import type { JSDOM } from "jsdom";
export interface JsonLdArticle {
    title?: string;
    text?: string;
}
export interface JsonLdScan {
    present: boolean;
    article: JsonLdArticle | null;
}
/**
 * Scan a document's JSON-LD blocks for Article schema, reporting presence and
 * extractable content as separate facts.
 *
 * They must stay separate. `applyPostExtract` previously reported presence as
 * `result.source === "json_ld"` — true only when JSON-LD supplied the page
 * body, which additionally requires an `articleBody` of at least
 * MIN_JSONLD_BODY_CHARS. Sites overwhelmingly publish headline and metadata
 * without a body, so the domain DB recorded json_ld_article present on 0 of 111
 * sampled pages while og:title, checked directly against the HTML, read 40% on
 * the same sample. shouldSkipJsonLdPostExtract then latched those domains off
 * after five samples, making the false negative self-sustaining.
 */
export declare function scanJsonLd(dom: JSDOM): JsonLdScan;
export declare function extractJsonLdArticle(dom: JSDOM): JsonLdArticle | null;