Issue #2959479: Make sitmemap structure pluggable

README.md

@@ -66,7 +66,7 @@ basis. Just head over to a bundle instance edit form (e.g. node/1/edit) to
 override its sitemap settings.
 
 If you wish for the sitemap to reflect the new configuration instantly, check
-'Regenerate sitemap after clicking save'. This setting only appears if a change
+'Regenerate sitemaps after clicking save'. This setting only appears if a change
 in the settings has been detected.
 
 As the sitemap is accessible to anonymous users, bear in mind that only links
@@ -77,7 +77,7 @@ To include custom links into the sitemap, visit
 /admin/config/search/simplesitemap/custom.
 
 The settings page can be found under admin/config/search/simplesitemap.
-Here the module can be configured and the sitemap can be manually regenerated.
+Here the module can be configured and the sitemaps can be manually regenerated.
 
 
 USAGE
@@ -85,7 +85,7 @@ USAGE
 
 The sitemap is accessible to the whole world under /sitemap.xml.
 
-If the cron generation is turned on, the sitemap will be regenerated according
+If the cron generation is turned on, the sitemaps will be regenerated according
 to the 'Sitemap generation interval' setting ranging from 'On every cron run' to
 'Once a week'.
 

simple_sitemap.drush.inc

@@ -10,7 +10,7 @@
  */
 function simple_sitemap_drush_command() {
   $items['simple_sitemap-generate'] = [
-    'description' => 'Regenerate the XML sitemap according to the module settings.',
+    'description' => 'Regenerate the XML sitemaps according to the module settings.',
     'callback' => 'drush_simple_sitemap_generate',
     'drupal dependencies' => ['simple_sitemap'],
   ];
@@ -20,7 +20,7 @@ function simple_sitemap_drush_command() {
 /**
  * Callback function for hook_drush_command().
  *
- * Regenerate the XML sitemap.
+ * Regenerate the XML sitemaps.
  */
 function drush_simple_sitemap_generate() {
   \Drupal::service('simple_sitemap.generator')->generateSitemap('drush');

simple_sitemap.install

@@ -28,7 +28,7 @@ function simple_sitemap_requirements($phase) {
     case 'runtime':
 
       $generator = \Drupal::service('simple_sitemap.generator');
-      $generated_ago = $generator->getGeneratedAgo();
+      $generated_ago = $generator->getGeneratedAgo('default'); //todo
       $cron_generation = $generator->getSetting('cron_generate');
 
       if (!$generated_ago) {
@@ -42,12 +42,11 @@ function simple_sitemap_requirements($phase) {
         $severity = REQUIREMENT_WARNING;
       }
       else {
-        $value = t('XML sitemap is available');
-        $description = t('The <a href="@sitemap">XML sitemap</a> was generated @ago ago.'
+        $value = t('XML sitemaps are available');
+        $description = t('The last generation finished @ago ago.'
           . ' ' . ($cron_generation
-            ? 'Run cron, or <a href="@generate">regenerate</a> the sitemap manually.'
-            : 'Generation on cron run is disabled. <a href="@generate">Regenerate</a> the sitemap manually.'), [
-              '@sitemap' => $GLOBALS['base_url'] . '/sitemap.xml',
+            ? 'Run cron, or <a href="@generate">regenerate</a> the sitemaps manually.'
+            : 'Generation on cron run is disabled. <a href="@generate">Regenerate</a> the sitemaps manually.'), [
               '@ago' => $generated_ago,
               '@generate' => $GLOBALS['base_url'] . '/admin/config/search/simplesitemap'
             ]
@@ -80,6 +79,20 @@ function simple_sitemap_schema() {
         'not null' => TRUE,
         'unsigned' => TRUE,
       ],
+      'type' => [
+        'description' => 'Type of sitemap this chunk belongs to.',
+        'type' => 'varchar',
+        'length' => 50,
+        'not null' => TRUE,
+        'default' => '',
+      ],
+      'delta' => [
+        'description' => 'Delta of the chunk within the type scope.',
+        'type' => 'int',
+        'size' => 'small',
+        'not null' => TRUE,
+        'unsigned' => TRUE,
+      ],
       'sitemap_string' => [
         'description' => 'XML sitemap chunk string.',
         'type' => 'text',
@@ -97,6 +110,26 @@ function simple_sitemap_schema() {
     'primary key' => ['id'],
   ];
 
