v.1.0.12

Author: findlabnet

README.md

@@ -1,8 +1,12 @@
 Easy XML sitemap
 ================
 
-Provides XML sitemap conforming to [Google's recommendations](https://support.google.com/webmasters/answer/2620865) 
-by using "hreflang" attribute in sitemap on multilingual site and [sitemaps.org protocol](http://www.sitemaps.org/protocol.html).
+Provides XML sitemap conforming to Google's recommendations and sitemaps.org protocol. 
+[Google's recommendations](https://support.google.com/webmasters/answer/2620865) 
+[sitemaps.org protocol](http://www.sitemaps.org/protocol.html)
+
+New in version 1.0.12 - integration with "SEO Meta Tags" module - exclude from sitemap 
+any page have "noindex" in meta-tag "robots". 
 
 Features
 --------
@@ -52,6 +56,8 @@ Under "Advanced settings" fieldset:
 - select sitemap rebuild frequency: manually, daily (default) or any cron run;
 - add exclusions for URLs you won't include in sitemap;
 - add exclusions for content of languages you won't include in sitemap.
+Note: if you have module "SEO Meta Tags" installed, you can exclude from sitemap 
+also all pages with "noindex" in meta-tag "robots".
 
 License
 -------

config/easy_xmlsitemap.settings.json

@@ -5,5 +5,6 @@
   "rebuild_frequency": "daily",
   "excluded_urls": "",
   "excluded_langs": [],
-  "force_https": 0
+  "force_https": 0,
+  "seo_meta_integration": 0
 } 

easy_xmlsitemap.admin.inc

@@ -76,25 +76,37 @@ function easy_xmlsitemap_settings_form() {
   );
   $form['settings']['excluded_urls'] = array(
     '#type' =>  'textarea',
-    '#title' => t('Excluded URLs list (optional)'),
+    '#title' => t('Excluded URLs list:'),
     '#default_value' => $config->get('excluded_urls'),
-    '#description' => t('If some URLs should be excluded from sitemap, place these URLs here, one URL per line without domain, see an example below:') 
+    '#description' => t('If URL should be excluded from sitemap, place one URL per line without domain part, see an example below:') 
     . '<br />403.html'
     . '<br />404.html'
     . '<br />node/123'
     . '<br />' . t('Multiple similar URLs can be selected by using asterisk wildcard (*), for example:')
     . '<br /><em>post-*</em>,'
-    . '<br />' . t('so all URLs starting with post- (such as post-123 or post-about-cat.html) will be excluded from sitemap.')
+    . '<br />' . t('so all URLs starting with post- (such as post-123 or post-about-cat.html) will be excluded from sitemap.'),
+    '#prefix' => '<b>' . t('Exclusions:') . '</b>'
+    
   );
   if (count($site_langs) > 1) {
     $form['settings']['excluded_langs'] = array(
       '#type' =>  'checkboxes',
-      '#title' => t('Exclude from sitemap all pages for selected language (optional):'),
+      '#title' => t('Exclude from sitemap all pages for selected language:'),
       '#options' => $site_langs,
       '#default_value' => $config->get('excluded_langs'),
-      '#description' => t('In some cases you might want do not include to sitemap links to pages for specific language.'), 
+      '#description' => t('You might want do not include to sitemap links to pages for specific language.'), 
     );
   }
+  if (module_exists('seo_meta')) {
+    $form['settings']['seo_meta_integration'] = array(
+      '#type' => 'checkbox',
+      '#title' => t('Exclude also pages with "noindex" value of meta tag "robots".'),
+      '#default_value' => $config->get('seo_meta_integration'),
+      '#description' => t('Your site have module "SEO Meta Tags" enabled, so "noindex" in "robots" can be used for excluding pages from sitemap.'),
+      '#prefix' => '<b>' . t('"SEO Meta Tags" module integration:') . '</b>',
+    );
+  }
+  
   $form['settings']['submit'] = array(
     '#type' => 'submit',
     '#value' => t('Save settings'),
@@ -122,6 +134,9 @@ function easy_xmlsitemap_settings_form_submit($form, &$form_state) {
     if (isset($form_state['values']['excluded_langs'])) {  
       $config->set('excluded_langs',   $form_state['values']['excluded_langs']);
     }
+    if (isset($form_state['values']['seo_meta_integration'])) {  
+      $config->set('seo_meta_integration',   $form_state['values']['seo_meta_integration']);
+    }
     $config->save();
     watchdog('easy_xmlsitemap', t('Settings changed by @user.', array('@user' => $user->name)));
     backdrop_set_message(t('Configuration settings have been saved.'));

easy_xmlsitemap.info

@@ -4,5 +4,5 @@ package = SEO
 backdrop = 1.x
 type = module
 configure = admin/config/metadata/easy_xmlsitemap
-version = 1.0.11
+version = 1.0.12
 

easy_xmlsitemap.install

@@ -52,4 +52,13 @@ function easy_xmlsitemap_update_1002() {
   $config = config('easy_xmlsitemap.settings');
   $config->set('force_https', 0);
   $config->save();  
+}
+
+/**
+ *  Add and initialize new variable
+ */
+function easy_xmlsitemap_update_1003() {
+  $config = config('easy_xmlsitemap.settings');
+  $config->set('seo_meta_integration', 0);
+  $config->save();  
 }

easy_xmlsitemap.module

@@ -3,6 +3,11 @@
  * @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().
  */
@@ -115,7 +120,7 @@ function easy_xmlsitemap_build() {
       $url = url('node/' . $node->nid, array('absolute' => FALSE, 'language' => (object) array('langcode' => $node->langcode)));
 
       // skip excluded URLs
-      if (easy_xmlsitemap_url_excluded($url)) {
+      if (easy_xmlsitemap_seo_meta_check_robots(SEO_META_TYPE_NODE, $node->nid) || easy_xmlsitemap_url_excluded($url)) {
         continue;
       }
 
@@ -176,6 +181,7 @@ function easy_xmlsitemap_url_excluded($url) {
       }
     }
   }
+  
   return $excluded;
 }
 
@@ -306,7 +312,7 @@ function easy_xmlsitemap_build_taxonomy_items($base_url) {
       if (!empty($array_of_nodes)) {
         $url = url('taxonomy/term/' . $term->tid, array('absolute' => FALSE, 'language' => (object) array('langcode' => $langcode)));
         // skip excluded URLs
-        if (easy_xmlsitemap_url_excluded($url)) {
+        if (easy_xmlsitemap_seo_meta_check_robots(SEO_META_TYPE_TAXONOMY, $term->tid) || easy_xmlsitemap_url_excluded($url)) {
           continue;
         }
 
@@ -346,7 +352,7 @@ function easy_xmlsitemap_build_views_items($base_url) {
         if (strpos($page_path, 'admin/') === FALSE) {
           $url = url('/' . $page_path, array('absolute' => FALSE, 'language' => (object) array('langcode' => $langcode)));
           // skip excluded URLs
-          if (easy_xmlsitemap_url_excluded($url)) {
+          if (easy_xmlsitemap_seo_meta_check_robots(SEO_META_TYPE_VIEW, str_replace('-', '_', $view_name)) || easy_xmlsitemap_url_excluded($url)) {
             continue;
           }
 
@@ -472,3 +478,49 @@ function easy_xmlsitemap_cron() {
     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;
+    }
+  }
+}