|
1
|
package eu.dnetlib.data.claims.migration;
|
|
2
|
|
|
3
|
import eu.dnetlib.data.claims.migration.entity.*;
|
|
4
|
import eu.dnetlib.data.claims.migration.handler.*;
|
|
5
|
import eu.dnetlib.data.claims.migration.parser.OafParser;
|
|
6
|
import eu.dnetlib.data.claimsDemo.ClaimUtils;
|
|
7
|
import eu.dnetlib.data.claimsDemo.SQLStoreException;
|
|
8
|
import org.apache.log4j.BasicConfigurator;
|
|
9
|
import org.apache.log4j.Level;
|
|
10
|
import org.apache.log4j.Logger;
|
|
11
|
import org.springframework.context.ApplicationContext;
|
|
12
|
import org.springframework.context.support.ClassPathXmlApplicationContext;
|
|
13
|
import org.springframework.context.support.PropertySourcesPlaceholderConfigurer;
|
|
14
|
import org.xml.sax.SAXException;
|
|
15
|
|
|
16
|
import javax.xml.parsers.ParserConfigurationException;
|
|
17
|
import javax.xml.transform.TransformerException;
|
|
18
|
import javax.xml.xpath.XPathExpressionException;
|
|
19
|
import java.io.*;
|
|
20
|
import java.math.BigInteger;
|
|
21
|
import java.security.MessageDigest;
|
|
22
|
import java.sql.Timestamp;
|
|
23
|
import java.util.ArrayList;
|
|
24
|
import java.util.Date;
|
|
25
|
import java.util.List;
|
|
26
|
|
|
27
|
/**
|
|
28
|
* Created by kiatrop on 4/2/2016.
|
|
29
|
*/
|
|
30
|
|
|
31
|
public class Migration {
|
|
32
|
private static final Logger logger = Logger.getLogger(Migration.class);
|
|
33
|
|
|
34
|
RelationHandler relationHandler = null;
|
|
35
|
ContextRelationHandler contextRelationHandler = null;
|
|
36
|
ProjectHandler projectHandler = null;
|
|
37
|
DMFContextHandler dmfContextHandler = null;
|
|
38
|
DMFResultHandler dmfResultHandler = null;
|
|
39
|
IndexResultHandler indexResultHandler = null;
|
|
40
|
ExternalRecordHandler externalRecordHandler = null;
|
|
41
|
ResultHandler resultHandler = null;
|
|
42
|
|
|
43
|
ClaimHandler claimHandler = null;
|
|
44
|
ClaimValidation claimValidation = new ClaimValidation();
|
|
45
|
|
|
46
|
|
|
47
|
public Migration() {
|
|
48
|
ApplicationContext context = new ClassPathXmlApplicationContext("eu/dnetlib/data/claims/migration/springContext-claimsDemo.xml");
|
|
49
|
relationHandler = context.getBean(RelationHandler.class);
|
|
50
|
contextRelationHandler = context.getBean(ContextRelationHandler.class);
|
|
51
|
projectHandler = context.getBean(ProjectHandler.class);
|
|
52
|
dmfContextHandler = context.getBean(DMFContextHandler.class);
|
|
53
|
dmfResultHandler = context.getBean(DMFResultHandler.class);
|
|
54
|
externalRecordHandler = context.getBean(ExternalRecordHandler.class);
|
|
55
|
indexResultHandler = context.getBean(IndexResultHandler.class);
|
|
56
|
resultHandler = context.getBean(ResultHandler.class);
|
|
57
|
claimHandler = context.getBean(ClaimHandler.class);
|
|
58
|
claimValidation = context.getBean(ClaimValidation.class);
|
|
59
|
|
|
60
|
}
|
|
61
|
|
|
62
|
|
|
63
|
public List<Claim> createRelationsClaims() throws Exception, SQLStoreException {
|
|
64
|
|
|
65
|
|
|
66
|
List<Relation> relations = relationHandler.fetchAllRelations();
|
|
67
|
List<Claim> claims = new ArrayList<Claim>();
|
|
68
|
for (Relation relation : relations) {
|
|
69
|
|
|
70
|
Claim claim= new Claim();
|
|
71
|
|
|
72
|
claim.setUserMail(relation.getClaimedBy());
|
|
73
|
claim.setDate(relation.getClaimDate());
|
|
74
|
claim.setId(relation.getClaimId());
|
|
75
|
System.out.println("Claim id:" + claim.getId());
|
|
76
|
OpenaireEntity source = null;
|
|
77
|
//first work with the Source
|
|
78
|
if (relation.getSourceType().equals(ClaimUtils.PROJECT)) {
|
|
79
|
source= buildProject(relation.getSourceId(),relation.getClaimId());
|
|
80
|
claim.setSourceType(ClaimUtils.PROJECT);
|
|
81
|
} else if (relation.getSourceType().equals(ClaimUtils.CONTEXT)) {
