islandora_piwik.module

<?php

/**
 * @file
 * The main Piwik module file.
 */

/**
 * Implements hook_menu().
 */
function islandora_piwik_menu() {
  $items = array();
  $items['admin/islandora/tools/piwik'] = array(
    'title' => 'Islandora Piwik',
    'description' => 'Configure Islandora Piwik.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('islandora_piwik_admin_settings'),
    'access arguments' => array('administer site configuration'),
    'type' => MENU_NORMAL_ITEM,
  );
  $items['admin/reports/islandora_piwik_reports'] = array(
    'title' => 'Islandora Piwik Reports',
    'description' => 'Makes Piwik reports available to authorized users.',
    'page callback' => 'islandora_piwik_reports',
    'access arguments' => array('view islandora piwik reports'),
    'type' => MENU_NORMAL_ITEM,
  );
  return $items;
}

/**
 * Implements hook_permission().
 */
function islandora_piwik_permission() {
  return array(
    'view islandora piwik reports' => array(
      'title' => t('View Islandora Piwik reports'),
      'description' => t('View reports managed by the Islandora Piwik module'),
    ),
  );
}

/**
 * Admin settings form builder.
 */
function islandora_piwik_admin_settings() {
  $form['islandora_piwik_endpoint'] = array(
    '#title' => t("Piwik API endpoint"),
    '#type' => 'textfield',
    '#size' => 60,
    '#default_value' => variable_get('islandora_piwik_endpoint', 'http://piwik.example.com/piwik.php'),
    '#description' => t("Your Piwik instance's API endpoint"),
    '#maxlength' => 255,
  );
  $form['islandora_piwik_site_id'] = array(
    '#title' => t('Site ID'),
    '#type' => 'textfield',
    '#size' => 5,
    '#default_value' => variable_get('islandora_piwik_site_id', '1'),
    '#description' => t("The Piwik site ID."),
    '#maxlength' => 5,
  );
  $form['islandora_piwik_collection_site_ids'] = array(
    '#title' => t('Collection-specific site IDs'),
    '#type' => 'textarea',
    '#size' => 5,
    '#default_value' => variable_get('islandora_piwik_collection_site_ids', ''),
    '#description' => t("Piwik site IDs for specific collections, one entry per
      row. Each row should be in PID, site ID format
      (e.g., \"islandora:mycollection, 5\"). Multiple collections can be
      associated with the same site ID."),
  );
  $form['islandora_piwik_use_javascript'] = array(
    '#type' => 'checkbox',
    '#title' => t('Use Inline Javascript to track'),
    '#default_value' => variable_get('islandora_piwik_use_javascript', 1),
    '#description' => t("Check this option if you want to track sites using inline Javascript instead of PIWIK PHP httpd call.
      If checked, this will automatically add inline PIWIK tracking Javascript to each drupal and Islandora page, needed
      if you run a front end caching mechanism like Varnish or if you want to avoid the httpd call to PIWIK blocks or delays your page load
      "),
  );
  $form['islandora_piwik_timeout'] = array(
    '#title' => t('Timeout'),
    '#type' => 'textfield',
    '#size' => 5,
    '#default_value' => variable_get('islandora_piwik_timeout', '2'),
    '#description' => t("Number of seconds to use as the timeout for requests to
      the Piwki endpoint."),
    '#maxlength' => 5,
  );
  $form['islandora_piwik_track_collection_usage'] = array(
    '#type' => 'checkbox',
    '#title' => t('Track collection usage'),
    '#default_value' => variable_get('islandora_piwik_track_collection_usage', 1),
    '#description' => t("Check this option if you want to track collection usage
      in addtion to object usage. If enabled, each object's parent collections
      are counted every time the object is viewed, and usage reports will be
      available under \"Visitors > Custom Variables\" in your Piwik server.
      High-traffic or large sites may want to disable this option to improve
      performance."),
  );
  $form['islandora_piwik_api_key'] = array(
    '#title' => t("Piwik API key"),
    '#type' => 'textfield',
    '#size' => 40,
    '#default_value' => variable_get('islandora_piwik_api_key', ''),
    '#description' => t("Your Piwik 's API key (a.k.a. token_auth). Required if you want to view reports generated by Piwik. It it available under your account on the Piwik server, in the API menu."),
    '#maxlength' => 40,
  );
  $form['islandora_piwik_javascript_tracking_code'] = array(
    '#title' => t('Piwik JavaScript tracking code'),
    '#type' => 'textarea',
    '#size' => 5,
    '#default_value' => variable_get('islandora_piwik_javascript_tracking_code', ''),
    '#description' => t("If you want to use Piwik to provide analytics for your entire
      Islandora site, enter the \"JavaScript Tracking Code\" generated by Piwik here.
      Note that if you update any of the options provided by Piwik, you will need to
      update this JavaScript code. Islandora objects are not counted twice."),
  );

