Struct RagPipeline 

pub struct RagPipeline { /* private fields */ }

Available on crate feature rag only.

The RAG pipeline orchestrator.

Coordinates document ingestion (chunk → embed → store) and query execution (embed → search → rerank → filter). Construct one via RagPipeline::builder().

Implementations

impl RagPipeline

pub fn builder() -> RagPipelineBuilder

Create a new RagPipelineBuilder.

pub fn config(&self) -> &RagConfig

Return a reference to the pipeline configuration.

pub fn embedding_provider(&self) -> &Arc<dyn EmbeddingProvider>

Return a reference to the embedding provider.

pub fn vector_store(&self) -> &Arc<dyn VectorStore>

Return a reference to the vector store.

pub async fn create_collection(&self, name: &str) -> Result<(), RagError>

Create a named collection in the vector store.

The collection is created with the dimensionality reported by the configured EmbeddingProvider.

Errors

Returns RagError::PipelineError if the vector store operation fails.

pub async fn delete_collection(&self, name: &str) -> Result<(), RagError>

Delete a named collection from the vector store.

Errors

Returns RagError::PipelineError if the vector store operation fails.

pub async fn ingest( &self, collection: &str, document: &Document, ) -> Result<Vec<Chunk>, RagError>

Ingest a single document: chunk → embed → store.

Returns the chunks that were stored (with embeddings attached).

Errors

Returns RagError::PipelineError if embedding or storage fails, including the document ID in the error message.

pub async fn ingest_batch( &self, collection: &str, documents: &[Document], ) -> Result<Vec<Chunk>, RagError>

Ingest multiple documents through the chunk → embed → store workflow.

Returns all chunks that were stored across all documents.

Errors

Returns RagError::PipelineError on the first document that fails, including the document ID in the error message.

pub async fn query( &self, collection: &str, query: &str, ) -> Result<Vec<SearchResult>, RagError>

Query the pipeline: embed → search → rerank → filter by threshold.

Returns search results ordered by descending relevance score. Results below the configured similarity_threshold are filtered out.

Errors

Returns RagError::PipelineError if embedding or search fails.