|
|
82
|
Context context = dmfContextHandler.fetchContextByIdFromDmf(relation.getSourceId(), false);
|
|
83
|
source = context;
|
|
84
|
claim.setSourceType(ClaimUtils.CONTEXT);
|
|
85
|
} else if (relation.getSourceType().equals(ClaimUtils.DATASET) || relation.getSourceType().equals(ClaimUtils.PUBLICATION)) {
|
|
86
|
Result result = buildResult(relation.getDmf(), relation.getCollectedFrom(), relation.getSourceId(), relation.getClaimId());
|
|
87
|
source = result;
|
|
88
|
claim.setSourceType(result.getResultType());
|
|
89
|
}
|
|
90
|
claim.setSource(source);
|
|
91
|
|
|
92
|
Result target = null;
|
|
93
|
if (relation.getTargetType().equals(ClaimUtils.DATASET) || relation.getTargetType().equals(ClaimUtils.PUBLICATION)) {
|
|
94
|
target = buildResult(relation.getDmf(), relation.getCollectedFrom(), relation.getTargetId(), relation.getClaimId());
|
|
95
|
|
|
96
|
}
|
|
97
|
if( target != null) {
|
|
98
|
claim.setTarget(target);
|
|
99
|
claim.setTargetType(target.getResultType());
|
|
100
|
}
|
|
101
|
if(claimValidation.validateClaim(claim)){ // is valid
|
|
102
|
// System.out.println(claim);
|
|
103
|
claims.add(claim);
|
|
104
|
}
|
|
105
|
|
|
106
|
}
|
|
107
|
return claims;
|
|
108
|
|
|
109
|
}
|
|
110
|
public void createAndSaveRelationsClaims(boolean save) throws Exception, SQLStoreException {
|
|
111
|
/*for stats*/
|
|
112
|
List<Claim> claimsOp= new ArrayList<Claim>();
|
|
113
|
List<Claim> claimsCr= new ArrayList<Claim>();
|
|
114
|
List<Claim> claimsDat= new ArrayList<Claim>();
|
|
115
|
List<Claim> claimsOrc= new ArrayList<Claim>();
|
|
116
|
Integer targetOp=0;
|
|
117
|
Integer targetCr=0;
|
|
118
|
Integer targetDat=0;
|
|
119
|
Integer targetOrc=0;
|
|
120
|
Integer sourceRes=0;
|
|
121
|
Integer sourceOp=0;
|
|
122
|
Integer sourceCr=0;
|
|
123
|
Integer sourceDat=0;
|
|
124
|
Integer sourceOrc=0;
|
|
125
|
Integer projects=0;
|
|
126
|
Integer contexts=0;
|
|
127
|
Integer totalClaims=0;
|
|
128
|
|
|
129
|
|
|
130
|
List<Relation> relations = relationHandler.fetchAllRelations();
|
|
131
|
List<Claim> claims = new ArrayList<Claim>();
|
|
132
|
logger.info("Relation size: "+relations.size());
|
|
133
|
for (Relation relation : relations) {
|
|
134
|
try {
|
|
135
|
Claim claim = new Claim();
|
|
136
|
|
|
137
|
claim.setUserMail(relation.getClaimedBy());
|
|
138
|
claim.setDate(relation.getClaimDate());
|
|
139
|
claim.setId(relation.getClaimId());
|
|
140
|
logger.info("Claim id:" + claim.getId()+" "+claim.getDate());
|
|
141
|
OpenaireEntity source = null;
|
|
142
|
//first work with the Source
|
|
143
|
if (relation.getSourceType().equals(ClaimUtils.PROJECT)) {
|
|
144
|
source = buildProject(relation.getSourceId(), relation.getClaimId());
|
|
145
|
claim.setSourceType(ClaimUtils.PROJECT);
|
|
146
|
} else if (relation.getSourceType().equals(ClaimUtils.CONTEXT)) {
|
|
147
|
Context context = dmfContextHandler.fetchContextByIdFromDmf(relation.getSourceId(), false);
|
|
148
|
source = context;
|
|
149
|
claim.setSourceType(ClaimUtils.CONTEXT);
|
|
150
|
} else if (relation.getSourceType().equals(ClaimUtils.DATASET) || relation.getSourceType().equals(ClaimUtils.PUBLICATION)) {
|
|
151
|
Result result = buildResult(relation.getDmf(), relation.getCollectedFrom(), relation.getSourceId(), relation.getClaimId());
|
|
152
|
source = result;
|
|
153
|
claim.setSourceType(result.getResultType());
|
|
154
|
}
|
|
155
|
|
|
156
|
claim.setSource(source);
|
|
157
|
|
|
158
|
Result target = null;
|
|
159
|
if (relation.getTargetType().equals(ClaimUtils.DATASET) || relation.getTargetType().equals(ClaimUtils.PUBLICATION)) {
|
|
160
|
target = buildResult(relation.getDmf(), relation.getCollectedFrom(), relation.getTargetId(), relation.getClaimId());
