forked from D-Net/dnet-hadoop
Merge branch 'master' of https://code-repo.d4science.org/D-Net/dnet-hadoop
This commit is contained in:
commit
79969d78b9
|
@ -1,66 +1,72 @@
|
|||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
|
||||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.2-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
<project xmlns="http://maven.apache.org/POM/4.0.0"
|
||||
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
|
||||
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
|
||||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.2-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
<artifactId>dhp-broker-events</artifactId>
|
||||
<artifactId>dhp-broker-events</artifactId>
|
||||
|
||||
<dependencies>
|
||||
<dependencies>
|
||||
|
||||
<dependency>
|
||||
<groupId>commons-io</groupId>
|
||||
<artifactId>commons-io</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>commons-io</groupId>
|
||||
<artifactId>commons-io</artifactId>
|
||||
</dependency>
|
||||
|
||||
<dependency>
|
||||
<groupId>org.apache.spark</groupId>
|
||||
<artifactId>spark-core_2.11</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>org.apache.spark</groupId>
|
||||
<artifactId>spark-sql_2.11</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>org.apache.spark</groupId>
|
||||
<artifactId>spark-hive_2.11</artifactId>
|
||||
<scope>test</scope>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>org.apache.spark</groupId>
|
||||
<artifactId>spark-core_2.11</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>org.apache.spark</groupId>
|
||||
<artifactId>spark-sql_2.11</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>org.apache.spark</groupId>
|
||||
<artifactId>spark-hive_2.11</artifactId>
|
||||
<scope>test</scope>
|
||||
</dependency>
|
||||
|
||||
<dependency>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp-common</artifactId>
|
||||
<version>${project.version}</version>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp-schemas</artifactId>
|
||||
<version>${project.version}</version>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp-common</artifactId>
|
||||
<version>${project.version}</version>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp-schemas</artifactId>
|
||||
<version>${project.version}</version>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>eu.dnetlib</groupId>
|
||||
<artifactId>dnet-pace-core</artifactId>
|
||||
</dependency>
|
||||
|
||||
<dependency>
|
||||
<groupId>com.jayway.jsonpath</groupId>
|
||||
<artifactId>json-path</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>dom4j</groupId>
|
||||
<artifactId>dom4j</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>jaxen</groupId>
|
||||
<artifactId>jaxen</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>com.jayway.jsonpath</groupId>
|
||||
<artifactId>json-path</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>dom4j</groupId>
|
||||
<artifactId>dom4j</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>jaxen</groupId>
|
||||
<artifactId>jaxen</artifactId>
|
||||
</dependency>
|
||||
|
||||
<dependency>
|
||||
<groupId>eu.dnetlib</groupId>
|
||||
<artifactId>dnet-openaire-broker-common</artifactId>
|
||||
<version>[2.0.1,3.0.0)</version>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>eu.dnetlib</groupId>
|
||||
<artifactId>dnet-openaire-broker-common</artifactId>
|
||||
<version>[2.0.1,3.0.0)</version>
|
||||
</dependency>
|
||||
|
||||
</dependencies>
|
||||
</dependencies>
|
||||
|
||||
|
||||
</project>
|
|
@ -63,6 +63,9 @@ import eu.dnetlib.dhp.schema.oaf.Publication;
|
|||
import eu.dnetlib.dhp.schema.oaf.Relation;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.dhp.schema.oaf.Software;
|
||||
import eu.dnetlib.dhp.utils.ISLookupClientFactory;
|
||||
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
import scala.Tuple2;
|
||||
|
||||
public class GenerateEventsApplication {
|
||||
|
@ -107,7 +110,10 @@ public class GenerateEventsApplication {
|
|||
private static final UpdateMatcher<Pair<Result, List<eu.dnetlib.dhp.schema.oaf.Dataset>>, ?> enrichMissingDatasetIsSupplementedBy =
|
||||
new EnrichMissingDatasetIsSupplementedBy();
|
||||
|
||||
public static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
|
||||
// Aggregators
|
||||
private static final TypedColumn<Tuple2<Result, Relation>, ResultGroup> resultAggrTypedColumn = new ResultAggregator().toColumn();
|
||||
|
||||
private static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
|
||||
|
||||
public static void main(final String[] args) throws Exception {
|
||||
final ArgumentApplicationParser parser = new ArgumentApplicationParser(
|
||||
|
@ -128,8 +134,16 @@ public class GenerateEventsApplication {
|
|||
final String eventsPath = parser.get("eventsPath");
|
||||
log.info("eventsPath: {}", eventsPath);
|
||||
|
||||
final String isLookupUrl = parser.get("isLookupUrl");
|
||||
log.info("isLookupUrl: {}", isLookupUrl);
|
||||
|
||||
final String dedupConfigProfileId = parser.get("dedupConfProfile");
|
||||
log.info("dedupConfigProfileId: {}", dedupConfigProfileId);
|
||||
|
||||
final SparkConf conf = new SparkConf();
|
||||
|
||||
final DedupConfig dedupConfig = loadDedupConfig(isLookupUrl, dedupConfigProfileId);
|
||||
|
||||
runWithSparkSession(conf, isSparkSessionManaged, spark -> {
|
||||
|
||||
removeOutputDir(spark, eventsPath);
|
||||
|
@ -137,10 +151,10 @@ public class GenerateEventsApplication {
|
|||
final Dataset<Event> all = spark.emptyDataset(Encoders.kryo(Event.class));
|
||||
|
||||
for (final Class<? extends Result> r1 : BrokerConstants.RESULT_CLASSES) {
|
||||
all.union(generateSimpleEvents(spark, graphPath, r1));
|
||||
all.union(generateSimpleEvents(spark, graphPath, r1, dedupConfig));
|
||||
|
||||
for (final Class<? extends Result> r2 : BrokerConstants.RESULT_CLASSES) {
|
||||
all.union(generateRelationEvents(spark, graphPath, r1, r2));
|
||||
all.union(generateRelationEvents(spark, graphPath, r1, r2, dedupConfig));
|
||||
}
|
||||
}
|
||||
|
||||
|
@ -155,38 +169,37 @@ public class GenerateEventsApplication {
|
|||
|
||||
private static <R extends Result> Dataset<Event> generateSimpleEvents(final SparkSession spark,
|
||||
final String graphPath,
|
||||
final Class<R> resultClazz) {
|
||||
final Class<R> resultClazz,
|
||||
final DedupConfig dedupConfig) {
|
||||
|
||||
final Dataset<Result> results = readPath(spark, graphPath + "/" + resultClazz.getSimpleName().toLowerCase(), Result.class)
|
||||
.filter(r -> r.getDataInfo().getDeletedbyinference());
|
||||
|
||||
final Dataset<Relation> rels = readPath(spark, graphPath + "/relation", Relation.class)
