easy_xmlsitemap.module

<?php
/**
 * @file main module file
 */

// Define same content entry types as in module "SEO Meta Tags" 
define('SEO_META_TYPE_NODE',     2);
define('SEO_META_TYPE_VIEW',     3);
define('SEO_META_TYPE_TAXONOMY', 4);

/**
 * Implements hook_config_info().
 */
function easy_xmlsitemap_config_info() {
  $prefixes['easy_xmlsitemap.settings'] = array(
    'label' => t('Easy XML Sitemap'),
    'group' => t('Configuration'),
  );
  
  return $prefixes;
}

/**
 * Implements hook_permission().
 *
 */
function easy_xmlsitemap_permission() {
  return array(
    'administer easy_xmlsitemap settings' => array(
      'title' => t('Administer Easy XML Sitemap settings')
    ),
  );
}

/**
 * Implements hook_menu().
 *
 */
function easy_xmlsitemap_menu() {

  $sub_path = 'search';
  $backdrop_version = explode('.', BACKDROP_VERSION);
  if ($backdrop_version[0] == 1 && $backdrop_version[1] > 5) {
    $sub_path = 'metadata';
  } 
  
  $items = array();

  $items['admin/config/' . $sub_path . '/easy_xmlsitemap'] = array(
    'type' => MENU_NORMAL_ITEM,
    'title' => 'Easy XML Sitemap',
    'page callback' => 'backdrop_get_form',
    'page arguments' => array('easy_xmlsitemap_settings_form'),
    'file' => 'easy_xmlsitemap.admin.inc',
    'access callback' => 'user_access',
    'access arguments' => array('administer easy_xmlsitemap settings'),
  );
  
  $items['sitemap.xml'] = array(
    'page callback' => 'easy_xmlsitemap_output',
    'access callback' => TRUE,
    'type' => MENU_CALLBACK,
  );

  return $items;
}

/**
 *  Wrapper for core language_list(), just remove excluded languages
 */
function easy_xmlsitemap_language_list() {
  $config         = config('easy_xmlsitemap.settings');
  $excluded_langs = $config->get('excluded_langs');
  $site_langs     = language_list(TRUE);  

  if (!empty($excluded_langs)) {
    foreach ($excluded_langs as $lang => $excluded) {
      if ($excluded !== 0) {
        unset($site_langs[$lang]);
      }
    }
  }
  return $site_langs;
}

/**
 * Main sitemap build function
 * @return callback save to file
 */
function easy_xmlsitemap_build() {
  set_time_limit(1800);
  
  $config            = config('easy_xmlsitemap.settings');
  $force_https       = $config->get('force_https');
  $proto             = ((!empty($_SERVER['HTTPS']) && $_SERVER['HTTPS'] !== 'off') || $force_https) ? "https://" : "http://";
  $host              = $_SERVER['HTTP_HOST']; 
  $config_base_url   = rtrim($config->get('base_url'), '/');
  $base_url          = !empty($config_base_url) ? $config_base_url : $proto . $host;

  $output = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
      . "<urlset xmlns=\"http://www.sitemaps.org/schemas/sitemap/0.9\" xmlns:xhtml=\"http://www.w3.org/1999/xhtml\">";
  $output .= easy_xmlsitemap_build_front_url($base_url);
  $output .= easy_xmlsitemap_build_node_items($base_url);
  $output .= easy_xmlsitemap_build_taxonomy_items($base_url);
  $output .= easy_xmlsitemap_build_views_items($base_url);
  $output .= "\n" . '</urlset>';
  
  return easy_xmlsitemap_file_save($output);
}

/**
 * Check if given URL should be excluded from sitemap
 * @param string $url (e.g. "/about.html" or "/node/123")
 * @return boolean
 */
function easy_xmlsitemap_url_excluded($url) {
  $config            = config('easy_xmlsitemap.settings');
  $excluded_raw_urls = $config->get('excluded_urls');
  $excluded_urls     = easy_xmlsitemap_textarea_to_array($excluded_raw_urls);
  $excluded          = FALSE;
  $masked            = array();
  
