www.aleph1.co.uk Git - yaffs-website/blob - web/modules/contrib/simple_sitemap/src/Batch/BatchUrlGenerator.php

   1 <?php
   2
   3 namespace Drupal\simple_sitemap\Batch;
   4
   5 use Drupal\Core\Url;
   6 use Drupal\Component\Utility\Html;
   7 use Drupal\Core\Cache\Cache;
   8 use Drupal\Core\StringTranslation\StringTranslationTrait;
   9 use Drupal\simple_sitemap\Logger;
  10 use Drupal\simple_sitemap\Simplesitemap;
  11 use Drupal\simple_sitemap\SitemapGenerator;
  12 use Drupal\Core\Language\LanguageManagerInterface;
  13 use Drupal\Core\Entity\EntityTypeManagerInterface;
  14 use Drupal\Core\Path\PathValidator;
  15 use Drupal\Core\Entity\Query\QueryFactory;
  16
  17 /**
  18  * Class BatchUrlGenerator
  19  * @package Drupal\simple_sitemap\Batch
  20  */
  21 class BatchUrlGenerator {
  22
  23   use StringTranslationTrait;
  24
  25   const ANONYMOUS_USER_ID = 0;
  26   const PATH_DOES_NOT_EXIST_OR_NO_ACCESS_MESSAGE = "The custom path @path has been omitted from the XML sitemap as it either does not exist, or it is not accessible to anonymous users. You can review custom paths <a href='@custom_paths_url'>here</a>.";
  27   const PROCESSING_PATH_MESSAGE = 'Processing path #@current out of @max: @path';
  28   const REGENERATION_FINISHED_MESSAGE = "The <a href='@url' target='_blank'>XML sitemap</a> has been regenerated for all languages.";
  29   const REGENERATION_FINISHED_ERROR_MESSAGE = 'The sitemap generation finished with an error.';
  30
  31   /**
  32    * @var \Drupal\simple_sitemap\Simplesitemap
  33    */
  34   protected $generator;
  35
  36   /**
  37    * @var \Drupal\simple_sitemap\SitemapGenerator
  38    */
  39   protected $sitemapGenerator;
  40
  41   /**
  42    * @var \Drupal\Core\Language\LanguageManagerInterface
  43    */
  44   protected $languageManager;
  45
  46   /**
  47    * @var \Drupal\Core\Language\LanguageInterface[]
  48    */
  49   protected $languages;
  50
  51   /**
  52    * @var string
  53    */
  54   protected $defaultLanguageId;
  55
  56   /**
  57    * @var \Drupal\Core\Entity\EntityTypeManagerInterface
  58    */
  59   protected $entityTypeManager;
  60
  61   /**
  62    * @var \Drupal\Core\Path\PathValidator
  63    */
  64   protected $pathValidator;
  65
  66   /**
  67    * @var \Drupal\Core\Entity\Query\QueryFactory
  68    */
  69   protected $entityQuery;
  70
  71   /**
  72    * @var \Drupal\simple_sitemap\Logger
  73    */
  74   protected $logger;
  75
  76   /**
  77    * @var \Drupal\Core\Entity\EntityInterface|null
  78    */
  79   protected $anonUser;
  80
  81   /**
  82    * @var array
  83    */
  84   protected $context;
  85
  86   /**
  87    * @var array
  88    */
  89   protected $batchInfo;
  90
  91   /**
  92    * BatchUrlGenerator constructor.
  93    * @param \Drupal\simple_sitemap\Simplesitemap $generator
  94    * @param \Drupal\simple_sitemap\SitemapGenerator $sitemap_generator
  95    * @param \Drupal\Core\Language\LanguageManagerInterface $language_manager
  96    * @param \Drupal\Core\Entity\EntityTypeManagerInterface $entity_type_manager
  97    * @param \Drupal\Core\Path\PathValidator $path_validator
  98    * @param \Drupal\Core\Entity\Query\QueryFactory $entity_query
  99    * @param \Drupal\simple_sitemap\Logger $logger
 100    */
 101   public function __construct(
 102     Simplesitemap $generator,
 103     SitemapGenerator $sitemap_generator, //todo: use $this->generator->sitemapGenerator instead?
 104     LanguageManagerInterface $language_manager,
 105     EntityTypeManagerInterface $entity_type_manager,
 106     PathValidator $path_validator,
 107     QueryFactory $entity_query,
 108     Logger $logger
 109   ) {
 110     $this->generator = $generator;
 111     // todo: using only one method, maybe make method static instead?