|
||||
final Dataset<Relation> mergedRels = readPath(spark, graphPath + "/relation", Relation.class)
|
||||
.filter(r -> r.getRelClass().equals(BrokerConstants.IS_MERGED_IN_CLASS));
|
||||
|
||||
final TypedColumn<Tuple2<Result, Relation>, ResultGroup> aggr = new ResultAggregator().toColumn();
|
||||
|
||||
return results.joinWith(rels, results.col("id").equalTo(rels.col("source")), "inner")
|
||||
return results.joinWith(mergedRels, results.col("id").equalTo(mergedRels.col("source")), "inner")
|
||||
.groupByKey((MapFunction<Tuple2<Result, Relation>, String>) t -> t._2.getTarget(), Encoders.STRING())
|
||||
.agg(aggr)
|
||||
.agg(resultAggrTypedColumn)
|
||||
.map((MapFunction<Tuple2<String, ResultGroup>, ResultGroup>) t -> t._2, Encoders.kryo(ResultGroup.class))
|
||||
.filter(ResultGroup::isValid)
|
||||
.map((MapFunction<ResultGroup, EventGroup>) g -> GenerateEventsApplication.generateSimpleEvents(g), Encoders.kryo(EventGroup.class))
|
||||
.map((MapFunction<ResultGroup, EventGroup>) g -> GenerateEventsApplication.generateSimpleEvents(g, dedupConfig), Encoders.kryo(EventGroup.class))
|
||||
.flatMap(group -> group.getData().iterator(), Encoders.kryo(Event.class));
|
||||
}
|
||||
|
||||
private static EventGroup generateSimpleEvents(final ResultGroup results) {
|
||||
private static EventGroup generateSimpleEvents(final ResultGroup results, final DedupConfig dedupConfig) {
|
||||
final List<UpdateInfo<?>> list = new ArrayList<>();
|
||||
|
||||
for (final Result target : results.getData()) {
|
||||
list.addAll(enrichMissingAbstract.searchUpdatesForRecord(target, results.getData()));
|
||||
list.addAll(enrichMissingAuthorOrcid.searchUpdatesForRecord(target, results.getData()));
|
||||
list.addAll(enrichMissingOpenAccess.searchUpdatesForRecord(target, results.getData()));
|
||||
list.addAll(enrichMissingPid.searchUpdatesForRecord(target, results.getData()));
|
||||
list.addAll(enrichMissingPublicationDate.searchUpdatesForRecord(target, results.getData()));
|
||||
list.addAll(enrichMissingSubject.searchUpdatesForRecord(target, results.getData()));
|
||||
list.addAll(enrichMoreOpenAccess.searchUpdatesForRecord(target, results.getData()));
|
||||
list.addAll(enrichMorePid.searchUpdatesForRecord(target, results.getData()));
|
||||
list.addAll(enrichMoreSubject.searchUpdatesForRecord(target, results.getData()));
|
||||
list.addAll(enrichMissingAbstract.searchUpdatesForRecord(target, results.getData(), dedupConfig));
|
||||
list.addAll(enrichMissingAuthorOrcid.searchUpdatesForRecord(target, results.getData(), dedupConfig));
|
||||
list.addAll(enrichMissingOpenAccess.searchUpdatesForRecord(target, results.getData(), dedupConfig));
|
||||
list.addAll(enrichMissingPid.searchUpdatesForRecord(target, results.getData(), dedupConfig));
|
||||
list.addAll(enrichMissingPublicationDate.searchUpdatesForRecord(target, results.getData(), dedupConfig));
|
||||
list.addAll(enrichMissingSubject.searchUpdatesForRecord(target, results.getData(), dedupConfig));
|
||||
list.addAll(enrichMoreOpenAccess.searchUpdatesForRecord(target, results.getData(), dedupConfig));
|
||||
list.addAll(enrichMorePid.searchUpdatesForRecord(target, results.getData(), dedupConfig));
|
||||
list.addAll(enrichMoreSubject.searchUpdatesForRecord(target, results.getData(), dedupConfig));
|
||||
}
|
||||
|
||||
final EventGroup events = new EventGroup();
|
||||
|
@ -197,9 +210,10 @@ public class GenerateEventsApplication {
|
|||
private static <SRC extends Result, TRG extends OafEntity> Dataset<Event> generateRelationEvents(final SparkSession spark,
|
||||
final String graphPath,
|
||||
final Class<SRC> sourceClass,
|
||||
final Class<TRG> targetClass) {
|
||||
final Class<TRG> targetClass,
|
||||
final DedupConfig dedupConfig) {
|
||||
|
||||
final Dataset<SRC> sources = readPath(spark, graphPath + "/" + sourceClass.getSimpleName().toLowerCase(), sourceClass)
|
||||
final Dataset<Result> sources = readPath(spark, graphPath + "/" + sourceClass.getSimpleName().toLowerCase(), Result.class)
|
||||
.filter(r -> r.getDataInfo().getDeletedbyinference());
|
||||
|
||||
final Dataset<TRG> targets = readPath(spark, graphPath + "/" + sourceClass.getSimpleName().toLowerCase(), targetClass);
|
||||
|
@ -210,6 +224,12 @@ public class GenerateEventsApplication {
|
|||
final Dataset<Relation> rels = readPath(spark, graphPath + "/relation", Relation.class)
|
||||
.filter(r -> !r.getRelClass().equals(BrokerConstants.IS_MERGED_IN_CLASS));
|
||||
|
||||
final Dataset<ResultGroup> duplicates = sources.joinWith(mergedRels, sources.col("id").equalTo(rels.col("source")), "inner")
|
||||
.groupByKey((MapFunction<Tuple2<Result, Relation>, String>) t -> t._2.getTarget(), Encoders.STRING())
|
||||
.agg(resultAggrTypedColumn)
|
||||
.map((MapFunction<Tuple2<String, ResultGroup>, ResultGroup>) t -> t._2, Encoders.kryo(ResultGroup.class))
|
||||
.filter(ResultGroup::isValid);
|
||||
|
||||
if (targetClass == Project.class) {
|
||||
// TODO join using: generateProjectsEvents
|
||||
} else if (targetClass == Software.class) {
|
||||
|
@ -223,29 +243,30 @@ public class GenerateEventsApplication {
|
|||
return null;
|
||||
}
|
||||
|
||||
private List<Event> generateProjectsEvents(final Collection<Pair<Result, List<Project>>> childrenWithProjects) {
|
||||
private List<Event> generateProjectsEvents(final Collection<Pair<Result, List<Project>>> childrenWithProjects, final DedupConfig dedupConfig) {
|
||||
final List<UpdateInfo<?>> list = new ArrayList<>();
|
||||
|
||||
for (final Pair<Result, List<Project>> target : childrenWithProjects) {
|
||||
list.addAll(enrichMissingProject.searchUpdatesForRecord(target, childrenWithProjects));
|
||||
list.addAll(enrichMoreProject.searchUpdatesForRecord(target, childrenWithProjects));
|
||||
list.addAll(enrichMissingProject.searchUpdatesForRecord(target, childrenWithProjects, dedupConfig));
|
||||
list.addAll(enrichMoreProject.searchUpdatesForRecord(target, childrenWithProjects, dedupConfig));
|
||||
}
|
||||
|
||||
return list.stream().map(EventFactory::newBrokerEvent).collect(Collectors.toList());
|
||||
}
|
||||
|
||||
private List<Event> generateSoftwareEvents(final Collection<Pair<Result, List<Software>>> childrenWithSoftwares) {
|
||||
private List<Event> generateSoftwareEvents(final Collection<Pair<Result, List<Software>>> childrenWithSoftwares, final DedupConfig dedupConfig) {
|
||||
final List<UpdateInfo<?>> list = new ArrayList<>();
|
||||
|
||||
for (final Pair<Result, List<Software>> target : childrenWithSoftwares) {
|
||||
list.addAll(enrichMissingSoftware.searchUpdatesForRecord(target, childrenWithSoftwares));