  if (!empty($excluded_urls)) {
    // Check for url groups masked by *, then full url 
    foreach ($excluded_urls as $excluded_url) {
      if (substr($excluded_url, -1) === '*') {
        $masked[] = substr($excluded_url, 0, -1);
      }
      if (!empty($masked)) {
        foreach ($masked as $found) {
          $position = strpos($url, $found);
          if ($position !== FALSE) {
            $excluded = TRUE;
            continue 2;
          }
          else {
            $excluded = easy_xml_found($url, $excluded_urls);
          }
        }
      }
      else {
        $excluded = easy_xml_found($url, $excluded_urls);
      }
    }
  }
  
  return $excluded;
}

/**
 * Found in an array
 * @param string $url
 * @param array $excluded_urls
 * @return boolean
 */
function easy_xml_found($url, $excluded_urls) {
  $found = FALSE;
  if (in_array(substr($url, 1), $excluded_urls)) {
    $found = TRUE;
  }

  return $found; 
}

/**
 * Return node nid if this node set as front page
 * @return string or 0
 */
function easy_xmlsitemap_get_frontpage_nid() {
  $frontpage_nid  = 0;
  $site_frontpage = config_get('system.core', 'site_frontpage');

  if (strpos($site_frontpage, 'node/') !== FALSE) {
    $frontpage_nid = substr($site_frontpage, 5);
  }
  
  return $frontpage_nid;
}

/**
 * Find and print front page(s) URL 
 * @param string $base_url
 * @return string
 */
function easy_xmlsitemap_build_front_url($base_url) {
  $site_langs = easy_xmlsitemap_language_list();
  $output = '';

  if (count($site_langs) == 1) {
    $output .= easy_xmlsitemap_wrap_url($base_url . '/', key($site_langs));
  }
  else {
    foreach ($site_langs as $lang) {
      $lang_suffix = $lang->default ? '/' : '/' . $lang->langcode . '/';
      $url = $base_url . $lang_suffix;

      $output .= "\n" . '  <url>' . "\n" .
        '    <loc>' . $url . '</loc>';

      foreach ($site_langs as $lang) {
        $hlang_suffix = $lang->default ? '/' : '/' . $lang->langcode . '/';
        $hurl = $base_url . $hlang_suffix;
        
        $output .= "\n" . 
          '    <xhtml:link rel="alternate" hreflang="'. $lang->langcode . '" href="' . $hurl . '" />';
      }
      
      $output .= "\n" . '  </url>';
    }
  }
  
  return $output;
}

/**
 * Find and print nodes URLs
 * @global object $language
 * @param string $base_url
 * @return string
 */
function easy_xmlsitemap_build_node_items($base_url) {
  $config           = config('easy_xmlsitemap.settings');
  $default_langcode = config_get('system.core', 'language_default');
  $site_langs       = easy_xmlsitemap_language_list();
  $urls_block       = array();
  $output           = '';
  $excluded_types   = $config->get('excluded_types');
  $types            = array_values($excluded_types);

  foreach (array_keys($types, 0, true) as $key) {
    unset($types[$key]);
  }
  // @todo - fix fake array
  if (empty($types)) {
    $types = array('nonexistent_type' => 'nonexistent_type');
  }
  
  $frontpage_nid = easy_xmlsitemap_get_frontpage_nid(); 
  
  $result = db_select('node', 'n')
    ->addTag('node_access') 
    ->fields('n')
    ->condition('n.type', $types, 'NOT IN')  
    ->condition('n.status', 0, '>')
    ->execute()
    ->fetchAll();
  
  foreach ($result as $node) {
    // skip node if set as frontpage
    if (!empty($frontpage_nid) && $node->nid == $frontpage_nid) {
      continue;
    }

    $url = url('node/' . $node->nid, array('absolute' => FALSE, 'language' => (object) array('langcode' => $node->langcode)));

    // skip excluded URLs
    if (easy_xmlsitemap_seo_meta_check_robots(SEO_META_TYPE_NODE, $node->nid) || easy_xmlsitemap_url_excluded($url)) {
      continue;
    }

    //skip nodes with disabled languages but with "und" 
    if ($node->langcode != 'und' && !array_key_exists($node->langcode, $site_langs)) {
      continue;
    }