|
|
161
|
|
|
162
|
}
|
|
163
|
if (target != null) {
|
|
164
|
claim.setTarget(target);
|
|
165
|
claim.setTargetType(target.getResultType());
|
|
166
|
}
|
|
167
|
|
|
168
|
if (claimValidation.validateClaim(claim)) { // is valid
|
|
169
|
if(save) {
|
|
170
|
claim = claimHandler.exportMedatataForClaim(claim);
|
|
171
|
claimHandler.saveClaim(claim);
|
|
172
|
}
|
|
173
|
/*for stats*/
|
|
174
|
if(((Result)claim.getTarget()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_CROSSREF)){
|
|
175
|
targetCr++;
|
|
176
|
}else if(((Result)claim.getTarget()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_DATACITE)){
|
|
177
|
targetDat++;
|
|
178
|
}else if(((Result)claim.getTarget()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_ORCID)){
|
|
179
|
targetOrc++;
|
|
180
|
}else {
|
|
181
|
targetOp++;
|
|
182
|
}
|
|
183
|
if(claim.getSourceType().equals(ClaimUtils.PUBLICATION)||claim.getSourceType().equals(ClaimUtils.DATASET)){
|
|
184
|
sourceRes++;
|
|
185
|
if(((Result)claim.getSource()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_CROSSREF)){
|
|
186
|
sourceCr++;
|
|
187
|
}else if(((Result)claim.getSource()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_DATACITE)){
|
|
188
|
sourceDat++;
|
|
189
|
}else if(((Result)claim.getSource()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_ORCID)){
|
|
190
|
sourceOrc++;
|
|
191
|
}else {
|
|
192
|
sourceOp++;
|
|
193
|
}
|
|
194
|
}else if(claim.getSourceType().equals(ClaimUtils.PROJECT)) {
|
|
195
|
projects++;
|
|
196
|
}else if(claim.getSourceType().equals(ClaimUtils.CONTEXT)){
|
|
197
|
contexts++;
|
|
198
|
}
|
|
199
|
|
|
200
|
}else{
|
|
201
|
claimHandler.saveOrphanClaimId(claim.getId());
|
|
202
|
}
|
|
203
|
}catch (Exception e){
|
|
204
|
e.printStackTrace();
|
|
205
|
logger.error("Exception: "+e.getMessage());
|
|
206
|
|
|
207
|
}
|
|
208
|
}
|
|
209
|
|
|
210
|
/*for stats*/
|
|
211
|
System.out.println("\n\nTargets:\n\nCrossref : "+targetCr+ " DataCite :"+targetDat+" Orcid : "+targetOrc+" OpenAire : "+targetOp);
|
|
212
|
System.out.println("\n\nSources:\n\nAll : "+sourceRes+" Crossref : "+sourceCr+ " DataCite :"+sourceDat+" Orcid : "+sourceOrc+" OpenAire : "+sourceOp);
|
|
213
|
System.out.println("\n\nTotalClaims: "+totalClaims+" Projects: "+projects+ " Contexts :"+contexts);
|
|
214
|
|
|
215
|
|
|
216
|
}
|
|
217
|
public List<Claim> createContextRelationClaims() throws Exception, SQLStoreException {
|
|
218
|
|
|
219
|
|
|
220
|
List<ContextRelation> relations = contextRelationHandler.fetchAllConceptRelations();
|
|
221
|
List<Claim> claims = new ArrayList<Claim>();
|
|
222
|
for (ContextRelation relation : relations) {
|
|
223
|
|
|
224
|
Claim claim= new Claim();
|
|
225
|
claim.setId(relation.getClaimId());
|
|
226
|
claim.setUserMail(relation.getClaimedBy());
|
|
227
|
claim.setDate(relation.getClaimDate());
|
|
228
|
System.out.println("Claim id:" + claim.getId());
|
|
229
|
|
|
230
|
//first work with the Source
|
|
231
|
Context context = dmfContextHandler.extractContextFromDMF(relation.getContextDmf(), false);
|
|
232
|
claim.setSourceType(ClaimUtils.CONTEXT);
|
|
233
|
claim.setSource(context);
|
|
234
|
|
|
235
|
Result target = buildResult(relation.getResultDmf(), relation.getCollectedFrom(), relation.getResultId(), relation.getClaimId());
|
|
236
|
if( target != null){
|
|
237
|
claim.setTarget(target);
|
|
238
|
claim.setTargetType(target.getResultType());
|
|
239
|
}
|
|
240
|
if(claimValidation.validateClaim(claim)){ // is valid
|
|
241
|
claims.add(claim);
|
|
242
|
}
|
|
243
|
|
|
244
|
}
|
|
245
|
return claims;
|
|
246
|
|
|
247
|
|
|
248
|
}
|
|
249
|
public void createAndSaveContextRelationClaims(Boolean save) throws Exception, SQLStoreException {
|
|
250
|