|
||||
list.addAll(enrichMoreSoftware.searchUpdatesForRecord(target, childrenWithSoftwares));
|
||||
list.addAll(enrichMissingSoftware.searchUpdatesForRecord(target, childrenWithSoftwares, dedupConfig));
|
||||
list.addAll(enrichMoreSoftware.searchUpdatesForRecord(target, childrenWithSoftwares, dedupConfig));
|
||||
}
|
||||
return list.stream().map(EventFactory::newBrokerEvent).collect(Collectors.toList());
|
||||
}
|
||||
|
||||
private List<Event> generatePublicationRelatedEvents(final String relType,
|
||||
final Collection<Pair<Result, Map<String, List<Publication>>>> childrenWithRels) {
|
||||
final Collection<Pair<Result, Map<String, List<Publication>>>> childrenWithRels,
|
||||
final DedupConfig dedupConfig) {
|
||||
|
||||
final List<UpdateInfo<?>> list = new ArrayList<>();
|
||||
|
||||
|
@ -258,15 +279,15 @@ public class GenerateEventsApplication {
|
|||
|
||||
for (final Pair<Result, List<Publication>> target : cleanedChildrens) {
|
||||
if (relType.equals("isRelatedTo")) {
|
||||
list.addAll(enrichMisissingPublicationIsRelatedTo.searchUpdatesForRecord(target, cleanedChildrens));
|
||||
list.addAll(enrichMisissingPublicationIsRelatedTo.searchUpdatesForRecord(target, cleanedChildrens, dedupConfig));
|
||||
} else if (relType.equals("references")) {
|
||||
list.addAll(enrichMissingPublicationReferences.searchUpdatesForRecord(target, cleanedChildrens));
|
||||
list.addAll(enrichMissingPublicationReferences.searchUpdatesForRecord(target, cleanedChildrens, dedupConfig));
|
||||
} else if (relType.equals("isReferencedBy")) {
|
||||
list.addAll(enrichMissingPublicationIsReferencedBy.searchUpdatesForRecord(target, cleanedChildrens));
|
||||
list.addAll(enrichMissingPublicationIsReferencedBy.searchUpdatesForRecord(target, cleanedChildrens, dedupConfig));
|
||||
} else if (relType.equals("isSupplementedTo")) {
|
||||
list.addAll(enrichMissingPublicationIsSupplementedTo.searchUpdatesForRecord(target, cleanedChildrens));
|
||||
list.addAll(enrichMissingPublicationIsSupplementedTo.searchUpdatesForRecord(target, cleanedChildrens, dedupConfig));
|
||||
} else if (relType.equals("isSupplementedBy")) {
|
||||
list.addAll(enrichMissingPublicationIsSupplementedBy.searchUpdatesForRecord(target, cleanedChildrens));
|
||||
list.addAll(enrichMissingPublicationIsSupplementedBy.searchUpdatesForRecord(target, cleanedChildrens, dedupConfig));
|
||||
}
|
||||
}
|
||||
|
||||
|
@ -275,7 +296,8 @@ public class GenerateEventsApplication {
|
|||
}
|
||||
|
||||
private List<Event> generateDatasetRelatedEvents(final String relType,
|
||||
final Collection<Pair<Result, Map<String, List<eu.dnetlib.dhp.schema.oaf.Dataset>>>> childrenWithRels) {
|
||||
final Collection<Pair<Result, Map<String, List<eu.dnetlib.dhp.schema.oaf.Dataset>>>> childrenWithRels,
|
||||
final DedupConfig dedupConfig) {
|
||||
|
||||
final List<UpdateInfo<?>> list = new ArrayList<>();
|
||||
|
||||
|
@ -288,15 +310,15 @@ public class GenerateEventsApplication {
|
|||
|
||||
for (final Pair<Result, List<eu.dnetlib.dhp.schema.oaf.Dataset>> target : cleanedChildrens) {
|
||||
if (relType.equals("isRelatedTo")) {
|
||||
list.addAll(enrichMisissingDatasetIsRelatedTo.searchUpdatesForRecord(target, cleanedChildrens));
|
||||
list.addAll(enrichMisissingDatasetIsRelatedTo.searchUpdatesForRecord(target, cleanedChildrens, dedupConfig));
|
||||
} else if (relType.equals("references")) {
|
||||
list.addAll(enrichMissingDatasetReferences.searchUpdatesForRecord(target, cleanedChildrens));
|
||||
list.addAll(enrichMissingDatasetReferences.searchUpdatesForRecord(target, cleanedChildrens, dedupConfig));
|
||||
} else if (relType.equals("isReferencedBy")) {
|
||||
list.addAll(enrichMissingDatasetIsReferencedBy.searchUpdatesForRecord(target, cleanedChildrens));
|
||||
list.addAll(enrichMissingDatasetIsReferencedBy.searchUpdatesForRecord(target, cleanedChildrens, dedupConfig));
|
||||
} else if (relType.equals("isSupplementedTo")) {
|
||||
list.addAll(enrichMissingDatasetIsSupplementedTo.searchUpdatesForRecord(target, cleanedChildrens));
|
||||
list.addAll(enrichMissingDatasetIsSupplementedTo.searchUpdatesForRecord(target, cleanedChildrens, dedupConfig));
|
||||
} else if (relType.equals("isSupplementedBy")) {
|
||||
list.addAll(enrichMissingDatasetIsSupplementedBy.searchUpdatesForRecord(target, cleanedChildrens));
|
||||
list.addAll(enrichMissingDatasetIsSupplementedBy.searchUpdatesForRecord(target, cleanedChildrens, dedupConfig));
|
||||
}
|
||||
}
|
||||
|
||||
|
@ -313,4 +335,20 @@ public class GenerateEventsApplication {
|
|||
.textFile(inputPath)
|
||||
.map((MapFunction<String, R>) value -> OBJECT_MAPPER.readValue(value, clazz), Encoders.bean(clazz));
|
||||
}
|
||||
|
||||
private static DedupConfig loadDedupConfig(final String isLookupUrl, final String profId) throws Exception {
|
||||
final ISLookUpService isLookUpService = ISLookupClientFactory.getLookUpService(isLookupUrl);
|
||||
|
||||
final String conf = isLookUpService.getResourceProfileByQuery(String
|
||||
.format("for $x in /RESOURCE_PROFILE[.//RESOURCE_IDENTIFIER/@value = '%s'] return $x//DEDUPLICATION/text()", profId));
|
||||
|
||||
final DedupConfig dedupConfig = new ObjectMapper().readValue(conf, DedupConfig.class);
|
||||
dedupConfig.getPace().initModel();
|
||||
dedupConfig.getPace().initTranslationMap();
|
||||
// dedupConfig.getWf().setConfigurationId("???");
|
||||
|
||||
return dedupConfig;
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -12,6 +12,7 @@ import org.apache.commons.lang3.StringUtils;
|
|||
|
||||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Field;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public abstract class UpdateMatcher<K, T> {
|
||||
|
||||
|
@ -21,16 +22,15 @@ public abstract class UpdateMatcher<K, T> {
|
|||
this.multipleUpdate = multipleUpdate;
|
||||
}
|
||||
|
||||
public Collection<UpdateInfo<T>> searchUpdatesForRecord(final K res, final Collection<K> others) {
|
||||
public Collection<UpdateInfo<T>> searchUpdatesForRecord(final K res, final Collection<K> others, final DedupConfig dedupConfig) {
|
||||
|
||||
final Map<String, UpdateInfo<T>> infoMap = new HashMap<>();
|
||||
|
||||
for (final K source : others) {
|
||||
if (source != res) {
|
||||
for (final UpdateInfo<T> info : findUpdates(source, res)) {
|
||||
for (final UpdateInfo<T> info : findUpdates(source, res, dedupConfig)) {
|
||||
final String s = DigestUtils.md5Hex(info.getHighlightValueAsString());
|
||||
if (!infoMap.containsKey(s) || infoMap.get(s).getTrust() < info.getTrust()) {
|
||||
} else {
|
||||
