GoogleSiteSearch.inc in Google Site Search 6

File

includes/GoogleSiteSearch.inc

<?php

// $Id$
// Defines the Google site search base url.
define('GSS_BASE_URL', 'http://www.google.com/search');

/**
 * Class for interaction with Google Site Search.
 */
class GoogleSiteSearch {

  /**
   * The Google key to use.
   * @var string
   */
  private $key = NULL;

  /**
   * The language to use in the search.
   * @var string
   */
  private $language = NULL;

  /**
   * The search query.
   * @var string
   */
  private $query = NULL;

  /**
   * Extra parameters to be passed onto Google CSE.
   * @var array
   */
  private $extraParams = NULL;

  /**
   * Page size, number of results per page.
   * @var int
   */
  private $pageSize = 20;

  /**
   * Pager size, number of pages to show in the pager.
   * @var int
   */
  private $pagerSize = 19;

  /**
   * Current page in the results.
   * @var int
   */
  private $currentPage = NULL;

  /**
   * Total results for a search.
   * @var int
   */
  private $totalResults = 0;

  /**
   * The search results return from Google.
   * @var SimpleXml
   */
  private $searchResultsXml = NULL;

  /**
   * The allowed tags in results from Google.
   * @var array
   */
  private $allowedTags = array(
    'a',
    'em',
    'strong',
    'cite',
    'blockquote',
    'code',
    'ul',
    'ol',
    'li',
    'dl',
    'dt',
    'dd',
    'b',
  );

  /**
   * Create a new instance of GoogleSiteSearch.
   *
   * @param string
   */
  public function __construct($query, $key = NULL, $pageSize = 20, $extraParams = NULL) {
    $this->query = $query;
    $this->key = $key;
    $this->pageSize = $pageSize;
    $this->extraParams = $extraParams;
  }

  /**
   * Set Google key.
   *
   * @param string $key The Google key.
   * @return void
   */
  public function SetKey($key) {
    $this->key = $key;
  }

  /**
   * Get Google key
   *
   * @return string The Google key.
   */
  public function GetKey() {
    return $this->key;
  }

  /**
   * Set the language.
   *
   * @param str $language The new language.
   * @return void
   */
  public function SetLanguage($language) {
    $this->language = $language;
  }

  /**
   * Set the page size.
   *
   * @param int $pageSize The new page size.
   * @return void
   */
  public function SetPageSize($pageSize) {
    $this->pageSize = $pageSize;
  }

  /**
   * Set the pager size.
   *
   * @param int $pagerSize The new pager size.
   * @return void
   */
  public function SetPagerSize($pagerSize) {
    $this->pagerSize = $pagerSize;
  }

  /**
   * Get current page size.
   *
   * @return int The current page size.
   */
  public function GetPageSize() {
    return $this->pageSize;
  }

  /**
   * Get current pager size.
   *
   * @return int The current pager size.
   */
  public function GetPagerSize() {
    return $this->pagerSize;
  }

  /**
   * Get current page in the results.
   *
   * @return int The current page.
   */
  public function GetCurrentPage() {
    return $this->currentPage;
  }

  /**
   * Get total results for current search.
   *
   * @return int Total number of results.
   */
  public function GetTotalResults() {
    return $this->totalResults;
  }

  /**
   * Get total number of pages for search.
   *
   * @return int Total number of pages.
   */
  public function GetTotalPages() {
    return $this->totalResults / $this->pageSize + 1;
  }

  /**
   * Get the allowed HTML tags.
   *
   * @return array The allowed HTML tags.
   */
  public function GetAllowedTags() {
    return $this->allowedTags;
  }

  /**
   * Get query current search.
   *
   * @return string The current query.
   */
  public function GetQuery() {
    return $this->query;
  }

  /**
   * Get query extra parameters array.
   *
   * @return array The extra parameters to CSE.
   */
  public function GetExtraParameters() {
    return $this->extraParams;
  }

