// Documentation: https://adssystem.store/scripts/final-url-health-check/
// Writes: Nothing. Read-only.

// Final URL health check
// Collects every unique final URL on enabled ads and keywords, requests each one
// following redirects, and reports anything that does not return 200 or that
// ends up on a different host than it started on.
// Read-only: it reports, it does not pause anything.

var EMAIL      = '';    // leave empty to log only
var MAX_URLS   = 300;   // scripts stop at 30 minutes; raise only if you have headroom
var USER_AGENT = 'Mozilla/5.0 (compatible; AdsScriptUrlCheck/1.0)';

function main() {
  var urls = {};

  collect(
    'SELECT ad_group_ad.ad.final_urls ' +
    'FROM ad_group_ad ' +
    'WHERE ad_group_ad.status = "ENABLED" ' +
    '  AND ad_group.status = "ENABLED" ' +
    '  AND campaign.status = "ENABLED"',
    function (r) { return r.adGroupAd && r.adGroupAd.ad ? r.adGroupAd.ad.finalUrls : null; },
    urls
  );

  collect(
    'SELECT ad_group_criterion.final_urls ' +
    'FROM keyword_view ' +
    'WHERE ad_group_criterion.status = "ENABLED" ' +
    '  AND ad_group.status = "ENABLED" ' +
    '  AND campaign.status = "ENABLED"',
    function (r) { return r.adGroupCriterion ? r.adGroupCriterion.finalUrls : null; },
    urls
  );

  var list = Object.keys(urls);
  Logger.log(list.length + ' unique final URLs found.');

  var bad     = [];
  var checked = 0;

  for (var i = 0; i < list.length && checked < MAX_URLS; i++) {
    var url = list[i];
    checked++;
    var result = check(url);
    if (result) { bad.push(result); }
  }

  if (bad.length === 0) {
    Logger.log('All ' + checked + ' URLs returned 200 on the same host.');
    return;
  }

  var lines = ['url\tstatus\tnote'];
  for (var k = 0; k < bad.length; k++) {
    lines.push(bad[k].url + '\t' + bad[k].status + '\t' + bad[k].note);
  }

  var body = AdsApp.currentAccount().getName() + '\n' +
             checked + ' URLs checked, ' + bad.length + ' need attention.\n\n' +
             lines.join('\n');

  Logger.log(body);
  if (EMAIL) {
    MailApp.sendEmail(EMAIL, 'Final URL check - ' + AdsApp.currentAccount().getName(), body);
  }
}

function collect(query, extract, into) {
  var rows = AdsApp.search(query);
  while (rows.hasNext()) {
    var final = extract(rows.next());
    if (!final || !final.length) { continue; }
    for (var i = 0; i < final.length; i++) {
      var u = String(final[i]).split('#')[0];
      if (u) { into[u] = true; }
    }
  }
}

function check(url) {
  try {
    var res = UrlFetchApp.fetch(url, {
      muteHttpExceptions: true,
      followRedirects:    true,
      validateHttpsCertificates: true,
      headers: { 'User-Agent': USER_AGENT }
    });

    var status = res.getResponseCode();
    if (status !== 200) {
      return { url: url, status: status, note: 'non-200 response' };
    }

    var landed = res.getAllHeaders()['Location'];
    if (landed && host(String(landed)) !== host(url)) {
      return { url: url, status: status, note: 'redirected to ' + host(String(landed)) };
    }

    return null;
  } catch (e) {
    return { url: url, status: 'error', note: String(e).substring(0, 120) };
  }
}

function host(url) {
  var m = url.match(/^https?:\/\/([^\/\?]+)/i);
  return m ? m[1].toLowerCase().replace(/^www\./, '') : '';
}