    $urls_block[$node->tnid][] = array(
      'nid' => $node->nid,
      'tnid' => $node->tnid,
      'langcode' => $node->langcode,
    );
  }

  foreach ($urls_block as $tnid => $urls) {
    if ($tnid == '0') {
      foreach ($urls as $node) {
        $url = $base_url . url('node/' . $node['nid'], array('absolute' => FALSE, 'language' => (object) array('langcode' => $node['langcode'])));
        $langcode = $node['langcode'] != 'und' ? $node['langcode'] : $default_langcode;
        $output .= easy_xmlsitemap_wrap_url($url, $langcode);
      }
    }
    else {
      foreach ($urls as $node) {
        $url = url('node/' . $node['nid'], array('absolute' => FALSE, 'language' => (object) array('langcode' => $node['langcode'])));
        
        $output .= "\n" . '  <url>' . "\n" .
          '    <loc>' . $base_url . $url . '</loc>';
        
        foreach ($urls as $tnid) {
          $hurl = url('node/' . $tnid['nid'], array('absolute' => FALSE, 'language' => (object) array('langcode' => $tnid['langcode'])));
          $output .= "\n" 
            . '    <xhtml:link rel="alternate" hreflang="'. $tnid['langcode'] . '" href="' . $base_url . $hurl . '" />';
        }
        $output .= "\n" . '  </url>';
      }
    }
  }

  return $output;
}

/**
 * Find and print non-empty taxonomy term pages
 * @global object $language
 * @param string $base_url
 * @return string
 */
function easy_xmlsitemap_build_taxonomy_items($base_url) {
  // Default language only for such URLs
  $langcode = config_get('system.core', 'language_default');
  $output = '';

  if (module_exists('taxonomy')) {
    $result = db_select('taxonomy_term_data', 'ttd')
      ->fields('ttd', array('tid'))
      ->execute();
    
    foreach ($result as $term) {
      // add term only if have nodes assigned to
      $array_of_nodes = taxonomy_select_nodes($term->tid);

      if (!empty($array_of_nodes)) {
        $url = url('taxonomy/term/' . $term->tid, array('absolute' => FALSE, 'language' => (object) array('langcode' => $langcode)));
        // skip excluded URLs
        if (easy_xmlsitemap_seo_meta_check_robots(SEO_META_TYPE_TAXONOMY, $term->tid) || easy_xmlsitemap_url_excluded($url)) {
          continue;
        }

        $output .= easy_xmlsitemap_wrap_url($base_url . $url, $langcode);
      }
    }
    return $output;
  }
}

/**
 * Find and print public available views pages URLs
 * @global object $language
 * @param string $base_url
 * @return string
 */
function easy_xmlsitemap_build_views_items($base_url) {
  // Default language only for such URLs
  $langcode = config_get('system.core', 'language_default');
  $output = '';
  $views = views_get_enabled_views();
  
  foreach ($views as $view) {
    if (empty($view->display)) {
      // Skip this view as it is broken.
      debug(t("Skipping broken view @view", array('@view' => $view->name)));
      continue;
    }
    
    foreach (array_keys($view->display) as $id) {
      $plugin = views_fetch_plugin_data('display', $view->display[$id]->display_plugin);

      if ($plugin['name'] == 'page') {
        $view_name = $view->name;
        $page_path = $view->display[$id]->display_options['path'];
        // Skip admin views
        if (strpos($page_path, 'admin/') === FALSE) {
          $url = url('/' . $page_path, array('absolute' => FALSE, 'language' => (object) array('langcode' => $langcode)));
          // skip excluded URLs
          if (easy_xmlsitemap_seo_meta_check_robots(SEO_META_TYPE_VIEW, str_replace('-', '_', $view_name)) || easy_xmlsitemap_url_excluded($url)) {
            continue;
          }
          
          $output .= easy_xmlsitemap_wrap_url($base_url . $url, $langcode);
        }
      }           
    }
  }
  return $output;  
}

/**
 * Wrap to tags a single URL
 * @param string $full_url
 * @param string $langcode
 * @return string
 */
function easy_xmlsitemap_wrap_url($full_url, $langcode) {
  $output = "\n" . '  <url>' . "\n" .
    '    <loc>' . $full_url . '</loc>';
  $output .= "\n" 
    . '    <xhtml:link rel="alternate" hreflang="'. $langcode . '" href="' . $full_url . '" />';
  $output .= "\n" . '  </url>';
  
  return $output;
}

/**
 * Save string to file
 * @param string $content
 * @return boolean 
 */
function easy_xmlsitemap_file_save($content) {
  if (empty($content)) {
    return FALSE;
  }
    
