1
0
Fork 0

Merge branch 'beta' into import_orps_fix

This commit is contained in:
Claudio Atzori 2024-02-15 15:01:56 +01:00
commit a63b091bae
18 changed files with 224 additions and 29 deletions

View File

@ -23,15 +23,18 @@ public class InstanceTypeMatch extends AbstractListComparator {
// jolly types
translationMap.put("Conference object", "*");
translationMap.put("Research", "*");
translationMap.put("Other literature type", "*");
translationMap.put("Unknown", "*");
translationMap.put("UNKNOWN", "*");
// article types
translationMap.put("Article", "Article");
translationMap.put("Journal", "Article");
translationMap.put("Data Paper", "Article");
translationMap.put("Software Paper", "Article");
translationMap.put("Preprint", "Article");
translationMap.put("Part of book or chapter of book", "Article");
// thesis types
translationMap.put("Thesis", "Thesis");

View File

@ -0,0 +1,39 @@
/*
* Copyright (c) 2024.
* SPDX-FileCopyrightText: © 2023 Consiglio Nazionale delle Ricerche
* SPDX-License-Identifier: AGPL-3.0-or-later
*/
package eu.dnetlib.dhp.actionmanager.promote;
/** Encodes the Actionset promotion strategies */
public class PromoteAction {
/** The supported actionset promotion strategies
*
* ENRICH: promotes only records in the actionset matching another record in the
* graph and enriches them applying the given MergeAndGet strategy
* UPSERT: promotes all the records in an actionset, matching records are updated
* using the given MergeAndGet strategy, the non-matching record as inserted as they are.
*/
public enum Strategy {
ENRICH, UPSERT
}
/**
* Returns the string representation of the join type implementing the given PromoteAction.
*
* @param strategy the strategy to be used to promote the Actionset contents
* @return the join type used to implement the promotion strategy
*/
public static String joinTypeForStrategy(PromoteAction.Strategy strategy) {
switch (strategy) {
case ENRICH:
return "left_outer";
case UPSERT:
return "full_outer";
default:
throw new IllegalStateException("unsupported PromoteAction: " + strategy.toString());
}
}
}

