Search in sources :

Example 1 with MediaSegmentDescriptor

use of org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor in project cineast by vitrivr.

the class CineastQueryService method getSimilar.

// TODO This has enormous code duplication with the TemporalQueryMessageHandler
@Override
public void getSimilar(CineastGrpc.TemporalQuery query, StreamObserver<CineastGrpc.QueryResult> responseObserver) {
    StopWatch watch = StopWatch.createStarted();
    MediaSegmentReader mediaSegmentReader = new MediaSegmentReader(Config.sharedConfig().getDatabase().getSelectorSupplier().get());
    MediaObjectReader mediaObjectReader = new MediaObjectReader(Config.sharedConfig().getDatabase().getSelectorSupplier().get());
    MediaSegmentMetadataReader segmentMetadataReader = new MediaSegmentMetadataReader(Config.sharedConfig().getDatabase().getSelectorSupplier().get());
    MediaObjectMetadataReader objectMetadataReader = new MediaObjectMetadataReader(Config.sharedConfig().getDatabase().getSelectorSupplier().get());
    Set<String> sentSegmentIds = new HashSet<>(), sentObjectIds = new HashSet<>();
    CineastGrpc.QueryConfig config = query.getQueryList().get(0).getConfig();
    ReadableQueryConfig rqconf = QueryContainerUtil.queryConfig(config);
    QueryConfig qconf = new QueryConfig(rqconf);
    /* Prepare QueryConfig (so as to obtain a QueryId). */
    final String uuid = qconf.getQueryId().toString();
    final int max = qconf.getMaxResults().orElse(Config.sharedConfig().getRetriever().getMaxResults());
    qconf.setMaxResults(max);
    final int resultsPerModule = qconf.getRawResultsPerModule() == -1 ? Config.sharedConfig().getRetriever().getMaxResultsPerModule() : qconf.getResultsPerModule();
    qconf.setResultsPerModule(resultsPerModule);
    List<Thread> metadataRetrievalThreads = new ArrayList<>();
    /* We iterate over all components independently, because they have a temporal context.*/
    for (int containerIdx = 0; containerIdx < query.getQueryCount(); containerIdx++) {
        List<QueryStage> stages = QueryContainerUtil.query(query.getQueryList().get(containerIdx));
        /* We make a new stagedQueryConfig per stage because the relevant segments will differ for each stage. This also resets the filter (relevant ids in the config)*/
        QueryConfig stageQConf = QueryConfig.clone(qconf);
        /* For the first stage, there will be no relevant segments when querying. This is ok because the retrieval engine handles this appropriately */
        HashSet<String> relevantSegments = new HashSet<>();
        /* Store for each queryterm per category all results to be sent at a later time */
        List<Map<String, List<StringDoublePair>>> cache = new ArrayList<>();
        /* For the terms of a stage, ordering matters. The assumption is that each term is used as a filter for its successor */
        for (int stageIndex = 0; stageIndex < stages.size(); stageIndex++) {
            /* Initalize stage with this hashmap */
            cache.add(stageIndex, new HashMap<>());
            QueryStage stage = stages.get(stageIndex);
            List<Thread> qtThreads = new ArrayList<>();
            /* We now iterate over all QueryTerms for this stage, simply adding their results to the list of relevant segments for the next querystage.
         * The list is only updated once we've iterated over all terms
         */
            for (int i = 0; i < stage.getQueryTerms().size(); i++) {
                QueryTerm qt = stage.getQueryTerms().get(i);
                final int finalContainerIdx = containerIdx;
                final int finalStageIndex = stageIndex;
                Thread qtRetrievalThread = new Thread(() -> {
                    /* Prepare QueryTerm and perform sanity-checks */
                    if (qt == null) {
                        /* In rare instances, it is possible to have null as query stage. If this happens to you, please report this to the developers so we can try to fix it. */
                        LOGGER.warn("QueryTerm was null for stage {}", stage);
                        return;
                    }
                    AbstractQueryTermContainer qc = qt.getContainer();
                    if (qc == null) {
                        LOGGER.warn("Likely an empty query, as it could not be converted to a query container. Ignoring it");
                        return;
                    }
                    List<Thread> categoryThreads = new ArrayList<>();
                    /* For each category of a specific queryterm, we actually go and retrieve. Be aware that we do not change the relevant ids after this call */
                    for (String category : qt.getCategories()) {
                        /* Merge partial results with score-map */
                        List<SegmentScoreElement> scores = continuousRetrievalLogic.retrieve(qc, category, stageQConf);
