Search in sources :

Example 11 with Entity

use of io.openk9.entity.manager.cache.model.Entity in project openk9 by smclab.

the class CreateEntitiesRunnable method run_.

@Override
public void run_() {
    _log.info("start CreateEntitiesRunnable");
    IMap<EntityKey, Entity> entityIMap = MapUtil.getEntityMap(_hazelcastInstance);
    IMap<AssociableEntityKey, Entity> associableEntityMap = MapUtil.getAssociableEntityMap(_hazelcastInstance);
    Set<EntityKey> entityKeys = entityIMap.localKeySet(Predicates.and(Predicates.equal("id", null), Predicates.equal("graphId", null)));
    EntityGraphConfig config = CDI.current().select(EntityGraphConfig.class).get();
    EntityNameCleanerProvider entityNameCleanerProvider = CDI.current().select(EntityNameCleanerProvider.class).get();
    EntityService entityService = CDI.current().select(EntityService.class).get();
    EntityGraphService entityGraphService = CDI.current().select(EntityGraphService.class).get();
    Map<EntityKey, Entity> localEntityMap = entityIMap.getAll(entityKeys);
    Collection<Entity> localEntityValues = localEntityMap.values();
    Set<EntityKey> localEntityKeys = localEntityMap.keySet();
    List<Member> collect = _hazelcastInstance.getCluster().getMembers().stream().filter(member -> !member.localMember()).collect(Collectors.toList());
    String[] ingestionIds = localEntityKeys.stream().map(EntityKey::getIngestionId).distinct().toArray(String[]::new);
    IExecutorService entityExecutor = _hazelcastInstance.getExecutorService("entityExecutor");
    Map<Member, Future<Map<EntityKey, Entity>>> memberFutureMap = entityExecutor.submitToMembers(new GetEntitiesCallable(ingestionIds), collect);
    Map<EntityKey, Entity> otherEntityKeyEntityMap = memberFutureMap.values().stream().map(FutureUtil::makeCompletableFuture).map(CompletableFuture::join).reduce((a, b) -> {
        Map<EntityKey, Entity> map = new HashMap<>();
        map.putAll(a);
        map.putAll(b);
        return map;
    }).orElseGet(Map::of);
    Stream<EntityMember> otherEntityMemberStream = otherEntityKeyEntityMap.values().stream().map(entity -> EntityMember.of(entity, false));
    Stream<EntityMember> localEntityMemberStream = localEntityValues.stream().map(entity -> EntityMember.of(entity, true));
    Map<String, List<EntityMember>> entitiesGroupingByIngestionId = Stream.concat(localEntityMemberStream, otherEntityMemberStream).collect(Collectors.groupingBy(entityMember -> entityMember.getEntity().getIngestionId()));
    Collection<List<EntityMember>> values = entitiesGroupingByIngestionId.values();
    Map<EntityKey, Entity> entityMap = new HashMap<>();
    for (List<EntityMember> ingestionIdEntities : values) {
        Map<AssociableEntityKey, Entity> localAssociableEntityMap = new HashMap<>();
        List<EntityCandidates> entityCandidateList = new ArrayList<>();
        for (EntityMember ingestionIdEntity : ingestionIdEntities) {
            Entity innerEntity = ingestionIdEntity.getEntity();
            entityCandidateList.add(getEntityCandidates(entityNameCleanerProvider, entityService, ingestionIdEntity, innerEntity));
        }
        List<Mono<Entity>> completableFutureList = entityCandidateList.stream().filter(entityCandidates -> entityCandidates.getEntity().isLocal()).map(entityCandidates -> Mono.fromSupplier(_getAndCreateEntityDisambiguate(config, entityNameCleanerProvider, entityService, entityGraphService, entityCandidateList, entityCandidates, entityCandidates.getEntity())).subscribeOn(Schedulers.boundedElastic())).collect(Collectors.toList());
        Mono<List<Entity>> zip = Mono.zip(completableFutureList, a -> {
            List<Entity> entities = new ArrayList<>();
            for (Object o : a) {
                entities.add((Entity) o);
            }
            return entities;
        }).defaultIfEmpty(List.of());
        for (Entity currentEntityRequest : zip.block()) {