View File

@ -67,8 +67,9 @@ public class PromoteActionPayloadForGraphTableJob {
String outputGraphTablePath = parser.get("outputGraphTablePath");
logger.info("outputGraphTablePath: {}", outputGraphTablePath);
MergeAndGet.Strategy strategy = MergeAndGet.Strategy.valueOf(parser.get("mergeAndGetStrategy").toUpperCase());
logger.info("strategy: {}", strategy);
MergeAndGet.Strategy mergeAndGetStrategy = MergeAndGet.Strategy
.valueOf(parser.get("mergeAndGetStrategy").toUpperCase());
logger.info("mergeAndGetStrategy: {}", mergeAndGetStrategy);
Boolean shouldGroupById = Optional
.ofNullable(parser.get("shouldGroupById"))
@ -76,6 +77,12 @@ public class PromoteActionPayloadForGraphTableJob {
.orElse(true);
logger.info("shouldGroupById: {}", shouldGroupById);
PromoteAction.Strategy promoteActionStrategy = Optional
.ofNullable(parser.get("promoteActionStrategy"))
.map(PromoteAction.Strategy::valueOf)
.orElse(PromoteAction.Strategy.UPSERT);
logger.info("promoteActionStrategy: {}", promoteActionStrategy);
@SuppressWarnings("unchecked")
Class<? extends Oaf> rowClazz = (Class<? extends Oaf>) Class.forName(graphTableClassName);
@SuppressWarnings("unchecked")
@ -97,7 +104,8 @@ public class PromoteActionPayloadForGraphTableJob {
inputGraphTablePath,
inputActionPayloadPath,
outputGraphTablePath,
strategy,
mergeAndGetStrategy,
promoteActionStrategy,
rowClazz,
actionPayloadClazz,
shouldGroupById);
@ -124,14 +132,16 @@ public class PromoteActionPayloadForGraphTableJob {
String inputGraphTablePath,
String inputActionPayloadPath,
String outputGraphTablePath,
MergeAndGet.Strategy strategy,
MergeAndGet.Strategy mergeAndGetStrategy,
PromoteAction.Strategy promoteActionStrategy,
Class<G> rowClazz,
Class<A> actionPayloadClazz, Boolean shouldGroupById) {
Dataset<G> rowDS = readGraphTable(spark, inputGraphTablePath, rowClazz);
Dataset<A> actionPayloadDS = readActionPayload(spark, inputActionPayloadPath, actionPayloadClazz);
Dataset<G> result = promoteActionPayloadForGraphTable(
rowDS, actionPayloadDS, strategy, rowClazz, actionPayloadClazz, shouldGroupById)
rowDS, actionPayloadDS, mergeAndGetStrategy, promoteActionStrategy, rowClazz, actionPayloadClazz,
shouldGroupById)
.map((MapFunction<G, G>) value -> value, Encoders.bean(rowClazz));
saveGraphTable(result, outputGraphTablePath);
@ -183,7 +193,8 @@ public class PromoteActionPayloadForGraphTableJob {
private static <G extends Oaf, A extends Oaf> Dataset<G> promoteActionPayloadForGraphTable(
Dataset<G> rowDS,
Dataset<A> actionPayloadDS,
MergeAndGet.Strategy strategy,
MergeAndGet.Strategy mergeAndGetStrategy,
PromoteAction.Strategy promoteActionStrategy,
Class<G> rowClazz,
Class<A> actionPayloadClazz,
Boolean shouldGroupById) {
@ -195,8 +206,9 @@ public class PromoteActionPayloadForGraphTableJob {
SerializableSupplier<Function<G, String>> rowIdFn = ModelSupport::idFn;
SerializableSupplier<Function<A, String>> actionPayloadIdFn = ModelSupport::idFn;
SerializableSupplier<BiFunction<G, A, G>> mergeRowWithActionPayloadAndGetFn = MergeAndGet.functionFor(strategy);
SerializableSupplier<BiFunction<G, G, G>> mergeRowsAndGetFn = MergeAndGet.functionFor(strategy);
SerializableSupplier<BiFunction<G, A, G>> mergeRowWithActionPayloadAndGetFn = MergeAndGet
.functionFor(mergeAndGetStrategy);
SerializableSupplier<BiFunction<G, G, G>> mergeRowsAndGetFn = MergeAndGet.functionFor(mergeAndGetStrategy);
SerializableSupplier<G> zeroFn = zeroFn(rowClazz);
SerializableSupplier<Function<G, Boolean>> isNotZeroFn = PromoteActionPayloadForGraphTableJob::isNotZeroFnUsingIdOrSourceAndTarget;
@ -207,6 +219,7 @@ public class PromoteActionPayloadForGraphTableJob {
rowIdFn,
actionPayloadIdFn,
mergeRowWithActionPayloadAndGetFn,
promoteActionStrategy,
rowClazz,
actionPayloadClazz);

View File

@ -34,6 +34,7 @@ public class PromoteActionPayloadFunctions {
* @param rowIdFn Function used to get the id of graph table row
* @param actionPayloadIdFn Function used to get id of action payload instance
* @param mergeAndGetFn Function used to merge graph table row and action payload instance
* @param promoteActionStrategy the Actionset promotion strategy
* @param rowClazz Class of graph table
* @param actionPayloadClazz Class of action payload
* @param <G> Type of graph table row
@ -46,6 +47,7 @@ public class PromoteActionPayloadFunctions {
SerializableSupplier<Function<G, String>> rowIdFn,
SerializableSupplier<Function<A, String>> actionPayloadIdFn,
SerializableSupplier<BiFunction<G, A, G>> mergeAndGetFn,
PromoteAction.Strategy promoteActionStrategy,
Class<G> rowClazz,
Class<A> actionPayloadClazz) {
if (!isSubClass(rowClazz, actionPayloadClazz)) {
@ -61,7 +63,7 @@ public class PromoteActionPayloadFunctions {
.joinWith(
actionPayloadWithIdDS,
rowWithIdDS.col("_1").equalTo(actionPayloadWithIdDS.col("_1")),
"full_outer")
PromoteAction.joinTypeForStrategy(promoteActionStrategy))
.map(
(MapFunction<Tuple2<Tuple2<String, G>, Tuple2<String, A>>, G>) value -> {
Optional<G> rowOpt = Optional.ofNullable(value._1()).map(Tuple2::_2);

