import type { ChatMessage, DarwinTrajectory, DpoRow, ExportOptions, ExportResult, SftRow } from './types.js';
/**
 * Rough token estimate for a message list. We deliberately do NOT pull in a
 * tokenizer dependency (the package is dependency-free, like darwin-mode). The
 * ~4-chars-per-token heuristic is conservative and stable; tool_calls
 * arguments count too (they cost context). This is a budget gate, not a billing
 * figure — over-estimating is the safe direction.
 */
export declare function estimateTokens(messages: ChatMessage[]): number;
/**
 * THE CONTAMINATION GUARD. Throws if any trajectory's instance_id appears in
 * the eval holdout. Call it the first thing the exporter does — fail loud, not
 * silently filter, when train/eval disjointness is violated.
 */
export declare function assertTrainEvalDisjoint(trajectories: DarwinTrajectory[], evalHoldout: Iterable<string>): void;
/**
 * Build the SFT and DPO sets from a Darwin trajectory archive.
 *
 * @param trajectories the input archive (already-excluded-of-holdout OR raw —
 *   the exporter excludes holdout members itself, but ASSERTS disjointness on
 *   what remains so a programming error can't slip eval data through).
 */
export declare function exportTrainingData(trajectories: DarwinTrajectory[], options: ExportOptions): ExportResult;
/** Serialize SFT rows to JSONL (one row per line). */
export declare function sftToJsonl(rows: SftRow[]): string;
/** Serialize DPO rows to JSONL (one row per line). */
export declare function dpoToJsonl(rows: DpoRow[]): string;
//# sourceMappingURL=export.d.ts.map