 112     $this->sitemapGenerator = $sitemap_generator;
 113     $this->languageManager = $language_manager;
 114     $this->languages = $language_manager->getLanguages();
 115     $this->defaultLanguageId = $language_manager->getDefaultLanguage()->getId();
 116     $this->entityTypeManager = $entity_type_manager;
 117     $this->pathValidator = $path_validator;
 118     $this->entityQuery = $entity_query;
 119     $this->logger = $logger;
 120     $this->anonUser = $this->entityTypeManager->getStorage('user')->load(self::ANONYMOUS_USER_ID);
 121   }
 122
 123   /**
 124    * @param $context
 125    * @return $this
 126    */
 127   public function setContext(&$context) {
 128     $this->context = &$context;
 129     return $this;
 130   }
 131
 132   /**
 133    * @param array $batch_info
 134    * @return $this
 135    */
 136   public function setBatchInfo(array $batch_info) {
 137     $this->batchInfo = $batch_info;
 138     return $this;
 139   }
 140
 141   /**
 142    * Batch callback function which generates urls to entity paths.
 143    *
 144    * @param array $entity_info
 145    */
 146   public function generateBundleUrls(array $entity_info) {
 147
 148     foreach ($this->getBatchIterationEntities($entity_info) as $entity_id => $entity) {
 149
 150       $this->setCurrentId($entity_id);
 151
 152       $entity_settings = $this->generator->getEntityInstanceSettings($entity_info['entity_type_name'], $entity_id);
 153
 154       if (empty($entity_settings['index'])) {
 155         continue;
 156       }
 157
 158       switch ($entity_info['entity_type_name']) {
 159         // Loading url object for menu links.
 160         case 'menu_link_content':
 161           if (!$entity->isEnabled()) {
 162             continue 2;
 163           }
 164           $url_object = $entity->getUrlObject();
 165           break;
 166
 167         // Loading url object for other entities.
 168         default:
 169           $url_object = $entity->toUrl();
 170       }
 171
 172       // Do not include external paths.
 173       if (!$url_object->isRouted()) {
 174         continue;
 175       }
 176
 177       $path = $url_object->getInternalPath();
 178
 179       // Do not include paths that have been already indexed.
 180       if ($this->batchInfo['remove_duplicates'] && $this->pathProcessed($path)) {
 181         continue;
 182       }
 183
 184       $url_object->setOption('absolute', TRUE);
 185
 186       $path_data = [
 187         'path' => $path,
 188         'entity_info' => ['entity_type' => $entity_info['entity_type_name'], 'id' => $entity_id],
 189         'lastmod' => method_exists($entity, 'getChangedTime') ? date_iso8601($entity->getChangedTime()) : NULL,
 190         'priority' => $entity_settings['priority'],
 191       ];
 192       $this->addUrlVariants($url_object, $path_data, $entity);
 193     }
 194     $this->processSegment();
 195   }
 196
 197   /**
 198    * Batch function which generates urls to custom paths.
 199    *
 200    * @param array $custom_paths
 201    */
 202   public function generateCustomUrls(array $custom_paths) {
 203
 204     $custom_paths = $this->getBatchIterationCustomPaths($custom_paths);
 205
 206     if ($this->needsInitialization()) {
 207       $this->initializeBatch(count($custom_paths));
 208     }
 209
 210     foreach ($custom_paths as $i => $custom_path) {
 211       $this->setCurrentId($i);
 212
 213       // todo: Change to different function, as this also checks if current user has access. The user however varies depending if process was started from the web interface or via cron/drush. Use getUrlIfValidWithoutAccessCheck()?