View File

@ -41,6 +41,12 @@
"paramDescription": "strategy for merging graph table objects with action payload instances, MERGE_FROM_AND_GET or SELECT_NEWER_AND_GET",
"paramRequired": true
},
{
"paramName": "pas",
"paramLongName": "promoteActionStrategy",
"paramDescription": "strategy for promoting the actionset contents into the graph tables, ENRICH or UPSERT (default)",
"paramRequired": false
},
{
"paramName": "sgid",
"paramLongName": "shouldGroupById",

View File

@ -115,6 +115,7 @@
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
<arg>--outputGraphTablePath</arg><arg>${workingDir}/dataset</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
<arg>--promoteActionStrategy</arg><arg>${promoteActionStrategy}</arg>
<arg>--shouldGroupById</arg><arg>${shouldGroupById}</arg>
</spark>
<ok to="DecisionPromoteResultActionPayloadForDatasetTable"/>
@ -167,6 +168,7 @@
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/dataset</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
<arg>--promoteActionStrategy</arg><arg>${promoteActionStrategy}</arg>
<arg>--shouldGroupById</arg><arg>${shouldGroupById}</arg>
</spark>
<ok to="End"/>

View File

@ -106,6 +106,7 @@
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Datasource</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/datasource</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
<arg>--promoteActionStrategy</arg><arg>${promoteActionStrategy}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>

View File

@ -106,6 +106,7 @@
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Organization</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/organization</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
<arg>--promoteActionStrategy</arg><arg>${promoteActionStrategy}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>

View File

@ -114,6 +114,7 @@
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
<arg>--outputGraphTablePath</arg><arg>${workingDir}/otherresearchproduct</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
<arg>--promoteActionStrategy</arg><arg>${promoteActionStrategy}</arg>
<arg>--shouldGroupById</arg><arg>${shouldGroupById}</arg>
</spark>
<ok to="DecisionPromoteResultActionPayloadForOtherResearchProductTable"/>
@ -166,6 +167,7 @@
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/otherresearchproduct</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
<arg>--promoteActionStrategy</arg><arg>${promoteActionStrategy}</arg>
<arg>--shouldGroupById</arg><arg>${shouldGroupById}</arg>
</spark>
<ok to="End"/>

View File

@ -106,6 +106,7 @@
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Project</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/project</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
<arg>--promoteActionStrategy</arg><arg>${promoteActionStrategy}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>

View File

@ -115,6 +115,7 @@
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
<arg>--outputGraphTablePath</arg><arg>${workingDir}/publication</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
<arg>--promoteActionStrategy</arg><arg>${promoteActionStrategy}</arg>
<arg>--shouldGroupById</arg><arg>${shouldGroupById}</arg>
</spark>
<ok to="DecisionPromoteResultActionPayloadForPublicationTable"/>
@ -167,6 +168,7 @@
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/publication</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
<arg>--promoteActionStrategy</arg><arg>${promoteActionStrategy}</arg>
<arg>--shouldGroupById</arg><arg>${shouldGroupById}</arg>
</spark>
<ok to="End"/>

View File

@ -107,6 +107,7 @@
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Relation</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/relation</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
<arg>--promoteActionStrategy</arg><arg>${promoteActionStrategy}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>

View File

@ -114,6 +114,7 @@
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
<arg>--outputGraphTablePath</arg><arg>${workingDir}/software</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
<arg>--promoteActionStrategy</arg><arg>${promoteActionStrategy}</arg>
<arg>--shouldGroupById</arg><arg>${shouldGroupById}</arg>
</spark>
<ok to="DecisionPromoteResultActionPayloadForSoftwareTable"/>
@ -166,6 +167,7 @@
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/software</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
<arg>--promoteActionStrategy</arg><arg>${promoteActionStrategy}</arg>
<arg>--shouldGroupById</arg><arg>${shouldGroupById}</arg>
</spark>
<ok to="End"/>

