D-Net
/
dnet-hadoop
14
5
Fork 11
Code Issues 14 Pull Requests 13 Projects 2 Releases 2 Wiki Activity

mapping by participant project contribution

This commit is contained in:
Claudio Atzori 2022-06-24 17:16:42 +02:00
parent 929b145130
commit 5130eac247
4 changed files with 114 additions and 152 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

58
dhp-common/src/main/java/eu/dnetlib/dhp/schema/oaf/utils/OafMapperUtils.java
View File

@ -419,4 +419,62 @@ public class OafMapperUtils {
m.setUnit(Arrays.asList(newKeyValueInstance(key, value, dataInfo))); m.setUnit(Arrays.asList(newKeyValueInstance(key, value, dataInfo)));
return m; return m;
} }
public static Relation getRelation(final String source,
final String target,
final String relType,
final String subRelType,
final String relClass,
final OafEntity entity) {
return getRelation(source, target, relType, subRelType, relClass, entity, null);
}
public static Relation getRelation(final String source,
final String target,
final String relType,
final String subRelType,
final String relClass,
final OafEntity entity,
final String validationDate) {
return getRelation(
source, target, relType, subRelType, relClass, entity.getCollectedfrom(), entity.getDataInfo(),
entity.getLastupdatetimestamp(), validationDate, null);
}
public static Relation getRelation(final String source,
final String target,
final String relType,
final String subRelType,
final String relClass,
final List<KeyValue> collectedfrom,
final DataInfo dataInfo,
final Long lastupdatetimestamp) {
return getRelation(
source, target, relType, subRelType, relClass, collectedfrom, dataInfo, lastupdatetimestamp, null, null);
}
public static Relation getRelation(final String source,
final String target,
final String relType,
final String subRelType,
final String relClass,
final List<KeyValue> collectedfrom,
final DataInfo dataInfo,
final Long lastupdatetimestamp,
final String validationDate,
final List<KeyValue> properties) {
final Relation rel = new Relation();
rel.setRelType(relType);
rel.setSubRelType(subRelType);
rel.setRelClass(relClass);
rel.setSource(source);
rel.setTarget(target);
rel.setCollectedfrom(collectedfrom);
rel.setDataInfo(dataInfo);
rel.setLastupdatetimestamp(lastupdatetimestamp);
rel.setValidated(StringUtils.isNotBlank(validationDate));
rel.setValidationDate(StringUtils.isNotBlank(validationDate) ? validationDate : null);
rel.setProperties(properties);
return rel;
}
} }

46
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java
View File

@ -267,10 +267,13 @@ public abstract class AbstractMdRecordToOafMapper {
res res
.add( .add(
getRelation( OafMapperUtils
.getRelation(
docId, projectId, RESULT_PROJECT, OUTCOME, IS_PRODUCED_BY, entity, validationdDate)); docId, projectId, RESULT_PROJECT, OUTCOME, IS_PRODUCED_BY, entity, validationdDate));
res res
.add(getRelation(projectId, docId, RESULT_PROJECT, OUTCOME, PRODUCES, entity, validationdDate)); .add(
OafMapperUtils
.getRelation(projectId, docId, RESULT_PROJECT, OUTCOME, PRODUCES, entity, validationdDate));
} }
} }
@ -303,11 +306,14 @@ public abstract class AbstractMdRecordToOafMapper {
final String targetId = createOpenaireId(targetType, target, true); final String targetId = createOpenaireId(targetType, target, true);
rels rels
.add( .add(
getRelation( OafMapperUtils
entity.getId(), targetId, relType, subRelType, relClass, entity, validationdDate)); .getRelation(
entity.getId(), targetId, relType, subRelType, relClass, entity,
validationdDate));
rels rels
.add( .add(
getRelation( OafMapperUtils
.getRelation(
targetId, entity.getId(), relType, subRelType, relClassInverse, entity, targetId, entity.getId(), relType, subRelType, relClassInverse, entity,
validationdDate)); validationdDate));
} }
@ -317,36 +323,6 @@ public abstract class AbstractMdRecordToOafMapper {
return rels; return rels;
} }
protected Relation getRelation(final String source,
final String target,
final String relType,
final String subRelType,
final String relClass,
final OafEntity entity) {
return getRelation(source, target, relType, subRelType, relClass, entity, null);
}
protected Relation getRelation(final String source,
final String target,
final String relType,
final String subRelType,
final String relClass,
final OafEntity entity,
final String validationDate) {
final Relation rel = new Relation();
rel.setRelType(relType);
rel.setSubRelType(subRelType);
rel.setRelClass(relClass);
rel.setSource(source);
rel.setTarget(target);
rel.setCollectedfrom(entity.getCollectedfrom());
rel.setDataInfo(entity.getDataInfo());
rel.setLastupdatetimestamp(entity.getLastupdatetimestamp());
rel.setValidated(StringUtils.isNotBlank(validationDate));
rel.setValidationDate(StringUtils.isNotBlank(validationDate) ? validationDate : null);
return rel;
}
protected abstract List<Oaf> addOtherResultRels( protected abstract List<Oaf> addOtherResultRels(
final Document doc, final Document doc,
final OafEntity entity); final OafEntity entity);

