/**
 * get_web_source: read a public web page (a blog post, newsletter or club or
 * vendor article) as text, with its Wayback Machine snapshots so a citation
 * can point at a fixed copy.
 *
 * Limits:
 * - Only public http(s) addresses. Names that resolve to loopback, private or
 *   link-local addresses are refused, so a page cannot steer the tool at a
 *   service on the user's machine or network.
 * - When a site answers with a bot check (Cloudflare and similar), the tool
 *   stops. It never tries to get past one.
 * - It does not ask the Wayback Machine to save a page; it only reads what is
 *   there and gives the save link.
 */
import { parseHTML } from "linkedom";
import { type PaperContext, type ServiceLog } from "./core.js";
import { type Section } from "./text.js";
export type Lookup = (host: string) => Promise<string[]>;
/** Throws when the URL is not a public http(s) address. */
export declare function checkPublicUrl(raw: string, lookup?: Lookup): Promise<URL>;
export declare function isBotChallenge(status: number, headers: Headers, body: string): boolean;
export type FetchedPage = {
    url: string;
    status: number;
    contentType: string;
    bytes: Uint8Array;
};
export declare class BotChallengeError extends Error {
}
/** GET with redirects followed by hand, so every hop is checked against the address rules. */
export declare function fetchPage(ctx: PaperContext, start: URL, lookup?: Lookup): Promise<FetchedPage>;
export type Snapshot = {
    timestamp: string;
    url: string;
};
export type Snapshots = {
    earliest?: Snapshot;
    latest?: Snapshot;
};
export declare function snapshotDate(timestamp: string): string;
/** The earliest and latest Wayback snapshots of a URL. */
export declare function waybackSnapshots(ctx: PaperContext, url: string): Promise<Snapshots>;
type Doc = ReturnType<typeof parseHTML>["document"];
export type WebPage = {
    title?: string;
    author?: string;
    published?: string;
    modified?: string;
    siteName?: string;
    canonical?: string;
    licence?: string;
    description?: string;
    sections: Section[];
    text: string;
};
/**
 * Replace each MathML formula with one text form. LaTeXML pages (arXiv HTML)
 * carry every formula twice, as rendered MathML and as a LaTeX annotation, so
 * their text read "13×1013\times 10" and a quote with maths in it could not
 * match. A simple formula becomes its rendered text ("13×10", "M=192"), which
 * is what a reader quotes; one with subscripts, fractions and the like
 * becomes its LaTeX source as inline code, which keeps its structure.
 */
export declare function flattenMath(document: Doc): void;
export declare function extractPage(html: string, pageUrl: string): WebPage;
/** The text of a fetched HTML page, PDF or plain-text file. */
export declare function pageText(page: FetchedPage): Promise<WebPage & {
    pdf: boolean;
}>;
export type WebSourceArgs = {
    url: string;
    section?: number;
};
export type LoadedWebSource = {
    url: URL;
    page: FetchedPage;
    content: WebPage & {
        pdf: boolean;
    };
    snapshots: Snapshots;
    snapshotError: unknown;
    fromArchive: boolean;
};
/**
 * Fetch a public page with its Wayback snapshots. When the live page fails
 * (but not at a bot check), read the latest archived copy instead. Returns an
 * error message instead of throwing.
 */
export declare function loadWebSource(ctx: PaperContext, rawUrl: string, log: ServiceLog, lookup?: Lookup): Promise<LoadedWebSource | {
    error: string;
}>;
export type WebSourceResult = {
    text: string;
    isError?: boolean;
};
export declare function readWebSource(ctx: PaperContext, args: WebSourceArgs, log: ServiceLog, lookup?: Lookup): Promise<WebSourceResult>;
export {};