+//  $schema['simple_sitemap_index'] = [
+//    'description' => 'Holds XML sitemaps indexes as strings for quick retrieval.',
+//    'fields' => [
+//      'id' => [
+//        'description' => 'Sitemap index unique identifier.',
+//        'type' => 'varchar',
+//        'length' => 50,
+//        'not null' => TRUE,
+//        'default' => '',
+//      ],
+//      'sitemap_string' => [
+//        'description' => 'XML sitemap index string.',
+//        'type' => 'text',
+//        'size' => 'big',
+//        'not null' => TRUE,
+//      ]
+//    ],
+//    'primary key' => ['id'],
+//  ];
+
   $schema['simple_sitemap_entity_overrides'] = [
     'description' => 'Holds sitemap settings overridden by entities.',
     'fields' => [
@@ -398,3 +431,64 @@ function simple_sitemap_update_8209() {
 
   return t('You may now want to configure your sitemap entities to include images.');
 }
+
+/**
+ * Adding simple_sitemap_index table and altering the simple_sitemap table.
+ */
+function simple_sitemap_update_8210() {
+
+  $database = \Drupal::database();
+
+  // Create simple_sitemap_index table.
+//  if (!$database->schema()->tableExists('simple_sitemap_index')) {
+//    $database->schema()->createTable('simple_sitemap_index', [
+//      'description' => 'Holds XML sitemaps indexes as strings for quick retrieval.',
+//      'fields' => [
+//        'id' => [
+//          'description' => 'Sitemap index unique identifier.',
+//          'type' => 'varchar',
+//          'length' => 50,
+//          'not null' => TRUE,
+//          'default' => '',
+//        ],
+//        'sitemap_string' => [
+//          'description' => 'XML sitemap index string.',
+//          'type' => 'text',
+//          'size' => 'big',
+//          'not null' => TRUE,
+//        ]
+//      ],
+//      'primary key' => ['id'],
+//    ]);
+//  }
+
+    // Alter simple_sitemap table.
+    $database->truncate('simple_sitemap')->execute();
+
+    if (!$database->schema()->fieldExists('simple_sitemap', 'type')) {
+      $database->schema()->addField(
+        'simple_sitemap',
+        'type', [
+          'description' => 'Type of sitemap this chunk belongs to.',
+          'type' => 'varchar',
+          'length' => 50,
+          'not null' => TRUE,
+          'default' => '',
+        ]
+      );
+    }
+    if (!$database->schema()->fieldExists('simple_sitemap', 'delta')) {
+      $database->schema()->addField(
+        'simple_sitemap',
+        'delta', [
+          'description' => 'Delta of the chunk within the type scope.',
+          'type' => 'int',
+          'size' => 'small',
+          'not null' => TRUE,
+          'unsigned' => TRUE,
+        ]
+      );
+    }
+}
+
+

simple_sitemap.module

@@ -240,9 +240,9 @@ function simple_sitemap_delete_bundle_config($entity_type_id, $bundle) {
     // Delete bundle settings.
     \Drupal::service('config.factory')->getEditable("simple_sitemap.bundle_settings.$entity_type_id.$bundle")->delete();
 
-    $message = 'You may want to <a href="@url">regenerate</a> your XML sitemap now.';
+    $message = 'You may want to <a href="@url">regenerate</a> your XML sitemaps now.';
     if ($generator->getSetting('cron_generate')) {
-      $message .= ' Otherwise the sitemap will be regenerated during a future cron run.';
+      $message .= ' Otherwise the sitemaps will be regenerated during a future cron run.';
     }
     drupal_set_message(t($message, ['@url' => $GLOBALS['base_url'] . '/admin/config/search/simplesitemap']));
   }
@@ -250,10 +250,13 @@ function simple_sitemap_delete_bundle_config($entity_type_id, $bundle) {
 