                        /* Transform raw results into list of StringDoublePairs (segmentId -> score) */
                        final List<StringDoublePair> results = scores.stream().map(elem -> new StringDoublePair(elem.getSegmentId(), elem.getScore())).filter(p -> p.value > 0d).sorted(StringDoublePair.COMPARATOR).limit(max).collect(Collectors.toList());
                        if (results.isEmpty()) {
                            LOGGER.warn("No results found for category {} and qt {} in stage with id {}. Full compoment: {}", category, qt, finalContainerIdx, stage);
                        }
                        if (cache.get(finalStageIndex).containsKey(category)) {
                            LOGGER.error("Category {} was used twice in stage {}. This erases the results of the previous category... ", category, finalStageIndex);
                        }
                        cache.get(finalStageIndex).put(category, results);
                        results.forEach(res -> relevantSegments.add(res.key));
                        LOGGER.trace("Category {} at stage {} executed @ {} ms", category, finalStageIndex, watch.getTime(TimeUnit.MILLISECONDS));
                        /* If this is the last stage, we can send relevant results per category back to the UI.
               * Otherwise, we cannot since we might send results to the UI which would be filtered at a later stage
               */
                        if (finalStageIndex == stages.size() - 1) {
                            /* Finalize and submit per-container results */
                            responseObserver.onNext(QueryContainerUtil.queryResult(QueryContainerUtil.similarityQueryResult(qt.getQueryConfig().getQueryId().toString(), category, results)));
                            List<String> segmentIds = results.stream().map(x -> x.key).filter(x -> !sentSegmentIds.contains(x)).collect(Collectors.toList());
                            if (segmentIds.isEmpty()) {
                                continue;
                            }
                            Map<String, MediaSegmentDescriptor> segments = mediaSegmentReader.lookUpSegments(segmentIds);
                            responseObserver.onNext(QueryContainerUtil.queryResult(CineastGrpc.MediaSegmentQueryResult.newBuilder().addAllSegments(segments.values().stream().map(MediaSegmentUtil::fromMediaSegmentDescriptor).collect(Collectors.toList())).build()));
                            List<MediaSegmentMetadataDescriptor> segmentMetaData = segmentMetadataReader.lookupMultimediaMetadata(segmentIds);
                            responseObserver.onNext(QueryContainerUtil.queryResult(CineastGrpc.MediaSegmentMetaDataQueryResult.newBuilder().addAllSegmentMetaData(segmentMetaData.stream().map(QueryContainerUtil::mediaSegmentMetaData).collect(Collectors.toList())).build()));
                            sentSegmentIds.addAll(segmentIds);
                            List<String> objectIds = segments.values().stream().map(MediaSegmentDescriptor::getObjectId).filter(x -> !sentObjectIds.contains(x)).collect(Collectors.toList());
                            if (objectIds.isEmpty()) {
                                continue;
                            }
                            Map<String, MediaObjectDescriptor> objects = mediaObjectReader.lookUpObjects(objectIds);
                            responseObserver.onNext(QueryContainerUtil.queryResult(CineastGrpc.MediaObjectQueryResult.newBuilder().addAllObjects(objects.values().stream().map(MediaObjectUtil::fromMediaObjectDescriptor).collect(Collectors.toList())).build()));
                            List<MediaObjectMetadataDescriptor> objectMetaData = objectMetadataReader.lookupMultimediaMetadata(objectIds);
                            responseObserver.onNext(QueryContainerUtil.queryResult(CineastGrpc.MediaObjectMetaDataQueryResult.newBuilder().addAllObjectMetaData(objectMetaData.stream().map(QueryContainerUtil::mediaObjectMetaData).collect(Collectors.toList())).build()));
                            sentObjectIds.addAll(objectIds);
                        }
                    }
                /* We're done for this querycontainer */
                });
                // TODO Better name
                qtRetrievalThread.setName("qt-stage" + stageIndex + "-" + qt.getCategories());
                qtThreads.add(qtRetrievalThread);
                qtRetrievalThread.start();
            }
            for (Thread thread : qtThreads) {
                try {
                    thread.join();
                } catch (InterruptedException e) {
                    e.printStackTrace();
                }
            }
            /* After we are done with a stage, we add all relevant segments to the config for the next stage. */
            if (relevantSegments.size() == 0) {
                LOGGER.warn("No relevant segments anymore, aborting staged querying");