  /**
   * Get search results.
   *
   * @param string $query The search query.
   * @param int $page The page to get results from.
   * @return array The search results.
   * @throws Exception
   */
  public function GetSearchResults($page = 1) {

    // set page
    $this->currentPage = $page;

    // calculate start postition based on page
    $startPos = ($this->currentPage - 1) * $this->pageSize;
    if ($this->language) {
      $language = "&lr=lang_{$this->language}";
    }
    else {
      $language = NULL;
    }

    // prepare query parameters for URL assembly
    if (count($this->extraParams) > 0) {
      $extraParamsQuery = '&' . http_build_query($this->extraParams);
    }
    else {
      $extraParamsQuery = NULL;
    }

    // fetch results from google
    $url_response = drupal_http_request(url(GSS_BASE_URL, array(
      'query' => drupal_query_string_encode(array(
        'start' => $startPos,
        'num' => $this->pageSize,
        $language,
        'client' => 'google-csbe',
        'output' => 'xml_no_dtd',
        'cx' => $this->key,
        'q' => $this->query,
        $extraParamsQuery,
      )),
    )));
    if ($url_response->error) {
      return NULL;
    }
    $results = simplexml_load_string($url_response->data);
    if ($results !== FALSE) {
      if (!isset($results->RES->M)) {

        // no results, return NULL
        return NULL;
      }
      else {

        // save total results
        $this->totalResults = intval(check_plain((string) $results->RES->M));

        // store faceted items in order to use them later in the header
        $categories = array();
        foreach ($results->Context->Facet as $facet) {
          $categories[] = array(
            'label' => check_plain((string) $facet->FacetItem->label),
            'anchor_text' => check_plain((string) $facet->FacetItem->anchor_text),
          );
        }
        $this->categories = $categories;

        // init results array
        $arr = array();

        // loop results
        foreach ($results->RES->R as $result) {

          // init result array
          $item = array();
          $item['title'] = filter_xss((string) $result->T, $this
            ->GetAllowedTags());
          $item['url'] = check_url((string) $result->UE);
          $item['description'] = filter_xss((string) $result->S, $this
            ->GetAllowedTags());

          // let's get the image thumbnail - present in PageMap->DataObject['type'] == cse_thumbnail
          if (isset($result->PageMap->DataObject)) {
            foreach ($result->PageMap->DataObject as $do) {
              switch ((string) $do['type']) {
                case 'cse_thumbnail':

                  //we are inside the thumbnail node, lets get the src attribute:
                  foreach ($do->Attribute as $att) {
                    switch ((string) $att['name']) {
                      case 'src':
                        $item['thumbnail_url'] = check_url((string) $att['value']);
                        break;
                    }
                  }
                  break;
              }
            }
          }
          $arr[] = $item;
        }

        // return results
        return $arr;
      }
    }
    else {
      throw new Exception('Could not load search results from Google.');
    }
  }

  /**
   * Get a pager for the search results.
   *
   * @return string The pager HTML code.
   */
  public function GetPager($link = '?page=') {

    // get total pages
    $totalPages = $this
      ->GetTotalPages();
    $currentPage = $this
      ->GetCurrentPage();
    $pagerSize = $this
      ->GetPagerSize();
    $halfPagerSize = $pagerSize / 2;
    if ($totalPages < 2) {
      return;
    }
    $html = "<div class=\"item-list google-search-pager\"><ul class=\"pager\">\n";

    // Limit the number of pages to 50
    $totalPages = $totalPages > 50 ? 50 : $totalPages;
    if ($currentPage > 1) {
      $link_page = $link . ($currentPage - 1);
      $html .= "\t<li class=\"pager-previous\"><a href=\"{$link_page}\">" . t('‹ previous') . "</a></li>\n";
    }

    // Defines the start and finish of the pager
    $pagerStart = $currentPage - (round($pagerSize / 2) - 1);
    $pagerFinish = $currentPage + (round($pagerSize / 2) - 1);

