Member Junction
    Preparing search index...

    Data access that can also count a source exactly, as complete reads require.

    interface ICountingDataAccess {
        CountRows(
            params: { EntityName: string; ExtraFilter?: string },
        ): Promise<CountRowsResult>;
        fetchEmbedding(
            entity: string,
            recordId: string,
            embeddingModelRef: string,
            dims: number,
        ): Promise<number[]>;
        fetchRows(params: FetchRowsParams): Promise<FetchRowsResult>;
    }

    Hierarchy (View Summary)

    Implemented by

    Index

    Methods

    • Fetch a persisted, version-pinned embedding vector for a single record of an entity (plan §5.2). Returns the dense numeric vector, or null when no persisted embedding exists for that record/model.

      The default implementation is a documented integration seam (see RunViewDataAccess.fetchEmbedding); tests supply vectors directly.

      Parameters

      • entity: string

        entity whose persisted embedding is requested

      • recordId: string

        primary-key value of the record

      • embeddingModelRef: string

        pinned embedding-model id/name (part of model lineage)

      • dims: number

        expected dimensionality

      Returns Promise<number[]>