            localAssociableEntityMap.put(AssociableEntityKey.of(currentEntityRequest.getCacheId(), currentEntityRequest.getIngestionId()), currentEntityRequest);
            entityMap.put(EntityKey.of(currentEntityRequest.getTenantId(), currentEntityRequest.getName(), currentEntityRequest.getType(), currentEntityRequest.getCacheId(), currentEntityRequest.getIngestionId()), currentEntityRequest);
        }
        associableEntityMap.setAll(localAssociableEntityMap);
    }
    entityIMap.setAll(entityMap);
}
Also used : Arrays(java.util.Arrays) Member(com.hazelcast.cluster.Member) FutureUtil(io.openk9.entity.manager.util.FutureUtil) EntityKey(io.openk9.entity.manager.cache.model.EntityKey) SymbolicName(org.neo4j.cypherdsl.core.SymbolicName) Future(java.util.concurrent.Future) Map(java.util.Map) ToString(lombok.ToString) Property(org.neo4j.cypherdsl.core.Property) Cypher(org.neo4j.cypherdsl.core.Cypher) EntityIndex(io.openk9.entity.manager.model.index.EntityIndex) EntityGraph(io.openk9.entity.manager.model.graph.EntityGraph) Collection(java.util.Collection) CDI(javax.enterprise.inject.spi.CDI) Set(java.util.Set) Collectors(java.util.stream.Collectors) Serializable(java.io.Serializable) Objects(java.util.Objects) List(java.util.List) Stream(java.util.stream.Stream) Builder(lombok.Builder) Optional(java.util.Optional) Cypher.literalOf(org.neo4j.cypherdsl.core.Cypher.literalOf) HazelcastInstanceAware(com.hazelcast.core.HazelcastInstanceAware) AssociableEntityKey(io.openk9.entity.manager.cache.model.AssociableEntityKey) EntityGraphConfig(io.openk9.entity.manager.config.EntityGraphConfig) Logger(org.jboss.logging.Logger) HashMap(java.util.HashMap) CompletableFuture(java.util.concurrent.CompletableFuture) Supplier(java.util.function.Supplier) ArrayList(java.util.ArrayList) Statement(org.neo4j.cypherdsl.core.Statement) Schedulers(reactor.core.scheduler.Schedulers) EntityNameCleanerProvider(io.openk9.entity.manager.cleaner.EntityNameCleanerProvider) Node(org.neo4j.cypherdsl.core.Node) HazelcastInstance(com.hazelcast.core.HazelcastInstance) QueryBuilder(org.elasticsearch.index.query.QueryBuilder) EntityNameCleaner(io.openk9.entity.manager.cleaner.EntityNameCleaner) Entity(io.openk9.entity.manager.cache.model.Entity) Functions(org.neo4j.cypherdsl.core.Functions) Mono(reactor.core.publisher.Mono) IExecutorService(com.hazelcast.core.IExecutorService) GetEntitiesCallable(io.openk9.entity.manager.action.GetEntitiesCallable) AliasedExpression(org.neo4j.cypherdsl.core.AliasedExpression) Predicates(com.hazelcast.query.Predicates) EntityService(io.openk9.entity.manager.service.index.EntityService) Data(lombok.Data) AllArgsConstructor(lombok.AllArgsConstructor) EntityGraphService(io.openk9.entity.manager.service.graph.EntityGraphService) Collections(java.util.Collections) MapUtil(io.openk9.entity.manager.util.MapUtil) IMap(com.hazelcast.map.IMap) Entity(io.openk9.entity.manager.cache.model.Entity) EntityGraphService(io.openk9.entity.manager.service.graph.EntityGraphService) HashMap(java.util.HashMap) ArrayList(java.util.ArrayList) ToString(lombok.ToString) IExecutorService(com.hazelcast.core.IExecutorService) EntityKey(io.openk9.entity.manager.cache.model.EntityKey) AssociableEntityKey(io.openk9.entity.manager.cache.model.AssociableEntityKey) CompletableFuture(java.util.concurrent.CompletableFuture) GetEntitiesCallable(io.openk9.entity.manager.action.GetEntitiesCallable) List(java.util.List) ArrayList(java.util.ArrayList) EntityGraphConfig(io.openk9.entity.manager.config.EntityGraphConfig) Member(com.hazelcast.cluster.Member) EntityService(io.openk9.entity.manager.service.index.EntityService) Mono(reactor.core.publisher.Mono) EntityNameCleanerProvider(io.openk9.entity.manager.cleaner.EntityNameCleanerProvider) Future(java.util.concurrent.Future) CompletableFuture(java.util.concurrent.CompletableFuture) AssociableEntityKey(io.openk9.entity.manager.cache.model.AssociableEntityKey) Map(java.util.Map) HashMap(java.util.HashMap) IMap(com.hazelcast.map.IMap)