/*for stats*/
|
|
251
|
List<Claim> claimsOp= new ArrayList<Claim>();
|
|
252
|
List<Claim> claimsCr= new ArrayList<Claim>();
|
|
253
|
List<Claim> claimsDat= new ArrayList<Claim>();
|
|
254
|
List<Claim> claimsOrc= new ArrayList<Claim>();
|
|
255
|
Integer targetOp=0;
|
|
256
|
Integer targetCr=0;
|
|
257
|
Integer targetDat=0;
|
|
258
|
Integer targetOrc=0;
|
|
259
|
Integer sourceRes=0;
|
|
260
|
Integer sourceOp=0;
|
|
261
|
Integer sourceCr=0;
|
|
262
|
Integer sourceDat=0;
|
|
263
|
Integer sourceOrc=0;
|
|
264
|
Integer projects=0;
|
|
265
|
Integer contexts=0;
|
|
266
|
Integer totalClaims=0;
|
|
267
|
|
|
268
|
List<ContextRelation> relations = contextRelationHandler.fetchAllConceptRelations();
|
|
269
|
// List<Claim> claims = new ArrayList<Claim>();
|
|
270
|
for (ContextRelation relation : relations) {
|
|
271
|
|
|
272
|
Claim claim= new Claim();
|
|
273
|
claim.setId(relation.getClaimId());
|
|
274
|
claim.setUserMail(relation.getClaimedBy());
|
|
275
|
claim.setDate(relation.getClaimDate());
|
|
276
|
logger.info("Claim id:" + claim.getId());
|
|
277
|
|
|
278
|
//first work with the Source
|
|
279
|
Context context = dmfContextHandler.extractContextFromDMF(relation.getContextDmf(),false);
|
|
280
|
claim.setSourceType(ClaimUtils.CONTEXT);
|
|
281
|
claim.setSource(context);
|
|
282
|
|
|
283
|
Result target = buildResult(relation.getResultDmf(), relation.getCollectedFrom(), relation.getResultId(), relation.getClaimId());
|
|
284
|
if( target != null){
|
|
285
|
claim.setTarget(target);
|
|
286
|
claim.setTargetType(target.getResultType());
|
|
287
|
}
|
|
288
|
if(claimValidation.validateClaim(claim)){ // is valid
|
|
289
|
if(save){
|
|
290
|
claim = claimHandler.exportMedatataForClaim(claim);
|
|
291
|
claimHandler.saveClaim(claim);
|
|
292
|
}
|
|
293
|
|
|
294
|
/*for stats*/
|
|
295
|
if(((Result)claim.getTarget()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_CROSSREF)){
|
|
296
|
targetCr++;
|
|
297
|
}else if(((Result)claim.getTarget()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_DATACITE)){
|
|
298
|
targetDat++;
|
|
299
|
}else if(((Result)claim.getTarget()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_ORCID)){
|
|
300
|
targetOrc++;
|
|
301
|
}else {
|
|
302
|
targetOp++;
|
|
303
|
}
|
|
304
|
if(claim.getSourceType().equals(ClaimUtils.PUBLICATION)||claim.getSourceType().equals(ClaimUtils.DATASET)){
|
|
305
|
sourceRes++;
|
|
306
|
if(((Result)claim.getSource()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_CROSSREF)){
|
|
307
|
sourceCr++;
|
|
308
|
}else if(((Result)claim.getSource()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_DATACITE)){
|
|
309
|
sourceDat++;
|
|
310
|
}else if(((Result)claim.getSource()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_ORCID)){
|
|
311
|
sourceOrc++;
|
|
312
|
}else {
|
|
313
|
sourceOp++;
|
|
314
|
}
|
|
315
|
}else if(claim.getSourceType().equals(ClaimUtils.PROJECT)) {
|
|
316
|
projects++;
|
|
317
|
}else if(claim.getSourceType().equals(ClaimUtils.CONTEXT)){
|
|
318
|
contexts++;
|
|
319
|
}
|
|
320
|
}else{
|
|
321
|
if(save){
|
|
322
|
claimHandler.saveOrphanClaimId(claim.getId());
|
|
323
|
}
|
|
324
|
|
|
325
|
}
|
|
326
|
|
|
327
|
|
|
328
|
}
|
|
329
|
|
|
330
|
/*for stats*/
|
|
331
|
System.out.println("\n\nTargets:\n\nCrossref : "+targetCr+ " DataCite :"+targetDat+" Orcid : "+targetOrc+" OpenAire : "+targetOp);
|
|
332
|
System.out.println("\n\nSources:\n\nAll : "+sourceRes+" Crossref : "+sourceCr+ " DataCite :"+sourceDat+" Orcid : "+sourceOrc+" OpenAire : "+sourceOp);
|
|
333
|
System.out.println("\n\nTotalClaims: "+totalClaims+" Projects: "+projects+ " Contexts :"+contexts);
|
|
334
|
|
|
335
|
}
|
|
336
|
|
|
337
|
/**
|
|
338
|
*Returns a project from Openaire.
