Detekcia 404

/**
 * @name Link Checker
 *
 * @overview The Link Checker script iterates through all the ads and keywords
 *     in your account and makes sure their URLs do not produce "Page not found"
 *     or other types of error responses. See
 *     https://developers.google.com/adwords/scripts/docs/solutions/link-checker
 *     for more details.
 *
 * @author AdWords Scripts Team [[email protected]]
 *
 * @version 1.2
 */

/**
 * The URL of the tracking spreadsheet. This should be a copy of
 * https://goo.gl/bMrbW5
 */
var SPREADSHEET_URL = 'https://docs.google.com/spreadsheets/d/1CaUSdtVeRW69jW4VFfdCx35w2_Opf0Em3Pe1B8sbmRw/edit#gid=0';

/**
 * Status code for items to be checked.
 */
var STATUSES_TO_CHECK = ['ENABLED', 'PAUSED'];

/**
 * Limit the number of ad and keyword URLs checked by this script per run.
 */
var MAX_URLS_TO_CHECK = 800;

var LABEL_NAME = 'link_checked';

var shelper = new SHelper();
var badUrls = 0;
var numUrlsChecked = 0;

function main() {
  var isFirstRun = isFirstRunOfTheDay();

  if (isFirstRun) {
    dealWithFirstRunOfTheDay();
  }

  if (shelper.config.email.length == 0 &&
      shelper.config.emailPreference != 'Never') {
    Logger.log('WARNING: no email specified, proceeding...');
  }
  if (!shelper.config.checkAds && !shelper.config.checkKeywords) {
    Logger.log('WARNING: requested no keywords and no ads checking. Exiting.');
    return;
  }
  createLinkCheckerLabel();

  var anythingChanged = shelper.config.checkKeywords &&
      checkKeywordUrls(AdWordsApp.keywords(), isFirstRun);
  anythingChanged = (shelper.config.checkAds &&
      checkAdUrls(AdWordsApp.ads(), isFirstRun)) || anythingChanged;

  if (anythingChanged) {
    shelper.flush();
    if (badUrls > 0 && shelper.config.email.length > 0 &&
        shelper.config.emailPreference == 'As soon as an error is discovered') {
      var bad = shelper.spreadsheet.getRangeByName('bad').getValue();
      var good = shelper.spreadsheet.getRangeByName('good').getValue();
      sendReportWithErrors(good, bad);
    }
  } else {
    shelper.spreadsheet.getRangeByName('finished').setValue(
        'All done for the day!');
  }
  Logger.log('Done');
}

function isFirstRunOfTheDay() {
  var date = new Date();
  var lastCheckDate = shelper.dataSheet.getRange(1, 3).getValue();
  return lastCheckDate.length == 0 ||
      date.getYear() != lastCheckDate.getYear() ||
      date.getMonth() != lastCheckDate.getMonth() ||
      date.getDay() != lastCheckDate.getDay();
}

function dealWithFirstRunOfTheDay() {
  var date = new Date();
  Logger.log('The script is running for the first time today...');

  // kill the label.
  var labels = AdWordsApp.labels().withCondition(
      "Name='" + LABEL_NAME + "'").get();
  if (labels.hasNext()) {
    if (AdWordsApp.getExecutionInfo().isPreview()) {
      Logger.log('WARNING: This script needs to remove a label named "%s" ' +
          'to work correctly, but this action cannot be performed in ' +
          'preview mode. Please run this script, or remove the label ' +
          'manually if you wish to preview the script.', LABEL_NAME);
    }

    labels.next().remove();
  }
  // send out yesterday's report
  if (shelper.config.email.length > 0 &&
      (shelper.config.emailPreference == 'Once a day' ||
      shelper.config.emailPreference == 'Once a day if there are errors')) {
    var bad = shelper.spreadsheet.getRangeByName('bad').getValue();
    var good = shelper.spreadsheet.getRangeByName('good').getValue();
    if (shelper.config.emailPreference == 'Once a day') {
      if (bad == 0) {
        MailApp.sendEmail(shelper.config.email,
            'AdWords Link Checker verified ' + good +
            ' URLs on account ' +
            AdWordsApp.currentAccount().getCustomerId() +
            ', all looking good!', '');
      } else {
        sendReportWithErrors(good, bad);
      }
    } else if (shelper.config.emailPreference ==
        'Once a day if there are errors' && bad > 0) {
      sendReportWithErrors(good, bad);
    }
  }
  // reset the spreadsheet
  shelper.spreadsheet.getRangeByName('account_id_dashboard').setValue(
      AdWordsApp.currentAccount().getCustomerId());
  shelper.spreadsheet.getRangeByName('account_id_report').setValue(
      AdWordsApp.currentAccount().getCustomerId());
  shelper.spreadsheet.getRangeByName('date').setValue(date);
  shelper.spreadsheet.getRangeByName('finished').setValue(
      'Checking links...');
  shelper.dataSheet.getRange(
      4, 1, shelper.dataSheet.getMaxRows() - 3, 6).clear();
}