                /* Clear relevant segments (there are none) */
                stageQConf.setRelevantSegmentIds(relevantSegments);
                break;
            }
            stageQConf.setRelevantSegmentIds(relevantSegments);
            relevantSegments.clear();
        }
        /* At this point, we have iterated over all stages. Now, we need to go back for all stages and send the results for the relevant ids. */
        for (int stageIndex = 0; stageIndex < stages.size() - 1; stageIndex++) {
            cache.get(stageIndex).forEach((category, results) -> {
                results.removeIf(pair -> !stageQConf.getRelevantSegmentIds().contains(pair.key));
                responseObserver.onNext(QueryContainerUtil.queryResult(QueryContainerUtil.similarityQueryResult(// TODO This assumes that all queries in a temporalquery have the same uuid
                uuid, category, results)));
            });
        }
    /* There should be no carry-over from this block since temporal queries are executed independently */
    }
    /* At this point, all StagedQueries have been executed for this TemporalQuery.
     * Since results have always been sent for the final stage or, when appropriate, in intermediate steps, there's nothing left to do.
     */
    responseObserver.onCompleted();
    mediaSegmentReader.close();
    mediaObjectReader.close();
    segmentMetadataReader.close();
    watch.stop();
    LOGGER.debug("Query executed in {} ms", watch.getTime(TimeUnit.MILLISECONDS));
}
Also used : MediaObjectUtil(org.vitrivr.cineast.api.grpc.util.MediaObjectUtil) AbstractQueryTermContainer(org.vitrivr.cineast.core.data.query.containers.AbstractQueryTermContainer) HashMap(java.util.HashMap) MediaSegmentUtil(org.vitrivr.cineast.api.grpc.util.MediaSegmentUtil) QueryStage(org.vitrivr.cineast.api.grpc.data.QueryStage) MediaObjectMetadataDescriptor(org.vitrivr.cineast.core.data.entities.MediaObjectMetadataDescriptor) MediaObjectDescriptor(org.vitrivr.cineast.core.data.entities.MediaObjectDescriptor) ArrayList(java.util.ArrayList) HashSet(java.util.HashSet) StreamObserver(io.grpc.stub.StreamObserver) MediaObjectReader(org.vitrivr.cineast.core.db.dao.reader.MediaObjectReader) QueryContainerUtil(org.vitrivr.cineast.api.grpc.util.QueryContainerUtil) Map(java.util.Map) MediaSegmentDescriptor(org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor) ContinuousRetrievalLogic(org.vitrivr.cineast.standalone.util.ContinuousRetrievalLogic) MediaSegmentReader(org.vitrivr.cineast.core.db.dao.reader.MediaSegmentReader) MediaObjectMetadataReader(org.vitrivr.cineast.core.db.dao.reader.MediaObjectMetadataReader) QueryConfig(org.vitrivr.cineast.core.config.QueryConfig) Set(java.util.Set) StopWatch(org.apache.commons.lang3.time.StopWatch) ReadableQueryConfig(org.vitrivr.cineast.core.config.ReadableQueryConfig) StringDoublePair(org.vitrivr.cineast.core.data.StringDoublePair) Collectors(java.util.stream.Collectors) TimeUnit(java.util.concurrent.TimeUnit) QueryUtil(org.vitrivr.cineast.api.util.QueryUtil) List(java.util.List) Logger(org.apache.logging.log4j.Logger) SegmentScoreElement(org.vitrivr.cineast.core.data.score.SegmentScoreElement) MediaSegmentMetadataDescriptor(org.vitrivr.cineast.core.data.entities.MediaSegmentMetadataDescriptor) MediaSegmentMetadataReader(org.vitrivr.cineast.core.db.dao.reader.MediaSegmentMetadataReader) LogManager(org.apache.logging.log4j.LogManager) QueryTerm(org.vitrivr.cineast.api.grpc.data.QueryTerm) Config(org.vitrivr.cineast.standalone.config.Config) MediaObjectDescriptor(org.vitrivr.cineast.core.data.entities.MediaObjectDescriptor) AbstractQueryTermContainer(org.vitrivr.cineast.core.data.query.containers.AbstractQueryTermContainer) ArrayList(java.util.ArrayList) MediaObjectReader(org.vitrivr.cineast.core.db.dao.reader.MediaObjectReader) QueryTerm(org.vitrivr.cineast.api.grpc.data.QueryTerm) MediaObjectMetadataReader(org.vitrivr.cineast.core.db.dao.reader.MediaObjectMetadataReader) StringDoublePair(org.vitrivr.cineast.core.data.StringDoublePair) MediaSegmentMetadataDescriptor(org.vitrivr.cineast.core.data.entities.MediaSegmentMetadataDescriptor) QueryStage(org.vitrivr.cineast.api.grpc.data.QueryStage) MediaObjectMetadataDescriptor(org.vitrivr.cineast.core.data.entities.MediaObjectMetadataDescriptor) HashSet(java.util.HashSet) MediaSegmentMetadataReader(org.vitrivr.cineast.core.db.dao.reader.MediaSegmentMetadataReader) QueryConfig(org.vitrivr.cineast.core.config.QueryConfig) ReadableQueryConfig(org.vitrivr.cineast.core.config.ReadableQueryConfig) StopWatch(org.apache.commons.lang3.time.StopWatch) ReadableQueryConfig(org.vitrivr.cineast.core.config.ReadableQueryConfig) MediaSegmentReader(org.vitrivr.cineast.core.db.dao.reader.MediaSegmentReader) SegmentScoreElement(org.vitrivr.cineast.core.data.score.SegmentScoreElement) MediaSegmentDescriptor(org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor) HashMap(java.util.HashMap) Map(java.util.Map)