155
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/MigrateDbEntitiesApplication.java
View File

@ -1,32 +1,7 @@
package eu.dnetlib.dhp.oa.graph.raw; package eu.dnetlib.dhp.oa.graph.raw;
import static eu.dnetlib.dhp.schema.common.ModelConstants.DATASET_DEFAULT_RESULTTYPE; import static eu.dnetlib.dhp.schema.common.ModelConstants.*;
import static eu.dnetlib.dhp.schema.common.ModelConstants.DATASOURCE_ORGANIZATION;
import static eu.dnetlib.dhp.schema.common.ModelConstants.DNET_PROVENANCE_ACTIONS;
import static eu.dnetlib.dhp.schema.common.ModelConstants.ENTITYREGISTRY_PROVENANCE_ACTION;
import static eu.dnetlib.dhp.schema.common.ModelConstants.HAS_PARTICIPANT;
import static eu.dnetlib.dhp.schema.common.ModelConstants.IS_MERGED_IN;
import static eu.dnetlib.dhp.schema.common.ModelConstants.IS_PARTICIPANT;
import static eu.dnetlib.dhp.schema.common.ModelConstants.IS_PRODUCED_BY;
import static eu.dnetlib.dhp.schema.common.ModelConstants.IS_PROVIDED_BY;
import static eu.dnetlib.dhp.schema.common.ModelConstants.IS_RELATED_TO;
import static eu.dnetlib.dhp.schema.common.ModelConstants.MERGES;
import static eu.dnetlib.dhp.schema.common.ModelConstants.ORG_ORG_RELTYPE;
import static eu.dnetlib.dhp.schema.common.ModelConstants.ORP_DEFAULT_RESULTTYPE;
import static eu.dnetlib.dhp.schema.common.ModelConstants.OUTCOME;
import static eu.dnetlib.dhp.schema.common.ModelConstants.PARTICIPATION;
import static eu.dnetlib.dhp.schema.common.ModelConstants.PRODUCES;
import static eu.dnetlib.dhp.schema.common.ModelConstants.PROJECT_ORGANIZATION;
import static eu.dnetlib.dhp.schema.common.ModelConstants.PROVIDES;
import static eu.dnetlib.dhp.schema.common.ModelConstants.PROVISION;
import static eu.dnetlib.dhp.schema.common.ModelConstants.PUBLICATION_DATASET;
import static eu.dnetlib.dhp.schema.common.ModelConstants.PUBLICATION_DEFAULT_RESULTTYPE;
import static eu.dnetlib.dhp.schema.common.ModelConstants.RELATIONSHIP;
import static eu.dnetlib.dhp.schema.common.ModelConstants.RESULT_PROJECT;
import static eu.dnetlib.dhp.schema.common.ModelConstants.RESULT_RESULT;
import static eu.dnetlib.dhp.schema.common.ModelConstants.SOFTWARE_DEFAULT_RESULTTYPE;
import static eu.dnetlib.dhp.schema.common.ModelConstants.USER_CLAIM;
import static eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils.*; import static eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils.*;
import java.io.Closeable; import java.io.Closeable;
@ -45,6 +20,8 @@ import org.apache.commons.lang3.StringUtils;
import org.slf4j.Logger; import org.slf4j.Logger;
import org.slf4j.LoggerFactory; import org.slf4j.LoggerFactory;
import com.google.common.collect.Lists;
import eu.dnetlib.dhp.application.ArgumentApplicationParser; import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.common.DbClient; import eu.dnetlib.dhp.common.DbClient;
import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup; import eu.dnetlib.dhp.common.vocabulary.VocabularyGroup;
@ -68,6 +45,7 @@ import eu.dnetlib.dhp.schema.oaf.Relation;
import eu.dnetlib.dhp.schema.oaf.Result; import eu.dnetlib.dhp.schema.oaf.Result;
import eu.dnetlib.dhp.schema.oaf.Software; import eu.dnetlib.dhp.schema.oaf.Software;
import eu.dnetlib.dhp.schema.oaf.StructuredProperty; import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
import eu.dnetlib.dhp.utils.ISLookupClientFactory; import eu.dnetlib.dhp.utils.ISLookupClientFactory;
