old-website/scripts/mdx-scripts/news-conversion.js

const fs = require("fs");
const jsdom = require("jsdom");
const { JSDOM } = jsdom;
const { window } = new JSDOM("");
global.window = window;
var showdown = require("showdown");
const converterShowdown = new showdown.Converter();
const libxmljs = require("libxmljs");
const getTerm = require("./getTerm.js");

fs.readFile("../../news.xml", "utf8", (err, data) => {
  if (err) {
    console.error(err);
    return;
  }
  data = data.replace(/<!DOCTYPE.*>/, "");
  parseXML(data);
});

fs.mkdir("./markdown-news/", { recursive: true }, (err) => {
  if (err) throw err;
});

const nodeChildrenToString = (nodeChildren) => {
  let string = "";
  for (let i = 0; i < nodeChildren.length; ++i) {
    if (nodeChildren[i].type() === "text") {
      string += nodeChildren[i];
    } else {
      string += nodeChildren[i].toString();
    }
  }
  return string;
};

const parseXML = (XML) => {
  const xmlDoc = libxmljs.parseHtml(XML);
  let newsdefsChildren = xmlDoc.get("//newsdefs").childNodes();
  let currentTerm = "";
  newsdefsChildren.forEach((newsItem) => {
    if (newsItem.type() === "element") {
      const author = newsItem.attr("author").value();
      const date = newsItem.attr("date").value();
      currentTerm = getTerm(date);
      const contentNodes = newsItem.childNodes();
      const content = converterShowdown.makeMarkdown(
        nodeChildrenToString(contentNodes)
      );

      const md = `---
author: '${author.replace(/'/g, "''")}'
date: '${date.replace(/'/g, "''")}'
---
      
${content}`;
      const mdTitle = `${date}-${author}`;
      fs.mkdirSync(
        `./markdown-news/${currentTerm.year}/${currentTerm.term}`,
        { recursive: true },
        (err) => {
          if (err) throw err;
        }
      );
      fs.writeFileSync(
        `./markdown-news/${currentTerm.year}/${currentTerm.term}/${mdTitle}.md`,
        md,
        (err) => {
          if (err) throw err;
        }
      );
    }
  });
};