find-classes.js

"use strict";

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

const jsdom = require("jsdom");
const { JSDOM } = jsdom;

/* get all ".html" files under the given directory into an array */
function getAllFiles(dirPath, files) {
  const dir = fs.readdirSync(dirPath);
  files = files || [];
  for (const file of dir) {
    if (fs.statSync(dirPath + "/" + file).isDirectory()) {
      files = getAllFiles(dirPath + "/" + file, files)
    } else {
      if (file.endsWith(".html")) {
        files.push(path.join(__dirname, dirPath, "/", file))
      }
    }
  }

  return files
}

const allDocPaths = getAllFiles(process.argv[2]);
const classToFind = process.argv[3];

/* scrape off the front matter.
We're considering this very roughly as the second occurrence of "---" in the file.*/
function removeFrontMatter(doc) {
  const regexp = /---/g;
  const matchArray = [...doc.matchAll(regexp)];
  let end = matchArray[1].index + 3;
  return doc.substr(end);
}

for (const docPath of allDocPaths) {
  let docHTML = fs.readFileSync(docPath, {encoding: "utf-8"});
  docHTML = removeFrontMatter(docHTML);
  const { window } = new JSDOM(docHTML);
  const found = window.document.querySelectorAll(`.${classToFind}`);
  for (const item of found) {
    console.log(docPath);
  }
  window.close();
}