  return system_settings_form($form);
}

/**
 * Implements hook_islandora_view_object().
 */
function islandora_piwik_islandora_view_object($object) {
  if (!variable_get('islandora_piwik_use_javascript', 0)) {
    $timeout = (int) variable_get('islandora_piwik_timeout', '2');
    $endpoint = variable_get('islandora_piwik_endpoint', 'http://piwik.example.com/piwik.php');
    $site_id = islandora_piwik_get_site_id($object);

    // Record usage of the object.
    global $base_url;
    $current_path = $base_url . '/' . current_path();
    $query = '?apiv=1&rec=1';
    $query .= '&idsite=' . $site_id;
    $query .= '&url=' . $current_path;
    // We urldecode() the referrer because URLS produced by Islandora
    // are usually encoded.
    $query .= '&urlref=' . urlencode(urldecode($_SERVER['HTTP_REFERER']));
    $uri = $endpoint . $query;
    try {
      $return = drupal_http_request($uri, array('timeout' => $timeout));
    }
    catch (Exception $e) {
      watchdog('islandora_piwik', "Error connecting to Piwik server at %ep:
        %message", array("%ep" => $endpoint, "%message" => $e->getMessage()),
        WATCHDOG_ERROR);
    }

    // Record usage of the object's collection(s).
    if (variable_get('islandora_piwik_track_collection_usage', 1)) {
      $collections = islandora_piwik_get_ancestor_collections($object);
      foreach ($collections as $collection) {
        // We don't want to count usage of the islandora:root collection.
        if ($collections != 'islandora:root') {
          $query = '?apiv=1&rec=1';
          $query .= '&idsite=' . $site_id;
          // Create the JSON object that holds the custom Piwik variable that
          // tracks the collection PID. We do not include the collection URL
          // in our request to the Piwik API since doing so would double-count
          // usage of the collection.
          $cvar_object = new stdClass();
          $cvar_object->{1} = array('collection_pid', $collection);
          $query .= '&cvar=' . urlencode(json_encode($cvar_object));
          $uri = $endpoint . $query;
          try {
            $return = drupal_http_request($uri, array('timeout' => $timeout));
          }
          catch (Exception $e) {
            watchdog('islandora_piwik', "Error connecting to Piwik server at %ep:
              %message", array("%ep" => $endpoint, "%message" => $e->getMessage()),
              WATCHDOG_ERROR);
          }
        }
      }
    }
  }
  return array();
}

/**
 * Implements hook_islandora_solr_query_result().
 */
function islandora_piwik_islandora_solr_query_result($result) {
  if (!variable_get('islandora_piwik_use_javascript', 0)) {
    $timeout = (int) variable_get('islandora_piwik_timeout', '2');
    $endpoint = variable_get('islandora_piwik_endpoint', 'http://piwik.example.com/piwik.php');
    $site_id = variable_get('islandora_piwik_site_id', '1');
    global $base_url;

