buildDatasource(OafEntity data) { |
|
178 |
//subjects |
|
179 |
String subj = new String(); |
|
180 |
for (StructuredProperty s : metadata.getSubjectsList()) { |
|
181 |
subj += s.getValue() + ';'; |
|
183 | 182 |
|
184 |
String buff = new String();
|
|
183 |
}
|
|
185 | 184 |
|
186 |
Datasource d = data.getDatasource(); |
|
185 |
//subjects |
|
186 |
buff += getStringField(subj); |
|
187 | 187 |
|
188 |
Metadata metadata = d.getMetadata(); |
|
188 |
//Number of items |
|
189 |
buff += getStringField(metadata.getOdnumberofitems().getValue()); |
|
189 | 190 |
|
190 |
//`original Id`
|
|
191 |
String originalId = new String();
|
|
191 |
//Date of number of items
|
|
192 |
buff += getStringField(metadata.getOdnumberofitemsdate().getValue());
|
|
192 | 193 |
|
193 |
for (String oid : data.getOriginalIdList()) { |
|
194 |
originalId += oid + ";"; |
|
195 |
} |
|
194 |
// Policies |
|
195 |
buff += getStringField(metadata.getOdpolicies().getValue()); |
|
196 | 196 |
|
197 |
buff += getStringField(originalId); |
|
197 |
//languages |
|
198 |
String languages = new String(); |
|
199 |
for (StringField lang : metadata.getOdlanguagesList()) { |
|
200 |
languages += lang.getValue() + ";"; |
|
201 |
} |
|
202 |
buff += getStringField(languages); |
|
198 | 203 |
|
199 |
//dateOfCollection |
|
200 |
buff += getStringDateField(data.getDateofcollection()); |
|
201 | 204 |
|
202 |
//Provider Type
|
|
203 |
if (metadata.hasDatasourcetype()) {
|
|
204 |
buff += getStringField(metadata.getDatasourcetype().getClassname().replaceFirst(".*::", ""));
|
|
205 |
} else {
|
|
206 |
buff += getStringField(null);
|
|
207 |
}
|
|
205 |
// Content type
|
|
206 |
String contentType = new String();
|
|
207 |
for (StringField c : metadata.getOdcontenttypesList()) {
|
|
208 |
contentType += c.getValue() + ";";
|
|
209 |
}
|
|
210 |
buff += getStringField(contentType);
|
|
208 | 211 |
|
209 |
//Compatibility |
|
210 |
buff += getStringField(metadata.getOpenairecompatibility().getClassname()); |
|
212 |
//Access info package |
|
213 |
String accessInfo = new String(); |
|
214 |
for (StringField c : metadata.getAccessinfopackageList()) { |
|
215 |
accessInfo += c.getValue() + ";"; |
|
216 |
} |
|
217 |
buff += getStringField(accessInfo); |
|
211 | 218 |
|
212 |
//Official Name
|
|
213 |
buff += getStringField(metadata.getOfficialname().getValue());
|
|
219 |
//Release start date
|
|
220 |
buff += getStringField(metadata.getReleasestartdate().getValue());
|
|
214 | 221 |
|
215 |
// English Name
|
|
216 |
buff += getStringField(metadata.getEnglishname().getValue());
|
|
222 |
//Release end date
|
|
223 |
buff += getStringField(metadata.getReleaseenddate().getValue());
|
|
217 | 224 |
|
218 |
//Website URL
|
|
219 |
buff += getStringField(metadata.getWebsiteurl().getValue());
|
|
225 |
//Mission statement url
|
|
226 |
buff += getStringField(metadata.getMissionstatementurl().getValue());
|
|
220 | 227 |
|
221 |
//LogoURL
|
|
222 |
buff += getStringField(metadata.getLogourl().getValue());
|
|
228 |
//Data provider
|
|
229 |
buff += getStringField(String.valueOf(metadata.getDataprovider().getValue()));
|
|
223 | 230 |
|
224 |
//Email
|
|
225 |
buff += getStringField(metadata.getContactemail().getValue());
|
|
231 |
//Service provider
|
|
232 |
buff += getStringField(String.valueOf(metadata.getServiceprovider().getValue()));
|
|
226 | 233 |
|
227 |
//Namespaceprefix
|
|
228 |
buff += getStringField(metadata.getNamespaceprefix().getValue());
|
|
234 |
//Database access type
|
|
235 |
buff += getStringField(metadata.getDatabaseaccessrestriction().getValue());
|
|
229 | 236 |
|
230 |
// latitude
|
|
231 |
buff += getStringField(metadata.getLatitude().getValue());
|
|
237 |
//Data upload type
|
|
238 |
buff += getStringField(metadata.getDatauploadtype().getValue());
|
|
232 | 239 |
|
233 |
// longtitude
|
|
234 |
buff += getStringField(metadata.getLongitude().getValue());
|
|
240 |
//Data upload restrictions
|
|
241 |
