link-checker-cli/dist/services/parserHTML.js

CLI tool to check for broken links in a website or project

import { parse } from "parse5";
import { isLink } from "../utils.js";
const IMAGE_EXTENSIONS = ["jpg", "jpeg", "png", "gif", "bmp", "svg", "webp"];
export class ParserHtml {
    constructor() {
        this.parsePage = (page, url) => {
            const document = parse(page);
            const links = [];
            const ids = [];
            this.getLinksFromNode(document, links, ids, url);
            return { links, ids };
        };
        this.getLinksFromNode = (node, links, ids, baseUrl) => {
            if ((node.nodeName === "a" ||
                node.nodeName === "img" ||
                node.nodeName === "link") &&
                node.attrs) {
                let link = node.attrs.find((attr) => attr.name === "href");
                const isImage = node.nodeName === "img";
                const isCss = node.nodeName === "link" &&
                    node.attrs.some((attr) => attr.name === "rel" && attr.value === "stylesheet");
                if (isImage) {
                    link = node.attrs.find((attr) => attr.name === "src");
                }
                if (link) {
                    let { value } = link;
                    value = value.replace(" ", "%20");
                    const isAnchor = value.startsWith("#");
                    if (baseUrl)
                        value = new URL(value, baseUrl).href;
                    if (isImage)
                        links.push({
                            value, type: "image", parent: `${JSON.stringify(node.attrs.find(attr => attr.name === 'class'))} `
                        });
                    else if (isAnchor) {
                        links.push({
                            value, type: "anchor", parent: `${JSON.stringify(node.attrs.find(attr => attr.name === 'class'))} `
                        });
                    }
                    else if (isCss)
                        links.push({
                            value, type: "style", parent: `${JSON.stringify(node.attrs.find(attr => attr.name === 'class'))} `
                        });
                    else if (isLink(value) && baseUrl) {
                        const type = value.startsWith(baseUrl.origin)
                            ? "internal"
                            : "external";
                        // eslint-disable-next-line @typescript-eslint/no-unused-expressions
                        this.isImageExtension(value)
                            ? links.push({
                                value, type: "image", parent: `${JSON.stringify(node.attrs.find(attr => attr.name === 'class'))} `
                            })
                            : links.push({
                                value, type, parent: `${JSON.stringify(node.attrs.find(attr => attr.name === 'class'))} `
                            });
                    }
                }
            }
            if ("attrs" in node) {
                const idAttr = node.attrs?.find((attr) => attr.name === "id");
                if (idAttr && idAttr.value) {
                    ids.push(idAttr.value);
                }
            }
            if ("childNodes" in node) {
                node.childNodes.forEach((node) => this.getLinksFromNode(node, links, ids, baseUrl));
            }
        };
        this.isImageExtension = (link) => {
            const url = new URL(link);
            const extension = url.pathname.split(".").pop()?.toLowerCase();
            return IMAGE_EXTENSIONS.includes(extension || "");
        };
    }
}
export const parserHtml = new ParserHtml();