|
|
339
|
* In case the project id has a "welcometrust" prefix changes it to "wt__________", before search for it in the index.
|
|
340
|
*
|
|
341
|
* @param projectId
|
|
342
|
* @param claimId
|
|
343
|
* @return Project
|
|
344
|
* @throws Exception
|
|
345
|
*/
|
|
346
|
public Project buildProject(String projectId, String claimId) throws Exception {
|
|
347
|
String id = projectId;
|
|
348
|
if(projectId != null && projectId.contains("welcometrust")){
|
|
349
|
id=projectId.replace("welcometrust","wt__________");
|
|
350
|
}
|
|
351
|
Project project = projectHandler.fetchProjectByID(id, false);
|
|
352
|
if (project == null) {
|
|
353
|
project = projectHandler.fetchProjectByID(id, true);
|
|
354
|
}
|
|
355
|
if (project == null) {
|
|
356
|
try{
|
|
357
|
PrintWriter out = new PrintWriter(new BufferedWriter(new FileWriter(claimValidation.getPathToSaveReport()+"missing_projects.txt", true)));
|
|
358
|
Date date= new java.util.Date();
|
|
359
|
out.println(new Timestamp(date.getTime())+" ProjectId: "+id+ " ClaimId: "+claimId);
|
|
360
|
out.close();
|
|
361
|
}catch (IOException e) {
|
|
362
|
logger.error("Couldn't write to file " + "missing_projects.txt",e);
|
|
363
|
}
|
|
364
|
}
|
|
365
|
return project;
|
|
366
|
|
|
367
|
}
|
|
368
|
/**
|
|
369
|
* *If there is a dmf, the result is extracted from the external sources (crossRef, Datacite, Orcid)
|
|
370
|
* Otherwise Result is extracted from Openaire
|
|
371
|
* @param dmf
|
|
372
|
* @param collectedFrom
|
|
373
|
* @param resultId
|
|
374
|
* @param claimId
|
|
375
|
* @return Result or null
|
|
376
|
*/
|
|
377
|
private Result buildResult(String dmf, String collectedFrom, String resultId, String claimId){
|
|
378
|
Result result = null;
|
|
379
|
String external_id= null; // testing
|
|
380
|
if((dmf !=null && collectedFrom != null)&& !collectedFrom.equals(ClaimUtils.COLLECTED_FROM_OPENAIRE)){
|
|
381
|
|
|
382
|
if (collectedFrom.equals(ClaimUtils.COLLECTED_FROM_CROSSREF)) {
|
|
383
|
try {
|
|
384
|
String doi= dmfResultHandler.fetchDoiByDMF(dmf);
|
|
385
|
external_id=doi;
|
|
386
|
if(doi != null) {
|
|
387
|
result = externalRecordHandler.fetchResultfromCrossref(doi);
|
|
388
|
}
|
|
389
|
} catch (Exception e) {
|
|
390
|
logger.error("Error fetching result from Crossref",e);
|
|
391
|
}
|
|
392
|
|
|
393
|
} else if (collectedFrom.equals(ClaimUtils.COLLECTED_FROM_ORCID)) {
|
|
394
|
try {
|
|
395
|
String orcidwork= dmfResultHandler.fetchOrcidWorkByDMF(dmf);
|
|
396
|
external_id=orcidwork;
|
|
397
|
if(orcidwork != null) {
|
|
398
|
result = externalRecordHandler.fetchResultfromOrcid(orcidwork);
|
|
399
|
}
|
|
400
|
} catch (Exception e) {
|
|
401
|
logger.error("Error fetching result from Orcid",e);
|
|
402
|
}
|
|
403
|
} else if (collectedFrom.equals(ClaimUtils.COLLECTED_FROM_DATACITE)) {
|
|
404
|
try {
|
|
405
|
String doi= dmfResultHandler.fetchDoiByDMF(dmf);
|
|
406
|
external_id=doi;
|
|
407
|
if(doi != null) {
|
|
408
|
result = externalRecordHandler.fetchResultfromDatacite(doi);
|
|
409
|
}
|
|
410
|
} catch (Exception e) {
|
|
411
|
logger.error("Error fetching result from Datacite",e);
|
|
412
|
}
|
|
413
|
}
|
|
414
|
if(result == null){
|
|
415
|
//report missing external results
|
|
416
|
try{
|
|
417
|
PrintWriter out = new PrintWriter(new BufferedWriter(new FileWriter(claimValidation.getPathToSaveReport()+"missing_results_external_sources.txt", true)));
|
|
418
|
Date date= new java.util.Date();
|
|
419
|
out.println(new Timestamp(date.getTime())+" OpenaireId: "+resultId+ " collectedFrom: "+collectedFrom +" externalId: "+external_id+" claimId: "+claimId);
