Embabel Agent Documentation/com.embabel.agent.rag.lucene/LuceneSearchOperations

LuceneSearchOperations

class LuceneSearchOperations constructor(val name: String, val enhancers: List<<Error class: unknown class>> = emptyList(), embeddingModel: <Error class: unknown class>? = null, keywordExtractor: KeywordExtractor? = null, vectorWeight: Double = 0.5, chunkerConfig: <Error class: unknown class> = ContentChunker.DefaultConfig(), indexPath: Path? = null) : Closeable

Lucene RAG facet with optional vector search support via an EmbeddingModel. Supports both in-memory and disk-based persistence. Implements WritableContentElementRepository so we can add to the store.

Parameters

name

Name of this RAG service

embeddingModel

Optional embedding model for vector search; if null, only text search is supported

keywordExtractor

Optional keyword extractor for keyword-based search; if null, keyword search is disabled

vectorWeight

Weighting for vector similarity in hybrid search (0.0 to 1.0)

chunkerConfig

Configuration for content chunking

indexPath

Optional path for disk-based index storage; if null, uses in-memory storage

Constructors

LuceneSearchOperations

constructor(name: String, enhancers: List<<Error class: unknown class>> = emptyList(), embeddingModel: <Error class: unknown class>? = null, keywordExtractor: KeywordExtractor? = null, vectorWeight: Double = 0.5, chunkerConfig: <Error class: unknown class> = ContentChunker.DefaultConfig(), indexPath: Path? = null)

Types

Companion

object Companion

Properties

enhancers

open val enhancers: List<<Error class: unknown class>>

luceneSyntaxNotes

open val luceneSyntaxNotes: String

name

open val name: String

Functions

clear

fun clear(): Int

Clear all stored content - useful for testing

open override fun close()

commit

open fun commit()

createInternalRelationships

open fun createInternalRelationships(root: <Error class: unknown class>)

deleteRootAndDescendants

open fun deleteRootAndDescendants(uri: String): <Error class: unknown class>?

expandResult

open fun expandResult(id: String, method: <Error class: unknown class>, elementsToAdd: Int): List<<Error class: unknown class>>

facets

open fun facets(): List<<Error class: unknown class><out <Error class: unknown class>>>

findAll

fun findAll(): List<<Error class: unknown class>>

Default to finding chunks

fun <T> findAll(clazz: Class<T>): List<T>

findAllChunksById

open fun findAllChunksById(chunkIds: List<String>): List<<Error class: unknown class>>

findById

open fun findById(id: String): <Error class: unknown class>?

findChunkIdsByKeywords

fun findChunkIdsByKeywords(keywords: Set<String>, minIntersection: Int = 1, maxResults: Int = 100): List<<Error class: unknown class><String, Int>>

Find chunk IDs by keyword intersection. Returns pairs of (chunkId, matchCount) sorted by match count descending.

findChunksForEntity

open fun findChunksForEntity(entityId: String): List<<Error class: unknown class>>

findContentRootByUri

open fun findContentRootByUri(uri: String): <Error class: unknown class>?

hybridSearch

fun hybridSearch(ragRequest: <Error class: unknown class>): <Error class: unknown class><<Error class: unknown class>>

info

open fun info(): LuceneStatistics

infoString

open fun infoString(verbose: Boolean?, indent: Int): String

keywordSearch

fun keywordSearch(ragRequest: <Error class: unknown class>): <Error class: unknown class><<Error class: unknown class>>

loadExistingChunksFromDisk

fun loadExistingChunksFromDisk(): LuceneSearchOperations

Manually trigger loading of existing chunks from disk. Useful for ensuring chunks are loaded immediately after startup.

onNewRetrievables

open fun onNewRetrievables(retrievables: List<<Error class: unknown class>>)

save

open fun save(element: <Error class: unknown class>): <Error class: unknown class>

textSearch

open fun <T> textSearch(request: <Error class: unknown class>, clazz: Class<T>): List<<Error class: unknown class><T>>

updateKeywords

fun updateKeywords(updates: Map<String, List<String>>)

Update keywords for existing chunks. This will re-index the chunks with new keywords.

vectorSearch

open fun <T> vectorSearch(request: <Error class: unknown class>, clazz: Class<T>): List<<Error class: unknown class><T>>