chore: Get meta data

html/sites/all/modules/hr/hr_core/hr_core.drush.inc

@@ -28,9 +28,126 @@ function hr_core_drush_command() {
     'aliases' => array(),
   );
 
+  $items['hr-core-rwr-392'] = array(
+    'description' => "Extract meta data",
+    'drupal dependencies' => array(),
+    'aliases' => array(),
+  );
+
   return $items;
 }
 
+/**
+ * Get the average file size of all files uploaded on HR.info.
+ */
+function drush_hr_core_rwr_392() {
+  $file = fopen('/var/www/rwr_links.csv', 'r');
+  $counter = 0;
+
+  // Skip header.
+  $row = fgetcsv($file);
+
+  while (($row = fgetcsv($file)) !== FALSE && $counter < 50) {
+    $counter++;
+    if ($row[8] == 'The linked content originates from a prior revision of a paragraph.') {
+      continue;
+    }
+
+    $url = $row[0];
+
+    // Check if it's a link to a node.
+    if (strpos($url, '/node/') !== FALSE) {
+      // Extract last part as nid.
+      $path = ltrim(parse_url($url, PHP_URL_PATH), '/');
+      $parts = explode('/', $path);
+      $nid = array_pop($parts);
+    }
+    // Lookup using alias.
+    else {
+      $url = parse_url($url, PHP_URL_PATH);
+      $parts = explode('/', $url);
+
+      // Remove language.
+      if ($parts[1] === 'en' || $parts[1] === 'es' || $parts[1] === 'fr' || $parts[1] === 'ru') {
+        unset($parts[1]);
+        $parts = array_values($parts);
+      }
+      $url = implode('/', $parts);
+
+      $nid = _hr_core_get_id_from_alias($url);
+
+      // If not found, search by title.
+      if (!$nid) {
+        // Extract last part as title.
+        $path = ltrim(parse_url($url, PHP_URL_PATH), '/');
+        $parts = explode('/', $path);
+        $title = array_pop($parts);
+        $title = urldecode($title);
+
+        $query = new EntityFieldQuery();
+        $query->entityCondition('entity_type', 'node')
+          ->propertyCondition('title', $title);
+        $result = $query->execute();
+
+        if ($result && isset($result['node'])) {
+          $nid = reset($result['node'])->nid;
+        }
+      }
+
+      // Replace dashes.
+      // If not found, search by title.
+      if (!$nid) {
+        // Extract last part as title.
+        $path = ltrim(parse_url($url, PHP_URL_PATH), '/');
+        $parts = explode('/', $path);
+        $title = array_pop($parts);
+        $title = urldecode($title);
+        $title = str_replace('-', ' ', $title);
+
+        $query = new EntityFieldQuery();
+        $query->entityCondition('entity_type', 'node')
+          ->propertyCondition('title', $title);
+        $result = $query->execute();
+
+        if ($result && isset($result['node'])) {
+          $nid = reset($result['node'])->nid;
+        }
+      }
+
+    }
+
+    drush_print_r([
+      $url,
+      $nid
+    ]);
+  }
+  fclose($file);
+}
+
+/**
+ * Get Id from alias.
+ */
+function _hr_core_get_id_from_alias($url) {
+  $path = ltrim(parse_url($url, PHP_URL_PATH), '/');
+
+  // Check redirects and get final path.
+  if (function_exists('redirect_load_by_source')) {
+    if ($redirect = redirect_load_by_source($path)) {
+      $path = $redirect['redirect'];
+    }
+  }
+
+  $node_path = drupal_lookup_path('source', $path);
+
+  if (!$node_path) {
+    return FALSE;
+  }
+
+  $node = menu_get_object('node', 1, $node_path);
+
+  return $node->nid;
+}
+
 /**
  * Get the average file size of all files uploaded on HR.info.
  */