Скрипт получает информацию от Google о сайте

Класс позволяет получать различную полезную информацию о сайте через сервисы Google.

/**
* This file contains the googledancer class that checks if Google is "dancing".
*
* @package googledancer
*/

/**
* googledancer class
*
* This class allows to check if Google is "dancing".
*
* @package googledancer
* @author Setec Astronomy
* @version 1.0
* @abstract Check if Google. is "dancing".
* @copyright 2004
* @example sample.php A sample code.
* @link http://setecastronomy.stufftoread.com
*/
class googledancer {
  
  /* @access private */
  var $main_server = '';
  var $alt_servers = array ();
  var $domains = array ();
  
  /* Default constructor of googledancer class. */
  // public function __construct ()
  function googledancer () {
    if (!function_exists ('fsockopen')) {
      die ('googledancer class needs the fsockopen function!');
    }
    $this->main_server = 'www.google.com';
    $this->alt_servers = array ('www.google.ch', 'www.google.at');
    $this->domains = array ('yahoo.com', 'cnn.com', 'amazon.com');
  }
  
  /* @access private */
  // private safe_set (&$var_true, $var_false = '') {
  function safe_set (&$var_true, $var_false = '') {
    if (!isset ($var_true)) {
      $var_true = $var_false;
    }
    return $var_true;
  }

  /* @access private */
  // private queryHTTP ($url, $timeout = 10) {
  function queryHTTP ($url, $timeout = 10) {
    $parsed = parse_url ($url);
    $this->safe_set ($parsed['host']);
    $this->safe_set ($parsed['port'], 80);
    $this->safe_set ($parsed['path']);
    $this->safe_set ($parsed['query']);
    $fp = fsockopen ($parsed['host'], $parsed['port'], $errno, $errstr, $timeout);
    if ($fp !== false) {
      fputs ($fp, "GET " . $parsed['path'] . "?" . $parsed['query'] . " HTTP/1.0\r\n");
      fputs ($fp, "Accept: */*\r\n");
      fputs ($fp, "Referer: http://setecastronomy.stufftoread.com/\r\n");
      fputs ($fp, "Accept-Language: en\r\n");
      fputs ($fp, "User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0)\r\n");
      fputs ($fp, "Host: " . $parsed['host'] . "\r\n");
      fputs($fp, "Connection: close\r\n\r\n");
      $result = '';
      while (!feof ($fp)) {
        $result .= fgets ($fp, 128);
      }
       fclose ($fp);
      return $result;
    }
    return false;
  }
  
  /**
  * The getBackLinks method
  *
  * This method returns an array with the backlinks count about some important websites.
  * @param integer $timeout the timeout value used in the fsockopen function.
  * @return mixed it returns an array or false if it fails
  */
  // public function getBackLinks ($timeout = 10) {
  function getBackLinks ($timeout = 10) {
    $result = array ();
    $servers[] = $this->main_server;
    $servers = array_merge ($servers, $this->alt_servers);
    foreach ($this->domains as $domain) {
      foreach ($servers as $server) {
        $url = 'http://' . $server . '/search?hl=en&ie=UTF-8&q=link%3A' . urlencode ($domain);
        $content = $this->queryHTTP ($url);
        if ($content !== false) {
          if (preg_match ('/ of about <b>([\d,]*?)<\/b> linking /si', $content, $matches)) {
            $result[$domain][$server] = str_replace (',', '', $matches[1]);
          } else {
            $result[$domain][$server] = 0;
          }  
        }
      }
    }
    return $result;
  }

  /**
  * The analizeBackLinks method
  *
  * This method analyse the {@link googledancer::getBackLinks() googledancer::getBackLinks()} return value and
  * returns a float with the percentage probability that Google is dancing.
  * @param array $results {@link googledancer::getBackLinks() googledancer::getBackLinks()} return value
  * @return mixed it returns a float value with the percentage or false if it fails
  */
  function analizeBackLinks ($results) {
    if (!is_array ($results)) {
      return false;
    }
    $diff_count = 0;
    foreach ($results as $domain => $values) {
      if (is_array ($values) && isset ($values[$this->main_server])) {
        foreach ($this->alt_servers as $server) {
          if (isset ($values[$server]) && is_numeric ($values[$server]) && ($values[$server] > 0)) {
            $diff_count += abs ($values[$this->main_server] - $values[$server]) / $values[$server];
          } else {
            return false;
          }
        }
      } else {
        return false;
      }
    }
    return $diff_count * 100;
  }