Example 12 with Entity

use of io.openk9.entity.manager.cache.model.Entity in project openk9 by smclab.

the class CreateEntitiesRunnable method cleanCandidates.

private List<EntityIndex> cleanCandidates(Entity entityRequest, List<EntityIndex> candidates, EntityNameCleanerProvider entityNameCleanerProvider, float scoreThreshold) {
    if (!candidates.isEmpty()) {
        EntityIndex documentEntityResponse = candidates.get(0);
        double bestScore;
        if (candidates.size() > 1) {
            double[] scores = candidates.stream().mapToDouble(EntityIndex::getScore).toArray();
            bestScore = _softmax(documentEntityResponse.getScore(), scores);
        } else {
            bestScore = _levenshteinDistance(entityNameCleanerProvider.get(documentEntityResponse.getType()).cleanEntityName(documentEntityResponse.getName()), entityNameCleanerProvider.get(entityRequest.getType()).cleanEntityName(entityRequest.getName()));
        }
        if (bestScore > scoreThreshold) {
            return Collections.singletonList(documentEntityResponse);
        }
    }
    return candidates;
}
Also used : EntityIndex(io.openk9.entity.manager.model.index.EntityIndex)

Example 13 with Entity

use of io.openk9.entity.manager.cache.model.Entity in project openk9 by smclab.

the class CreateRelationRunnable method run_.

@Override
public void run_() {
    IMap<EntityKey, Entity> entityIMap = MapUtil.getEntityMap(_hazelcastInstance);
    Set<EntityKey> entityKeys = entityIMap.localKeySet(Predicates.and(Predicates.notEqual("id", null), Predicates.notEqual("graphId", null)));
    _log.info("entityKeys: " + entityKeys.size());
    Map<EntityKey, Entity> entityIMapAll = entityIMap.getAll(entityKeys);
    _log.info("entityIMapAll: " + entityIMapAll.size());
    IMap<EntityRelationKey, EntityRelation> entityRelationMap = MapUtil.getEntityRelationMap(_hazelcastInstance);
    Map<String, String> collect = entityIMapAll.values().stream().collect(Collectors.toMap(Entity::getCacheId, Entity::getId));
    String[] cacheIds = entityKeys.stream().map(EntityKey::getCacheId).distinct().toArray(String[]::new);
    Map<EntityRelationKey, EntityRelation> entries = entityRelationMap.getAll(entityRelationMap.keySet(Predicates.in("__key.entityId", cacheIds)));
    _log.info("entityRelations: " + entries.size());
    EntityGraphService entityGraphService = CDI.current().select(EntityGraphService.class).get();
    List<EntityRelationKey> entityRelationKeysToDelete = new ArrayList<>();
    for (Map.Entry<EntityRelationKey, EntityRelation> entry : entries.entrySet()) {
        EntityRelationKey key = entry.getKey();
        EntityRelation value = entry.getValue();
        String from = collect.get(value.getEntityCacheId());
        String to = collect.get(value.getTo());
        if (from != null && to != null) {
            try {
                entityGraphService.createRelationship(from, to, value.getName());
                entityRelationKeysToDelete.add(key);
            } catch (Exception e) {
                _log.error(e.getMessage(), e);
            }
        }
    }
    try {
        Pipelining pipelining = new Pipelining<>(10);
        for (EntityRelationKey entityRelationKey : entityRelationKeysToDelete) {
            pipelining.add(entityRelationMap.removeAsync(entityRelationKey));
        }
        pipelining.results();
    } catch (Exception e) {
        _log.error(e.getMessage(), e);
    }
}
Also used : Entity(io.openk9.entity.manager.cache.model.Entity) EntityGraphService(io.openk9.entity.manager.service.graph.EntityGraphService) ArrayList(java.util.ArrayList) EntityRelationKey(io.openk9.entity.manager.cache.model.EntityRelationKey) EntityKey(io.openk9.entity.manager.cache.model.EntityKey) EntityRelation(io.openk9.entity.manager.cache.model.EntityRelation) Pipelining(com.hazelcast.core.Pipelining) Map(java.util.Map) IMap(com.hazelcast.map.IMap)

Example 14 with Entity

use of io.openk9.entity.manager.cache.model.Entity in project openk9 by smclab.

the class EntityManagerConsumer method consume.

