blogger-to-hugo.js

blogger-to-hugo.js

const { XMLParser } = require("fast-xml-parser");
const TurndownService = require("turndown");
const moment = require("moment");
const shell = require("shelljs");
const fetch = require("node-fetch");
const yaml = require("js-yaml");

const fs = require("fs");
const path = require("path");

const OUT_DIR = "out";

const tds = new TurndownService({ codeBlockStyle: "fenced", fence: "```" });
tds.addRule("wppreblock", {
  filter: ["pre"],
  replacement: function (content) {
    return "```\n" + content + "\n```";
  },
});

const parser = new XMLParser({
  ignoreAttributes: false,
});
// const parseString = util.promisify(parser.parseString);

function normalizeFilename(title) {
  const specialCharsMap = {
    "<": "〈",
    ">": "〉",
    ":": "：",
    '"': "〃",
    "/": "／",
    "\\": "＼",
    "|": "｜",
    "?": "？",
    "*": "＊",
    ".": "．",
    "\0": "",
  };

  return title.replace(
    /[<>:"\/\\|?.\*\x00-\x1F]/g,
    (match) => specialCharsMap[match]
  );
}

async function transformImages(articleFolder, htmlContent) {
  let modifiedHtml = htmlContent;
  shell.mkdir("-p", `${articleFolder}/images`);
  const pattern = /<img.*?src="(.*?)"/g;
  let matched;
  let i = 0;
  for (; (matched = pattern.exec(htmlContent)) !== null; i++) {
    const imageUrl = matched[1];
    try {
      const response = await fetch(imageUrl);
      const buffer = await response.buffer();
      const ext = path.extname(imageUrl);
      const filename = `${i}${ext}`;
      const imagePath = `${articleFolder}/images/${filename}`;
      fs.writeFileSync(imagePath, buffer);
      modifiedHtml = modifiedHtml.replace(imageUrl, `images/${filename}`);
    } catch (e) {
      console.error(e.message);
    }
  }

  if (i === 0) {
    shell.rm("-rf", `${articleFolder}/images`);
  }

  return modifiedHtml;
}

async function main() {
  shell.rm("-rf", OUT_DIR);
  shell.mkdir("-p", OUT_DIR);
  const result = await parser.parse(shell.cat("blog-backup.xml"));
  const posts = result.feed.entry
    .filter((entry) => {
      return (
        entry.id.indexOf(".post-") > 0 &&
        !entry.hasOwnProperty("thr:in-reply-to")
      );
    })
    .filter(
      (entry) =>
        !(entry["app:control"] && entry["app:control"]["app:draft"] == "yes")
    );
  console.log(`found ${posts.length} posts`);
  for (let i = 0; i < posts.length; i++) {
    const post = posts[i];
    const date = moment(post.published).format("YYYY-MM-DD");
    const filenameTitle = normalizeFilename(post.title["#text"] || date);
    const articleFolderName = `${date}_${filenameTitle}`;
    const articleFolder = `${OUT_DIR}/${articleFolderName}`;
    console.log(i, articleFolderName);
    shell.mkdir("-p", articleFolder);
    const modifiedContent = await transformImages(
      articleFolder,
      post.content["#text"]
    );
    let tags = [];
    if (Array.isArray(post.category)) {
      tags = post.category
        .filter(
          (category) =>
            category["@_scheme"] === "http://www.blogger.com/atom/ns#"
        )
        .map((category) => category["@_term"]);
    }
    const markdown = tds.turndown(modifiedContent);
    const title = post.title["#text"] || date;

    const categories = ["tech"];

    const header = yaml.dump({ title, date, tags, categories });
    const content = `---\n${header}---\n${markdown}`;
    fs.writeFileSync(`${articleFolder}/index.md`, content);
  }
}

main();