    $current_path = $base_url . '/' . current_path();
    $query = '?apiv=1&rec=1';
    $query .= '&idsite=' . $site_id;
    $search_words = $result['responseHeader']['params']['q'];
    $search_words = urlencode($search_words);
    $query .= '&search=' . $search_words;
    if ($result['response']['numFound'] === 0) {
      $query .= '&search_count=0';
    }
    $query .= '&url=' . $current_path;
    $uri = $endpoint . $query;
    try {
      $return = drupal_http_request($uri, array('timeout' => $timeout));
    }
    catch (Exception $e) {
      watchdog('islandora_piwik', "Error connecting to Piwik server at %ep:
      %message", array("%ep" => $endpoint, "%message" => $e->getMessage()),
      WATCHDOG_ERROR);
    }
  }
}

/**
 * Returns the PIDs of all collections that the Islandora object belongs to.
 *
 * @param object $islandora_object
 *   The Islandora object being inspected.
 *
 * @return array $collection_pids
 *   A list of PIDS.
 */
function islandora_piwik_get_ancestor_collections($islandora_object) {
  $collection_pids = array();
  // islandora:is_member_of_collection property.
  if ($islandora_object) {
    $collections = $islandora_object->relationships->get(FEDORA_RELS_EXT_URI,
      'isMemberOfCollection');
    // The root collection PID will be a member of 0 collections, so we need
    // to check for it.
    if (count($collections) > 0) {
      foreach ($collections as $collection) {
        $collection_pids[] = $collection['object']['value'];
      }
    }
    else {
      // If the object is a book page or a newspaper issue, query the RI
      // to get its ancestor collection.
      $object_cmodel_pids = $islandora_object->models;
      $pid = $islandora_object->id;
      if ($object_cmodel_pids[0] == 'islandora:newspaperIssueCModel'
        || $object_cmodel_pids[0] == 'islandora:pageCModel') {
        $query = <<<EOQ
          select ?collection from <#ri> where {
            # Get the collection that book pages and newspaper issues belong to.
            <info:fedora/$pid> <fedora-rels-ext:isMemberOf> ?parent .
            ?parent <fedora-rels-ext:isMemberOfCollection> ?collection .
          }
EOQ;
        $connection = islandora_get_tuque_connection();
        $results = $connection->repository->ri->sparqlQuery($query, 1);
        if (count($results)) {
          $collection_pids[] = $results[0]['collection']['value'];
        }
      }
      // If the object is a newspaper page, query the RI to get the
      // its ancestor collection.
      if ($object_cmodel_pids[0] == 'islandora:newspaperPageCModel') {
        $query = <<<EOQ
          select ?collection from <#ri> where {
            # Get the collection that newspaper pages belong to.
            <info:fedora/$pid> <fedora-rels-ext:isMemberOf> ?issue .
            ?issue <fedora-rels-ext:isMemberOf> ?newspaper .
            ?newspaper <fedora-rels-ext:isMemberOfCollection> ?collection .
          }
EOQ;
        $connection = islandora_get_tuque_connection();
        $results = $connection->repository->ri->sparqlQuery($query, 1);
        if (count($results)) {
          $collection_pids[] = $results[0]['collection']['value'];
        }
      }
    }
    return $collection_pids;
  }
}

/**
 * Returns the PIDs via solr of all collections an Islandora object belongs to.
 *
 * @param object $islandora_object
 *   The Islandora object being inspected.
 *
 * @return array $collection_pids
 *   A list of PIDS.
 */
function islandora_piwik_get_ancestorviasolr_collections($islandora_object) {
  module_load_include('inc', 'islandora_solr', 'query_processor');
  $collection_pids = array();
  // islandora:is_member_of_collection property.
  if ($islandora_object) {
    global $user;
    $query_processor = new IslandoraSolrQueryProcessor();
    $query = '*:*';
    $query_processor->buildQuery($query);
    // Todo make this solr field configurable.
    $query_processor->solrParams['fl'] = 'PID, ancestors_ms';
    $query_processor->solrParams['fq'][] = 'PID:' . "\"{$islandora_object->id}\"";
    try {
      $query_processor->executeQuery(FALSE);
      $solr_results = $query_processor->islandoraSolrResult['response'];
      if ($solr_results['numFound'] == 1) {
        if (isset($solr_results['objects'][0]['solr_doc']['ancestors_ms'])) {
          $collection_pids = (array) $solr_results['objects'][0]['solr_doc']['ancestors_ms'];
        }
      }
    }
    catch (Exception $e) {
      watchdog(t('Islandora PIWIK fetch Solr ancestors Error', nl2br(check_plain($e->getMessage()))), NULL, WATCHDOG_WARNING);
    }
  }
  return $collection_pids;
}

