D-Net
/
dnet-hadoop
13
4
Fork 10
Code Issues 14 Pull Requests 8 Projects 2 Releases 1 Wiki Activity

fixed dedup record id prefix, set the correct dataInfo in the DedupRecordFactory

This commit is contained in:
Claudio Atzori 2020-04-23 07:50:18 +02:00
parent ade4cb97af
commit 9ddafd46ca
4 changed files with 54 additions and 32 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

53
dhp-workflows/dhp-dedup-openaire/src/main/java/eu/dnetlib/dhp/oa/dedup/DedupRecordFactory.java
View File

@ -25,6 +25,7 @@ public class DedupRecordFactory {
public static <T extends OafEntity> Dataset<T> createDedupRecord(
final SparkSession spark,
final DataInfo dataInfo,
final String mergeRelsInputPath,
final String entitiesInputPath,
final Class<T> clazz) {
@ -67,41 +68,39 @@ public class DedupRecordFactory {
Encoders.STRING())
.mapGroups(
(MapGroupsFunction<String, Tuple2<String, T>, T>)
(key, values) -> entityMerger(key, values, ts, clazz),
(key, values) -> entityMerger(key, values, ts, dataInfo),
Encoders.bean(clazz));
}
private static <T extends OafEntity> T entityMerger(
String id, Iterator<Tuple2<String, T>> entities, long ts, Class<T> clazz) {
try {
T entity = clazz.newInstance();
entity.setId(id);
entity.setDataInfo(new DataInfo());
entity.getDataInfo().setTrust("0.9");
entity.setLastupdatetimestamp(ts);
String id, Iterator<Tuple2<String, T>> entities, long ts, DataInfo dataInfo) {
final Collection<String> dates = Lists.newArrayList();
entities.forEachRemaining(
t -> {
T duplicate = t._2();
entity.mergeFrom(duplicate);
if (ModelSupport.isSubClass(duplicate, Result.class)) {
Result r1 = (Result) duplicate;
Result er = (Result) entity;
er.setAuthor(DedupUtility.mergeAuthor(er.getAuthor(), r1.getAuthor()));
T entity = entities.next()._2();
if (er.getDateofacceptance() != null) {
dates.add(r1.getDateofacceptance().getValue());
}
final Collection<String> dates = Lists.newArrayList();
entities.forEachRemaining(
t -> {
T duplicate = t._2();
entity.mergeFrom(duplicate);
if (ModelSupport.isSubClass(duplicate, Result.class)) {
Result r1 = (Result) duplicate;
Result er = (Result) entity;
er.setAuthor(DedupUtility.mergeAuthor(er.getAuthor(), r1.getAuthor()));
if (r1.getDateofacceptance() != null) {
dates.add(r1.getDateofacceptance().getValue());
}
});
}
});
if (ModelSupport.isSubClass(entity, Result.class)) {
((Result) entity).setDateofacceptance(DatePicker.pick(dates));
}
return entity;
} catch (IllegalAccessException | InstantiationException e) {
throw new RuntimeException(e);
if (ModelSupport.isSubClass(entity, Result.class)) {
((Result) entity).setDateofacceptance(DatePicker.pick(dates));
}
entity.setId(id);
entity.setLastupdatetimestamp(ts);
entity.setDataInfo(dataInfo);
return entity;
}
}

29
dhp-workflows/dhp-dedup-openaire/src/main/java/eu/dnetlib/dhp/oa/dedup/SparkCreateDedupRecord.java
View File

@ -3,7 +3,9 @@ package eu.dnetlib.dhp.oa.dedup;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.schema.common.EntityType;
import eu.dnetlib.dhp.schema.common.ModelSupport;
import eu.dnetlib.dhp.schema.oaf.DataInfo;
import eu.dnetlib.dhp.schema.oaf.OafEntity;
import eu.dnetlib.dhp.schema.oaf.Qualifier;
import eu.dnetlib.dhp.utils.ISLookupClientFactory;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
@ -21,6 +23,10 @@ public class SparkCreateDedupRecord extends AbstractSparkAction {
private static final Logger log = LoggerFactory.getLogger(SparkCreateDedupRecord.class);
public static final String ROOT_TRUST = "0.8";
public static final String PROVENANCE_ACTION_CLASS = "sysimport:dedup";
public static final String PROVENANCE_ACTIONS = "dnet:provenanceActions";
public SparkCreateDedupRecord(ArgumentApplicationParser parser, SparkSession spark) {
super(parser, spark);
}
@ -67,13 +73,30 @@ public class SparkCreateDedupRecord extends AbstractSparkAction {
DedupUtility.createMergeRelPath(workingPath, actionSetId, subEntity);
final String entityPath = DedupUtility.createEntityPath(graphBasePath, subEntity);
Class<OafEntity> clazz = ModelSupport.entityTypes.get(EntityType.valueOf(subEntity));
DedupRecordFactory.createDedupRecord(spark, mergeRelPath, entityPath, clazz)
final Class<OafEntity> clazz =
ModelSupport.entityTypes.get(EntityType.valueOf(subEntity));
final DataInfo dataInfo = getDataInfo(dedupConf);
DedupRecordFactory.createDedupRecord(spark, dataInfo, mergeRelPath, entityPath, clazz)
.write()
.mode(SaveMode.Overwrite)
.option("compression", "gzip")
.json(outputPath);
}
}
private static DataInfo getDataInfo(DedupConfig dedupConf) {
DataInfo info = new DataInfo();
info.setDeletedbyinference(false);
info.setInferred(true);
info.setInvisible(false);
info.setTrust(ROOT_TRUST);
info.setInferenceprovenance(dedupConf.getWf().getConfigurationId());
Qualifier provenance = new Qualifier();
provenance.setClassid(PROVENANCE_ACTION_CLASS);
provenance.setClassname(PROVENANCE_ACTION_CLASS);
provenance.setSchemeid(PROVENANCE_ACTIONS);
provenance.setSchemename(PROVENANCE_ACTIONS);
info.setProvenanceaction(provenance);
return info;
}
}

2
dhp-workflows/dhp-dedup-openaire/src/main/java/eu/dnetlib/dhp/oa/dedup/graph/ConnectedComponent.java
View File

@ -25,7 +25,7 @@ public class ConnectedComponent implements Serializable {
if (docIds.size() > 1) {
final String s = getMin();
String prefix = s.split("\\|")[0];
ccId = prefix + "|dedup_______::" + DedupUtility.md5(s);
ccId = prefix + "|dedup_wf_001::" + DedupUtility.md5(s);
return ccId;
} else {
return docIds.iterator().next();

2
dhp-workflows/dhp-dedup-scholexplorer/src/main/java/eu/dnetlib/dedup/graph/ConnectedComponent.java
View File

@ -25,7 +25,7 @@ public class ConnectedComponent implements Serializable {
if (docIds.size() > 1) {
final String s = getMin();
String prefix = s.split("\\|")[0];
ccId = prefix + "|dedup_______::" + DedupUtility.md5(s);
ccId = prefix + "|dedup_wf_001::" + DedupUtility.md5(s);
return ccId;
} else {
return docIds.iterator().next();