Count tags in HTML files using Node

count-tags.js

'use strict';

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

console.log('🛠  Scanning HTML files...');

const files = getAllFiles('./', ['.html']);
const tagCount = countTags(files);

console.log(`tags:`, tagCount);

console.log(`found ${Object.keys(tagCount).length} unique tags in ${files.length} files`);

function getAllFiles(initialPath, fileExtensions) {
    var fullPaths = [];
    var files = fs.readdirSync(initialPath);

    files.forEach((file) => {
      var joinedPath = path.join(initialPath, file);

      if (fs.lstatSync(joinedPath).isDirectory()) {
          fullPaths = fullPaths.concat(getAllFiles(joinedPath, fileExtensions));
      } else {
          fileExtensions.forEach((fileExtension) => {
              if (file.indexOf(fileExtension) >= 0 || fileExtension == null) {
                  fullPaths.push(joinedPath);
              }
          });
      }
    });

    return fullPaths;
};

function countTags(files) {
    const tagCount = {};

    for (let index in files) {
        if (files.hasOwnProperty(index)) {
            const file = files[index];
            const htmlString = fs.readFileSync(file, 'utf8');
            const htmlTagRe = RegExp('<[a-z\-]+((\/>)|( |>))','gi');
            const allTags = [];

            console.log(`finding tags in ${file}`);
            const matches = htmlString.matchAll(htmlTagRe);

            for (const match of matches) {
                let newMatch = match[0].replace(/</g, '').replace(/>/g, '').trim();
                allTags.push(newMatch);
            }

            allTags.forEach(tag => {
                tagCount[tag] = (tagCount[tag] || 0) + 1;
            });
        }
    };

    // Sort and return
    return Object.entries(tagCount).sort(([,a],[,b]) => b-a).reduce((r, [k, v]) => ({ ...r, [k]: v }), {});;
};