/**
 * Returns the Piwik site ID to use for the current object using.
 *
 * @param object $islandora_object
 *   The Islandora object being inspected.
 *
 * @return string $site_id
 *   The site ID to use.
 */
function islandora_piwik_get_site_id($islandora_object) {
  $general_site_id = variable_get('islandora_piwik_site_id', '1');
  $collection_site_ids_setting = variable_get('islandora_piwik_collection_site_ids', '');
  // If there are no collection-specific site IDs, use the general one.
  if (!strlen($collection_site_ids_setting)) {
    return $general_site_id;
  }
  else {
    $collection_ids = array();
    $entries = array_filter(preg_split('/\r\n|[\r\n]/', $collection_site_ids_setting));
    foreach ($entries as &$entry) {
      list($pid, $site_id) = explode(',', $entry);
      $pid = trim($pid);
      $collection_ids[$pid] = trim($site_id);
    }
  }

  // If the object is a collection object, its PID might be in the
  // list of collection-specific site IDs. Test this first since
  // doing so is less expensive then the following operation.
  if (array_key_exists($islandora_object->id, $collection_ids)) {
    return $collection_ids[$islandora_object->id];
  }
  else {
    return $general_site_id;
  }
}

/**
 * Returns the Piwik site ID to use for the current object using.
 *
 * @param string $pid
 *   The Islandora object PID being inspected.
 *
 * @return array $site_ids
 *   The PIWIK IDs to use for this PID.
 */
function islandora_piwik_get_site_ids_bypid($pid) {
  $general_site_id = variable_get('islandora_piwik_site_id', '1');
  $collection_site_ids_setting = variable_get('islandora_piwik_collection_site_ids', '');
  // If there are no collection-specific site IDs, use the general one.
  if (!strlen(trim($collection_site_ids_setting))) {
    return array($general_site_id);
  }
  else {
    $collection_ids = array();
    $entries = array_filter(preg_split('/\r\n|[\r\n]/', $collection_site_ids_setting));
    foreach ($entries as $entry) {
      list($pid2, $site_id) = explode(',', $entry);
      if ($pid == trim($pid2)) {
        $collection_ids[] = trim($site_id);
      }
    }
  }

  // If the object is a collection object, its PID might be in the
  // list of collection-specific site IDs. Test this first since
  // doing so is less expensive then the following operation.
  if (count($collection_ids) > 0) {
    //Add also the default one to keep general track
    $collection_ids[] = $general_site_id;
    //In case of duplicates, remove
    $collection_ids = array_unique($collection_ids);
    return $collection_ids;
  }
  else {
    return array($general_site_id);
  }
}

