CavalliumDBEngine/src/main/java/it/cavallium/dbengine/lucene/searcher/UnsortedUnscoredStreamingMultiSearcher.java

120 lines
4.3 KiB
Java
Raw Normal View History

package it.cavallium.dbengine.lucene.searcher;
import io.net5.buffer.api.Send;
import it.cavallium.dbengine.client.query.current.data.TotalHitsCount;
import it.cavallium.dbengine.database.LLKeyScore;
import it.cavallium.dbengine.database.LLUtils;
import it.cavallium.dbengine.database.disk.LLIndexSearchers;
import it.cavallium.dbengine.lucene.LuceneUtils;
2021-11-08 23:53:53 +01:00
import it.cavallium.dbengine.lucene.collector.ReactiveCollectorMultiManager;
import it.cavallium.dbengine.lucene.searcher.LLSearchTransformer.TransformerInput;
2021-11-08 18:52:52 +01:00
import java.util.concurrent.CancellationException;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.atomic.AtomicInteger;
import org.apache.lucene.search.IndexSearcher;
import org.apache.lucene.search.ScoreDoc;
2021-11-08 18:52:52 +01:00
import org.warp.commonutils.type.ShortNamedThreadFactory;
import reactor.core.publisher.Flux;
2021-11-08 16:33:41 +01:00
import reactor.core.publisher.FluxSink.OverflowStrategy;
import reactor.core.publisher.Mono;
public class UnsortedUnscoredStreamingMultiSearcher implements MultiSearcher {
2021-11-08 18:52:52 +01:00
private static final ExecutorService EXECUTOR_SERVICE
= Executors.newCachedThreadPool(new ShortNamedThreadFactory("StreamingExecutor"));
@Override
2021-11-08 11:17:52 +01:00
public Mono<LuceneSearchResult> collectMulti(Mono<Send<LLIndexSearchers>> indexSearchersMono,
LocalQueryParams queryParams,
String keyFieldName,
LLSearchTransformer transformer) {
return LLUtils.usingSendResource(indexSearchersMono, indexSearchers -> {
Mono<LocalQueryParams> queryParamsMono;
if (transformer == LLSearchTransformer.NO_TRANSFORMATION) {
queryParamsMono = Mono.just(queryParams);
} else {
queryParamsMono = transformer.transform(Mono
.fromCallable(() -> new TransformerInput(indexSearchers, queryParams)));
}
return queryParamsMono
.flatMap(queryParams2 -> {
var localQueryParams = getLocalQueryParams(queryParams2);
2021-10-15 22:03:53 +02:00
if (queryParams2.isSorted() && queryParams2.limitLong() > 0) {
return Mono.error(new UnsupportedOperationException("Sorted queries are not supported"
+ " by UnsortedUnscoredContinuousLuceneMultiSearcher"));
}
2021-10-15 22:03:53 +02:00
if (queryParams2.needsScores() && queryParams2.limitLong() > 0) {
return Mono.error(new UnsupportedOperationException("Scored queries are not supported"
+ " by UnsortedUnscoredContinuousLuceneMultiSearcher"));
}
return Mono.fromCallable(() -> {
LLUtils.ensureBlocking();
var shards = indexSearchers.shards();
2021-11-08 16:33:41 +01:00
Flux<ScoreDoc> scoreDocsFlux = Flux.create(scoreDocsSink -> {
2021-11-08 23:53:53 +01:00
var cmm = new ReactiveCollectorMultiManager(scoreDocsSink);
2021-11-08 16:33:41 +01:00
AtomicInteger runningTasks = new AtomicInteger(0);
runningTasks.addAndGet(shards.size());
int mutableShardIndex = 0;
for (IndexSearcher shard : shards) {
int shardIndex = mutableShardIndex++;
2021-11-08 18:52:52 +01:00
EXECUTOR_SERVICE.execute(() -> {
2021-11-08 16:33:41 +01:00
try {
2021-11-08 23:53:53 +01:00
var collector = cmm.get(shardIndex);
assert queryParams.complete() == cmm.scoreMode().isExhaustive();
2021-11-08 16:33:41 +01:00
assert queryParams
.getScoreModeOptional()
2021-11-08 23:53:53 +01:00
.map(scoreMode -> scoreMode == cmm.scoreMode())
2021-11-08 16:33:41 +01:00
.orElse(true);
shard.search(localQueryParams.query(), collector);
} catch (Throwable e) {
2021-11-08 18:52:52 +01:00
if (!(e instanceof CancellationException)) {
scoreDocsSink.error(e);
}
2021-11-08 16:33:41 +01:00
} finally {
if (runningTasks.decrementAndGet() <= 0) {
scoreDocsSink.complete();
}
}
2021-11-08 16:33:41 +01:00
});
}
}, OverflowStrategy.BUFFER);
2021-11-08 16:33:41 +01:00
Flux<LLKeyScore> resultsFlux = LuceneUtils.convertHits(scoreDocsFlux, shards, keyFieldName, false);
var totalHitsCount = new TotalHitsCount(0, false);
Flux<LLKeyScore> mergedFluxes = resultsFlux
2021-10-15 22:03:53 +02:00
.skip(queryParams2.offsetLong())
.take(queryParams2.limitLong(), true);
2021-11-08 11:17:52 +01:00
return new LuceneSearchResult(totalHitsCount, mergedFluxes, indexSearchers::close);
});
});
}, false);
}
private LocalQueryParams getLocalQueryParams(LocalQueryParams queryParams) {
return new LocalQueryParams(queryParams.query(),
2021-10-15 22:03:53 +02:00
0L,
queryParams.offsetLong() + queryParams.limitLong(),
queryParams.pageLimits(),
queryParams.minCompetitiveScore(),
queryParams.sort(),
queryParams.complete()
);
}
@Override
public String getName() {
return "unsorted unscored streaming multi";
}
}