if (!infoMap.containsKey(s) || infoMap.get(s).getTrust() < info.getTrust()) {} else {
|
||||
infoMap.put(s, info);
|
||||
}
|
||||
}
|
||||
|
@ -51,11 +51,7 @@ public abstract class UpdateMatcher<K, T> {
|
|||
}
|
||||
}
|
||||
|
||||
protected abstract List<UpdateInfo<T>> findUpdates(K source, K target);
|
||||
|
||||
protected abstract UpdateInfo<T> generateUpdateInfo(final T highlightValue,
|
||||
final K source,
|
||||
final K target);
|
||||
protected abstract List<UpdateInfo<T>> findUpdates(K source, K target, DedupConfig dedupConfig);
|
||||
|
||||
protected static boolean isMissing(final List<Field<String>> list) {
|
||||
return list == null || list.isEmpty() || StringUtils.isBlank(list.get(0).getValue());
|
||||
|
|
|
@ -13,6 +13,7 @@ import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
|
|||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Dataset;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public abstract class AbstractEnrichMissingDataset
|
||||
extends UpdateMatcher<Pair<Result, List<Dataset>>, eu.dnetlib.broker.objects.Dataset> {
|
||||
|
@ -27,7 +28,8 @@ public abstract class AbstractEnrichMissingDataset
|
|||
@Override
|
||||
protected final List<UpdateInfo<eu.dnetlib.broker.objects.Dataset>> findUpdates(
|
||||
final Pair<Result, List<Dataset>> source,
|
||||
final Pair<Result, List<Dataset>> target) {
|
||||
final Pair<Result, List<Dataset>> target,
|
||||
final DedupConfig dedupConfig) {
|
||||
|
||||
final Set<String> existingDatasets = target
|
||||
.getRight()
|
||||
|
@ -40,21 +42,22 @@ public abstract class AbstractEnrichMissingDataset
|
|||
.stream()
|
||||
.filter(d -> !existingDatasets.contains(d.getId()))
|
||||
.map(ConversionUtils::oafDatasetToBrokerDataset)
|
||||
.map(i -> generateUpdateInfo(i, source, target))
|
||||
.map(i -> generateUpdateInfo(i, source, target, dedupConfig))
|
||||
.collect(Collectors.toList());
|
||||
|
||||
}
|
||||
|
||||
@Override
|
||||
protected final UpdateInfo<eu.dnetlib.broker.objects.Dataset> generateUpdateInfo(
|
||||
final eu.dnetlib.broker.objects.Dataset highlightValue,
|
||||
final Pair<Result, List<Dataset>> source,
|
||||
final Pair<Result, List<Dataset>> target) {
|
||||
final Pair<Result, List<Dataset>> target,
|
||||
final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
getTopic(),
|
||||
highlightValue, source.getLeft(), target.getLeft(),
|
||||
(p, rel) -> p.getDatasets().add(rel),
|
||||
rel -> rel.getInstances().get(0).getUrl());
|
||||
rel -> rel.getInstances().get(0).getUrl(),
|
||||
dedupConfig);
|
||||
}
|
||||
|
||||
public Topic getTopic() {
|
||||
|
|
|
@ -13,6 +13,7 @@ import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
|
|||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Project;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMissingProject
|
||||
extends UpdateMatcher<Pair<Result, List<Project>>, eu.dnetlib.broker.objects.Project> {
|
||||
|
@ -23,7 +24,8 @@ public class EnrichMissingProject
|
|||
|
||||
@Override
|
||||
protected List<UpdateInfo<eu.dnetlib.broker.objects.Project>> findUpdates(final Pair<Result, List<Project>> source,
|
||||
final Pair<Result, List<Project>> target) {
|
||||
final Pair<Result, List<Project>> target,
|
||||
final DedupConfig dedupConfig) {
|
||||
|
||||
if (source.getRight().isEmpty()) {
|
||||
return Arrays.asList();
|
||||
|
@ -32,21 +34,21 @@ public class EnrichMissingProject
|
|||
.getRight()
|
||||
.stream()
|
||||
.map(ConversionUtils::oafProjectToBrokerProject)
|
||||
.map(p -> generateUpdateInfo(p, source, target))
|
||||
.map(p -> generateUpdateInfo(p, source, target, dedupConfig))
|
||||
.collect(Collectors.toList());
|
||||
}
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<eu.dnetlib.broker.objects.Project> generateUpdateInfo(
|
||||
final eu.dnetlib.broker.objects.Project highlightValue,
|
||||
final Pair<Result, List<Project>> source,
|
||||
final Pair<Result, List<Project>> target) {
|
||||
final Pair<Result, List<Project>> target,
|
||||
final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
Topic.ENRICH_MISSING_PROJECT,
|
||||
highlightValue, source.getLeft(), target.getLeft(),
|
||||
(p, prj) -> p.getProjects().add(prj),
|
||||
prj -> prj.getFunder() + "::" + prj.getFundingProgram() + prj.getCode());
|
||||
prj -> prj.getFunder() + "::" + prj.getFundingProgram() + prj.getCode(), dedupConfig);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -13,6 +13,7 @@ import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
|
|||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Project;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMoreProject extends UpdateMatcher<Pair<Result, List<Project>>, eu.dnetlib.broker.objects.Project> {
|
||||
|
||||
|
@ -22,7 +23,8 @@ public class EnrichMoreProject extends UpdateMatcher<Pair<Result, List<Project>>
|
|||
|
||||
@Override
|
||||
protected List<UpdateInfo<eu.dnetlib.broker.objects.Project>> findUpdates(final Pair<Result, List<Project>> source,
|
||||
final Pair<Result, List<Project>> target) {
|
||||
final Pair<Result, List<Project>> target,
|
||||
final DedupConfig dedupConfig) {
|
||||
|
||||
final Set<String> existingProjects = source
|
||||
.getRight()
|
||||
|
@ -35,20 +37,20 @@ public class EnrichMoreProject extends UpdateMatcher<Pair<Result, List<Project>>
|
|||
.stream()
|
||||
.filter(p -> !existingProjects.contains(p.getId()))
|
||||
.map(ConversionUtils::oafProjectToBrokerProject)
|
||||
.map(p -> generateUpdateInfo(p, source, target))
|
||||
.map(p -> generateUpdateInfo(p, source, target, dedupConfig))
|
||||
.collect(Collectors.toList());
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<eu.dnetlib.broker.objects.Project> generateUpdateInfo(
|
||||
final eu.dnetlib.broker.objects.Project highlightValue,
|
||||
final Pair<Result, List<Project>> source,
|
||||
final Pair<Result, List<Project>> target) {
|
||||
final Pair<Result, List<Project>> target,
|
||||
final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
Topic.ENRICH_MORE_PROJECT,
|
||||
highlightValue, source.getLeft(), target.getLeft(),
|
||||
(p, prj) -> p.getProjects().add(prj),
|
||||
prj -> prj.getFunder() + "::" + prj.getFundingProgram() + prj.getCode());
|
||||
prj -> prj.getFunder() + "::" + prj.getFundingProgram() + prj.getCode(), dedupConfig);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -13,6 +13,7 @@ import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
|
|||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Publication;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public abstract class AbstractEnrichMissingPublication
|
||||