Example 2 with MediaSegmentDescriptor

use of org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor in project cineast by vitrivr.

the class CineastQueryService method getMediaSegments.

@Override
public void getMediaSegments(CineastGrpc.MediaSegmentIdList request, StreamObserver<CineastGrpc.MediaSegmentQueryResult> responseObserver) {
    Set<String> ids = request.getIdsList().stream().map(CineastGrpc.MediaSegmentId::getId).collect(Collectors.toSet());
    MediaSegmentReader reader = new MediaSegmentReader(Config.sharedConfig().getDatabase().getSelectorSupplier().get());
    Map<String, MediaSegmentDescriptor> segments = reader.lookUpSegments(ids);
    CineastGrpc.MediaSegmentQueryResult result = CineastGrpc.MediaSegmentQueryResult.newBuilder().addAllSegments(segments.values().stream().map(MediaSegmentUtil::fromMediaSegmentDescriptor).collect(Collectors.toList())).build();
    responseObserver.onNext(result);
    responseObserver.onCompleted();
}
Also used : MediaSegmentReader(org.vitrivr.cineast.core.db.dao.reader.MediaSegmentReader) MediaSegmentDescriptor(org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor)

Example 3 with MediaSegmentDescriptor

use of org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor in project cineast by vitrivr.

the class FindSegmentsByIdGetHandler method doGet.