    // Start values, if $pagerStart is negative
    if ($pagerStart < 1) {
      $pagerStart = 1;
      $pagerFinish = $pagerSize > $totalPages ? $totalPages : $pagerSize;
    }

    // Limit $pagerFinish to the same as $totalPages
    $pagerFinish = $pagerFinish > $totalPages ? $totalPages : $pagerFinish;
    if ($pagerStart > 1) {
      $html .= "\t<li class=\"pager-ellipsis\">...</li>\n";
    }
    for ($i = $pagerStart; $i <= $pagerFinish; $i++) {
      if ($i != $currentPage) {
        $link_page = $link . $i;
        $html .= "\t<li class=\"pager-item\"><a href=\"{$link_page}\">{$i}</a></li>\n";
      }
      else {
        $html .= "\t<li class=\"pager-current\">{$i}</li>\n";
      }
    }
    if ($pagerFinish < $totalPages) {
      $html .= "\t<li class=\"pager-ellipsis\">...</li>\n";
    }
    if ($currentPage < intval($totalPages)) {
      $link_page = $link . ($currentPage + 1);
      $html .= "\t<li class=\"pager-next\"><a href=\"{$link_page}\">" . t('next ›') . "</a></li>\n";
    }

    // close ol
    $html .= "</ul></div>\n";

    // return pager
    return $html;
  }

  /**
   * Get a head to the search results.
   *
   * @return string The search head HTML code.
   */
  public function GetSearchHead() {

    // get total pages
    $totalResults = $this
      ->GetTotalResults();
    $pageSize = $this
      ->GetPageSize();
    $currentPage = $this
      ->GetCurrentPage();
    $show_start = $pageSize * $currentPage - $pageSize + 1;
    $show_end = $pageSize * $currentPage;
    if ($show_end > $totalResults) {
      $show_end = $totalResults;
    }
    $html = '<div class="searchhead">';
    if (variable_get('gss_labels', TRUE) == 1) {

      // adding the tabs for the search labels
      $html .= '<span>' . t("Show only results of type:") . '</span><ul>';

      // checking which one is the active tab
      $active_html_first_element = "";
      $active_html_others = "";
      if (strpos($this->query, "more:") === FALSE) {

        //there is no faceted search
        $active_html_first_element = 'class="active"';

        //then the first element is active
        $term_from_query = "";
      }
      else {
        $term_from_query = end(explode(":", $this->query));

        //in this case we store the category to check later
      }
      $this->query = current(explode(" more:", $this->query));

      //resume $this->query to its original value (no categories)

      // first item will always be the original query
      $html .= '<li>' . '<a ' . $active_html_first_element . ' href=' . urlencode($this->query) . '>' . t('All results') . '</a>' . '</li>';

      // loop to create all remaining categorized tabs / queries
      if ($totalResults > 0) {
        foreach ($this->categories as $categ) {
          $active_html_others = "";

          //do not inherit from previous loop...
          if ($categ['label'] == $term_from_query) {
            $active_html_others = 'class="active"';
          }
          $new_query = urlencode($this->query) . '+more%3A' . $categ['label'];

          //builds up the path to the tab links
          $html .= '<li>' . '<a ' . $active_html_others . ' href=' . $new_query . '>' . $categ['anchor_text'] . '</a>' . '</li>';
        }
      }
      $html .= '</ul>';
    }
    if ($totalResults !== 0) {
      $html .= t('Shows @show_start to @show_end of approximately @totalResults hits', array(
        '@show_start' => $show_start,
        '@show_end' => $show_end,
        '@totalResults' => $totalResults,
      ));
    }
    $html .= '</div>';

    // return search head
    return $html;
  }

}

Constants

Name	Description
GSS_BASE_URL

Classes

Name	Description
GoogleSiteSearch	Class for interaction with Google Site Search.