Examples with EntityNameCleanerProvider io.openk9.entity.manager.api.EntityNameCleanerProvider used on opensource projects

Example 1 with EntityNameCleanerProvider

use of io.openk9.entity.manager.api.EntityNameCleanerProvider in project openk9 by smclab.

the class CreateEntitiesRunnable method getEntityCandidates.

private EntityCandidates getEntityCandidates(EntityNameCleanerProvider entityNameCleanerProvider, EntityService entityService, EntityMember ingestionIdEntityMember, Entity ingestionIdEntity) {
    EntityNameCleaner entityNameCleaner = entityNameCleanerProvider.get(ingestionIdEntity.getType());
    QueryBuilder queryBuilder = entityNameCleaner.cleanEntityName(ingestionIdEntity.getTenantId(), ingestionIdEntity.getName());
    List<EntityIndex> candidates = entityService.search(ingestionIdEntity.getTenantId(), queryBuilder, 0, 10);
    return EntityCandidates.of(ingestionIdEntityMember, candidates);
}

Example 2 with EntityNameCleanerProvider

use of io.openk9.entity.manager.api.EntityNameCleanerProvider 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);
}

Example 3 with EntityNameCleanerProvider

use of io.openk9.entity.manager.api.EntityNameCleanerProvider 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;
}