@Override
public MediaSegmentQueryResult doGet(Context ctx) {
    final Map<String, String> parameters = ctx.pathParamMap();
    final String segmentId = parameters.get(ID_NAME);
    final MediaSegmentReader sl = new MediaSegmentReader(Config.sharedConfig().getDatabase().getSelectorSupplier().get());
    final List<MediaSegmentDescriptor> list = sl.lookUpSegment(segmentId).map(s -> {
        final List<MediaSegmentDescriptor> segments = new ArrayList<>(1);
        segments.add(s);
        return segments;
    }).orElse(new ArrayList<>(0));
    sl.close();
    return new MediaSegmentQueryResult("", list);
}
Also used : List(java.util.List) GetRestHandler(org.vitrivr.cineast.api.rest.handlers.interfaces.GetRestHandler) MediaSegmentReader(org.vitrivr.cineast.core.db.dao.reader.MediaSegmentReader) Context(io.javalin.http.Context) Map(java.util.Map) MediaSegmentQueryResult(org.vitrivr.cineast.api.messages.result.MediaSegmentQueryResult) MediaSegmentDescriptor(org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor) OpenApiDocumentation(io.javalin.plugin.openapi.dsl.OpenApiDocumentation) OpenApiBuilder(io.javalin.plugin.openapi.dsl.OpenApiBuilder) ArrayList(java.util.ArrayList) Config(org.vitrivr.cineast.standalone.config.Config) MediaSegmentReader(org.vitrivr.cineast.core.db.dao.reader.MediaSegmentReader) MediaSegmentDescriptor(org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor) MediaSegmentQueryResult(org.vitrivr.cineast.api.messages.result.MediaSegmentQueryResult) List(java.util.List) ArrayList(java.util.ArrayList)

Example 4 with MediaSegmentDescriptor

use of org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor in project cineast by vitrivr.

the class EvaluationRuntime method objectDescriptorForId.

/**
 * Returns a MediaObjectDescriptor for the provided docID. This method uses a cache to speedup lookup of objects.
 *
 * @param docID ID of the segment for which the MediaObjectDescriptor is required.
 * @return MediaObjectDescriptor
 */
private MediaObjectDescriptor objectDescriptorForId(String docID) {
    if (this.cache.containsKey(docID)) {
        return this.cache.get(docID);
    }
    Optional<MediaSegmentDescriptor> descriptor = this.mediaSegmentReader.lookUpSegment(docID);
    if (descriptor.isPresent()) {
        MediaObjectDescriptor object = this.mediaObjectReader.lookUpObjectById(descriptor.get().getObjectId());
        this.cache.put(docID, object);
        return object;
    } else {
        return null;
    }
}
Also used : MediaObjectDescriptor(org.vitrivr.cineast.core.data.entities.MediaObjectDescriptor) MediaSegmentDescriptor(org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor)

Example 5 with MediaSegmentDescriptor

use of org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor in project cineast by vitrivr.

the class LSC21TemporalUpdateCommand method run.