/**
 * Implements hook_page_alter().
 */
function islandora_piwik_page_alter(&$page) {
  module_load_include('inc', 'islandora', 'includes/utilities');
  global $_islandora_solr_queryclass;
  $endpoint = variable_get('islandora_piwik_endpoint', 'http://piwik.example.com/piwik.php');
  $general_site_id = $siteId = variable_get('islandora_piwik_site_id', '1');
  $script = 'var _paq = _paq || [];';
  $siteIds = array();
  $script2 = '';
  $inOtherIds = false;
  // Asume this will be true for all Solr Search pages.
  if (!empty($_islandora_solr_queryclass)) {
    // Searching in a particular collection?
    // Cp? Could be anything since it is user provided data.
    if (isset($_islandora_solr_queryclass->internalSolrParams['cp']) && islandora_is_valid_pid($_islandora_solr_queryclass->internalSolrParams['cp'])) {
      $object = islandora_object_load($_islandora_solr_queryclass->internalSolrParams['cp']);
      $collectionsearch = $_islandora_solr_queryclass->internalSolrParams['cp'];
      $siteIds = is_object($object) ? islandora_piwik_get_site_ids_bypid($object->id) : array($general_site_id);
    }
    // Check if we are using facets.
    $keywordlist = isset($_islandora_solr_queryclass->internalSolrParams['f']) ? (array) $_islandora_solr_queryclass->internalSolrParams['f'] : array();
    array_unshift($keywordlist, $_islandora_solr_queryclass->solrQuery);
    if (isset($collectionsearch)) {
      array_unshift($keywordlist, $collectionsearch);
    }
    $searchtype = isset($_islandora_solr_queryclass->internalSolrParams['type']) ? $_islandora_solr_queryclass->internalSolrParams['type'] : 'advanced';
    if ($searchtype == 'edismax' || $searchtype == 'dismax') {
      $searchtype = "simple";
    }
    $keywords = urlencode(stripslashes(implode(' ', $keywordlist)));
    $script .= "_paq.push(['trackSiteSearch',
    \"{$keywords}\",
    \"{$searchtype}\",
     {$_islandora_solr_queryclass->islandoraSolrResult['response']['numFound']}
    ]);";
    foreach($siteIds as $siteId) {
      if ($siteId != $general_site_id) {
        $inOtherIds = true;
        $script2 .= "var piwikTracker{$siteId} = Piwik.getTracker('{$endpoint}', {$siteId});";
        $script2 .=  "piwikTracker{$siteId}.trackSiteSearch(\"{$keywords}\",
        \"{$searchtype}\",
        {$_islandora_solr_queryclass->islandoraSolrResult['response']['numFound']});";
      }
    }
    if ($inOtherIds) {
      $script .= 'window.piwikAsyncInit = function () { try {';
      $script .=  $script2;
      $script .= '} catch( err ) {console.log(err);}};';
    }
    drupal_add_js($script,
    array(
      'type' => 'inline',
      'scope' => 'header',
      'weight' => 50,
    )
    );
  }
  else {
    if ((arg(0) == 'islandora') && (arg(1) == 'object')) {
      $object = menu_get_object('islandora_object', 2);
      if ($object) {
        $collections = array();
        $idtocollectionmapping = array();
        // Record usage of the object's collection(s).
        // Check first if the visited object is a collection,
        // because it may not be in the list of ancestors
        if (in_array('islandora:collectionCModel', $object->models)) {
           $collections[]=$object->id;
        }
        if (variable_get('islandora_piwik_track_collection_usage', 1)) {
          $collections += islandora_piwik_get_ancestorviasolr_collections($object);
          foreach ($collections as $collection) {
            // We don't want to count usage of the islandora:root collection.
            if (($collection != 'islandora:root') && ($collection != 'islandora:featured_collection')) {
              $siteIds2 = islandora_piwik_get_site_ids_bypid($collection);
              foreach ($siteIds2 as $siteId2) {
                $idtocollectionmapping[$siteId2][] = $collection;
              }
            }
          }
          $script .= 'window.piwikAsyncInit = function () { try {';
          foreach($idtocollectionmapping as $siteId => $collectionpids) {
               $script .= "var piwikTracker{$siteId} = Piwik.getTracker('{$endpoint}', {$siteId});";
               $collectionpid_unique = array_unique($collectionpids);
               $script .=  "piwikTracker{$siteId}.setCustomDimension(2, '{$object->id}');";
               foreach($collectionpid_unique as $collectionpid) {
                 $collectionpidURL = urlencode($collectionpid);
                 $script .=  "piwikTracker{$siteId}.setCustomDimension(1, '{$collectionpidURL}');";
               }
               $script .=  "piwikTracker{$siteId}.trackPageView();";
          }
          $script .= '} catch( err ) {console.log(err);}}';
        }
        drupal_add_js($script,
        array(
          'type' => 'inline',
          'scope' => 'header',
          'weight' => 50,
        )
        );
      }
    }
  }
  if (arg(0) != 'islandora' && arg(1) != 'object' && isset($page['footer'])) {
    $tracking_code = variable_get('islandora_piwik_javascript_tracking_code', '');
    $page['footer']['islandora_piwik']['#markup'] = $tracking_code;
  }
}