public class MigrateDbEntitiesApplication extends AbstractMigrationApplication implements Closeable { public class MigrateDbEntitiesApplication extends AbstractMigrationApplication implements Closeable {
@ -437,25 +415,14 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
final List<KeyValue> collectedFrom = listKeyValues( final List<KeyValue> collectedFrom = listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")); createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final Relation r1 = new Relation(); final Relation r1 = OafMapperUtils
r1.setRelType(DATASOURCE_ORGANIZATION); .getRelation(
r1.setSubRelType(PROVISION); dsId, orgId, DATASOURCE_ORGANIZATION, PRODUCES, IS_PROVIDED_BY, collectedFrom, info,
r1.setRelClass(IS_PROVIDED_BY); lastUpdateTimestamp);
r1.setSource(dsId);
r1.setTarget(orgId);
r1.setCollectedfrom(collectedFrom);
r1.setDataInfo(info);
r1.setLastupdatetimestamp(lastUpdateTimestamp);
final Relation r2 = new Relation(); final Relation r2 = OafMapperUtils
r2.setRelType(DATASOURCE_ORGANIZATION); .getRelation(
r2.setSubRelType(PROVISION); orgId, dsId, DATASOURCE_ORGANIZATION, PRODUCES, PROVIDES, collectedFrom, info, lastUpdateTimestamp);
r2.setRelClass(PROVIDES);
r2.setSource(orgId);
r2.setTarget(dsId);
r2.setCollectedfrom(collectedFrom);
r2.setDataInfo(info);
r2.setLastupdatetimestamp(lastUpdateTimestamp);
return Arrays.asList(r1, r2); return Arrays.asList(r1, r2);
} catch (final Exception e) { } catch (final Exception e) {
@ -471,25 +438,20 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
final List<KeyValue> collectedFrom = listKeyValues( final List<KeyValue> collectedFrom = listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")); createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final Relation r1 = new Relation(); final List<KeyValue> properties = Lists
r1.setRelType(PROJECT_ORGANIZATION); .newArrayList(
r1.setSubRelType(PARTICIPATION); keyValue("contribution", String.valueOf(rs.getDouble("totalcost"))),
r1.setRelClass(HAS_PARTICIPANT); keyValue("currency", rs.getString("currency")));
r1.setSource(projectId);
r1.setTarget(orgId);
r1.setCollectedfrom(collectedFrom);
r1.setDataInfo(info);
r1.setLastupdatetimestamp(lastUpdateTimestamp);
final Relation r2 = new Relation(); final Relation r1 = OafMapperUtils
r2.setRelType(PROJECT_ORGANIZATION); .getRelation(
r2.setSubRelType(PARTICIPATION); projectId, orgId, PROJECT_ORGANIZATION, PARTICIPATION, HAS_PARTICIPANT, collectedFrom, info,
r2.setRelClass(IS_PARTICIPANT); lastUpdateTimestamp, null, properties);
r2.setSource(orgId);
r2.setTarget(projectId); final Relation r2 = OafMapperUtils
r2.setCollectedfrom(collectedFrom); .getRelation(
r2.setDataInfo(info); orgId, projectId, PROJECT_ORGANIZATION, PARTICIPATION, IS_PARTICIPANT, collectedFrom, info,
r2.setLastupdatetimestamp(lastUpdateTimestamp); lastUpdateTimestamp, null, properties);
return Arrays.asList(r1, r2); return Arrays.asList(r1, r2);
} catch (final Exception e) { } catch (final Exception e) {
@ -703,25 +665,12 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
final List<KeyValue> collectedFrom = listKeyValues( final List<KeyValue> collectedFrom = listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")); createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final Relation r1 = new Relation(); final Relation r1 = OafMapperUtils
r1.setRelType(ORG_ORG_RELTYPE); .getRelation(orgId1, orgId2, ORG_ORG_RELTYPE, DEDUP, MERGES, collectedFrom, info, lastUpdateTimestamp);
r1.setSubRelType(ModelConstants.DEDUP);
r1.setRelClass(MERGES);
r1.setSource(orgId1);
