RevMob PHP revenue scraper

<?
// Fetches your RevMob revenue & impressions statistics so you can update your own local admin reports.
//
// author: Jeff Marshall (10coders)
// contact: [email protected]

require_once("simple_html_dom.php");


// define your console.revmob.com login information here:
$REVMOB_EMAIL = "[email protected]";
$REVMOB_PASSWORD = "yourpassword";

// list out your revmob application ids here.  I group them by project name.
$REVMOB_APPS = array("game1" => array("ios" => "XXXXXXXXXXXXXXXXXXXXXXXX", "android" => "XXXXXXXXXXXXXXXXXXXXXXXX"),
                     "game2" => array("ios" => "XXXXXXXXXXXXXXXXXXXXXXXX", "android" => "XXXXXXXXXXXXXXXXXXXXXXXX"),
                     "game3" => array("ios" => "XXXXXXXXXXXXXXXXXXXXXXXX", "android" => "XXXXXXXXXXXXXXXXXXXXXXXX"));



class Net {
  function FetchURL($url, $postargs=array(), $options=array()) {
    if (!$url) return array();
    $use_get = array_key_exists('use_get', $options) && $options['use_get'] ? $options['use_get'] : null;
    $response_headers = array_key_exists('response_headers', $options) ? $options['response_headers'] : false;
    $cookiefile = array_key_exists('cookiefile', $options) ? $options['cookiefile'] : null;

    $chunks = parse_url($url);
    $port = array_key_exists('port', $chunks) ? $chunks['port'] : null;

    $ch = curl_init();
    $qstr = "";
    if ($use_get && $postargs)
      $qstr = "?".http_build_query($postargs);

    curl_setopt($ch, CURLOPT_URL,$url.$qstr);
    if ($port)
      curl_setopt($ch, CURLOPT_PORT, $port);

    curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
    curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
    curl_setopt($ch, CURLOPT_TIMEOUT, 60);
    curl_setopt($ch, CURLOPT_USERAGENT,"Mozilla/4.0");
    curl_setopt($ch, CURLOPT_AUTOREFERER, true);
    if ($cookiefile) {
      curl_setopt($ch, CURLOPT_COOKIEJAR, $cookiefile);
      curl_setopt($ch, CURLOPT_COOKIEFILE, $cookiefile);
    }
    if (!$use_get) {
      curl_setopt($ch, CURLOPT_POST, 1);
      curl_setopt($ch, CURLOPT_POSTFIELDS, $postargs);
    }

    if ($response_headers) {
      curl_setopt($ch, CURLOPT_HEADER, 1);
    }

    $data = curl_exec($ch);
    $info = curl_getinfo($ch);
    $error = curl_error($ch);

    if ($response_headers) {
      $header_size = curl_getinfo($ch, CURLINFO_HEADER_SIZE);
      $res['response_headers'] = substr($data, 0, $header_size);
      $res['data'] = substr($data, $header_size);
    } else {
      $res['data'] = $data;
    }
    $res['info'] = $info;
    $res['error'] = $error;
    return $res;
  }
}


class RevenueFetcher {

  var $revenues = array();


  // grab revenue data from revmob
  public function grab_revmob() {
    // step 1: fetch the login page to get our session and the login form's authenticity token
    // returns: _mobile_ads_session in cookie and authenticity_token in login form
    global $REVMOB_EMAIL, $REVMOB_PASSWORD;
    $cookiefile = tempnam("/tmp", "curlcookie");
    $res = Net::FetchURL("https://console.revmob.com/",
                         array(),
                         array("response_headers" => true, "cookiefile" => $cookiefile));

    // parse out authenticity_token
    // <input name="authenticity_token" type="hidden" value="uFHD/bgBV4J8RiQ/rfa99UXw5+0qn4FH0fsbFkdyV2w=" />
    $matches = array();
    $authenticity_token = null;
    if (preg_match("/<input\s+name=\"authenticity_token\" .*value=\"(.*)\"/", $res['data'], $matches))
      $authenticity_token = $matches[1];

    if (!$authenticity_token) {
      print "revmob login: error fetching authenticity_token\n";
      return;
    }

