2022-11-25 21:05:58 +00:00
|
|
|
import { Parser } from "htmlparser2";
|
|
|
|
import { DomHandler } from "domhandler";
|
2022-11-25 21:47:57 +00:00
|
|
|
import { readdir, readFile } from "fs/promises";
|
|
|
|
import { join } from "path";
|
2022-11-25 21:05:58 +00:00
|
|
|
|
2022-11-25 21:47:57 +00:00
|
|
|
const { argv } = process;
|
|
|
|
const dirPath = argv[2] || ".";
|
2022-11-25 21:05:58 +00:00
|
|
|
|
2022-11-25 21:47:57 +00:00
|
|
|
function recurseElement(report, el) {
|
2022-11-25 21:05:58 +00:00
|
|
|
if (el.name === "a") {
|
|
|
|
if (el.attribs.href) {
|
2022-11-25 21:47:57 +00:00
|
|
|
checkUrl(report, "link", el, el.attribs.href);
|
2022-11-25 21:05:58 +00:00
|
|
|
} else {
|
2022-11-25 21:47:57 +00:00
|
|
|
report.warnings.push(`Link with no href: ${getHtml(el)}`);
|
2022-11-25 21:05:58 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
if (["audio", "video", "img"].includes(el.name)) {
|
|
|
|
if (el.attribs.src) {
|
2022-11-25 21:47:57 +00:00
|
|
|
checkUrl(report, el.name, el, el.attribs.src);
|
2022-11-25 21:05:58 +00:00
|
|
|
} else {
|
2022-11-25 21:47:57 +00:00
|
|
|
report.warnings.push(`${el.name} with no src: ${getHtml(el)}`);
|
2022-11-25 21:05:58 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
for (const child of el.children) {
|
|
|
|
if (child.type === "tag") {
|
2022-11-25 21:47:57 +00:00
|
|
|
recurseElement(report, child);
|
2022-11-25 21:05:58 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-11-25 21:47:57 +00:00
|
|
|
function checkUrl(report, type, el, url) {
|
2022-11-25 21:08:54 +00:00
|
|
|
if (isHttp(url)) {
|
2022-11-25 21:47:57 +00:00
|
|
|
report.warnings.push(`HTTP/S ${type}: ${getHtml(el)}`);
|
2022-11-25 21:08:54 +00:00
|
|
|
} else if (isAbsolute(url)) {
|
2022-11-25 21:47:57 +00:00
|
|
|
report.warnings.push(`Absolute ${type}: ${getHtml(el)}`);
|
2022-11-25 21:08:54 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
function isHttp(url) {
|
|
|
|
const r = /^(https?:\/\/|\/\/)/;
|
|
|
|
return r.test(url);
|
|
|
|
}
|
|
|
|
function isAbsolute(url) {
|
|
|
|
return url.startsWith("/");
|
|
|
|
}
|
|
|
|
|
|
|
|
function getHtml(el) {
|
|
|
|
// return rawHtml.slice(el.startIndex, el.endIndex);
|
|
|
|
let text = "";
|
|
|
|
for (const child of el.children) {
|
|
|
|
if (child.type === "text") text += child.data.trim();
|
|
|
|
else if (child.type === "tag") text += getHtml(child);
|
|
|
|
}
|
|
|
|
return text;
|
|
|
|
}
|
|
|
|
|
2022-11-25 21:47:57 +00:00
|
|
|
function processFile(content) {
|
2022-11-25 21:08:54 +00:00
|
|
|
return new Promise((resolve, reject) => {
|
2022-11-25 21:47:57 +00:00
|
|
|
const handler = new DomHandler(
|
|
|
|
(error, dom) => {
|
|
|
|
if (error) {
|
|
|
|
reject(error);
|
|
|
|
} else {
|
|
|
|
let report = {
|
|
|
|
warnings: [],
|
|
|
|
};
|
|
|
|
for (const el of dom) {
|
|
|
|
if (el.type === "tag") {
|
|
|
|
recurseElement(report, el);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
resolve(report);
|
|
|
|
}
|
|
|
|
},
|
|
|
|
{ withEndIndices: true, withStartIndices: true }
|
|
|
|
);
|
|
|
|
const parser = new Parser(handler);
|
|
|
|
parser.parseComplete(content);
|
2022-11-25 21:08:54 +00:00
|
|
|
});
|
|
|
|
}
|
2022-11-25 21:47:57 +00:00
|
|
|
|
|
|
|
async function recurseDirectory(path) {
|
|
|
|
const dir = await readdir(path, { withFileTypes: true });
|
|
|
|
for (const file of dir) {
|
|
|
|
const filePath = join(path, file.name);
|
|
|
|
if (file.isDirectory()) recurseDirectory(filePath);
|
|
|
|
else {
|
|
|
|
if (!file.name.endsWith(".html")) continue;
|
|
|
|
const content = await readFile(filePath, "utf-8");
|
|
|
|
console.time(filePath);
|
|
|
|
await processFile(content);
|
|
|
|
console.timeEnd(filePath);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
await recurseDirectory(dirPath);
|