extends UpdateMatcher<Pair<Result, List<Publication>>, eu.dnetlib.broker.objects.Publication> {
|
||||
|
@ -27,7 +28,8 @@ public abstract class AbstractEnrichMissingPublication
|
|||
@Override
|
||||
protected final List<UpdateInfo<eu.dnetlib.broker.objects.Publication>> findUpdates(
|
||||
final Pair<Result, List<Publication>> source,
|
||||
final Pair<Result, List<Publication>> target) {
|
||||
final Pair<Result, List<Publication>> target,
|
||||
final DedupConfig dedupConfig) {
|
||||
|
||||
final Set<String> existingPublications = target
|
||||
.getRight()
|
||||
|
@ -40,21 +42,21 @@ public abstract class AbstractEnrichMissingPublication
|
|||
.stream()
|
||||
.filter(d -> !existingPublications.contains(d.getId()))
|
||||
.map(ConversionUtils::oafResultToBrokerPublication)
|
||||
.map(i -> generateUpdateInfo(i, source, target))
|
||||
.map(i -> generateUpdateInfo(i, source, target, dedupConfig))
|
||||
.collect(Collectors.toList());
|
||||
|
||||
}
|
||||
|
||||
@Override
|
||||
protected final UpdateInfo<eu.dnetlib.broker.objects.Publication> generateUpdateInfo(
|
||||
final eu.dnetlib.broker.objects.Publication highlightValue,
|
||||
final Pair<Result, List<Publication>> source,
|
||||
final Pair<Result, List<Publication>> target) {
|
||||
final Pair<Result, List<Publication>> target,
|
||||
final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
getTopic(),
|
||||
highlightValue, source.getLeft(), target.getLeft(),
|
||||
(p, rel) -> p.getPublications().add(rel),
|
||||
rel -> rel.getInstances().get(0).getUrl());
|
||||
rel -> rel.getInstances().get(0).getUrl(), dedupConfig);
|
||||
}
|
||||
|
||||
public Topic getTopic() {
|
||||
|
|
|
@ -13,6 +13,7 @@ import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
|
|||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.dhp.schema.oaf.Software;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMissingSoftware
|
||||
extends UpdateMatcher<Pair<Result, List<Software>>, eu.dnetlib.broker.objects.Software> {
|
||||
|
@ -24,7 +25,8 @@ public class EnrichMissingSoftware
|
|||
@Override
|
||||
protected List<UpdateInfo<eu.dnetlib.broker.objects.Software>> findUpdates(
|
||||
final Pair<Result, List<Software>> source,
|
||||
final Pair<Result, List<Software>> target) {
|
||||
final Pair<Result, List<Software>> target,
|
||||
final DedupConfig dedupConfig) {
|
||||
|
||||
if (source.getRight().isEmpty()) {
|
||||
return Arrays.asList();
|
||||
|
@ -33,21 +35,21 @@ public class EnrichMissingSoftware
|
|||
.getRight()
|
||||
.stream()
|
||||
.map(ConversionUtils::oafSoftwareToBrokerSoftware)
|
||||
.map(p -> generateUpdateInfo(p, source, target))
|
||||
.map(p -> generateUpdateInfo(p, source, target, dedupConfig))
|
||||
.collect(Collectors.toList());
|
||||
}
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<eu.dnetlib.broker.objects.Software> generateUpdateInfo(
|
||||
final eu.dnetlib.broker.objects.Software highlightValue,
|
||||
final Pair<Result, List<Software>> source,
|
||||
final Pair<Result, List<Software>> target) {
|
||||
final Pair<Result, List<Software>> target,
|
||||
final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
Topic.ENRICH_MISSING_SOFTWARE,
|
||||
highlightValue, source.getLeft(), target.getLeft(),
|
||||
(p, s) -> p.getSoftwares().add(s),
|
||||
s -> s.getName());
|
||||
s -> s.getName(), dedupConfig);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -13,6 +13,7 @@ import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
|
|||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.dhp.schema.oaf.Software;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMoreSoftware
|
||||
extends UpdateMatcher<Pair<Result, List<Software>>, eu.dnetlib.broker.objects.Software> {
|
||||
|
@ -24,7 +25,8 @@ public class EnrichMoreSoftware
|
|||
@Override
|
||||
protected List<UpdateInfo<eu.dnetlib.broker.objects.Software>> findUpdates(
|
||||
final Pair<Result, List<Software>> source,
|
||||
final Pair<Result, List<Software>> target) {
|
||||
final Pair<Result, List<Software>> target,
|
||||
final DedupConfig dedupConfig) {
|
||||
|
||||
final Set<String> existingSoftwares = source
|
||||
.getRight()
|
||||
|
@ -37,20 +39,20 @@ public class EnrichMoreSoftware
|
|||
.stream()
|
||||
.filter(p -> !existingSoftwares.contains(p.getId()))
|
||||
.map(ConversionUtils::oafSoftwareToBrokerSoftware)
|
||||
.map(p -> generateUpdateInfo(p, source, target))
|
||||
.map(p -> generateUpdateInfo(p, source, target, dedupConfig))
|
||||
.collect(Collectors.toList());
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<eu.dnetlib.broker.objects.Software> generateUpdateInfo(
|
||||
final eu.dnetlib.broker.objects.Software highlightValue,
|
||||
final Pair<Result, List<Software>> source,
|
||||
final Pair<Result, List<Software>> target) {
|
||||
final Pair<Result, List<Software>> target,
|
||||
final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
Topic.ENRICH_MORE_SOFTWARE,
|
||||
highlightValue, source.getLeft(), target.getLeft(),
|
||||
(p, s) -> p.getSoftwares().add(s),
|
||||
s -> s.getName());
|
||||
s -> s.getName(), dedupConfig);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -9,6 +9,7 @@ import eu.dnetlib.dhp.broker.model.Topic;
|
|||
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
|
||||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMissingAbstract extends UpdateMatcher<Result, String> {
|
||||
|
||||
|
@ -17,22 +18,22 @@ public class EnrichMissingAbstract extends UpdateMatcher<Result, String> {
|
|||
}
|
||||
|
||||
@Override
|
||||
protected List<UpdateInfo<String>> findUpdates(final Result source, final Result target) {
|
||||
protected List<UpdateInfo<String>> findUpdates(final Result source, final Result target, final DedupConfig dedupConfig) {
|
||||
if (isMissing(target.getDescription()) && !isMissing(source.getDescription())) {
|
||||
return Arrays.asList(generateUpdateInfo(source.getDescription().get(0).getValue(), source, target));
|
||||
return Arrays.asList(generateUpdateInfo(source.getDescription().get(0).getValue(), source, target, dedupConfig));
|
||||
}
|
||||
return new ArrayList<>();
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<String> generateUpdateInfo(final String highlightValue,
|
||||
final Result source,
|
||||
final Result target) {
|
||||
final Result target,
|
||||
final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
Topic.ENRICH_MISSING_ABSTRACT,
|
||||
highlightValue, source, target,
|
||||
(p, s) -> p.getAbstracts().add(s),
|
||||
s -> s);
|
||||
s -> s, dedupConfig);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -10,6 +10,7 @@ import eu.dnetlib.dhp.broker.model.Topic;
|
