/**
 * Per-reel target shape for `ReelSet.setResult` and
 * `ReelSetBuilder.initialFrame`. One object per reel.
 *
 * Use this for every result grid that crosses a worker, network, or
 * serializer boundary. The shape survives `structuredClone`, JSON, and
 * `postMessage` round-trips.
 */
export interface ColumnTarget {
    /** Visible-area target symbols, indexed `0 ... visibleRows-1`. */
    visible: string[];
    /**
     * Buffer-above target symbols. `bufferAbove[0]` is the slot closest to the
     * visible top row; later indices go further above. Up to `bufferSymbols`
     * entries are honored.
     *
     * Big-symbol anchors may sit here. Place a multi-cell symbol id (one whose
     * `SymbolData.size.h > 1`) at any `bufferAbove[i]` and the coordinator
     * paints OCCUPIED stubs across the rest of the block, including any cells
     * that fall in visible. The block must fit on the strip end-to-end
     * (`anchor.row + h <= visibleRows + bufferBelow`); the portion above
     * visible is clipped by the reel mask. This is the "tail-visible"
     * partial-landing pattern.
     */
    bufferAbove?: (string | undefined)[];
    /**
     * Buffer-below target symbols. `bufferBelow[0]` is the slot closest to the
     * visible bottom row; later indices go further below. Up to `bufferSymbols`
     * entries are honored.
     *
     * Big-symbol stubs may sit here. A block anchored at the last visible row
     * with `h > 1` will have its non-anchor cells spill into `bufferBelow`
     * automatically. You can also place an anchor here, but the block then
     * lies entirely off-screen (legal but invisible).
     */
    bufferBelow?: (string | undefined)[];
}
/**
 * Materialize a `ColumnTarget` into the internal `string[]` form the
 * engine pipeline runs on. Buffer-above entries map to negative-index
 * string properties (`arr[-1]`, `arr[-2]`, ...); buffer-below entries
 * map to indices `>= visible.length`.
 */
export declare function columnTargetToArray(target: ColumnTarget): string[];
/**
 * Validate that a target grid does not carry more `bufferAbove` / `bufferBelow`
 * entries than the engine can consume. Throws a `RangeError` with a
 * column-pointing message if it does; otherwise a no-op.
 *
 * Background: without this check the failure is silent. `columnTargetToArray`
 * materializes `bufferAbove[k]` as `arr[-1-k]` and `bufferBelow[k]` as
 * `arr[visible.length + k]`, but downstream the pipeline only reads the first
 * `bufferAbove` negative-index slots and the first `bufferBelow` post-visible
 * slots. Extra entries land in the array, are dropped at the next clone, and
 * never reach the reel. Failing here at the entry point is cheaper than a
 * "why did not my target land" debugging session.
 *
 * `callerLabel` shows up in the thrown message so the caller knows which
 * public API surfaced the error.
 */
export declare function assertBufferCountsInRange(grid: ColumnTarget[], bufferAbovePerReel: ReadonlyArray<number>, bufferBelowPerReel: ReadonlyArray<number>, callerLabel: string): void;
//# sourceMappingURL=ColumnTarget.d.ts.map