  /**
  * The getIndexedPages method
  *
  * This method returns an array with the indexed pages count about some important websites.
  * @param integer $timeout the timeout value used in the fsockopen function.
  * @return mixed it returns an array or false if it fails
  */
  // public function getIndexedPages ($timeout = 10) {
  function getIndexedPages ($timeout = 10) {
    $result = array ();
    $servers[] = $this->main_server;
    $servers = array_merge ($servers, $this->alt_servers);
    foreach ($this->domains as $domain) {
      foreach ($servers as $server) {
        $url = 'http://' . $server . '/search?hl=en&ie=UTF-8&q=site%3A' . urlencode ($domain);
        $content = $this->queryHTTP ($url);
        if ($content !== false) {
          if (preg_match ('/ of about <b>([\d,]*?)<\/b> from /si', $content, $matches)) {
            $result[$domain][$server] = str_replace (',', '', $matches[1]);
          } else {
            $result[$domain][$server] = 0;
          }  
        }
      }
    }
    return $result;
  }

  /**
  * The analizeIndexedPages method
  *
  * This method analyse the {@link googledancer::getIndexedPages() googledancer::getIndexedPages()} return value and
  * returns a float with the percentage probability that Google is dancing.
  * @param array $results {@link googledancer::getIndexedPages() googledancer::getIndexedPages()} return value
  * @return mixed it returns a float value with the percentage or false if it fails
  */
  function analizeIndexedPages ($results) {
    if (!is_array ($results)) {
      return false;
    }
    $diff_count = 0;
    foreach ($results as $domain => $values) {
      if (is_array ($values) && isset ($values[$this->main_server])) {
        foreach ($this->alt_servers as $server) {
          if (isset ($values[$server]) && is_numeric ($values[$server]) && ($values[$server] > 0)) {
            $diff_count += abs ($values[$this->main_server] - $values[$server]) / $values[$server];
          } else {
            return false;
          }
        }
      } else {
        return false;
      }
    }
    return $diff_count * 100;
  }
  
  /**
  * The getResults method
  *
  * This method returns an array with the results of the given query.
  * @param string $keywords the keyword used for the query.
  * @param integer $resultcount the number fo results to grab (10, 20, 50, 100).
  * @param integer $timeout the timeout value used in the fsockopen function.
  * @return mixed it returns an array or false if it fails
  */
  function getResults ($keywords = '', $resultcount = 10, $timeout = 10) {
    $result = array ();
    $servers[] = $this->main_server;
    $servers = array_merge ($servers, $this->alt_servers);
    foreach ($servers as $server) {
      $url = 'http://' . $server . '/search?hl=en&ie=UTF-8&q=' . urlencode ($keywords);
      $content = $this->queryHTTP ($url);
      if ($content !== false) {
        if (preg_match_all ('/<p class=g><a href=(.*?) /si', $content, $matches)) {
          $result[$server] = $matches[1];
        } else {
          $result[$server] = array ();
        }  
      }
    }
    return $result;
  }
  
  /**
  * The analizeResults method
  *
  * This method analyse the {@link googledancer::getResults() googledancer::getResults()} return value and
  * returns a float with the percentage probability that Google is dancing.
  * @param array $results {@link googledancer::getResults() googledancer::getResults()} return value
  * @return mixed it returns a float value with the percentage or false if it fails
  */
  function analizeResults ($results) {
    if (!is_array ($results) || !isset ($results[$this->main_server])) {
      return false;
    }
    $result_count = 0;
    $diff_result_count = 0;
    $diff_index_count = 0;
    foreach ($this->alt_servers as $server) {
      if (isset ($results[$server])) {
        $diff_result_count += count ($results[$server]) - count (array_intersect ($results[$this->main_server], $results[$server]));
        $diff_index_count += count ($results[$server]) - count (array_intersect_assoc ($results[$this->main_server], $results[$server]));
        $result_count += count ($results[$server]) * 2;
      } else {
        return false;
      }
    }    
    if ($result_count == 0) {
      $result = 0;
    } else {
      $result = ($diff_result_count + $diff_index_count) / $result_count * 100;
    }
    return $result;
  }
}