buff += getStringField(metadata.getDatauploadrestriction().getValue());
|
|
235 | 242 |
|
236 |
// dateofvalidation,
|
|
237 |
buff += getStringField(metadata.getDateofvalidation().getValue());
|
|
243 |
//Versioning
|
|
244 |
buff += getStringField(String.valueOf(metadata.getVersioning().getValue()));
|
|
238 | 245 |
|
239 |
//Description
|
|
240 |
buff += getStringField(metadata.getDescription().getValue());
|
|
246 |
//Citation guideline url
|
|
247 |
buff += getStringField(String.valueOf(metadata.getCitationguidelineurl().getValue()));
|
|
241 | 248 |
|
242 |
//subjects |
|
243 |
String subj = new String(); |
|
244 |
for (StructuredProperty s : metadata.getSubjectsList()) { |
|
245 |
subj += s.getValue() + ';'; |
|
249 |
//Quality management kind |
|
250 |
buff += getStringField(String.valueOf(metadata.getQualitymanagementkind().getValue())); |
|
246 | 251 |
|
247 |
} |
|
252 |
//PID systems |
|
253 |
buff += getStringField(metadata.getPidsystems().getValue()); |
|
248 | 254 |
|
249 |
//subjects
|
|
250 |
buff += getStringField(subj);
|
|
255 |
//Certificates
|
|
256 |
buff += getStringField(metadata.getCertificates().getValue());
|
|
251 | 257 |
|
252 |
//Number of items |
|
253 |
buff += getStringField(metadata.getOdnumberofitems().getValue()); |
|
258 |
//Policies |
|
259 |
String policies = new String(); |
|
260 |
for (FieldTypeProtos.KeyValue property : metadata.getPoliciesList()) { |
|
261 |
policies += property.getValue() + ";"; |
|
262 |
} |
|
263 |
buff += getStringField(policies); |
|
254 | 264 |
|
255 |
//Date of number of items
|
|
256 |
buff += getStringField(metadata.getOdnumberofitemsdate().getValue());
|
|
265 |
return buff;
|
|
266 |
}
|
|
257 | 267 |
|
258 |
// Policies |
|
259 |
buff += getStringField(metadata.getOdpolicies().getValue()); |
|
260 | 268 |
|
261 |
//languages |
|
262 |
String languages = new String(); |
|
263 |
for (StringField lang : metadata.getOdlanguagesList()) { |
|
264 |
languages += lang.getValue() + ";"; |
|
265 |
} |
|
266 |
buff += getStringField(languages); |
|
269 |
private String buildOrganization(OafEntity data) { |
|
267 | 270 |
|
271 |
String buff = new String(); |
|
268 | 272 |
|
269 |
// Content type |
|
270 |
String contentType = new String(); |
|
271 |
for (StringField c : metadata.getOdcontenttypesList()) { |
|
272 |
contentType += c.getValue() + ";"; |
|
273 |
} |
|
274 |
buff += getStringField(contentType); |
|
273 |
Organization organization = data.getOrganization(); |
|
274 |
eu.dnetlib.data.proto.OrganizationProtos.Organization.Metadata metadata = organization.getMetadata(); |
|
275 | 275 |
|
276 |
//Access info package |
|
277 |
String accessInfo = new String(); |
|
278 |
for (StringField c : metadata.getAccessinfopackageList()) { |
|
279 |
accessInfo += c.getValue() + ";"; |
|
280 |
} |
|
281 |
buff += getStringField(accessInfo); |
|
276 |
//`original Id` |
|
277 |
String originalId = new String(); |
|
278 |
for (String oid : data.getOriginalIdList()) { |
|
279 |
originalId += oid + ";"; |
|
280 |
} |
|
282 | 281 |
|
283 |
//Release start date |
|
284 |
buff += getStringField(metadata.getReleasestartdate().getValue()); |
|
285 | 282 |
|
286 |
//Release end date |
|
287 |
buff += getStringField(metadata.getReleaseenddate().getValue()); |
|
283 |
buff += getStringField(originalId); |
|
284 |
//dateOfCollection |
|
285 |
buff += getStringDateField(data.getDateofcollection()); |
|
288 | 286 |
|
289 |
//Mission statement url |
|
290 |
buff += getStringField(metadata.getMissionstatementurl().getValue()); |
|
291 | 287 |
|
292 |
//Data provider |
|
293 |
buff += getStringField(String.valueOf(metadata.getDataprovider().getValue())); |
|
288 |
//getLegalshortname |
|
289 |
buff += getStringField(metadata.getLegalshortname().getValue()); |
|
290 |
// `name`, |
|
291 |
buff += getStringField(metadata.getLegalname().getValue()); |
|
292 |
//website URL |
|
293 |
buff += getStringField(metadata.getWebsiteurl().getValue()); |
|
294 |
//logourl |
Also available in: Unified diff