bookworms/src/convert-bookmarks.js

A cli tool for centralising and generating bookmarks

// TODO this is a bit messy - will rewrite when I'm not feeling sick
// This is a utility to convert existing browser bookmarks into YAML
// It can't pull descriptions as they are no required in browsers
import events from "events";
import readline from "readline";
import stringify from "json-stringify-safe";
import { createReadStream } from "fs";
import { v4 as uuidv4 } from "uuid";
import { xml2json } from "xml2json-light";
import yaml from "js-yaml";
import { writeBookmark } from "./save-bookmarks.js";

const parseHTMLtoJSON = async (path) => {
  let bookmarks = {};
  let currentNode = bookmarks;
  let latestNode = bookmarks;
  let idQueue = [];
  let subId = uuidv4().replace(/-/g, "");

  try {
    const rl = readline.createInterface({
      input: createReadStream(path),
      crlfDelay: Infinity,
    });

    // rewrite this to make it cleaner and more testable
    // This was heavily borrowed from https://github.com/CCharlieLi/bookmark-parser
    rl.on("line", (line) => {
      line = line
        .replace(/(<DT>.+)/g, "$1</DT>")
        .replace(/(<!DOCTYPE.+)/g, "")
        .replace(/(<META.+)/g, "")
        .replace(/<p>/g, "")
        .replace(/<HR>/g, "");

      let str;

      if (line.match(/^<TITLE>(.+)<\/TITLE>$/)) {
        str = line.match(/^<TITLE>(.+)<\/TITLE>$/);
        latestNode = currentNode[str[1]] = {};
      }

      if (/<DL>/.test(line)) {
        const prevNode = currentNode;
        currentNode = latestNode;
        currentNode.prevNode = prevNode;
        currentNode.children = [];
        idQueue.push(subId);
        subId = 0;
      }

      if (/<\/DL>/.test(line)) {
        currentNode = currentNode.prevNode;
        subId = parseInt(idQueue[idQueue.length - 1], 10);
        idQueue.pop();
      }

      if (line.match(/<DT>(<H3.+<\/H3>)<\/DT>/)) {
        str = line.match(/<DT>(<H3.+<\/H3>)<\/DT>/);
        const tmp = xml2json(str[1]);
        currentNode.children.push({
          label: tmp["H3"]["_@ttribute"],
          type: "folder",
        });
        latestNode = currentNode.children[currentNode.children.length - 1];
      }

      if (line.match(/<DT>(<A.+<\/A>)<\/DT>/)) {
        str = line.match(/<DT>(<A.+<\/A>)<\/DT>/);
        const tmp = xml2json(str[1].replace(/&/g, "%26"));
        currentNode.children.push({
          label: tmp["A"]["_@ttribute"],
          type: "bookmark",
          href: tmp["A"]["HREF"],
        });
        latestNode = currentNode.children[currentNode.children.length - 1];
      }

      if (line.match(/<DD>(.+)/)) {
        str = line.match(/<DD>(.+)/);
        latestNode.description = str[1];
      }
    });

    await events.once(rl, "close");

    // this returns the entire bookmarks object
    // so far only really tested chrome
    return JSON.parse(stringify(bookmarks)).Bookmarks.children[0].children;
  } catch (err) {
    // need to write tests here
    console.error(err);
  }
};

const generateBookwormJSON = (json) => {
  return {
    label: "Bookworms",
    description: "These bookmarks were generated by a browser export",
    ...traverseStructure(json),
  };
};

// I need to clean this up but it was so gross making it work
// will come back to it
const traverseStructure = (json) => {
  const arr = {};
  json.forEach((entry) => {
    const { type, label } = entry;
    if (type === "folder") {
      if (!arr.folders) {
        arr.folders = [];
      }
      const response = traverseStructure(entry.children);
      const folder = {
        label,
      };
      if (response.folders) {
        folder.folders = response.folders;
      }
      if (response.bookmarks) {
        folder.bookmarks = response.bookmarks;
      }
      arr.folders.push(folder);
    }
    if (type === "bookmark") {
      if (!arr.bookmarks) {
        arr.bookmarks = [];
      }
      arr.bookmarks.push({
        label,
        href: entry.href,
      });
    }
  });
  return arr;
};

const convertHTMLtoYAML = async (path) => {
  const parsedJSON = await parseHTMLtoJSON(path);
  const bookwormJSON = generateBookwormJSON(parsedJSON);
  return yaml.dump(bookwormJSON);
};

const createBookmarks = async (path, directory, filename) => {
  console.log(`Writing ${filename} to ${directory}`);
  writeBookmark(await convertHTMLtoYAML(path), directory, filename);
};

export {
  parseHTMLtoJSON,
  generateBookwormJSON,
  traverseStructure,
  convertHTMLtoYAML,
  createBookmarks,
};