/**
 * Structured Data Extractor
 *
 * Comprehensive extraction of structured metadata from HTML pages.
 * Inspired by Python's extruct library, supports multiple formats:
 *
 * - JSON-LD (Google's preferred format)
 * - Microdata (schema.org HTML5 attributes)
 * - OpenGraph (Facebook meta tags)
 * - Twitter Cards (Twitter meta tags)
 * - Dublin Core (Document metadata)
 * - RDFa Lite (Resource Description Framework)
 *
 * @see https://github.com/scrapinghub/extruct
 * @see https://schema.org/
 * @see https://ogp.me/
 */
/**
 * Extracted structured data result
 */
export interface StructuredDataResult {
    /** JSON-LD structured data */
    jsonLd: JsonLdItem[];
    /** Microdata items */
    microdata: MicrodataItem[];
    /** OpenGraph metadata */
    openGraph: OpenGraphData;
    /** Twitter Card metadata */
    twitterCard: TwitterCardData;
    /** Dublin Core metadata */
    dublinCore: DublinCoreData;
    /** RDFa Lite items */
    rdfa: RdfaItem[];
    /** Combined/normalized product data (if applicable) */
    product?: ProductData;
    /** Combined/normalized article data (if applicable) */
    article?: ArticleData;
    /** Combined/normalized organization data (if applicable) */
    organization?: OrganizationData;
    /** Raw meta tags */
    meta: Record<string, string>;
}
export interface JsonLdItem {
    '@context'?: string;
    '@type': string;
    [key: string]: unknown;
}
export interface MicrodataItem {
    type: string;
    properties: Record<string, unknown>;
}
export interface OpenGraphData {
    title?: string;
    type?: string;
    url?: string;
    image?: string | string[];
    description?: string;
    siteName?: string;
    locale?: string;
    [key: string]: unknown;
}
export interface TwitterCardData {
    card?: string;
    site?: string;
    creator?: string;
    title?: string;
    description?: string;
    image?: string;
    [key: string]: unknown;
}
export interface DublinCoreData {
    title?: string;
    creator?: string;
    subject?: string;
    description?: string;
    publisher?: string;
    contributor?: string;
    date?: string;
    type?: string;
    format?: string;
    identifier?: string;
    source?: string;
    language?: string;
    relation?: string;
    coverage?: string;
    rights?: string;
}
export interface RdfaItem {
    typeof: string;
    properties: Record<string, unknown>;
}
/**
 * Normalized product data from any structured data source
 */
export interface ProductData {
    name?: string;
    description?: string;
    brand?: string;
    sku?: string;
    gtin?: string;
    mpn?: string;
    price?: number;
    priceCurrency?: string;
    availability?: string;
    condition?: string;
    image?: string | string[];
    url?: string;
    rating?: {
        value: number;
        count: number;
        best?: number;
    };
    reviews?: Array<{
        author?: string;
        rating?: number;
        text?: string;
        date?: string;
    }>;
    category?: string;
    offers?: Array<{
        price: number;
        currency: string;
        seller?: string;
        availability?: string;
    }>;
}
/**
 * Normalized article data from any structured data source
 */
export interface ArticleData {
    headline?: string;
    description?: string;
    author?: string | string[];
    datePublished?: string;
    dateModified?: string;
    publisher?: string;
    image?: string | string[];
    wordCount?: number;
    articleBody?: string;
    section?: string;
    keywords?: string[];
}
/**
 * Normalized organization data
 */
export interface OrganizationData {
    name?: string;
    url?: string;
    logo?: string;
    description?: string;
    address?: string;
    phone?: string;
    email?: string;
    sameAs?: string[];
}
/**
 * Extract all structured data from HTML
 */
export declare function extractStructuredData(html: string, baseUrl?: string): StructuredDataResult;
/**
 * Check if page has any structured data
 */
export declare function hasStructuredData(html: string): boolean;
/**
 * Quick check for product structured data
 */
export declare function hasProductData(html: string): boolean;
/**
 * Quick check for article structured data
 */
export declare function hasArticleData(html: string): boolean;
export default extractStructuredData;
//# sourceMappingURL=structured-data-extractor.d.ts.map