getNetworkLogs.js

import { browserStackFetch } from './lib/auth.js';
import { isMainModule, requireArg } from './lib/cli.js';

const IGNORED_EXTENSIONS = /\.(png|jpg|jpeg|gif|svg|webp|ico|css|js|woff|woff2|ttf|otf)$/i;

/**
 * Filter HAR entries down to main HTML page navigations.
 * @param {object[]} entries
 * @returns {object[]}
 */
export function extractMainPages(entries) {
  return entries.filter((entry) => {
    const url = entry.request?.url || '';
    const mimeType = entry.response?.content?.mimeType || '';

    if (IGNORED_EXTENSIONS.test(url)) return false;
    if (mimeType && !mimeType.includes('text/html')) return false;

    return true;
  });
}

/**
 * Fetch network logs for a session and return main page entries.
 * @param {string} sessionId
 * @returns {Promise<object[]>}
 */
export async function getNetworkLogs(sessionId) {
  const data = await browserStackFetch(
    `https://api.browserstack.com/automate/sessions/${sessionId}/networklogs`
  );
  const entries = data.log?.entries || [];
  return extractMainPages(entries);
}

/**
 * Convenience: return only page URLs from a session's network logs.
 * @param {string} sessionId
 * @returns {Promise<string[]>}
 */
export async function getPageUrls(sessionId) {
  const pages = await getNetworkLogs(sessionId);
  return pages.map((entry) => entry.request?.url).filter(Boolean);
}

async function main() {
  const sessionId = requireArg(
    'sessionId',
    process.argv[2] || process.env.BROWSERSTACK_SESSION_ID,
    'node getNetworkLogs.js <sessionId>'
  );

  const urls = await getPageUrls(sessionId);
  console.log(`Retrieved ${urls.length} page URL(s) for session ${sessionId}:`);
  console.log(JSON.stringify(urls, null, 2));
}

if (isMainModule(import.meta.url)) {
  main().catch((error) => {
    console.error('Error fetching network logs:', error.message);
    process.exit(1);
  });
}