A script that imports Disqus posts into a format usable by Staticman

import_disqus.mjs

// import_disqus.mjs
// =================
//
// Converts Disqus XML export to a format compatible with Staticman (https://staticman.net/)
//
// Requires Node.js and the following packages:
//
// npm install xml2js crypto-js strip-indent
//
// (c) 2024 Carlos Duarte Do Nascimento (https://chester.me)
// Released under the MIT License (https://opensource.org/licenses/MIT)

import fs from 'fs';
import { Parser } from 'xml2js';
import path from 'path';
import stripIndent from 'strip-indent';
import CryptoJS from 'crypto-js';

var parser = new Parser();
fs.readFile('data.xml', function(err, data) {
    parser.parseString(data, function (err, result) {
        // Yes, Disqus calls a post a "thread" and a comment a "post" 🤦;
        // let's first untangle this mess
        const posts = result.disqus.thread;
        const comments = result.disqus.post;

        // All we need from posts is the URL and slug, so let's build a
        // dictionary to quickly retrieve them from the post ID
        const postInfo = posts.reduce((obj, post) => {
            const slug = post.link[0].replace(/\/$/, "").split('/').slice(-1)[0].replace(/\.html$/, "");
            obj[post['$']['dsq:id']] = {
                "slug": slug,
                "url": post.link[0]
            }
            return obj;
        }, {});

        // Now we can create one file for each comment
        comments.forEach(comment => {
            if (comment.isDeleted[0] === 'true' || comment.isSpam[0] === 'true') {
                return;
            }

            const disqus_comment_id = comment['$']['dsq:id'];
            const disqus_unix_timestamp = new Date(comment.createdAt[0]).getTime() / 1000;
            const post = postInfo[comment.thread[0]['$']['dsq:id']];
            const username = comment.author[0].username ? comment.author[0].username[0] : "";

            const yml = stripIndent(`
                _id: ${disqus_comment_id}
                _parent: ${post.url}
                replying_to_uid: '${comment.parent ? comment.parent[0]['$']['dsq:id'] : ''}'
                message: ${singleQuote(comment.message[0])}
                name: ${singleQuote(comment.author[0].name[0])}
                email: '${username == "chesterbr" ? "3a49ee98333d753103cf708e40d36984" : CryptoJS.MD5(username+"@chester.me").toString()}'
                hp: ''
                date: ${disqus_unix_timestamp}
            `).replace(/^\s+/, '') // Nix first line
              .replace(/[\x00-\x09\x0B\x0C\x0E-\x1F\x7F-\x9F]/g, ""); // 🔪 invalid Unicode characters

            // Use comment ID as milisseconds in filename for uniqueness
            const dirname = "comments/" + post.slug;
            const filename =  `entry${disqus_unix_timestamp * 1000 + disqus_comment_id % 1000}.yml`;

            if (!fs.existsSync(dirname)){
                fs.mkdirSync(dirname, { recursive: true });
            }
            fs.writeFileSync(path.join(dirname, filename), yml);
        });
    });
});

function singleQuote(str) {
    return `'${str.replace(/'/g, "''")}'`;
}