function sendReportWithErrors(good, bad) {
  var emailBody = [];
  emailBody.push('Summary for account ' +
      AdWordsApp.currentAccount().getCustomerId() +
      ': ' + good + ' good URLs, ' + bad + ' bad ones\n');
  emailBody.push('Full report available at ' + shelper.spreadsheet.getUrl() +
      '\n');
  shelper.reset();
  var row = shelper.readRow();
  while (row != null && emailBody.length < 200) {
    if (row[1] >= 300) {
      var entityType = row[4].length > 0 ? 'Keyword: ' : 'Ad: ';
      var entityText = row[4].length > 0 ? row[4] : row[5];
      emailBody.push('Campaign: ' + row[2] + ', Ad Group: ' + row[3] + ', ' +
          entityType + entityText);
      emailBody.push(row[0] + ' - ' + row[1] + ' response code.\n');
    }
    row = shelper.readRow();
  }
  if (emailBody.length >= 200) {
    emailBody.push('Further URLs omitted. Check the report at ' +
        shelper.spreadsheet.getUrl());
  }
  shelper.reset();
  MailApp.sendEmail(shelper.config.email,
    'AdWords Link Checker verified found ' + bad +
    ' bad URLs on account ' + AdWordsApp.currentAccount().getCustomerId() + '',
    emailBody.join('\n'));
}

function checkAdUrls(selector, isFirstRun) {

  var iterator = selector
      .withCondition('CreativeFinalUrls STARTS_WITH_IGNORE_CASE "h"')
      .withCondition('Status IN [' + STATUSES_TO_CHECK.join(',') + ']')
      .withCondition('CampaignStatus IN [' + STATUSES_TO_CHECK.join(',') + ']')
      .withCondition('AdGroupStatus IN [' + STATUSES_TO_CHECK.join(',') + ']')
      .withCondition('LabelNames CONTAINS_NONE ["' + LABEL_NAME + '"]')
      .get();
  Logger.log('Checking %s ads...', iterator.totalNumEntities());

  if (iterator.totalNumEntities() == 0 && isFirstRun) {
    Logger.log('WARNING: The script is not checking any ad URLs. If this is ' +
        'not expected, then ensure that you have enabled ads in your ' +
        'account. Also check the logs for any earlier errors.');
  }

  return checkUrls(iterator);
}

function checkKeywordUrls(selector, isFirstRun) {
  var iterator = selector
      .withCondition('FinalUrls STARTS_WITH_IGNORE_CASE "h"')
      .withCondition('Status IN [' + STATUSES_TO_CHECK.join(',') + ']')
      .withCondition('CampaignStatus IN [' + STATUSES_TO_CHECK.join(',') + ']')
      .withCondition('AdGroupStatus IN [' + STATUSES_TO_CHECK.join(',') + ']')
      .withCondition('LabelNames CONTAINS_NONE ["' + LABEL_NAME + '"]')
      .get();
  Logger.log('Checking %s keywords...', iterator.totalNumEntities());

  if (iterator.totalNumEntities() == 0 && isFirstRun) {
    Logger.log('WARNING: The script is not checking any keyword URLs. This ' +
        'may happen if none of your keywords have a final URL. If this not ' +
         'expected, then ensure that you have enabled keywords in your ' +
         'account. Also check the logs for any earlier errors.');
  }

  return checkUrls(iterator);
}

function checkUrls(iterator) {
  if (!iterator.hasNext()) {
    return false;
  }

  var urlMap = {};

  while (iterator.hasNext()) {
    var entity = iterator.next();

    if (numUrlsChecked > MAX_URLS_TO_CHECK) {
      Logger.log('Checked %s urls. Will resume in next run.', numUrlsChecked);
      break;
    }

    var urls = [entity.urls().getFinalUrl(), entity.urls().getMobileFinalUrl()];
    for (var i = 0; i < urls.length; i++) {
      if (urls[i] == null) {
        continue;
      }
      var lastUrl = encodeURI(urls[i]);
      if (lastUrl in urlMap) {
        continue;
      }
      urlMap[lastUrl] = true;

      var now = new Date().getTime();
      var responseCode = 0;
      try {
        numUrlsChecked++;
        var response = UrlFetchApp.fetch(lastUrl, {muteHttpExceptions: true});
        responseCode = response.getResponseCode();
      } catch (e) {
        // Something went wrong. Since this a script error, let's mark it as
        // 500.
        Logger.log('Could not fetch %s due to an internal error : "%s". ' +
            'Marking this URL as failed, with an error code 500.', lastUrl, e);
        responseCode = 500;
      }
      var then = new Date().getTime();
      Utilities.sleep(then - now);
      if (responseCode < 300) {
        shelper.writeRow(lastUrl, responseCode);
      } else {
        badUrls++;
        if (typeof(entity['getHeadline']) != 'undefined') {
          var adText = entity.getType() == 'TEXT_AD' ?
            entity.getHeadline() + '\n' + entity.getDescription1() + '\n' +
            entity.getDescription2() : entity.getType();
          shelper.writeRow(lastUrl, responseCode,
                           entity.getCampaign().getName(),
                           entity.getAdGroup().getName(),
                           null, adText);
        } else {
          shelper.writeRow(lastUrl, responseCode,
                           entity.getCampaign().getName(),
                           entity.getAdGroup().getName(),
                           entity.getText());
        }
      }

