data-sampler

/**
 * @file dataset.ts
 * @description HuggingFace Datasets Server REST API client.
 *
 * Uses the HF Datasets Server (datasets-server.huggingface.co) to discover
 * dataset configs/splits and fetch sample rows. Uses only Node.js built-in
 * fetch() — no additional npm dependencies.
 *
 * Endpoints used:
 *   - GET /splits  → discover available configs and splits
 *   - GET /rows    → fetch consecutive rows by offset/length
 *   - GET /search  → BM25 full-text search across rows
 *
 * @module dataset
 */
import type { LoadDatasetParams, LoadDatasetResult } from "./types";
interface SplitsResponseSplit {
    readonly dataset: string;
    readonly config: string;
    readonly split: string;
}
interface SplitsResponse {
    readonly splits: readonly SplitsResponseSplit[];
}
interface RowEntry {
    readonly row_idx: number;
    readonly row: Record<string, unknown>;
    readonly truncated_cells: readonly string[];
}
/**
 * Fetch available configs and splits for a given dataset.
 *
 * @param dataset - HF dataset name (e.g. "squad", "ibm/duorc").
 * @param signal - Optional AbortSignal.
 * @returns The parsed splits response.
 */
export declare function fetchSplits(dataset: string, signal?: AbortSignal): Promise<SplitsResponse>;
/**
 * Fetch a consecutive block of rows from a dataset split.
 *
 * @param dataset - HF dataset name.
 * @param config - Dataset config/subset name.
 * @param split - Split name ("train", "test", "validation").
 * @param offset - Row offset to start from (0-based).
 * @param length - Number of rows to fetch (max 100).
 * @param signal - Optional AbortSignal.
 * @returns An object with the rows array and total row count.
 */
export declare function fetchRows(dataset: string, config: string, split: string, offset: number, length: number, signal?: AbortSignal): Promise<{
    readonly rows: readonly RowEntry[];
    readonly totalRows: number;
}>;
/**
 * BM25 full-text search across rows in a dataset split.
 *
 * @param dataset - HF dataset name.
 * @param config - Dataset config/subset name.
 * @param split - Split name.
 * @param query - Search query text.
 * @param length - Number of results to return (max 100).
 * @param signal - Optional AbortSignal.
 * @returns An object with matching rows and an estimate of total matches.
 */
export declare function searchRows(dataset: string, config: string, split: string, query: string, length: number, signal?: AbortSignal): Promise<{
    readonly rows: readonly RowEntry[];
    readonly totalRows: number;
}>;
/**
 * Main entry point: load samples from a HuggingFace dataset.
 *
 * Orchestration flow:
 * 1. If no config was provided, call `/splits` to discover available configs.
 *    - Exactly one config → use it automatically.
 *    - Multiple configs → throw with a message listing choices.
 *    - No configs → throw with available splits info.
 * 2. If a query is provided, call `/search`; otherwise call `/rows`.
 * 3. Format the returned rows into the standard result structure.
 *
 * @param params - LoadDatasetParams from the tool invocation.
 * @param signal - Optional AbortSignal from LM Studio for cancellation.
 * @returns A LoadDatasetResult with samples, metadata, and formatted text.
 */
export declare function loadDataset(params: LoadDatasetParams, signal?: AbortSignal): Promise<LoadDatasetResult>;
export {};
//# sourceMappingURL=dataset.d.ts.map