@Incoming("entity-manager-request")
@Outgoing("entity-manager-response")
@Blocking
public Message<JsonObject> consume(Object obj) throws InterruptedException {
    JsonObject jsonObject = obj instanceof JsonObject ? (JsonObject) obj : new JsonObject(new String((byte[]) obj));
    Payload payload = jsonObject.mapTo(Payload.class);
    _entityManagerQueue.offer(payload, 45, TimeUnit.SECONDS);
    String replyTo = payload.getReplyTo();
    return Message.of(jsonObject, Metadata.of(new OutgoingRabbitMQMetadata.Builder().withRoutingKey(replyTo).withTimestamp(ZonedDateTime.now()).build()));
}
Also used : OutgoingRabbitMQMetadata(io.smallrye.reactive.messaging.rabbitmq.OutgoingRabbitMQMetadata) JsonObject(io.vertx.core.json.JsonObject) Payload(io.openk9.entity.manager.dto.Payload) Incoming(org.eclipse.microprofile.reactive.messaging.Incoming) Blocking(io.smallrye.reactive.messaging.annotations.Blocking) Outgoing(org.eclipse.microprofile.reactive.messaging.Outgoing)

Example 15 with Entity

use of io.openk9.entity.manager.cache.model.Entity in project openk9 by smclab.

the class EntityService method search.

public List<EntityIndex> search(long tenantId, QueryBuilder queryBuilder, int from, int size) {
    SearchRequest searchRequest = new SearchRequest(tenantId + "-entity");
    SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder();
    searchSourceBuilder.query(queryBuilder);
    searchSourceBuilder.size(from);
    searchSourceBuilder.size(size);
    searchRequest.source(searchSourceBuilder);
    try {
        SearchResponse searchResponse = _restHighLevelClient.search(searchRequest, RequestOptions.DEFAULT);
        SearchHits hits = searchResponse.getHits();
        List<EntityIndex> results = new ArrayList<>(hits.getHits().length);
        for (SearchHit hit : hits.getHits()) {
            String sourceAsString = hit.getSourceAsString();
            JsonObject json = new JsonObject(sourceAsString);
            EntityIndex entityIndex = json.mapTo(EntityIndex.class);
            entityIndex.setScore(hit.getScore());
            results.add(entityIndex);
        }
        return results;
    } catch (Exception e) {
        _logger.error(e.getMessage());
    }
    return List.of();
}
Also used : SearchRequest(org.elasticsearch.action.search.SearchRequest) SearchHit(org.elasticsearch.search.SearchHit) ArrayList(java.util.ArrayList) JsonObject(io.vertx.core.json.JsonObject) EntityIndex(io.openk9.entity.manager.model.index.EntityIndex) SearchHits(org.elasticsearch.search.SearchHits) IOException(java.io.IOException) SearchSourceBuilder(org.elasticsearch.search.builder.SearchSourceBuilder) SearchResponse(org.elasticsearch.action.search.SearchResponse)

Aggregations

ArrayList (java.util.ArrayList)11 List (java.util.List)11 Mono (reactor.core.publisher.Mono)10 Collectors (java.util.stream.Collectors)9 Node (org.neo4j.cypherdsl.core.Node)8 Statement (org.neo4j.cypherdsl.core.Statement)8 JsonFactory (io.openk9.json.api.JsonFactory)7 Map (java.util.Map)7 Cypher (org.neo4j.cypherdsl.core.Cypher)7 Cypher.literalOf (org.neo4j.cypherdsl.core.Cypher.literalOf)7 Functions (org.neo4j.cypherdsl.core.Functions)7 Activate (org.osgi.service.component.annotations.Activate)7 Component (org.osgi.service.component.annotations.Component)7 Reference (org.osgi.service.component.annotations.Reference)7 Entity (io.openk9.entity.manager.cache.model.Entity)6 Entity (io.openk9.entity.manager.model.Entity)6 EntityIndex (io.openk9.entity.manager.model.index.EntityIndex)6 Logger (org.slf4j.Logger)6 LoggerFactory (org.slf4j.LoggerFactory)6 EntityKey (io.openk9.entity.manager.cache.model.EntityKey)5