View File

@ -54,7 +54,7 @@ public class PromoteActionPayloadFunctionsTest {
RuntimeException.class,
() -> PromoteActionPayloadFunctions
.joinGraphTableWithActionPayloadAndMerge(
null, null, null, null, null, OafImplSubSub.class, OafImpl.class));
null, null, null, null, null, null, OafImplSubSub.class, OafImpl.class));
}
@Test
@ -104,6 +104,7 @@ public class PromoteActionPayloadFunctionsTest {
rowIdFn,
actionPayloadIdFn,
mergeAndGetFn,
PromoteAction.Strategy.UPSERT,
OafImplSubSub.class,
OafImplSubSub.class)
.collectAsList();
@ -183,6 +184,7 @@ public class PromoteActionPayloadFunctionsTest {
rowIdFn,
actionPayloadIdFn,
mergeAndGetFn,
PromoteAction.Strategy.UPSERT,
OafImplSubSub.class,
OafImplSub.class)
.collectAsList();

View File

@ -122,22 +122,41 @@ public class DedupRecordFactory {
}
return Stream
.concat(Stream.of(agg.getDedupId()), agg.aliases.stream())
.map(id -> {
try {
OafEntity res = (OafEntity) BeanUtils.cloneBean(agg.entity);
res.setId(id);
res.setDataInfo(dataInfo);
res.setLastupdatetimestamp(ts);
return res;
} catch (Exception e) {
throw new RuntimeException(e);
}
})
.concat(
Stream
.of(agg.getDedupId())
.map(id -> createDedupOafEntity(id, agg.entity, dataInfo, ts)),
agg.aliases
.stream()
.map(id -> createMergedDedupAliasOafEntity(id, agg.entity, dataInfo, ts)))
.iterator();
}, beanEncoder);
}
private static OafEntity createDedupOafEntity(String id, OafEntity base, DataInfo dataInfo, long ts) {
try {
OafEntity res = (OafEntity) BeanUtils.cloneBean(base);
res.setId(id);
res.setDataInfo(dataInfo);
res.setLastupdatetimestamp(ts);
return res;
} catch (Exception e) {
throw new RuntimeException(e);
}
}
private static OafEntity createMergedDedupAliasOafEntity(String id, OafEntity base, DataInfo dataInfo, long ts) {
try {
OafEntity res = createDedupOafEntity(id, base, dataInfo, ts);
DataInfo ds = (DataInfo) BeanUtils.cloneBean(dataInfo);
ds.setDeletedbyinference(true);
res.setDataInfo(ds);
return res;
} catch (Exception e) {
throw new RuntimeException(e);
}
}
private static OafEntity reduceEntity(OafEntity entity, OafEntity duplicate) {
if (duplicate == null) {

View File

@ -238,11 +238,23 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
(Element) doc
.selectSingleNode(
"//*[local-name()='metadata']/*[local-name() = 'resource']/*[local-name() = 'resourceType']"))
.map(element -> {
final String resourceTypeURI = element.attributeValue("uri");
final String resourceTypeAnyURI = element.attributeValue("anyURI");
final String resourceTypeTxt = element.getText();
final String resourceTypeGeneral = element.attributeValue("resourceTypeGeneral");
.map(e -> {
final String resourceTypeURI = Optional
.ofNullable(e.attributeValue("uri"))
.filter(StringUtils::isNotBlank)
.orElse(null);
final String resourceTypeAnyURI = Optional
.ofNullable(e.attributeValue("anyURI"))
.filter(StringUtils::isNotBlank)
.orElse(null);
final String resourceTypeTxt = Optional
.ofNullable(e.getText())
.filter(StringUtils::isNotBlank)
.orElse(null);
final String resourceTypeGeneral = Optional
.ofNullable(e.attributeValue("resourceTypeGeneral"))
.filter(StringUtils::isNotBlank)
.orElse(null);
return ObjectUtils
.firstNonNull(resourceTypeURI, resourceTypeAnyURI, resourceTypeTxt, resourceTypeGeneral);