|||
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
|
||||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMissingAuthorOrcid extends UpdateMatcher<Result, Pair<String, String>> {
|
||||
|
||||
|
@ -18,19 +19,21 @@ public class EnrichMissingAuthorOrcid extends UpdateMatcher<Result, Pair<String,
|
|||
}
|
||||
|
||||
@Override
|
||||
protected List<UpdateInfo<Pair<String, String>>> findUpdates(final Result source, final Result target) {
|
||||
protected List<UpdateInfo<Pair<String, String>>> findUpdates(final Result source, final Result target, final DedupConfig dedupConfig) {
|
||||
// TODO
|
||||
// return Arrays.asList(new EnrichMissingAbstract("xxxxxxx", 0.9f));
|
||||
return Arrays.asList();
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<Pair<String, String>> generateUpdateInfo(final Pair<String, String> highlightValue,
|
||||
final Result source,
|
||||
final Result target) {
|
||||
final Result target,
|
||||
final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
Topic.ENRICH_MISSING_AUTHOR_ORCID,
|
||||
highlightValue, source, target,
|
||||
(p, pair) -> p.getCreators().add(pair.getLeft() + " - ORCID: " + pair.getRight()),
|
||||
pair -> pair.getLeft() + "::" + pair.getRight());
|
||||
pair -> pair.getLeft() + "::" + pair.getRight(),
|
||||
dedupConfig);
|
||||
}
|
||||
}
|
||||
|
|
|
@ -12,6 +12,7 @@ import eu.dnetlib.dhp.broker.oa.util.BrokerConstants;
|
|||
import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
|
||||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMissingOpenAccess extends UpdateMatcher<Result, Instance> {
|
||||
|
||||
|
@ -20,7 +21,7 @@ public class EnrichMissingOpenAccess extends UpdateMatcher<Result, Instance> {
|
|||
}
|
||||
|
||||
@Override
|
||||
protected List<UpdateInfo<Instance>> findUpdates(final Result source, final Result target) {
|
||||
protected List<UpdateInfo<Instance>> findUpdates(final Result source, final Result target, final DedupConfig dedupConfig) {
|
||||
final long count = target
|
||||
.getInstance()
|
||||
.stream()
|
||||
|
@ -28,9 +29,7 @@ public class EnrichMissingOpenAccess extends UpdateMatcher<Result, Instance> {
|
|||
.filter(right -> right.equals(BrokerConstants.OPEN_ACCESS))
|
||||
.count();
|
||||
|
||||
if (count > 0) {
|
||||
return Arrays.asList();
|
||||
}
|
||||
if (count > 0) { return Arrays.asList(); }
|
||||
|
||||
return source
|
||||
.getInstance()
|
||||
|
@ -38,19 +37,19 @@ public class EnrichMissingOpenAccess extends UpdateMatcher<Result, Instance> {
|
|||
.filter(i -> i.getAccessright().getClassid().equals(BrokerConstants.OPEN_ACCESS))
|
||||
.map(ConversionUtils::oafInstanceToBrokerInstances)
|
||||
.flatMap(List::stream)
|
||||
.map(i -> generateUpdateInfo(i, source, target))
|
||||
.map(i -> generateUpdateInfo(i, source, target, dedupConfig))
|
||||
.collect(Collectors.toList());
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<Instance> generateUpdateInfo(final Instance highlightValue,
|
||||
final Result source,
|
||||
final Result target) {
|
||||
final Result target,
|
||||
final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
Topic.ENRICH_MISSING_OA_VERSION,
|
||||
highlightValue, source, target,
|
||||
(p, i) -> p.getInstances().add(i),
|
||||
Instance::getUrl);
|
||||
Instance::getUrl, dedupConfig);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -11,6 +11,7 @@ import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
|
|||
import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
|
||||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMissingPid extends UpdateMatcher<Result, Pid> {
|
||||
|
||||
|
@ -19,28 +20,25 @@ public class EnrichMissingPid extends UpdateMatcher<Result, Pid> {
|
|||
}
|
||||
|
||||
@Override
|
||||
protected List<UpdateInfo<Pid>> findUpdates(final Result source, final Result target) {
|
||||
protected List<UpdateInfo<Pid>> findUpdates(final Result source, final Result target, final DedupConfig dedupConfig) {
|
||||
final long count = target.getPid().size();
|
||||
|
||||
if (count > 0) {
|
||||
return Arrays.asList();
|
||||
}
|
||||
if (count > 0) { return Arrays.asList(); }
|
||||
|
||||
return source
|
||||
.getPid()
|
||||
.stream()
|
||||
.map(ConversionUtils::oafPidToBrokerPid)
|
||||
.map(i -> generateUpdateInfo(i, source, target))
|
||||
.map(i -> generateUpdateInfo(i, source, target, dedupConfig))
|
||||
.collect(Collectors.toList());
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<Pid> generateUpdateInfo(final Pid highlightValue, final Result source, final Result target) {
|
||||
public UpdateInfo<Pid> generateUpdateInfo(final Pid highlightValue, final Result source, final Result target, final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
Topic.ENRICH_MISSING_PID,
|
||||
highlightValue, source, target,
|
||||
(p, pid) -> p.getPids().add(pid),
|
||||
pid -> pid.getType() + "::" + pid.getValue());
|
||||
pid -> pid.getType() + "::" + pid.getValue(), dedupConfig);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -9,6 +9,7 @@ import eu.dnetlib.dhp.broker.model.Topic;
|
|||
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
|
||||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMissingPublicationDate extends UpdateMatcher<Result, String> {
|
||||
|
||||
|
@ -17,22 +18,22 @@ public class EnrichMissingPublicationDate extends UpdateMatcher<Result, String>
|
|||
}
|
||||
|
||||
@Override
|
||||
protected List<UpdateInfo<String>> findUpdates(final Result source, final Result target) {
|
||||
protected List<UpdateInfo<String>> findUpdates(final Result source, final Result target, final DedupConfig dedupConfig) {
|
||||
if (isMissing(target.getDateofacceptance()) && !isMissing(source.getDateofacceptance())) {
|
||||
return Arrays.asList(generateUpdateInfo(source.getDateofacceptance().getValue(), source, target));
|
||||
return Arrays.asList(generateUpdateInfo(source.getDateofacceptance().getValue(), source, target, dedupConfig));
|
||||
}
|
||||
return new ArrayList<>();
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<String> generateUpdateInfo(final String highlightValue,
|
||||
final Result source,
|
||||
final Result target) {
|
||||
final Result target,
|
||||
final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
Topic.ENRICH_MISSING_PUBLICATION_DATE,
|
||||
highlightValue, source, target,
|
||||
(p, date) -> p.setPublicationdate(date),
|
||||
s -> s);
|
||||
s -> s, dedupConfig);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -14,6 +14,7 @@ import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
|||