|
|
420
|
out.close();
|
|
421
|
}catch (IOException e) {
|
|
422
|
logger.error("Couldn't write to file " + "missing_results_external_sources.txt",e);
|
|
423
|
}
|
|
424
|
//give a second chance - search index
|
|
425
|
result = buildOpenaireResult(resultId,claimId);
|
|
426
|
if(result!=null) {
|
|
427
|
try {
|
|
428
|
PrintWriter out = new PrintWriter(new BufferedWriter(new FileWriter(claimValidation.getPathToSaveReport()+"external_results_found_inOpenaire.txt", true)));
|
|
429
|
Date date = new java.util.Date();
|
|
430
|
out.println(new Timestamp(date.getTime()) + " OpenaireId: " + resultId + " collectedFrom: " + collectedFrom + " externalId: " + external_id + " claimId: " + claimId);
|
|
431
|
out.close();
|
|
432
|
} catch (IOException e) {
|
|
433
|
logger.error("Couldn't write to file " + "external_results_found_inOpenaire.txt",e);
|
|
434
|
}
|
|
435
|
}
|
|
436
|
|
|
437
|
}
|
|
438
|
if(result!=null){
|
|
439
|
// if result found enriched with access rights/ embargo date from DMF
|
|
440
|
try {
|
|
441
|
result.setAccessRights(dmfResultHandler.fetchAccessRights(dmf));
|
|
442
|
result.setEmbargoEndDate(dmfResultHandler.fetchEmbargoEndDateByDMF(dmf));
|
|
443
|
} catch (Exception e) {
|
|
444
|
logger.error("Error fetching Access or Embargo end date from DMF",e);
|
|
445
|
}
|
|
446
|
}
|
|
447
|
} else { //If dmf/ collecteFrom is null or is collected from openaire search in the index
|
|
448
|
result = buildOpenaireResult(resultId,claimId);
|
|
449
|
}
|
|
450
|
|
|
451
|
|
|
452
|
return result;
|
|
453
|
}
|
|
454
|
|
|
455
|
/**
|
|
456
|
*
|
|
457
|
* @param resultId
|
|
458
|
* @param claimId
|
|
459
|
* @return target Result from Openaire or null
|
|
460
|
*/
|
|
461
|
public Result buildOpenaireResult(String resultId, String claimId ){
|
|
462
|
|
|
463
|
Result result = null;
|
|
464
|
try {
|
|
465
|
result = indexResultHandler.fetchPublicationById(resultId, false);
|
|
466
|
} catch (Exception e) {
|
|
467
|
logger.error("Error fetching result from Openaire",e);
|
|
468
|
System.out.println("Error fetching result from Openaire");
|
|
469
|
}
|
|
470
|
|
|
471
|
//If it is not found in the index
|
|
472
|
if (result == null) {
|
|
473
|
//give a second chance as dedup
|
|
474
|
// try {
|
|
475
|
// result = indexResultHandler.fetchDedupResultById(resultId);
|
|
476
|
// } catch (Exception e) {
|
|
477
|
// logger.error("Error fetching dedup Result",e);
|
|
478
|
// }
|
|
479
|
//give a second chance in production
|
|
480
|
try {
|
|
481
|
result = indexResultHandler.fetchPublicationById(resultId, true);
|
|
482
|
} catch (Exception e) {
|
|
483
|
logger.error("Error fetching result from Openaire",e);
|
|
484
|
System.out.println("Error fetching result from Openaire");
|
|
485
|
}
|
|
486
|
if (result == null) {
|
|
487
|
try {
|
|
488
|
result = indexResultHandler.fetchDatasetById(resultId, false);
|
|
489
|
} catch (Exception e) {
|
|
490
|
logger.error("Error fetching result from Openaire",e);
|
|
491
|
System.out.println("Error fetching result from Openaire");
|
|
492
|
}
|
|
493
|
}
|
|
494
|
if (result == null) {
|
|
495
|
try {
|
|
496
|
result = indexResultHandler.fetchDatasetById(resultId, true);
|
|
497
|
} catch (Exception e) {
|
|
498
|
logger.error("Error fetching result from Openaire",e);
|
|
499
|
System.out.println("Error fetching result from Openaire");
|
|
500
|
}
|
|
501
|
}
|
|
502
|
if (result == null) {
|
|
503
|
try {
|
|
504
|
result = indexResultHandler.fetchDedupById(resultId, false);
|
|
505
|
} catch (Exception e) {
|
|
506
|
