Retriever class

Wires a Chunker, an Embedder, and a VectorStore into a retrieval pipeline: add texts once, then retrieve the passages most relevant to a query.

Constructors

Retriever({required Embedder embedder, required VectorStore store, Chunker? chunker}): Creates a retriever.

chunker → Chunker: Splits added texts into chunks.
final
embedder → Embedder: Embeds batches of chunk texts and single queries.
final
hashCode → int: The hash code for this object.
no setterinherited
runtimeType → Type: A representation of the runtime type of the object.
no setterinherited
store → VectorStore: Holds the embedded chunks.
final

addText(String text, {String? sourceId, Map<String, Object?> metadata = const {}}) → Future<void>: Chunks text, embeds all chunks in a single embedder call, and upserts them into store.
buildContext(String query, {int topK = 5, int? maxChars, String separator = '\n\n---\n\n', double? minScore, bool where(Document document)?, bool diverse = false, double lambda = 0.5}) → Future<String>: Retrieves the chunks most relevant to query and joins their texts with separator into a single string, ready to paste into an LLM prompt.
noSuchMethod(Invocation invocation) → dynamic: Invoked when a nonexistent method or property is accessed.
inherited
retrieve(String query, {int topK = 5, double? minScore, bool where(Document document)?}) → Future<List<ScoredChunk>>: Embeds query and returns the most similar stored chunks, best first.
retrieveDiverse(String query, {int topK = 5, int? fetchK, double lambda = 0.5, double? minScore, bool where(Document document)?}) → Future<List<ScoredChunk>>: Retrieves chunks that are relevant to query and unlike each other.
toString() → String: A string representation of this object.
inherited