View File

@ -1171,6 +1171,34 @@ class MappersTest {
}
@Test
void test_Zenodo2() throws IOException {
final String xml = IOUtils.toString(Objects.requireNonNull(getClass().getResourceAsStream("odf_zenodo2.xml")));
final List<Oaf> list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml);
assertEquals(3, list.size());
Publication p = cleanup((Publication) list.get(0), vocs);
assertNotNull(p.getInstance());
assertEquals(1, p.getInstance().size());
final Instance instance = p.getInstance().get(0);
assertNotNull(instance.getInstanceTypeMapping());
assertEquals(1, instance.getInstanceTypeMapping().size());
Optional<InstanceTypeMapping> coarType = instance
.getInstanceTypeMapping()
.stream()
.filter(itm -> ModelConstants.OPENAIRE_COAR_RESOURCE_TYPES_3_1.equals(itm.getVocabularyName()))
.findFirst();
assertTrue(coarType.isPresent());
assertNotNull(coarType.get().getOriginalType());
assertNull(coarType.get().getTypeCode());
assertNull(coarType.get().getTypeLabel());
}
@Test
void testROHub2() throws IOException {
final String xml = IOUtils
@ -1229,7 +1257,7 @@ class MappersTest {
}
@Test
public void testD4ScienceTraining() throws IOException {
void testD4ScienceTraining() throws IOException {
final String xml = IOUtils
.toString(Objects.requireNonNull(getClass().getResourceAsStream("d4science-1-training.xml")));
final List<Oaf> list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml);
@ -1240,7 +1268,7 @@ class MappersTest {
}
@Test
public void testD4ScienceDataset() throws IOException {
void testD4ScienceDataset() throws IOException {
final String xml = IOUtils
.toString(Objects.requireNonNull(getClass().getResourceAsStream("d4science-2-dataset.xml")));
final List<Oaf> list = new OdfToOafMapper(vocs, false, true).processMdRecord(xml);

View File

@ -0,0 +1,59 @@
<?xml version="1.0" encoding="UTF-8"?>
<record xmlns:oaf="http://namespace.openaire.eu/oaf"
xmlns:oai="http://www.openarchives.org/OAI/2.0/"
xmlns:datacite="http://datacite.org/schema/kernel-3"
xmlns:dr="http://www.driver-repository.eu/namespace/dr"
xmlns:dri="http://www.driver-repository.eu/namespace/dri">
<header xmlns="http://www.openarchives.org/OAI/2.0/">
<identifier>oai:zenodo.org:1596086</identifier>
<datestamp>2020-01-20T13:50:28Z</datestamp>
<setSpec>openaire</setSpec>
<dr:dateOfTransformation>2024-02-08T11:03:10.994Z</dr:dateOfTransformation>
<dri:objIdentifier>od______2659::036d5555a6688ed00c8d0da97bdece3b</dri:objIdentifier>
<dri:dateOfCollection>2024-02-08T11:03:10.994Z</dri:dateOfCollection>
<dri:dateOfTransformation>2024-02-08T11:03:10.994Z</dri:dateOfTransformation>
</header>
<metadata>
<resource xmlns="http://datacite.org/schema/kernel-4"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://datacite.org/schema/kernel-4 http://schema.datacite.org/meta/kernel-4.1/metadata.xsd">
<identifier identifierType="URL">https://zenodo.org/record/1596086</identifier>
<alternateIdentifiers xmlns="http://datacite.org/schema/kernel-3"/>
<creators>
<creator>
<creatorName>Bonney, T. G.</creatorName>
<givenName>T. G.</givenName>
<familyName>Bonney</familyName>
</creator>
</creators>
<titles>
<title>Ice Blocks on a Moraine</title>
</titles>
<publisher>Zenodo</publisher>
<publicationYear>1889</publicationYear>
<dates>
<date dateType="Issued">1889-08-22</date>
</dates>
<resourceType resourceTypeGeneral="JournalArticle"/>
<relatedIdentifiers>
<relatedIdentifier relatedIdentifierType="DOI" relationType="IsIdenticalTo"
>10.1038/040391a0</relatedIdentifier>
</relatedIdentifiers>
<rightsList>
<rights rightsURI="https://creativecommons.org/publicdomain/zero/1.0/legalcode"
>Creative Commons Zero v1.0 Universal</rights>
<rights rightsURI="info:eu-repo/semantics/openAccess">Open Access</rights>
</rightsList>
<descriptions>
<description descriptionType="Abstract">n/a</description>
</descriptions>
</resource>
<dr:CobjCategory type="publication">0001</dr:CobjCategory>
<oaf:dateAccepted>1889-08-22</oaf:dateAccepted>
<oaf:accessrights>OPEN</oaf:accessrights>
<oaf:license>http://creativecommons.org/publicdomain/zero/1.0/legalcode</oaf:license>
<oaf:language/>
<oaf:hostedBy name="ZENODO" id="opendoar____::2659"/>
<oaf:collectedFrom name="ZENODO" id="opendoar____::2659"/>
</metadata>
</record>