import { GetContextClientParams, BaseVectorContext, ResetOptions, VectorPayload, AddContextPayload, SaveOperationResult } from './types.js';
import { Index } from '@upstash/vector';
import '@upstash/redis';
import 'ai';

declare const getContextClient: (params?: GetContextClientParams) => BaseVectorContext | undefined;

declare class VectorDB {
    private index;
    constructor(index: Index);
    reset(options?: ResetOptions): Promise<void>;
    delete({ ids, namespace }: {
        ids: string[];
        namespace?: string;
    }): Promise<void>;
    /**
     * A method that allows you to query the vector database with plain text.
     * It takes care of the text-to-embedding conversion by itself.
     * Additionally, it lets consumers pass various options to tweak the output.
     */
    retrieve<TMetadata>({ question, similarityThreshold, topK, namespace, contextFilter, queryMode, }: VectorPayload): Promise<{
        data: string;
        id: string;
        metadata: TMetadata;
    }[]>;
    /**
     * A method that allows you to add various data types into a vector database.
     * It supports plain text, embeddings, PDF, HTML, Text file and CSV. Additionally, it handles text-splitting for CSV, PDF and Text file.
     */
    save(input: AddContextPayload): Promise<SaveOperationResult>;
}

type ExtVectorConfig = {
    url: string;
    token: string;
};
declare class ExtVector {
    private vectorDB;
    private namespace;
    constructor(config: ExtVectorConfig, namespace: string);
    addContext(input: AddContextPayload): Promise<SaveOperationResult>;
    removeContext(ids: string[]): Promise<void>;
    getContext<TMetadata>(payload: Omit<VectorPayload, "namespace">): Promise<{
        data: string;
        id: string;
        metadata: TMetadata;
    }[]>;
    resetContext(): Promise<void>;
}

export { ExtVector, VectorDB, getContextClient };