import eu.dnetlib.dhp.schema.oaf.Qualifier;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.dhp.schema.oaf.StructuredProperty;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMissingSubject extends UpdateMatcher<Result, Pair<String, String>> {
|
||||
|
||||
|
@ -22,7 +23,7 @@ public class EnrichMissingSubject extends UpdateMatcher<Result, Pair<String, Str
|
|||
}
|
||||
|
||||
@Override
|
||||
protected List<UpdateInfo<Pair<String, String>>> findUpdates(final Result source, final Result target) {
|
||||
protected List<UpdateInfo<Pair<String, String>>> findUpdates(final Result source, final Result target, final DedupConfig dedupConfig) {
|
||||
final Set<String> existingTypes = target
|
||||
.getSubject()
|
||||
.stream()
|
||||
|
@ -35,20 +36,20 @@ public class EnrichMissingSubject extends UpdateMatcher<Result, Pair<String, Str
|
|||
.stream()
|
||||
.filter(pid -> !existingTypes.contains(pid.getQualifier().getClassid()))
|
||||
.map(ConversionUtils::oafSubjectToPair)
|
||||
.map(i -> generateUpdateInfo(i, source, target))
|
||||
.map(i -> generateUpdateInfo(i, source, target, dedupConfig))
|
||||
.collect(Collectors.toList());
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<Pair<String, String>> generateUpdateInfo(final Pair<String, String> highlightValue,
|
||||
final Result source,
|
||||
final Result target) {
|
||||
final Result target,
|
||||
final DedupConfig dedupConfig) {
|
||||
|
||||
return new UpdateInfo<>(
|
||||
Topic.fromPath("ENRICH/MISSING/SUBJECT/" + highlightValue.getLeft()),
|
||||
highlightValue, source, target,
|
||||
(p, pair) -> p.getSubjects().add(pair.getRight()),
|
||||
pair -> pair.getLeft() + "::" + pair.getRight());
|
||||
pair -> pair.getLeft() + "::" + pair.getRight(), dedupConfig);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -12,6 +12,7 @@ import eu.dnetlib.dhp.broker.oa.util.BrokerConstants;
|
|||
import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
|
||||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMoreOpenAccess extends UpdateMatcher<Result, Instance> {
|
||||
|
||||
|
@ -20,7 +21,7 @@ public class EnrichMoreOpenAccess extends UpdateMatcher<Result, Instance> {
|
|||
}
|
||||
|
||||
@Override
|
||||
protected List<UpdateInfo<Instance>> findUpdates(final Result source, final Result target) {
|
||||
protected List<UpdateInfo<Instance>> findUpdates(final Result source, final Result target, final DedupConfig dedupConfig) {
|
||||
final Set<String> urls = target
|
||||
.getInstance()
|
||||
.stream()
|
||||
|
@ -36,19 +37,19 @@ public class EnrichMoreOpenAccess extends UpdateMatcher<Result, Instance> {
|
|||
.map(ConversionUtils::oafInstanceToBrokerInstances)
|
||||
.flatMap(List::stream)
|
||||
.filter(i -> !urls.contains(i.getUrl()))
|
||||
.map(i -> generateUpdateInfo(i, source, target))
|
||||
.map(i -> generateUpdateInfo(i, source, target, dedupConfig))
|
||||
.collect(Collectors.toList());
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<Instance> generateUpdateInfo(final Instance highlightValue,
|
||||
final Result source,
|
||||
final Result target) {
|
||||
final Result target,
|
||||
final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
Topic.ENRICH_MORE_OA_VERSION,
|
||||
highlightValue, source, target,
|
||||
(p, i) -> p.getInstances().add(i),
|
||||
Instance::getUrl);
|
||||
Instance::getUrl, dedupConfig);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -11,6 +11,7 @@ import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
|
|||
import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
|
||||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMorePid extends UpdateMatcher<Result, Pid> {
|
||||
|
||||
|
@ -19,7 +20,7 @@ public class EnrichMorePid extends UpdateMatcher<Result, Pid> {
|
|||
}
|
||||
|
||||
@Override
|
||||
protected List<UpdateInfo<Pid>> findUpdates(final Result source, final Result target) {
|
||||
protected List<UpdateInfo<Pid>> findUpdates(final Result source, final Result target, final DedupConfig dedupConfig) {
|
||||
final Set<String> existingPids = target
|
||||
.getPid()
|
||||
.stream()
|
||||
|
@ -31,17 +32,16 @@ public class EnrichMorePid extends UpdateMatcher<Result, Pid> {
|
|||
.stream()
|
||||
.filter(pid -> !existingPids.contains(pid.getQualifier().getClassid() + "::" + pid.getValue()))
|
||||
.map(ConversionUtils::oafPidToBrokerPid)
|
||||
.map(i -> generateUpdateInfo(i, source, target))
|
||||
.map(i -> generateUpdateInfo(i, source, target, dedupConfig))
|
||||
.collect(Collectors.toList());
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<Pid> generateUpdateInfo(final Pid highlightValue, final Result source, final Result target) {
|
||||
public UpdateInfo<Pid> generateUpdateInfo(final Pid highlightValue, final Result source, final Result target, final DedupConfig dedupConfig) {
|
||||
return new UpdateInfo<>(
|
||||
Topic.ENRICH_MORE_PID,
|
||||
highlightValue, source, target,
|
||||
(p, pid) -> p.getPids().add(pid),
|
||||
pid -> pid.getType() + "::" + pid.getValue());
|
||||
pid -> pid.getType() + "::" + pid.getValue(), dedupConfig);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -12,6 +12,7 @@ import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
|
|||
import eu.dnetlib.dhp.broker.oa.util.ConversionUtils;
|
||||
import eu.dnetlib.dhp.broker.oa.util.UpdateInfo;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
|
||||
public class EnrichMoreSubject extends UpdateMatcher<Result, Pair<String, String>> {
|
||||
|
||||
|
@ -20,7 +21,7 @@ public class EnrichMoreSubject extends UpdateMatcher<Result, Pair<String, String
|
|||
}
|
||||
|
||||
@Override
|
||||
protected List<UpdateInfo<Pair<String, String>>> findUpdates(final Result source, final Result target) {
|
||||
protected List<UpdateInfo<Pair<String, String>>> findUpdates(final Result source, final Result target, final DedupConfig dedupConfig) {
|
||||
final Set<String> existingSubjects = target
|
||||
.getSubject()
|
||||
.stream()
|
||||
|
@ -32,20 +33,20 @@ public class EnrichMoreSubject extends UpdateMatcher<Result, Pair<String, String
|
|||
.stream()
|
||||
.filter(pid -> !existingSubjects.contains(pid.getQualifier().getClassid() + "::" + pid.getValue()))
|
||||
.map(ConversionUtils::oafSubjectToPair)
|
||||
.map(i -> generateUpdateInfo(i, source, target))
|
||||
.map(i -> generateUpdateInfo(i, source, target, dedupConfig))
|
||||
.collect(Collectors.toList());
|
||||
}
|
||||
|
||||
@Override
|
||||
public UpdateInfo<Pair<String, String>> generateUpdateInfo(final Pair<String, String> highlightValue,
|
||||
final Result source,
|
||||
final Result target) {
|
||||
final Result target,
|
||||
final DedupConfig dedupConfig) {
|
||||
|
||||
