Search in sources :

Example 1 with ReferenceTermContext

use of eu.europeana.enrichment.api.internal.ReferenceTermContext in project metis-framework by europeana.

the class MetisRecordParser method parseReferences.

@Override
public Set<ReferenceTermContext> parseReferences(RDF rdf) {
    // Get all direct references (also look in Europeana proxy as it may have been dereferenced - we
    // use this below to follow sameAs links).
    final List<ProxyType> proxies = Optional.ofNullable(rdf.getProxyList()).stream().flatMap(Collection::stream).filter(Objects::nonNull).collect(Collectors.toList());
    final Map<String, Set<ProxyFieldType>> directReferences = new HashMap<>();
    for (ProxyFieldType field : ProxyFieldType.values()) {
        final Set<String> directLinks = proxies.stream().map(field::extractFieldLinksForEnrichment).flatMap(Set::stream).collect(Collectors.toSet());
        for (String directLink : directLinks) {
            directReferences.computeIfAbsent(directLink, key -> new HashSet<>()).add(field);
        }
    }
    // Get all sameAs links from the directly referenced contextual entities.
    final Map<String, Set<ProxyFieldType>> indirectReferences = new HashMap<>();
    final Consumer<AboutType> contextualTypeProcessor = contextualClass -> {
        final Set<ProxyFieldType> linkTypes = Optional.ofNullable(directReferences.get(contextualClass.getAbout())).orElseGet(Collections::emptySet);
        if (!linkTypes.isEmpty()) {
            for (String sameAsLink : getSameAsLinks(contextualClass)) {
                indirectReferences.computeIfAbsent(sameAsLink, key -> new HashSet<>()).addAll(linkTypes);
            }
        }
    };
    Optional.ofNullable(rdf.getAgentList()).orElseGet(Collections::emptyList).forEach(contextualTypeProcessor);
    Optional.ofNullable(rdf.getConceptList()).orElseGet(Collections::emptyList).forEach(contextualTypeProcessor);
    Optional.ofNullable(rdf.getPlaceList()).orElseGet(Collections::emptyList).forEach(contextualTypeProcessor);
    Optional.ofNullable(rdf.getTimeSpanList()).orElseGet(Collections::emptyList).forEach(contextualTypeProcessor);
    // Merge the two maps.
    final Map<String, Set<ProxyFieldType>> resultMap = mergeMapInto(directReferences, indirectReferences);
    // Clean up the result: no null values. But objects we already have need to
    // stay: maybe they are matched using a sameAs link.
    resultMap.remove(null);
    // Convert and done
    final Set<ReferenceTermContext> result = new HashSet<>();
    for (Map.Entry<String, Set<ProxyFieldType>> entry : resultMap.entrySet()) {
        ReferenceTermContext value;
        try {
            value = new ReferenceTermContext(new URL(entry.getKey()), entry.getValue());
            result.add(value);
        } catch (MalformedURLException e) {
            LOGGER.debug("Invalid enrichment reference found: {}", entry.getKey());
        }
    }
    return result;
}
Also used : Concept(eu.europeana.metis.schema.jibx.Concept) URL(java.net.URL) AggregationFieldType(eu.europeana.enrichment.api.internal.AggregationFieldType) TimeSpanType(eu.europeana.metis.schema.jibx.TimeSpanType) LoggerFactory(org.slf4j.LoggerFactory) RecordParser(eu.europeana.enrichment.api.internal.RecordParser) HashMap(java.util.HashMap) RDF(eu.europeana.metis.schema.jibx.RDF) StringUtils(org.apache.commons.lang3.StringUtils) ProxyFieldType(eu.europeana.enrichment.api.internal.ProxyFieldType) ResourceType(eu.europeana.metis.schema.jibx.ResourceType) HashSet(java.util.HashSet) AboutType(eu.europeana.metis.schema.jibx.AboutType) RdfEntityUtils(eu.europeana.enrichment.utils.RdfEntityUtils) FieldType(eu.europeana.enrichment.api.internal.FieldType) SearchTermContext(eu.europeana.enrichment.api.internal.SearchTermContext) Map(java.util.Map) FieldValue(eu.europeana.enrichment.api.internal.FieldValue) Logger(org.slf4j.Logger) MalformedURLException(java.net.MalformedURLException) PlaceType(eu.europeana.metis.schema.jibx.PlaceType) ReferenceTermContext(eu.europeana.enrichment.api.internal.ReferenceTermContext) Collection(java.util.Collection) Set(java.util.Set) Collectors(java.util.stream.Collectors) AgentType(eu.europeana.metis.schema.jibx.AgentType) Objects(java.util.Objects) Consumer(java.util.function.Consumer) List(java.util.List) ProxyType(eu.europeana.metis.schema.jibx.ProxyType) Optional(java.util.Optional) Collections(java.util.Collections) MalformedURLException(java.net.MalformedURLException) HashSet(java.util.HashSet) Set(java.util.Set) HashMap(java.util.HashMap) ReferenceTermContext(eu.europeana.enrichment.api.internal.ReferenceTermContext) URL(java.net.URL) ProxyFieldType(eu.europeana.enrichment.api.internal.ProxyFieldType) Collection(java.util.Collection) ProxyType(eu.europeana.metis.schema.jibx.ProxyType) HashMap(java.util.HashMap) Map(java.util.Map) HashSet(java.util.HashSet) AboutType(eu.europeana.metis.schema.jibx.AboutType)

Aggregations

AggregationFieldType (eu.europeana.enrichment.api.internal.AggregationFieldType)1 FieldType (eu.europeana.enrichment.api.internal.FieldType)1 FieldValue (eu.europeana.enrichment.api.internal.FieldValue)1 ProxyFieldType (eu.europeana.enrichment.api.internal.ProxyFieldType)1 RecordParser (eu.europeana.enrichment.api.internal.RecordParser)1 ReferenceTermContext (eu.europeana.enrichment.api.internal.ReferenceTermContext)1 SearchTermContext (eu.europeana.enrichment.api.internal.SearchTermContext)1 RdfEntityUtils (eu.europeana.enrichment.utils.RdfEntityUtils)1 AboutType (eu.europeana.metis.schema.jibx.AboutType)1 AgentType (eu.europeana.metis.schema.jibx.AgentType)1 Concept (eu.europeana.metis.schema.jibx.Concept)1 PlaceType (eu.europeana.metis.schema.jibx.PlaceType)1 ProxyType (eu.europeana.metis.schema.jibx.ProxyType)1 RDF (eu.europeana.metis.schema.jibx.RDF)1 ResourceType (eu.europeana.metis.schema.jibx.ResourceType)1 TimeSpanType (eu.europeana.metis.schema.jibx.TimeSpanType)1 MalformedURLException (java.net.MalformedURLException)1 URL (java.net.URL)1 Collection (java.util.Collection)1 Collections (java.util.Collections)1