@wabarc/cairn/dist/html.d.ts

Node package and CLI tool for saving web page as single HTML file

/// <reference types="cheerio" />
import { Options } from './types';
/**
 * @see https://html.spec.whatwg.org/multipage/semantics.html
 */
export declare class HTML {
    opt: Options;
    private rx;
    constructor(opt?: Options);
    /**
     * Process assets within webpage
     *
     * @param {Object} page if error will be thrown
     * @return {cheerio.Cheerio} [cheerio.Cheerio] call .html() parse as html string
     * @api public
     */
    process(page: {
        uri: string;
        html: string;
    }): Promise<cheerio.Root | null>;
    /**
     * setContentSecurityPolicy prevent browsers from requesting any remote
     * resources by setting Content-Security-Policy to only allow from
     * inline element and data URL.
     *
     * @param {Document} $ cheerio.Root
     * @api private
     */
    setContentSecurityPolicy($: cheerio.Root): void;
    /**
     * Removes or replace elements following the configuration.
     *
     * @param {Document} $ cheerio.Root
     * @api private
     */
    applyConfiguration($: cheerio.Root): void;
    /**
     * Convert all noscript to div element.
     *
     * @param {Document} $ cheerio.Root
     * @param {boolean} [markNewDiv] mark to noscript
     * @api private
     */
    convertNoScriptToDiv($: cheerio.Root, markNewDiv?: boolean): void;
    /**
     * Find all comments in document then remove it.
     *
     * @param {Document} $ cheerio.Root
     * @api private
     */
    removeComments($: cheerio.Root): void;
    /**
     * Convert attributes data-src and data-srcset which often found
     * in lazy-loaded images and pictures, into basic attribute
     * src and srcset, so images that can be loaded without JS.
     *
     * @param {Document} $ cheerio.Root
     * @api private
     */
    convertLazyImageAttrs($: cheerio.Root): void;
    /**
     * Converts all relative URL in document into absolute URL.
     * We do this for a, img, picture, figure, video, audio, source, link,
     * embed, iframe and object.
     *
     * @param {Document} $ cheerio.Root
     * @param {string} [url] original request url
     * @api private
     */
    convertRelativeURLs($: cheerio.Root, url: string): void;
    /**
     * Removes integrity attributes from link tags.
     *
     * @param {Document} $ cheerio.Root
     * @api private
     */
    removeLinkIntegrityAttr($: cheerio.Root): void;
    /**
     * Set og:title to title when it empty.
     *
     * @param {Document} $ cheerio.Root
     * @api private
     */
    convertOpenGraph($: cheerio.Root): void;
    /**
     * Set webpage charset as UTF-8
     *
     * @param {Document} $ cheerio.Root
     * @api private
     */
    setCharset($: cheerio.Root): void;
    /**
     * Set source webpage url
     *
     * @param {Document} $ cheerio.Root
     * @param {string} url
     * @api private
     */
    setSource($: cheerio.Root, url: string): void;
    processStyleAttr(node: cheerio.Cheerio, baseURL?: string): Promise<void>;
    processStyleNode(node: cheerio.Cheerio, baseURL?: string): Promise<void>;
    processLinkNode(node: cheerio.Cheerio, baseURL?: string): Promise<void>;
    processURLNode(node: cheerio.Cheerio, attrName: string, baseURL: string): Promise<void>;
    processScriptNode(node: cheerio.Cheerio, baseURL: string): Promise<void>;
    processEmbedNode(node: cheerio.Cheerio, baseURL: string): Promise<void>;
    processMediaNode(node: cheerio.Cheerio, baseURL: string): Promise<void>;
    revertConvertedNoScript($: cheerio.Root): void;
}