Statcounter + BCD cleanup tools

brwoser-releases-to-csv.mjs

import { query } from '../src/index.js';  // You'll have to sort this bit out yourself. It's the pre-TS version of https://github.com/mdn/browser-compat-data/blob/main/utils/query.ts
import TSV from 'tsv';

const browsers = [
  'chrome',
  'edge',
  'firefox',
  'opera',
  'safari',
  'samsunginternet_android',
].map(q => query(`browsers.${q}`));

const lines = [];

for (const data of browsers) {
  for (const [version, details] of Object.entries(data.releases)) {
    lines.push({
      browser: data.name,
      version,
      concatenate: data.name + ' ' + version,
      release_date: details.release_date,
    });
  }
}

console.log(TSV.stringify(lines));

clean-statcounter.mjs

import { parse } from 'csv-parse/sync';
import TSV from 'tsv';
import { readFileSync } from 'node:fs';

// e.g., https://gs.statcounter.com/browser-version-market-share
const PATH_TO_STATCOUNTER_CSV = 'statcounter-july-2023.csv';

const re = {
  edge: /(Edge) ([\d.]+)/,
  chrome: /(Chrome) ([\d.]+)/,
  safari: /(Safari) ([\d.]+)/,
  firefox: /(Firefox) ([\d.]+)/,
  samsunginternet_android: /(Samsung Internet) ([\d.]+)/,
  opera: /(Opera) ([\d.]+)/,
};

function cleanVersionNumber(browser, version) {
  if (version === -1) return '-1';
  if (browser === 'Samsung Internet') return version;
  if (version.endsWith('.0')) return version.match(/(\d+)[.]0/)[1]; // BCD doesn't do .0 version numbers, except for Samsung Internet
  return version;
}

const errBrowser = 'NOT IN BCD';

function splitBrowserVersions(str) {
  // Browsers where Statcounter does something odd or combines releases
  if (
    [
      'Chrome for Android',
      'Safari iPhone',
      'Chrome for iPhone',
      'Safari iPad',
      'Android 0',
      'Chrome iPad',
      '360 Safe Browser 0',
      'Mozilla 0',
      'Sogou Explorer 0',
      'Sony PS4 0',
      'Unknown 0',
      'Other',
    ].includes(str)
  ) {
    return [str, errBrowser, '-1'];
  }

  // Browsers unknown to BCD
  if (
    str.startsWith('UC Browser ') ||
    str.startsWith('Yandex Browser ') ||
    str.startsWith('QQ Browser ') ||
    str.startsWith('Coc Coc ') ||
    str.startsWith('KaiOS ') ||
    str.startsWith('Whale ') ||
    str.startsWith('Opera Mini ') ||
    str.startsWith('Instabridge ') ||
    str.startsWith('IE ') ||
    str.startsWith('Puffin ')
  ) {
    return [str, errBrowser, '-1'];
  }

  for (const exp of Object.values(re)) {
    const result = str.match(exp);
    if (Array.isArray(result)) {
      return [str, ...result.slice(1)];
    }
  }

  throw Error(`unhandled string: ${str}`);
}

const statcounter = parse(
  readFileSync(PATH_TO_STATCOUNTER_CSV),
  {
    columns: true,
    skip_empty_lines: true,
  },
);

const lines = [];

for (const row of statcounter) {
  const browserVersion = row['Browser Version'];
  const marketSharePerc = row['Market Share Perc. (July 2023)'];
  const [original, browser, version] = splitBrowserVersions(browserVersion);
  lines.push({
    original,
    browser,
    version: cleanVersionNumber(browser, version),
    marketSharePerc,
  });
}

console.log(TSV.stringify(lines));