import type { MergedOutputItem, RequestQueueData } from "./types.js";
/**
 * Deduplicate a list of merged-output items by their `meta.uniqueKey`.
 *
 * In distributed mode every worker writes its own dataset shard and each item
 * already carries a deterministic `meta.uniqueKey` (see {@link generateUniqueKey}).
 * When those shards are concatenated for the consolidated webhook / run output,
 * identical items (same uniqueKey, scraped by more than one worker, or re-scraped
 * on retry) must collapse to a single record.
 *
 * Strategy is **last-wins** by default: when keys collide, the item appearing
 * later in `items` replaces the earlier one. Callers that concatenate worker
 * shards in a deterministic order therefore get a deterministic result.
 *
 * Items without a `meta.uniqueKey` are passed through unchanged (never dropped).
 */
export declare function dedupeByUniqueKey(items: MergedOutputItem[], strategy?: "last" | "first"): MergedOutputItem[];
/**
 * Build the merged, deduplicated output for a run from raw dataset items and the
 * request-queue data that produced them.
 *
 * Each dataset item is wrapped into a {@link MergedOutputItem} with a
 * `meta.uniqueKey` computed via {@link generateUniqueKey}, then the whole set is
 * deduplicated (last-wins by default).
 *
 * @param datasets - Raw scraped dataset items.
 * @param requestQueueData - Request-queue records keyed by url (provides the
 *   request uniqueKey + source metadata for each item, when available).
 * @param uniqueKeySelector - Optional field names that define identity; passed
 *   through to {@link generateUniqueKey}.
 */
export declare function buildMergedOutput(datasets: any[], requestQueueData?: RequestQueueData[], uniqueKeySelector?: string[]): MergedOutputItem[];
/**
 * Build merged, deduplicated output without request-queue metadata.
 *
 * Same as {@link buildMergedOutput} but for callers that only have the dataset
 * items themselves. The unique key is derived from item content (and the
 * optional selector) alone.
 */
export declare function buildSimpleMergedOutput(datasets: any[], uniqueKeySelector?: string[]): MergedOutputItem[];
//# sourceMappingURL=merged-output.d.ts.map