/**
 * Domain Selector Presets - Pre-configured selectors for common government sites
 *
 * These presets help extract content more reliably from known government
 * websites by specifying the correct selectors for their structure.
 */
/**
 * Pagination configuration for domain presets
 */
export interface PaginationPresetConfig {
    /** Pagination type */
    type: 'query_param' | 'path_segment' | 'date_range' | 'reference_based';
    /** Parameter name for pagination (e.g., 'page', 'offset', 'cursor') */
    paramName?: string;
    /** Starting value for first page */
    startValue?: number | string;
    /** Increment for page/offset types */
    increment?: number;
    /** CSS selector for next button (for button-based pagination) */
    nextButtonSelector?: string;
    /** API endpoint for paginated results (if different from page URL) */
    apiEndpoint?: string;
    /** Path to data array in API response */
    responseDataPath?: string;
    /** Path to total count in API response */
    totalCountPath?: string;
    /** Path to has-more indicator in API response */
    hasMorePath?: string;
    /** Path to next cursor/token in API response */
    nextCursorPath?: string;
    /** Items per page (for calculating total pages) */
    itemsPerPage?: number;
    /** Date-based pagination config (for legal document registries) */
    dateConfig?: {
        /** Date parameter name */
        paramName: string;
        /** Date format (e.g., 'YYYY-MM-DD', 'YYYYMMDD') */
        format: string;
        /** Start from newest or oldest */
        direction: 'newest_first' | 'oldest_first';
    };
    /** Notes about pagination behavior */
    notes?: string;
}
export interface DomainPreset {
    domain: string;
    name: string;
    selectors: {
        content: string;
        title?: string;
        navigation?: string;
        footer?: string;
        sidebar?: string;
        breadcrumb?: string;
        lastUpdated?: string;
        tables?: string;
    };
    waitStrategy?: 'load' | 'domcontentloaded' | 'networkidle';
    cookies?: {
        name: string;
        value: string;
        domain: string;
    }[];
    /** Pagination configuration for multi-page content */
    pagination?: PaginationPresetConfig;
    notes?: string;
}
/**
 * Spanish government site presets
 */
export declare const SPAIN_PRESETS: DomainPreset[];
/**
 * US government site presets
 */
export declare const US_PRESETS: DomainPreset[];
/**
 * EU and international presets
 */
export declare const EU_PRESETS: DomainPreset[];
/**
 * Legal document site presets with specialized pagination patterns
 */
export declare const LEGAL_PRESETS: DomainPreset[];
/**
 * All presets combined
 */
export declare const ALL_PRESETS: DomainPreset[];
/**
 * Find the best preset for a URL
 */
export declare function findPreset(url: string): DomainPreset | undefined;
/**
 * Get content selector for a URL (falls back to common selectors)
 */
export declare function getContentSelector(url: string): string;
/**
 * Get wait strategy for a URL
 */
export declare function getWaitStrategy(url: string): 'load' | 'domcontentloaded' | 'networkidle';
/**
 * Get pagination preset configuration for a URL
 */
export declare function getPaginationPreset(url: string): PaginationPresetConfig | undefined;
/**
 * Check if a domain has a pagination preset configured
 */
export declare function hasPaginationPreset(url: string): boolean;
/**
 * Get all domains that have pagination presets configured
 */
export declare function getDomainsWithPagination(): string[];
/**
 * Get elements to remove for cleaner extraction
 */
export declare function getRemovalSelectors(url: string): string[];
/**
 * Registry for adding custom presets at runtime
 */
declare class PresetRegistry {
    private customPresets;
    add(preset: DomainPreset): void;
    remove(domain: string): boolean;
    find(url: string): DomainPreset | undefined;
    list(): DomainPreset[];
}
export declare const presetRegistry: PresetRegistry;
export {};
//# sourceMappingURL=domain-presets.d.ts.map