    }
    entity.applyLabel(LABEL_NAME);
  }
  return true;
}

function createLinkCheckerLabel() {
  var labels = AdWordsApp.labels().withCondition(
      "Name='" + LABEL_NAME + "'").get();
  if (!labels.hasNext()) {
    if (AdWordsApp.getExecutionInfo().isPreview()) {
      Logger.log('WARNING: This script needs to create a label named "%s" to ' +
          'work correctly, but this action cannot be performed in preview ' +
          'mode. Please run this script, or create the label manually if you ' +
          'wish to preview the script.', LABEL_NAME);
    }
    AdWordsApp.createLabel(LABEL_NAME,
        "Managed by Link Checker, please don't modify!", '#60e020');
  }
}

// Spreadsheet helper
function SHelper() {
  this.MAX_ROWS = 20000;
  this.BATCH_SIZE = 50;

  Logger.log('Using spreadsheet - %s.', SPREADSHEET_URL);
  this.spreadsheet = SpreadsheetApp.openByUrl(SPREADSHEET_URL);

  this.dataSheet = this.spreadsheet.getSheets()[1];
  this.config = {
    checkAds: this.spreadsheet.getRangeByName('check_ads').getValue() == 'Yes',
    checkKeywords: this.spreadsheet.getRangeByName('check_keywords').
        getValue() == 'Yes',
    email: this.spreadsheet.getRangeByName('email_address').getValue(),
    emailPreference: this.spreadsheet.getRangeByName('email_preference').
        getValue()
  };
  this.globalRow = 4;
  this.cells = null;
  this.localRow = 0;

  this.reset = function() {
    this.globalRow = 4;
    this.cells = null;
    this.localRow = 0;
  };
  this.readRow = function() {
    initCells(this);
    if (this.localRow == this.cells.length) {
      this.globalRow += this.cells.length;
      if (this.globalRow >= this.dataSheet.getMaxRows()) {
        return null;
      }
      this.cells = this.dataSheet.getRange(
          this.globalRow, 2, this.BATCH_SIZE, 6).getValues();
      this.localRow = 0;
    }
    if (this.cells[this.localRow][0].length > 0) {
      return this.cells[this.localRow++];
    } else {
      return null;
    }
  };
  this.writeRow = function() {
    fetchCells(this);
    for (var i = 0; i < arguments.length; i++) {
      this.cells[this.localRow][i] = arguments[i];
    }
  };
  this.flush = function() {
    if (this.cells) {
      this.dataSheet.getRange(this.globalRow, 2, this.cells.length, 6).
          setValues(this.cells);
      this.dataSheet.getRange(1, 1).copyFormatToRange(
          this.dataSheet,
          3,
          3,
          this.globalRow,
          this.globalRow + this.cells.length);
    }
  };
  function initCells(instance) {
    if (instance.cells == null) {
      instance.globalRow = 4;
      instance.cells = instance.dataSheet.getRange(
          instance.globalRow, 2, instance.BATCH_SIZE, 6).getValues();
      instance.localRow = 0;
    }
  }
  function fetchCells(instance) {
    initCells(instance);
    while (!findEmptyRow(instance) && instance.globalRow < instance.MAX_ROWS) {
      if (instance.dataSheet.getMaxRows() <
          instance.globalRow + this.BATCH_SIZE) {
        instance.dataSheet.insertRowsAfter(
            instance.dataSheet.getMaxRows(), instance.BATCH_SIZE);
      }
      instance.flush();
      instance.globalRow += instance.cells.length;
      instance.cells = instance.dataSheet.getRange(
          instance.globalRow, 2, instance.BATCH_SIZE, 6).getValues();
      instance.localRow = 0;
    }
    if (instance.globalRow >= instance.MAX_ROWS) {
      Logger.log('WARNING: maximum length of the spreadsheet exceeded. ' +
          'Exiting.');
      throw '';
    }
  }
  function findEmptyRow(instance) {
    for (; instance.localRow < instance.cells.length &&
        !(instance.cells[instance.localRow][0] == null ||
        instance.cells[instance.localRow][0].length == 0); instance.localRow++);
    return instance.localRow < instance.cells.length;
  }
}