HTML to Markdown: @mozilla/readability + pandoc

dropper.ts

#!/usr/bin/env -S deno run -A --ext ts
import $ from "jsr:@david/[email protected]";
import { parseArgs } from "node:util";
import { Readability } from "npm:@mozilla/[email protected]";
import { JSDOM } from "npm:[email protected]";

const { code } = await $`/bin/which pandoc`.noThrow().quiet();
if (code !== 0) {
    console.error("Command not found: pandoc\nPlease install pandoc first.");
    Deno.exit(1);
}

const { positionals } = parseArgs({
    args: Deno.args,
    allowPositionals: true,
    options: {},
});

const url = positionals[0];
const html = await (await fetch(url)).text();
const doc = new JSDOM(html).window.document;

const article = new Readability(doc).parse();
if (!article) {
    console.error(`Failed to parse article: ${url}`);
    Deno.exit(1);
}

const buf = `<header><h1>${article.title}</h1></header>${article.content}`;
const md = await $`pandoc -f html -t gfm-raw_html`.stdinText(buf).text();
console.log(md);