logger.error("Error fetching result from Openaire",e);
|
|
507
|
System.out.println("Error fetching result from Openaire");
|
|
508
|
}
|
|
509
|
}
|
|
510
|
if (result == null) {
|
|
511
|
try {
|
|
512
|
result = indexResultHandler.fetchDedupById(resultId, true);
|
|
513
|
} catch (Exception e) {
|
|
514
|
logger.error("Error fetching result from Openaire",e);
|
|
515
|
System.out.println("Error fetching result from Openaire");
|
|
516
|
}
|
|
517
|
}
|
|
518
|
if (result != null) {
|
|
519
|
try {
|
|
520
|
PrintWriter out = new PrintWriter(new BufferedWriter(new FileWriter(claimValidation.getPathToSaveReport()+"dedup_results_found.txt", true)));
|
|
521
|
Date date = new java.util.Date();
|
|
522
|
out.println(new Timestamp(date.getTime()) + " OpenaireId: " + resultId + " claimId: " + claimId);
|
|
523
|
out.close();
|
|
524
|
} catch (IOException e) {
|
|
525
|
logger.error("Couldn't write to file " + "external_results_found_inOpenaire.txt",e);
|
|
526
|
}
|
|
527
|
}else {
|
|
528
|
//report it as missing
|
|
529
|
try {
|
|
530
|
PrintWriter out = new PrintWriter(new BufferedWriter(new FileWriter(claimValidation.getPathToSaveReport()+"missing_results.txt", true)));
|
|
531
|
Date date = new java.util.Date();
|
|
532
|
out.println(new Timestamp(date.getTime()) + " OpenaireId: " + resultId + " ClaimId: " + claimId);
|
|
533
|
out.close();
|
|
534
|
} catch (IOException e) {
|
|
535
|
logger.error("Couldn't write to file " + "missing_results.txt",e);
|
|
536
|
}
|
|
537
|
}
|
|
538
|
|
|
539
|
}else{
|
|
540
|
//FOUND report results that came from datacite but are publications
|
|
541
|
//TODO deal with them as externals?
|
|
542
|
|
|
543
|
if(result.getProvenanceaction() != null && result.getProvenanceaction().equals("user:claim:datacite")&&result.getOai()!=null){
|
|
544
|
try{
|
|
545
|
PrintWriter out = new PrintWriter(new BufferedWriter(new FileWriter(claimValidation.getPathToSaveReport()+"datacite_claim_results.txt", true)));
|
|
546
|
Date date= new java.util.Date();
|
|
547
|
out.println(new Timestamp(date.getTime())+" openaireId:"+resultId+" aoi: " +result.getOai() +" provenanceactions: "+result.getProvenanceaction()+" type: "+result.getResultType()+ " ClaimId: "+claimId);
|
|
548
|
out.close();
|
|
549
|
try {
|
|
550
|
result= externalRecordHandler.fetchResultfromDatacite(result.getOai());
|
|
551
|
result.setOpenaireId(resultId);
|
|
552
|
}catch(Exception e){
|
|
553
|
logger.error("Couldn't get Result from Datacite "+result.getOai(),e);
|
|
554
|
}
|
|
555
|
}catch (IOException e) {
|
|
556
|
logger.error("Couldn't write to file " + "datacite_claim_results.txt",e);
|
|
557
|
}
|
|
558
|
|
|
559
|
}
|
|
560
|
}
|
|
561
|
|
|
562
|
return result;
|
|
563
|
}
|
|
564
|
|
|
565
|
|
|
566
|
/**
|
|
567
|
* For each claim of the list
|
|
568
|
* if there are results in the relation exports their metadaa file
|
|
569
|
* Save the claim in DB
|
|
570
|
* @param claims
|
|
571
|
* @throws Exception
|
|
572
|
*/
|
|
573
|
private void saveClaims(List<Claim> claims) throws Exception, SQLStoreException {
|
|
574
|
for(Claim claim : claims){
|
|
575
|
/* if(claim.getTargetType().equals(ClaimUtils.DATASET)||claim.getTargetType().equals(ClaimUtils.PUBLICATION)){
|
|
576
|
String path = resultHandler.exportMetadataFileForResult((Result)claim.getTarget());
|
|
577
|
((Result) claim.getTarget()).setRecordPath(path);
|
|
578
|
}
|
|
579
|
if(claim.getSourceType().equals(ClaimUtils.DATASET)||claim.getSourceType().equals(ClaimUtils.PUBLICATION)){
|
|
580
|
String path = resultHandler.exportMetadataFileForResult((Result)claim.getSource());
|
|
581
|