return new UpdateInfo<>(
|
||||
Topic.fromPath("ENRICH/MORE/SUBJECT/" + highlightValue.getLeft()),
|
||||
highlightValue, source, target,
|
||||
(p, pair) -> p.getSubjects().add(pair.getRight()),
|
||||
pair -> pair.getLeft() + "::" + pair.getRight());
|
||||
pair -> pair.getLeft() + "::" + pair.getRight(), dedupConfig);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
|
@ -15,6 +15,9 @@ public class BrokerConstants {
|
|||
public static final String OPEN_ACCESS = "OPEN";
|
||||
public static final String IS_MERGED_IN_CLASS = "isMergedIn";
|
||||
|
||||
public static final float MIN_TRUST = 0.25f;
|
||||
public static final float MAX_TRUST = 1.00f;
|
||||
|
||||
public static final List<Class<? extends Result>> RESULT_CLASSES = Arrays
|
||||
.asList(Publication.class, Dataset.class, Software.class, OtherResearchProduct.class);
|
||||
|
||||
|
|
|
@ -0,0 +1,15 @@
|
|||
package eu.dnetlib.dhp.broker.oa.util;
|
||||
|
||||
public class TrustUtils {
|
||||
|
||||
public static float rescale(final double score, final double threshold) {
|
||||
if (score >= BrokerConstants.MAX_TRUST) { return BrokerConstants.MAX_TRUST; }
|
||||
|
||||
final double val = (score - threshold) * (BrokerConstants.MAX_TRUST - BrokerConstants.MIN_TRUST) / (BrokerConstants.MAX_TRUST - threshold);
|
||||
|
||||
if (val < BrokerConstants.MIN_TRUST) { return BrokerConstants.MIN_TRUST; }
|
||||
if (val > BrokerConstants.MAX_TRUST) { return BrokerConstants.MAX_TRUST; }
|
||||
|
||||
return (float) val;
|
||||
}
|
||||
}
|
|
@ -5,6 +5,11 @@ import java.util.List;
|
|||
import java.util.function.BiConsumer;
|
||||
import java.util.function.Function;
|
||||
|
||||
import org.slf4j.Logger;
|
||||
import org.slf4j.LoggerFactory;
|
||||
|
||||
import com.fasterxml.jackson.databind.ObjectMapper;
|
||||
|
||||
import eu.dnetlib.broker.objects.OpenAireEventPayload;
|
||||
import eu.dnetlib.broker.objects.Provenance;
|
||||
import eu.dnetlib.broker.objects.Publication;
|
||||
|
@ -12,6 +17,10 @@ import eu.dnetlib.dhp.broker.model.Topic;
|
|||
import eu.dnetlib.dhp.schema.oaf.Instance;
|
||||
import eu.dnetlib.dhp.schema.oaf.KeyValue;
|
||||
import eu.dnetlib.dhp.schema.oaf.Result;
|
||||
import eu.dnetlib.pace.config.DedupConfig;
|
||||
import eu.dnetlib.pace.model.MapDocument;
|
||||
import eu.dnetlib.pace.tree.support.TreeProcessor;
|
||||
import eu.dnetlib.pace.util.MapDocumentUtil;
|
||||
|
||||
public final class UpdateInfo<T> {
|
||||
|
||||
|
@ -29,16 +38,19 @@ public final class UpdateInfo<T> {
|
|||
|
||||
private final float trust;
|
||||
|
||||
private static final Logger log = LoggerFactory.getLogger(UpdateInfo.class);
|
||||
|
||||
public UpdateInfo(final Topic topic, final T highlightValue, final Result source, final Result target,
|
||||
final BiConsumer<Publication, T> compileHighlight,
|
||||
final Function<T, String> highlightToString) {
|
||||
final Function<T, String> highlightToString,
|
||||
final DedupConfig dedupConfig) {
|
||||
this.topic = topic;
|
||||
this.highlightValue = highlightValue;
|
||||
this.source = source;
|
||||
this.target = target;
|
||||
this.compileHighlight = compileHighlight;
|
||||
this.highlightToString = highlightToString;
|
||||
this.trust = calculateTrust(source, target);
|
||||
this.trust = calculateTrust(dedupConfig, source, target);
|
||||
}
|
||||
|
||||
public T getHighlightValue() {
|
||||
|
@ -53,9 +65,20 @@ public final class UpdateInfo<T> {
|
|||
return target;
|
||||
}
|
||||
|
||||
private float calculateTrust(final Result source, final Result target) {
|
||||
// TODO
|
||||
return 0.9f;
|
||||
private float calculateTrust(final DedupConfig dedupConfig, final Result r1, final Result r2) {
|
||||
try {
|
||||
final ObjectMapper objectMapper = new ObjectMapper();
|
||||
final MapDocument doc1 = MapDocumentUtil.asMapDocumentWithJPath(dedupConfig, objectMapper.writeValueAsString(r1));
|
||||
final MapDocument doc2 = MapDocumentUtil.asMapDocumentWithJPath(dedupConfig, objectMapper.writeValueAsString(r2));
|
||||
|
||||
final double score = new TreeProcessor(dedupConfig).computeScore(doc1, doc2);
|
||||
final double threshold = dedupConfig.getWf().getThreshold();
|
||||
|
||||
return TrustUtils.rescale(score, threshold);
|
||||
} catch (final Exception e) {
|
||||
log.error("Error computing score between results", e);
|
||||
return BrokerConstants.MIN_TRUST;
|
||||
}
|
||||
}
|
||||
|
||||
protected Topic getTopic() {
|
||||
|
@ -95,8 +118,7 @@ public final class UpdateInfo<T> {
|
|||
.map(Instance::getUrl)
|
||||
.flatMap(List::stream)
|
||||
.findFirst()
|
||||
.orElse(null);
|
||||
;
|
||||
.orElse(null);;
|
||||
|
||||
final Provenance provenance = new Provenance().setId(provId).setRepositoryName(provRepo).setUrl(provUrl);
|
||||
|
||||
|
|
|
@ -0,0 +1,72 @@
|
|||
package eu.dnetlib.dhp.broker.oa.util;
|
||||
|
||||
import static org.junit.jupiter.api.Assertions.assertTrue;
|
||||
|
||||
import org.junit.jupiter.api.Test;
|
||||
|
||||
public class TrustUtilsTest {
|
||||
|
||||
private static final double THRESHOLD = 0.95;
|
||||
|
||||
@Test
|
||||
public void rescaleTest_1() {
|
||||
verifyValue(-0.3, BrokerConstants.MIN_TRUST);
|
||||
}
|
||||
|
||||
@Test
|
||||
public void rescaleTest_2() {
|
||||
verifyValue(0.0, BrokerConstants.MIN_TRUST);
|
||||
}
|
||||
|
||||
@Test
|
||||
public void rescaleTest_3() {
|
||||
verifyValue(0.5, BrokerConstants.MIN_TRUST);
|
||||
}
|
||||
|
||||
@Test
|
||||
public void rescaleTest_4() {
|
||||
verifyValue(0.95, BrokerConstants.MIN_TRUST);
|
||||
}
|
||||
|
||||
@Test
|
||||
public void rescaleTest_5() {
|
||||
verifyValue(0.96, BrokerConstants.MIN_TRUST);
|
||||
}
|
||||
|
||||
@Test
|
||||
public void rescaleTest_6() {
|
||||
verifyValue(0.97, 0.3f);
|
||||
}
|
||||
|
||||
@Test
|
||||
public void rescaleTest_7() {
|
||||
verifyValue(0.98, 0.45f);
|
||||
}
|
||||
|
||||
@Test
|
||||
public void rescaleTest_8() {
|
||||
verifyValue(0.99, 0.6f);
|
||||
}
|
||||
|
||||
@Test
|
||||
public void rescaleTest_9() {
|
||||
verifyValue(1.00, BrokerConstants.MAX_TRUST);
|
||||
}
|
||||
|
||||
@Test
|
||||
public void rescaleTest_10() {
|
||||
verifyValue(1.01, BrokerConstants.MAX_TRUST);
|
||||
}
|
||||
|
||||
@Test
|
||||
public void rescaleTest_11() {
|
||||
verifyValue(2.00, BrokerConstants.MAX_TRUST);
|
||||
}
|
||||
|
||||
private void verifyValue(final double originalScore, final float expectedTrust) {
|
||||
final float trust = TrustUtils.rescale(originalScore, THRESHOLD);
|
||||
System.out.println(trust);
|
||||
assertTrue(Math.abs(trust - expectedTrust) < 0.01);
|
||||
}
|
||||
|
||||
}
|
Loading…
Reference in New Issue