 214       if (!$this->pathValidator->isValid($custom_path['path'])) {
 215 //        if (!(bool) $this->pathValidator->getUrlIfValidWithoutAccessCheck($custom_path['path'])) {
 216         $this->logger->m(self::PATH_DOES_NOT_EXIST_OR_NO_ACCESS_MESSAGE,
 217           ['@path' => $custom_path['path'], '@custom_paths_url' => $GLOBALS['base_url'] . '/admin/config/search/simplesitemap/custom'])
 218           ->display('warning', 'administer sitemap settings')
 219           ->log('warning');
 220         continue;
 221       }
 222       $url_object = Url::fromUserInput($custom_path['path'], ['absolute' => TRUE]);
 223
 224       $path = $url_object->getInternalPath();
 225       if ($this->batchInfo['remove_duplicates'] && $this->pathProcessed($path)) {
 226         continue;
 227       }
 228
 229       $entity = $this->getEntityFromUrlObject($url_object);
 230
 231       $path_data = [
 232         'path' => $path,
 233         'lastmod' => method_exists($entity, 'getChangedTime') ? date_iso8601($entity->getChangedTime()) : NULL,
 234         'priority' => isset($custom_path['priority']) ? $custom_path['priority'] : NULL,
 235       ];
 236       if (NULL !== $entity) {
 237         $path_data['entity_info'] = ['entity_type' => $entity->getEntityTypeId(), 'id' => $entity->id()];
 238       }
 239       $this->addUrlVariants($url_object, $path_data, $entity);
 240     }
 241     $this->processSegment();
 242   }
 243
 244   /**
 245    * @return bool
 246    */
 247   protected function isBatch() {
 248     return $this->batchInfo['from'] != 'nobatch';
 249   }
 250
 251   /**
 252    * @param string $path
 253    * @return bool
 254    */
 255   protected function pathProcessed($path) {
 256     $path_pool = isset($this->context['results']['processed_paths']) ? $this->context['results']['processed_paths'] : [];
 257     if (in_array($path, $path_pool)) {
 258       return TRUE;
 259     }
 260     $this->context['results']['processed_paths'][] = $path;
 261     return FALSE;
 262   }
 263
 264   /**
 265    * @param $entity_info
 266    * @return mixed
 267    */
 268   protected function getBatchIterationEntities($entity_info) {
 269     $query = $this->entityQuery->get($entity_info['entity_type_name']);
 270
 271     if (!empty($entity_info['keys']['id'])) {
 272       $query->sort($entity_info['keys']['id'], 'ASC');
 273     }
 274     if (!empty($entity_info['keys']['bundle'])) {
 275       $query->condition($entity_info['keys']['bundle'], $entity_info['bundle_name']);
 276     }
 277     if (!empty($entity_info['keys']['status'])) {
 278       $query->condition($entity_info['keys']['status'], 1);
 279     }
 280
 281     if ($this->needsInitialization()) {
 282       $count_query = clone $query;
 283       $this->initializeBatch($count_query->count()->execute());
 284     }
 285
 286     if ($this->isBatch()) {
 287       $query->range($this->context['sandbox']['progress'], $this->batchInfo['batch_process_limit']);
 288     }
 289
 290     return $this->entityTypeManager
 291       ->getStorage($entity_info['entity_type_name'])
 292       ->loadMultiple($query->execute());
 293   }
 294
 295   /**
 296    * @param array $custom_paths
 297    * @return array
 298    */
 299   protected function getBatchIterationCustomPaths(array $custom_paths) {
 300
 301     if ($this->needsInitialization()) {
 302       $this->initializeBatch(count($custom_paths));
 303     }
 304
 305     if ($this->isBatch()) {
 306       $custom_paths = array_slice($custom_paths, $this->context['sandbox']['progress'], $this->batchInfo['batch_process_limit']);
 307     }
 308
 309     return $custom_paths;
 310   }
 311
 312   /**
 313    * @param $url_object
 314    * @param $path_data
 315    * @param $entity
 316    */
 317   protected function addUrlVariants($url_object, $path_data, $entity) {
 318     $alternate_urls = [];
 319
 320     $translation_languages = NULL !== $entity && $this->batchInfo['skip_untranslated']
 321       ? $entity->getTranslationLanguages() : $this->languages;
 322
 323     // Entity is not translated.
 324     if (NULL !== $entity && isset($translation_languages['und'])) {
 325       if ($url_object->access($this->anonUser)) {
 326         $url_object->setOption('language', $this->languages[$this->defaultLanguageId]);
 327         $alternate_urls[$this->defaultLanguageId] = $this->replaceBaseUrlWithCustom($url_object->toString());
 328       }
 329     }
 330     else {
 331       // Including only translated variants of entity.
 332       if (NULL !== $entity && $this->batchInfo['skip_untranslated']) {
 333         foreach ($translation_languages as $language) {
 334           $translation = $entity->getTranslation($language->getId());
 335           if ($translation->access('view', $this->anonUser)) {
 336             $url_object->setOption('language', $language);
 337             $alternate_urls[$language->getId()] = $this->replaceBaseUrlWithCustom($url_object->toString());
 338           }
 339         }
 340       }
 341
 342       // Not an entity or including all untranslated variants.
