import { type EmbeddingProvider, type VectorDocument, type VectorSearchOptions, type VectorSearchResult, type VectorStore } from '../hallucination/retrieval.js';
/** Options for the Qdrant store. */
export interface QdrantVectorStoreOptions {
    /** The Qdrant server, such as `http://localhost:6333` or a Qdrant Cloud cluster URL. */
    url: string;
    /** The collection chunks are kept in. */
    collection: string;
    /** Width of every vector. Must match the embedding function's output. */
    dimensions: number;
    /** Sent as the `api-key` header, for Qdrant Cloud or a secured server. */
    apiKey?: string;
    /** Embeds chunks that arrive without a vector, and search queries. Defaults to hashed term vectors. */
    embed?: EmbeddingProvider;
    /** Replaces the global `fetch`, for a proxy, retries, or tests. */
    fetch?: typeof globalThis.fetch;
    /** Headers added to every request. */
    headers?: Record<string, string>;
    /** Aborts a request that takes longer, in milliseconds. Defaults to 30 seconds. */
    timeoutMs?: number;
}
/** Raised when Qdrant answers with an error. */
export declare class QdrantError extends Error {
    /** The HTTP status Qdrant answered with. */
    readonly status: number;
    /** Qdrant's response body, as text. */
    readonly body?: string | undefined;
    constructor(message: string, 
    /** The HTTP status Qdrant answered with. */
    status: number, 
    /** Qdrant's response body, as text. */
    body?: string | undefined);
}
/**
 * Retrieval chunks in a Qdrant collection, through its REST API.
 *
 * No Qdrant client is a dependency: the store speaks HTTP through `fetch`, so it runs wherever
 * `fetch` and Web Crypto do, edge runtimes included. The same contract as `MemoryVectorStore`,
 * checked by the same tests. Qdrant point ids must be UUIDs or integers, so each chunk id is mapped
 * to a stable UUID derived from it, and the original id travels in the payload.
 */
export declare class QdrantVectorStore implements VectorStore {
    private readonly options;
    private readonly base;
    private readonly embed;
    private readonly fetchImpl;
    constructor(options: QdrantVectorStoreOptions);
    /**
     * Creates the collection with cosine distance unless it already exists, then indexes the
     * `metadata` fields you filter on. A list indexes them as keywords; a map names each field's type.
     * Never runs implicitly.
     */
    migrate(options?: {
        filterFields?: readonly string[] | Record<string, 'keyword' | 'integer' | 'float' | 'bool'>;
    }): Promise<void>;
    /** Adds chunks, or replaces those whose id exists, embedding those without a vector in one batch. */
    add(documents: VectorDocument[]): Promise<void>;
    /** The chunks most similar to a query, best first. */
    search(query: string, options?: VectorSearchOptions): Promise<VectorSearchResult[]>;
    /** The chunks most similar to a vector, best first. */
    searchVector(vector: number[], options?: VectorSearchOptions): Promise<VectorSearchResult[]>;
    /** Removes chunks by id. */
    delete(ids: readonly string[]): Promise<void>;
    private assertWidth;
    private request;
}