Package org.apache.lucene.search

Class CachingCollectorManager<C extends Collector,R>

java.lang.Object

org.apache.lucene.search.CachingCollectorManager<C,R>

All Implemented Interfaces:

CollectorManager<CachingCollector,R>

public class CachingCollectorManager<C extends Collector,R>
extends Object
implements CollectorManager<CachingCollector,R>

A CollectorManager that wraps a delegate CollectorManager and caches all collected documents (and optionally scores) per slice, so they can be replayed to a second-pass CollectorManager without re-running the query.

One CachingCollector is created per slice. During replay(org.apache.lucene.search.CollectorManager<C2, R2>), each cached slice is replayed into a fresh second-pass collector, and all second-pass collectors are reduced together. This works correctly with both sequential and concurrent search.

Example usage:

 CachingCollectorManager<C1, R1> caching = new CachingCollectorManager<>(
     firstPassManager, cacheScores, maxRAMMB, null);
 R1 firstResult = searcher.search(query, caching);

 if (caching.isCached()) {
   R2 secondResult = caching.replay(secondPassManager);
 } else {
   // cache overflowed — re-run the query
   R2 secondResult = searcher.search(query, secondPassManager);
 }
 

WARNING: This API is experimental and might change in incompatible ways in the next release.

Constructor Summary

Constructors

Constructor	Description
CachingCollectorManager(CollectorManager<C,R> delegate, boolean cacheScores, Double maxRAMMB, Integer maxDocsToCache)

Method Summary

Modifier and Type	Method	Description
boolean	isCached()	Returns true if the search has been run and all per-slice caches are intact (none overflowed their RAM/doc budget).
CachingCollector	newCollector()	Return a new Collector.
R	reduce(Collection<CachingCollector> collectors)	Reduce the results of individual collectors into a meaningful result.
<C2 extends Collector, R2> R2	replay(CollectorManager<C2,R2> secondPassManager)	Replays each per-slice cache into a fresh second-pass collector, then reduces all results.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

CachingCollectorManager

public CachingCollectorManager(CollectorManager<C,R> delegate,
 boolean cacheScores,
 Double maxRAMMB,
 Integer maxDocsToCache)

Parameters:

delegate - the first-pass CollectorManager

cacheScores - whether to cache scores in addition to document IDs

maxRAMMB - the maximum RAM in MB to use per slice cache, or null if using maxDocsToCache

maxDocsToCache - the maximum number of documents to cache per slice, or null if using maxRAMMB

Method Details

newCollector

public CachingCollector newCollector()
                              throws IOException

Description copied from interface: CollectorManager

Return a new Collector. This must return a different instance on each call.

Specified by:

newCollector in interface CollectorManager<C extends Collector,R>

Throws:

IOException

reduce

public R reduce(Collection<CachingCollector> collectors)
         throws IOException

Description copied from interface: CollectorManager

Reduce the results of individual collectors into a meaningful result. For instance a TopDocsCollector would compute the top docs of each collector and then merge them using TopDocs.merge(int, TopDocs[]). This method must be called after collection is finished on all provided collectors.

Specified by:

reduce in interface CollectorManager<C extends Collector,R>

Throws:

IOException

isCached

public boolean isCached()

Returns true if the search has been run and all per-slice caches are intact (none overflowed their RAM/doc budget). Returns false if the search has not yet been run or any cache overflowed.

replay

public <C2 extends Collector,
R2> R2 replay(CollectorManager<C2,R2> secondPassManager)
                                     throws IOException

Replays each per-slice cache into a fresh second-pass collector, then reduces all results.

Throws:

IllegalStateException - if isCached() returns false

IOException