Crate lc_rag 

RAG (Retrieval-Augmented Generation) module for LangChainRust.

Provides document loaders, text splitters, retrievers, BM25 search, hybrid retrieval, GraphRAG, HyDE, multi-query, reranking, multimodal (image+text) retrieval, and a full RAG pipeline builder.

Re-exports

pub use adapter::RagRunnable;

pub use contextual::ContextualConfig;

pub use contextual::ContextualEnhancer;

pub use contextual::ContextualError;

pub use contextual::CONTEXTUAL_METADATA_KEY;

pub use parent_document::ParentDocumentRetriever;

pub use pipeline::RAGPipeline;

pub use pipeline::RAGPipelineBuilder;

pub use pipeline::RAGQueryResult;

pub use retriever_runnable::RetrieverRunnable;

pub use loaders::CSVLoader;

pub use loaders::DocumentLoader;

pub use loaders::DocxLoader;

pub use loaders::HTMLLoader;

pub use loaders::JSONLoader;

pub use loaders::LoaderError;

pub use loaders::MarkdownLoader;

pub use loaders::PDFLoader;

pub use loaders::SitemapLoader;

pub use loaders::TextLoader;

pub use loaders::WebScraperLoader;

pub use retriever::Retriever;

pub use retriever::RetrieverError;

pub use retriever::RetrieverTrait;

pub use retriever::SimilarityRetriever;

pub use self_query::SelfQueryArgs;

pub use self_query::SelfQueryRetriever;

pub use semantic_cache::CacheHitKind;

pub use semantic_cache::CachedRetriever;

pub use semantic_cache::SemanticCacheConfig;

pub use semantic_cache::SemanticCacheCore;

pub use semantic_splitter::SemanticSplitter;

pub use bm25::AutoMergingConfig;

pub use bm25::BM25Params;

pub use bm25::BM25Retriever;

pub use bm25::ChunkedBM25Retriever;

pub use bm25::ChunkedSearchResult;

pub use bm25::Tokenizer;

pub use hybrid::filter_by_score;

pub use hybrid::reciprocal_rank_fusion;

pub use hybrid::RetrievalSource;

pub use hybrid::RetrievedDocument;

pub use unified_hybrid::HybridIndexConfig;

pub use unified_hybrid::HybridSearchResult;

pub use unified_hybrid::UnifiedHybridIndex;

pub use multi_query::MultiQueryConfig;

pub use multi_query::MultiQueryError;

pub use multi_query::MultiQueryRetriever;

pub use multi_query::StaticQueryGenerator;

pub use multimodal::ImageAsset;

pub use multimodal::MediaBlock;

pub use multimodal::ModalityFilter;

pub use multimodal::MultimodalChunkConfig;

pub use multimodal::MultimodalChunker;

pub use multimodal::MultimodalRetriever;

pub use multimodal::MM_CAPTION_KEY;

pub use multimodal::MM_KIND_IMAGE;

pub use multimodal::MM_KIND_KEY;

pub use multimodal::MM_KIND_TEXT;

pub use multimodal::MM_MIME_KEY;

pub use multimodal::MM_URL_KEY;

pub use hyde::HyDEConfig;

pub use hyde::HyDEError;

pub use hyde::HyDERetriever;

pub use late_chunking::late_chunk;

pub use late_chunking::pool_tokens;

pub use late_chunking::LateChunk;

pub use late_chunking::LateChunkConfig;

pub use reranking::BM25Reranker;

pub use reranking::KeywordReranker;

pub use reranking::Reranker;

pub use reranking::RerankingConfig;

pub use reranking::RerankingError;

pub use reranking::RerankingExecutor;

pub use graph_rag::Community as GraphCommunity;

pub use graph_rag::Entity as GraphEntity;

pub use graph_rag::GraphStore;

pub use graph_rag::Relation as GraphRelation;

pub use graph_rag::GlobalLevel;

pub use graph_rag::GraphRAG;

pub use graph_rag::GraphRAGConfig;

pub use graph_rag::GraphRAGError;

pub use graph_rag::GraphRAGResult;

pub use graph_rag::QueryMode;

Modules

adapter

RagRunnable adapter - bridges RAGPipeline to the Runnable trait.

bm25

BM25 retrieval module

contextual

Contextual Retrieval: index-time context injection for chunks (0.21.0 S4.1).

graph_rag

GraphRAG (Knowledge Graph RAG) module.

hybrid

Hybrid retrieval module

hyde

HyDE (Hypothetical Document Embedding) Retriever implementation

late_chunking

Late chunking: embed the whole document token by token first, then pool along chunk boundaries (0.21.0 S5.2).

loaders

Document loader implementations

multi_query

MultiQueryRetriever implementation

multimodal

Multimodal (image + text) chunking and retrieval (B7, v0.22.4).

parent_document

ParentDocumentRetriever — a parent/child document retriever

pipeline

RAGPipeline & RAGPipelineBuilder — a complete RAG pipeline in one line

reranking

Reranking implementation

retriever

Retriever implementations

retriever_runnable

RetrieverRunnable — a Runnable adapter to bring any retriever into an LCEL chain

self_query

SelfQueryRetriever — a self-querying retriever

semantic_cache

Semantic cache for retrieval results (0.21.0 S4.2).

semantic_splitter

Semantic chunking

splitter

Text splitter implementation

unified_hybrid

Unified Hybrid Index

Structs

ChunkDocument

Chunk document (split document fragment).

ChunkedVectorStore

Chunked Vector Store

Document

Document structure.

InMemoryDocumentStore

In-memory document store

InMemoryVectorStore

In-memory vector store

MockEmbeddings

Mock embeddings for testing purposes.

OpenAIEmbeddings

OpenAI Embeddings client

RecursiveCharacterSplitter

Recursive character splitter

SearchResult

Search result.

Enums

EmbeddingError

Embedding error type

ImageInput

Provider-neutral image input for vision embedding models.

VectorStoreError

Vector store error types.

Traits

ChunkedDocumentStoreTrait

Document store trait supporting a parent-child structure

DocumentStore

Document store trait

Embeddings

Embedding model trait

TextSplitter

Text splitter trait

VectorStore

Vector store trait.

VisionEmbeddings

Cross-modal embedding model: images and text mapped to one shared space.

Functions

cosine_similarity

Compute cosine similarity between two vectors.

Type Aliases

ChunkedDocumentStore

Type alias for InMemoryChunkedDocumentStore