    // step 2: log in
    // requires: email, password
    // returns: _mobile_ads_session in cookie
    $res = Net::FetchURL("https://console.revmob.com/users/sessions",
                         array("email" => $REVMOB_EMAIL, "password" => $REVMOB_PASSWORD,
                               "authenticity_token" => $authenticity_token,
                               "utf8" => "&#x2713;",
                               "commit" => "Login",
                               "vendor" => ""),
                         array("cookiefile" => $cookiefile));

    // step 3: fetch revenue by app
    // requires: app id in URL
    // returns: javascript table inserted into dom
    global $REVMOB_APPS, $ENGINE_FILEROOT;
    foreach ($REVMOB_APPS as $project=>$app_ids) {
      foreach ($app_ids as $platform=>$app_id) {
        $res = Net::FetchURL("https://console.revmob.com/mobile_apps/$app_id",
                             array(),
                             array("use_get" => true, "cookiefile" => $cookiefile));
        // pull the revenue table out of the javascript dom append call
        $data = str_replace("\/", "/", $res['data']);

        $html_start = strpos($data, "$('body').append(\"");
        $html_end = strpos($data, "\");");
        if ($html_start === false || $html_end === false) {
          print "revmob: unknown javascript revenue table format\n";
          continue;
        }
        $revtable = substr($data, $html_start, $html_end - $html_start);
        $html =  str_get_html($revtable);

        // get headers rows
        $thead = $html->find('thead', 0);
        $headers = array();
        $i = 0;
        foreach($thead->find('th') as $th) {
          $headers[strtolower($th->plaintext)] = $i;
          $i++;
        }

        // walk the rows of data
        foreach($html->find('tr') as $tr) {
          $day = $tr->find('td', $headers["day"]);
          $impressions = $tr->find('td', $headers["impressions"]);
          $cpm = $tr->find('td', $headers["ecpm"]);
          $revenue = $tr->find('td', $headers["revenue"]);
          if (!$day || !$day->innertext || !$revenue || !$revenue->innertext)
            continue;

          $day_dt = new DateTime($day->innertext);

          $rev = str_replace(",", "", str_replace("$", "", $revenue->innertext));
          $impressions = str_replace(",", "", $impressions->innertext);
          $cpm = str_replace(",", "", str_replace("$", "", $cpm->innertext));

          $this->_add_revenues($project, "revmob", $platform, $rev, $day_dt, $impressions, $cpm);
        }
      }
    }
  }


  // add this bit of revenue to the list of revenue we are tracking
  protected function _add_revenues($project, $source, $platform, $revenue, $day, $impressions=null, $cpm=null) {
    $this->revenues[] = array('source' => $source,
                              'platform' => $platform,
                              'project' => $project,
                              'revenue' => $revenue,
                              'day' => $day,
                              'impressions' => $impressions,
                              'cpm' => $cpm);
  }


  // prints the fetched revenue information for debugging
  public function show() {
    foreach ($this->revenues as $r)
      print "Revenue [" . $r['day']->format("Y-m-d") . "] {$r['source']}.{$r['platform']} on {$r['project']}: \$" . number_format($r['revenue'], 2) . "   impressions=" . number_format($r['impressions']) . "\n";
  }


  // exercise for the reader: commit this data to your local admin revenue database
  public function save() {
    foreach ($this->revenues as $r) {
      //$r['source']
      //$r['platform']
      //$r['project']
      //$r['revenue']
      //$r['day']
      //$r['impressions']
      //$r['cpm']
    }
  }

}



// fetch the revenue data
$fetcher = new RevenueFetcher();
$fetcher->grab_revmob();
$fetcher->show();
$fetcher->save();