r1.setTarget(orgId2);
r1.setCollectedfrom(collectedFrom);
r1.setDataInfo(info);
r1.setLastupdatetimestamp(lastUpdateTimestamp);
final Relation r2 = new Relation(); final Relation r2 = OafMapperUtils
r2.setRelType(ORG_ORG_RELTYPE); .getRelation(
r2.setSubRelType(ModelConstants.DEDUP); orgId2, orgId1, ORG_ORG_RELTYPE, DEDUP, IS_MERGED_IN, collectedFrom, info, lastUpdateTimestamp);
r2.setRelClass(IS_MERGED_IN);
r2.setSource(orgId2);
r2.setTarget(orgId1);
r2.setCollectedfrom(collectedFrom);
r2.setDataInfo(info);
r2.setLastupdatetimestamp(lastUpdateTimestamp);
return Arrays.asList(r1, r2); return Arrays.asList(r1, r2);
} catch (final Exception e) { } catch (final Exception e) {
throw new RuntimeException(e); throw new RuntimeException(e);
@ -738,17 +687,12 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
final List<KeyValue> collectedFrom = listKeyValues( final List<KeyValue> collectedFrom = listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")); createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final Relation r = new Relation(); return Arrays
r.setRelType(ORG_ORG_RELTYPE); .asList(
r.setSubRelType(ModelConstants.RELATIONSHIP); OafMapperUtils
r.setRelClass(rs.getString("type")); .getRelation(
r.setSource(orgId1); orgId1, orgId2, ORG_ORG_RELTYPE, RELATIONSHIP, rs.getString("type"), collectedFrom, info,
r.setTarget(orgId2); lastUpdateTimestamp));
r.setCollectedfrom(collectedFrom);
r.setDataInfo(info);
r.setLastupdatetimestamp(lastUpdateTimestamp);
return Arrays.asList(r);
} catch (final Exception e) { } catch (final Exception e) {
throw new RuntimeException(e); throw new RuntimeException(e);
} }
@ -765,29 +709,12 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
final List<KeyValue> collectedFrom = listKeyValues( final List<KeyValue> collectedFrom = listKeyValues(
createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname")); createOpenaireId(10, rs.getString("collectedfromid"), true), rs.getString("collectedfromname"));
final Relation r1 = new Relation(); return Arrays
r1.setRelType(ORG_ORG_RELTYPE); .asList(
r1.setSubRelType(ModelConstants.DEDUP); OafMapperUtils
r1.setRelClass(relClass); .getRelation(
r1.setSource(orgId1); orgId1, orgId2, ORG_ORG_RELTYPE, DEDUP, relClass, collectedFrom, info,
r1.setTarget(orgId2); lastUpdateTimestamp));
r1.setCollectedfrom(collectedFrom);
r1.setDataInfo(info);
r1.setLastupdatetimestamp(lastUpdateTimestamp);
// removed because there's no difference between two sides //TODO
// final Relation r2 = new Relation();
// r2.setRelType(ORG_ORG_RELTYPE);
// r2.setSubRelType(ORG_ORG_SUBRELTYPE);
// r2.setRelClass(relClass);
// r2.setSource(orgId2);
// r2.setTarget(orgId1);
// r2.setCollectedfrom(collectedFrom);
// r2.setDataInfo(info);
// r2.setLastupdatetimestamp(lastUpdateTimestamp);
// return Arrays.asList(r1, r2);
return Arrays.asList(r1);
} catch (final Exception e) { } catch (final Exception e) {
throw new RuntimeException(e); throw new RuntimeException(e);
} }

1
dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/oa/graph/sql/queryProjectOrganization.sql
View File

@ -3,6 +3,7 @@ SELECT
po.resporganization AS resporganization, po.resporganization AS resporganization,
po.participantnumber AS participantnumber, po.participantnumber AS participantnumber,
po.contribution AS contribution, po.contribution AS contribution,
po.currency AS currency,
NULL AS startdate, NULL AS startdate,
NULL AS enddate, NULL AS enddate,
false AS inferred, false AS inferred,