  $config    = config('easy_xmlsitemap.settings');
  $file_dir  = 'public://' . $config->get('sitemap_dir');
  $file_path = $file_dir . '/' . 'sitemap.xml';
  
  if (!file_exists($file_dir)) {
    backdrop_mkdir($file_dir);
  }
  
  // make backup for previous version of file
  if (file_exists($file_path)) {
    rename($file_path, $file_path . '.bak');
  }

  $content = $content . "\n" . '<!-- ' . format_date(time(), 'short') . ' -->';
  file_put_contents($file_path, $content);
  
  $success = file_exists($file_path);
  
  if ($success) {
    state_set('easy_xmlsitemap_last_build_date', time());
    watchdog('easy_xmlsitemap', 'Sitemap file saved.');
    backdrop_set_message(t('File "sitemap.xml" has been saved successfully.'));
  }
  return $success;
}

/**
 * Output sitemap from file
 * 
 */
function easy_xmlsitemap_output() {
  $config          = config('easy_xmlsitemap.settings');
  $file_dir        = 'public://' . $config->get('sitemap_dir');
  $file_path       = $file_dir . '/' . 'sitemap.xml';
  
  if (!file_exists($file_path) || !is_readable($file_path)) {
    backdrop_not_found();
    backdrop_exit();
  }
  else {
    $file = file_get_contents($file_path);
    backdrop_add_http_header('Content-type', 'text/xml; charset=utf-8');
    backdrop_add_http_header('X-Robots-Tag', 'noindex, follow');
    echo $file;
    exit();
  }
}

/**
 * Utility function
 * @param string $raw_string
 * @return array of strings from textarea
 */
function easy_xmlsitemap_textarea_to_array($raw_string = '') {
  $array_of_strings = array();
  
  if (!empty($raw_string)) {
    $raw_array = preg_split("/\\r\\n|\\r|\\n/", $raw_string);
    
    $array_of_strings = array_map('trim', $raw_array);
  }
  return $array_of_strings;
}

/**
 * Implementation of hook_cron().
*/
function easy_xmlsitemap_cron() {
  $config            = config('easy_xmlsitemap.settings');
  $last_build_date   = state_get('easy_xmlsitemap_last_build_date', 0);  
  $rebuild_frequency = $config->get('rebuild_frequency');
  
  if ($rebuild_frequency == 'manually') {
    return;
  }
  
  if ($rebuild_frequency == 'daily') {
    $today = date("Y-m-d");
    $day   = date("Y-m-d", $last_build_date);
    
    if ($day != $today) {
      easy_xmlsitemap_build();
    }
  }
  else {
    easy_xmlsitemap_build();
  }
}

/**
 * Check given page for meta-tag "robots"
 * @param string $ce_type
 * @param string $item_id
 * @return boolean TRUE if found value "noindex"
 */

function easy_xmlsitemap_seo_meta_check_robots($ce_type, $item_id) {
  if (!module_exists('seo_meta')) {
    return FALSE;
  }
  else {
    $seo_meta_integration = config_get('easy_xmlsitemap.settings', 'seo_meta_integration');
    if ($seo_meta_integration == 1) {
      $found = FALSE;

      if ($ce_type == SEO_META_TYPE_TAXONOMY) {
        $term = taxonomy_term_load($item_id);
        $ce_id = $term->name;
      }
      else {
        $ce_id = $item_id;
      }
      
      $result = db_select('seo_meta', 'sm')
          ->fields('sm', array(
            'meta_robots',
            )
          )
          ->condition('ce_type', $ce_type, '=')
          ->condition('ce_id', $ce_id, '=')                
          ->execute()
          ->fetchAssoc();

      if ($result) {
        $meta_robots = unserialize($result['meta_robots']);
        if ($meta_robots['NOINDEX'] !== 0) {
          $found = TRUE;
        }
      }

      return $found;
    }
  }
}