 343       elseif ($url_object->access($this->anonUser)) {
 344         foreach ($translation_languages as $language) {
 345           $url_object->setOption('language', $language);
 346           $alternate_urls[$language->getId()] = $this->replaceBaseUrlWithCustom($url_object->toString());
 347         }
 348       }
 349     }
 350
 351     foreach ($alternate_urls as $langcode => $url) {
 352       $this->context['results']['generate'][] = $path_data + ['langcode' => $langcode, 'url' => $url, 'alternate_urls' => $alternate_urls];
 353     }
 354   }
 355
 356   /**
 357    * @return bool
 358    */
 359   protected function needsInitialization() {
 360     return empty($this->context['sandbox']);
 361   }
 362
 363   /**
 364    * @param $max
 365    */
 366   protected function initializeBatch($max) {
 367     $this->context['results']['generate'] = !empty($this->context['results']['generate']) ? $this->context['results']['generate'] : [];
 368     if ($this->isBatch()) {
 369       $this->context['sandbox']['progress'] = 0;
 370       $this->context['sandbox']['current_id'] = 0;
 371       $this->context['sandbox']['max'] = $max;
 372       $this->context['results']['processed_paths'] = !empty($this->context['results']['processed_paths'])
 373         ? $this->context['results']['processed_paths'] : [];
 374     }
 375   }
 376
 377   /**
 378    * @param $id
 379    */
 380   protected function setCurrentId($id) {
 381     if ($this->isBatch()) {
 382       $this->context['sandbox']['progress']++;
 383       $this->context['sandbox']['current_id'] = $id;
 384     }
 385   }
 386
 387   /**
 388    *
 389    */
 390   protected function processSegment() {
 391     if ($this->isBatch()) {
 392       $this->setProgressInfo();
 393     }
 394     if (!empty($this->batchInfo['max_links']) && count($this->context['results']['generate']) >= $this->batchInfo['max_links']) {
 395       $chunks = array_chunk($this->context['results']['generate'], $this->batchInfo['max_links']);
 396       foreach ($chunks as $i => $chunk_links) {
 397         if (count($chunk_links) == $this->batchInfo['max_links']) {
 398           $remove_sitemap = empty($this->context['results']['chunk_count']);
 399           $this->sitemapGenerator->generateSitemap($chunk_links, $remove_sitemap);
 400           $this->context['results']['chunk_count'] = !isset($this->context['results']['chunk_count'])
 401             ? 1 : $this->context['results']['chunk_count'] + 1;
 402           $this->context['results']['generate'] = array_slice($this->context['results']['generate'], count($chunk_links));
 403         }
 404       }
 405     }
 406   }
 407
 408   /**
 409    *
 410    */
 411   protected function setProgressInfo() {
 412     if ($this->context['sandbox']['progress'] != $this->context['sandbox']['max']) {
 413       // Providing progress info to the batch API.
 414       $this->context['finished'] = $this->context['sandbox']['progress'] / $this->context['sandbox']['max'];
 415       // Adding processing message after finishing every batch segment.
 416       end($this->context['results']['generate']);
 417       $last_key = key($this->context['results']['generate']);
 418       if (!empty($this->context['results']['generate'][$last_key]['path'])) {
 419         $this->context['message'] = $this->t(self::PROCESSING_PATH_MESSAGE, [
 420           '@current' => $this->context['sandbox']['progress'],
 421           '@max' => $this->context['sandbox']['max'],
 422           '@path' => HTML::escape($this->context['results']['generate'][$last_key]['path']),
 423         ]);
 424       }
 425     }
 426   }
 427
 428   /**
 429    * @param $url_object
 430    * @return object|null
 431    */
 432   protected function getEntityFromUrlObject($url_object) {
 433     $route_parameters = $url_object->getRouteParameters();
 434     return !empty($route_parameters) && $this->entityTypeManager
 435       ->getDefinition($entity_type_id = key($route_parameters), FALSE)
 436       ? $this->entityTypeManager->getStorage($entity_type_id)
 437         ->load($route_parameters[$entity_type_id])
 438       : NULL;
 439   }
 440
 441   /**
 442    * @param string $url
 443    * @return string
 444    */
 445   protected function replaceBaseUrlWithCustom($url) {
 446     return !empty($this->batchInfo['base_url'])
 447       ? str_replace($GLOBALS['base_url'], $this->batchInfo['base_url'], $url)
 448       : $url;
 449   }
 450
 451   /**
 452    * Callback function called by the batch API when all operations are finished.
 453    *
 454    * @see https://api.drupal.org/api/drupal/core!includes!form.inc/group/batch/8
 455    */
 456   public function finishGeneration($success, $results, $operations) {
 457     if ($success) {
 458       $remove_sitemap = empty($results['chunk_count']);
 459       if (!empty($results['generate']) || $remove_sitemap) {
 460         $this->sitemapGenerator->generateSitemap($results['generate'], $remove_sitemap);
 461       }
 462       Cache::invalidateTags(['simple_sitemap']);
 463       $this->logger->m(self::REGENERATION_FINISHED_MESSAGE,
 464         ['@url' => $GLOBALS['base_url'] . '/sitemap.xml'])
 465 //        ['@url' => $this->sitemapGenerator->getCustomBaseUrl() . '/sitemap.xml']) //todo: Use actual base URL for message.
 466         ->display('status')
 467         ->log('info');
 468     }
 469     else {
 470       $this->logger->m(self::REGENERATION_FINISHED_ERROR_MESSAGE)
 471         ->display('error', 'administer sitemap settings')
 472         ->log('error');
 473     }
 474   }
 475 }