/**
 * Top-level validation: load a dataset root (Layer 1) and run the semantic
 * rules (Layer 2), aggregating every violation.
 */
import { ValidationResult, Violation } from './model';
import { World } from './world';
/**
 * Validate a dataset root.
 *
 * @param rootDir Path to the dataset root.
 * @returns The aggregate result. `ok` is false when any error-level violation exists.
 * @throws Error When `rootDir` is not an existing directory.
 */
export declare function validateRoot(rootDir: string): ValidationResult;
/**
 * Format a single violation as one human-readable line:
 * `LEVEL [CODE] schema.table (field): message`. The location and field
 * segments are omitted when absent.
 *
 * @param violation The violation to format.
 * @returns The formatted line (no leading or trailing whitespace).
 */
export declare function formatViolation(violation: Violation): string;
/** Options for {@link loadValidatedRoot}. */
export interface LoadValidatedOptions {
    /**
     * Called once for each surviving warning-level violation, in load order.
     * Warnings never fail the load; this is the hook for surfacing them.
     */
    readonly onWarning?: (warning: Violation) => void;
}
/**
 * Load a dataset root, run full validation (structural + semantic), and return
 * the validated world. Throws when any error-level violation exists; warnings
 * do not fail the load and are delivered to `options.onWarning` if provided.
 *
 * @param rootDir Path to the dataset root.
 * @param options Optional callbacks, including `onWarning`.
 * @returns The validated World.
 * @throws Error When `rootDir` is not a directory, or any error-level violation is found.
 */
export declare function loadValidatedRoot(rootDir: string, options?: LoadValidatedOptions): World;