/**
 * Page callback that dumbly renders Piwik widget iframe markup.
 */
function islandora_piwik_reports() {
  $endpoint = variable_get('islandora_piwik_endpoint', 'http://piwik.example.com/piwik.php');
  $index_page = preg_replace('/piwik\.php/', 'index.php', $endpoint);
  $general_site_id = variable_get('islandora_piwik_site_id', '1');
  $token_auth = variable_get('islandora_piwik_api_key', '');

  $markup = <<<MARKUP
<div>Using data collected today.</div>
<h3>Real-time map</h3>
<div id="widgetIframe"><iframe width="100%" height="350" src="$index_page?module=Widgetize&action=iframe&widget=1&moduleToWidgetize=UserCountryMap&actionToWidgetize=realtimeMap&idSite=$general_site_id&period=day&date=today&disableLink=1&widget=1&token_auth=$token_auth" scrolling="yes" frameborder="0" marginheight="0" marginwidth="0"></iframe></div>

<h3>Visitor map</h3>
<div id="widgetIframe"><iframe width="100%" height="350" src="$index_page?module=Widgetize&action=iframe&widget=1&moduleToWidgetize=UserCountryMap&actionToWidgetize=visitorMap&idSite=$general_site_id&period=day&date=today&disableLink=1&widget=1&token_auth=$token_auth" scrolling="yes" frameborder="0" marginheight="0" marginwidth="0"></iframe></div>

<h3>Site pages</h3>
<div id="widgetIframe"><iframe width="100%" height="350" src="$index_page?module=Widgetize&action=iframe&widget=1&moduleToWidgetize=Actions&actionToWidgetize=getPageUrls&idSite=$general_site_id&period=day&date=today&disableLink=1&widget=1&token_auth=$token_auth" scrolling="yes" frameborder="0" marginheight="0" marginwidth="0"></iframe></div>

<h3>Islandora collection usage</h3>
<div id="widgetIframe"><iframe width="100%" height="350" src="$index_page?module=Widgetize&action=iframe&widget=1&moduleToWidgetize=CustomDimensions&actionToWidgetize=getCustomDimension&idDimension=1&idSite=$general_site_id&period=day&date=today&disableLink=1&widget=1&token_auth=$token_auth" scrolling="yes" frameborder="0" marginheight="0" marginwidth="0"></iframe></div>

<h3>Searches</h3>
<div id="widgetIframe"><iframe width="100%" height="350" src="$index_page?module=Widgetize&action=iframe&widget=1&moduleToWidgetize=Actions&actionToWidgetize=getSiteSearchKeywords&idSite=$general_site_id&period=day&date=today&disableLink=1&widget=1&token_auth=$token_auth" scrolling="yes" frameborder="0" marginheight="0" marginwidth="0"></iframe></div>

<h3>Searches returning no results</h3>
<div id="widgetIframe"><iframe width="100%" height="350" src="$index_page?module=Widgetize&action=iframe&widget=1&moduleToWidgetize=Actions&actionToWidgetize=getSiteSearchNoResultKeywords&idSite=$general_site_id&period=day&date=today&disableLink=1&widget=1&token_auth=$token_auth" scrolling="yes" frameborder="0" marginheight="0" marginwidth="0"></iframe></div>
MARKUP;
  return $markup;
}