@Override
@SuppressWarnings("unchecked")
public void run() {
    if (Config.sharedConfig().getDatabase().getSelector() != DataSource.COTTONTAIL) {
        System.out.println("Other DB than Cottontail DB not supported (yet). Aborting");
        return;
    }
    /* Preparation. */
    final CottontailWrapper cottontail = new CottontailWrapper(Config.sharedConfig().getDatabase().getHost(), Config.sharedConfig().getDatabase().getPort());
    long txId = cottontail.client.begin();
    final Query query = new Query(ENTITY_NAME).select("*", null).txId(txId);
    final TupleIterator ti = cottontail.client.query(query);
    final LinkedList<Update> updates = new LinkedList<>();
    int counter = 0;
    int totalCounter = 0;
    /* Prepare updates. */
    while (ti.hasNext()) {
        final Tuple t = ti.next();
        final MediaSegmentDescriptor segment = convert(t);
        try {
            final Optional<LocalDateTime> oldt = extractTimeInformation(segment.getSegmentId());
            if (!oldt.isPresent()) {
                continue;
            }
            final LocalDateTime ldt = oldt.get();
            final long msOfTheDay = ldt.getHour() * 60 * 60 * 1000 + ldt.getMinute() * 60 * 1000 + ldt.getSecond() * 1000;
            final long sAbs = ldt.toInstant(ZoneOffset.UTC).toEpochMilli() / 1000L;
            final Update update = new Update(ENTITY_NAME).values(new Pair<>(MediaSegmentDescriptor.SEGMENT_START_COL_NAME, (double) msOfTheDay), new Pair<>(MediaSegmentDescriptor.SEGMENT_END_COL_NAME, (double) msOfTheDay + 1), new Pair<>(MediaSegmentDescriptor.SEGMENT_STARTABS_COL_NAME, resetAbsTime ? 0.0 : sAbs), new Pair<>(MediaSegmentDescriptor.SEGMENT_ENDABS_COL_NAME, resetAbsTime ? 0.0 : (sAbs + 1))).where(new org.vitrivr.cottontail.client.language.extensions.Literal(CineastConstants.SEGMENT_ID_COLUMN_QUALIFIER, "=", segment.getSegmentId())).txId(txId);
            updates.add(update);
        } catch (Exception e) {
            LOGGER.warn("Could not update " + segment.getSegmentId() + " due to exception: " + e.getMessage());
        }
    }
    /* Execute updates. */
    Update update;
    while ((update = updates.poll()) != null) {
        cottontail.client.update(update);
        totalCounter++;
        if (counter++ > 99) {
            if (progress) {
                System.out.println("Updated " + totalCounter + " rows.");
            }
            counter = 0;
        }
    }
    cottontail.client.commit(txId);
    System.out.println("Done.");
}
Also used : LocalDateTime(java.time.LocalDateTime) CottontailWrapper(org.vitrivr.cineast.core.db.cottontaildb.CottontailWrapper) Query(org.vitrivr.cottontail.client.language.dql.Query) Update(org.vitrivr.cottontail.client.language.dml.Update) LinkedList(java.util.LinkedList) Literal(org.vitrivr.cottontail.grpc.CottontailGrpc.Literal) TupleIterator(org.vitrivr.cottontail.client.iterators.TupleIterator) MediaSegmentDescriptor(org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor) Tuple(org.vitrivr.cottontail.client.iterators.Tuple)

Aggregations

MediaSegmentDescriptor (org.vitrivr.cineast.core.data.entities.MediaSegmentDescriptor)23 ArrayList (java.util.ArrayList)12 List (java.util.List)10 HashMap (java.util.HashMap)9 StringDoublePair (org.vitrivr.cineast.core.data.StringDoublePair)8 MediaSegmentReader (org.vitrivr.cineast.core.db.dao.reader.MediaSegmentReader)8 Map (java.util.Map)7 TemporalObject (org.vitrivr.cineast.core.data.TemporalObject)7 Collectors (java.util.stream.Collectors)5 HashSet (java.util.HashSet)4 MediaSegmentQueryResult (org.vitrivr.cineast.api.messages.result.MediaSegmentQueryResult)4 Config (org.vitrivr.cineast.standalone.config.Config)4 Set (java.util.Set)3 IntStream (java.util.stream.IntStream)3 LogManager (org.apache.logging.log4j.LogManager)3 Logger (org.apache.logging.log4j.Logger)3 MediaObjectDescriptor (org.vitrivr.cineast.core.data.entities.MediaObjectDescriptor)3 FileNotFoundException (java.io.FileNotFoundException)2 Comparator (java.util.Comparator)2 TreeMap (java.util.TreeMap)2