((Result) claim.getSource()).setRecordPath(path);
|
|
582
|
}*/
|
|
583
|
claim = claimHandler.exportMedatataForClaim(claim);
|
|
584
|
claimHandler.saveClaim(claim);
|
|
585
|
}
|
|
586
|
|
|
587
|
}
|
|
588
|
|
|
589
|
|
|
590
|
|
|
591
|
public static void main(String[] args) throws IOException {
|
|
592
|
BasicConfigurator.configure();
|
|
593
|
logger.setLevel(Level.DEBUG);
|
|
594
|
Migration migration = new Migration();
|
|
595
|
try {
|
|
596
|
// migration.claimHandler.getQueryGenerator().setMigrationTable("claims_view");
|
|
597
|
// migration.createAndSaveRelationsClaims(true);
|
|
598
|
// migration.createAndSaveContextRelationClaims(true);
|
|
599
|
|
|
600
|
migration.claimHandler.getQueryGenerator().setMigrationTable("claims_clean");
|
|
601
|
migration.createAndSaveRelationsClaims(true);
|
|
602
|
migration.createAndSaveContextRelationClaims(true);
|
|
603
|
|
|
604
|
|
|
605
|
} catch (Exception e) {
|
|
606
|
logger.error("Error in migration",e);
|
|
607
|
} catch (SQLStoreException e) {
|
|
608
|
e.printStackTrace();
|
|
609
|
}
|
|
610
|
|
|
611
|
}
|
|
612
|
public void printStatistics(List<Claim> claims){
|
|
613
|
|
|
614
|
//testing
|
|
615
|
List<Claim> claimsOp= new ArrayList<Claim>();
|
|
616
|
List<Claim> claimsCr= new ArrayList<Claim>();
|
|
617
|
List<Claim> claimsDat= new ArrayList<Claim>();
|
|
618
|
List<Claim> claimsOrc= new ArrayList<Claim>();
|
|
619
|
Integer targetOp=0;
|
|
620
|
Integer targetCr=0;
|
|
621
|
Integer targetDat=0;
|
|
622
|
Integer targetOrc=0;
|
|
623
|
Integer sourceRes=0;
|
|
624
|
Integer sourceOp=0;
|
|
625
|
Integer sourceCr=0;
|
|
626
|
Integer sourceDat=0;
|
|
627
|
Integer sourceOrc=0;
|
|
628
|
Integer projects=0;
|
|
629
|
Integer contexts=0;
|
|
630
|
Integer totalClaims=0;
|
|
631
|
for(Claim claim: claims){
|
|
632
|
// System.out.println(claim.toString());
|
|
633
|
if(((Result)claim.getTarget()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_CROSSREF)){
|
|
634
|
targetCr++;
|
|
635
|
}else if(((Result)claim.getTarget()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_DATACITE)){
|
|
636
|
targetDat++;
|
|
637
|
}else if(((Result)claim.getTarget()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_ORCID)){
|
|
638
|
targetOrc++;
|
|
639
|
}else {
|
|
640
|
targetOp++;
|
|
641
|
}
|
|
642
|
if(claim.getSourceType().equals(ClaimUtils.PUBLICATION)||claim.getSourceType().equals(ClaimUtils.DATASET)){
|
|
643
|
sourceRes++;
|
|
644
|
if(((Result)claim.getSource()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_CROSSREF)){
|
|
645
|
sourceCr++;
|
|
646
|
}else if(((Result)claim.getSource()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_DATACITE)){
|
|
647
|
sourceDat++;
|
|
648
|
}else if(((Result)claim.getSource()).getCollectedFrom().equals(ClaimUtils.COLLECTED_FROM_ORCID)){
|
|
649
|
sourceOrc++;
|
|
650
|
}else {
|
|
651
|
sourceOp++;
|
|
652
|
}
|
|
653
|
}else if(claim.getSourceType().equals(ClaimUtils.PROJECT)) {
|
|
654
|
projects++;
|
|
655
|
}else if(claim.getSourceType().equals(ClaimUtils.CONTEXT)){
|
|
656
|
contexts++;
|
|
657
|
}
|
|
658
|
}
|
|
659
|
totalClaims=claims.size();
|
|
660
|
System.out.println("\n\nTargets:\n\nCrossref : "+targetCr+ " DataCite :"+targetDat+" Orcid : "+targetOrc+" OpenAire : "+targetOp);
|
|
661
|
System.out.println("\n\nSources:\n\nAll : "+sourceRes+" Crossref : "+sourceCr+ " DataCite :"+sourceDat+" Orcid : "+sourceOrc+" OpenAire : "+sourceOp);
|
|
662
|
System.out.println("\n\nTotalClaims: "+totalClaims+" Projects: "+projects+ " Contexts :"+contexts);
|
|
663
|
|
|
664
|
}
|
|
665
|
|
|
666
|
|
|
667
|
}
|