 /**
  * Implements hook_robotstxt().
+ *
+ * @todo Use sitemap base URL setting?
+ * @todo include non-default sitemaps as well?
  */
 function simple_sitemap_robotstxt() {
   return [
-    '# XML sitemap',
-    'Sitemap: ' . $GLOBALS['base_url'] . '/sitemap.xml', //todo: Use base URL setting?
+    '# XML sitemaps',
+    'Sitemap: ' . $GLOBALS['base_url'] . '/sitemap.xml',
   ];
 }

simple_sitemap.routing.yml

-simple_sitemap.sitemap:
+simple_sitemap.sitemap_default:
   path: '/sitemap.xml'
   defaults:
     _controller: '\Drupal\simple_sitemap\Controller\SimplesitemapController::getSitemap'
   requirements:
     _access: 'TRUE'
 
-simple_sitemap.sitemaps:
-  path: '/sitemaps'
+simple_sitemap.sitemap_type:
+  path: '/sitemaps/{type}/sitemap.xml'
+  defaults:
+    _controller: '\Drupal\simple_sitemap\Controller\SimplesitemapController::getSitemap'
+  requirements:
+    _access: 'TRUE'
+    
+simple_sitemap.sitemap_type_fallback:
+  path: '/sitemaps/{type}'
   defaults:
     _controller: '\Drupal\simple_sitemap\Controller\SimplesitemapController::getSitemap'
   requirements:
     _access: 'TRUE'
 
-simple_sitemap.chunk_fallback:
-  path: '/sitemaps/{chunk_id}'
+simple_sitemap.sitemap_type_delta:
+  path: '/sitemaps/{type}/{delta}/sitemap.xml'
   defaults:
     _controller: '\Drupal\simple_sitemap\Controller\SimplesitemapController::getSitemap'
   requirements:
     _access: 'TRUE'
 
-simple_sitemap.chunk:
-  path: '/sitemaps/{chunk_id}/sitemap.xml'
+simple_sitemap.sitemap_type_delta_fallback:
+  path: '/sitemaps/{type}/{delta}'
   defaults:
     _controller: '\Drupal\simple_sitemap\Controller\SimplesitemapController::getSitemap'
   requirements:

simple_sitemap.services.yml

@@ -3,7 +3,6 @@ services:
     class: Drupal\simple_sitemap\Simplesitemap
     public: true
     arguments:
-      - '@simple_sitemap.sitemap_generator'
       - '@simple_sitemap.entity_helper'
       - '@config.factory'
       - '@database'
@@ -13,21 +12,11 @@ services:
       - '@datetime.time'
       - '@simple_sitemap.batch'
       - '@plugin.manager.simple_sitemap.url_generator'
-
-  simple_sitemap.sitemap_generator:
-    class: Drupal\simple_sitemap\SitemapGenerator
-    public: true
-    arguments:
-      - '@simple_sitemap.entity_helper'
-      - '@database'
-      - '@module_handler'
-      - '@language_manager'
-      - '@datetime.time'
-      - '@simple_sitemap.sitemap_writer'
+      - '@plugin.manager.simple_sitemap.sitemap_generator'
 
   simple_sitemap.sitemap_writer:
-    class: Drupal\simple_sitemap\SitemapWriter
-    public: false
+    class: Drupal\simple_sitemap\Plugin\simple_sitemap\SitemapGenerator\SitemapWriter
+    public: true
 
   simple_sitemap.entity_helper:
     class: Drupal\simple_sitemap\EntityHelper
@@ -64,3 +53,7 @@ services:
   plugin.manager.simple_sitemap.url_generator:
     class: Drupal\simple_sitemap\Plugin\simple_sitemap\UrlGenerator\UrlGeneratorManager
     parent: default_plugin_manager
+
+  plugin.manager.simple_sitemap.sitemap_generator:
+    class: Drupal\simple_sitemap\Plugin\simple_sitemap\SitemapGenerator\SitemapGeneratorManager
+    parent: default_plugin_manager

src/Annotation/SitemapGenerator.php

+<?php
+
+namespace Drupal\simple_sitemap\Annotation;
+
+use Drupal\Component\Annotation\Plugin;
+
+/**
+ * Defines a SitemapGenerator item annotation object.
+ *
+ * @package Drupal\simple_sitemap\Annotation
+ *
+ * @see \Drupal\simple_sitemap\Plugin\simple_sitemap\SitemapGenerator\SitemapGeneratorManager
+ * @see plugin_api
+ *
+ * @Annotation
+ */
+class SitemapGenerator extends Plugin {
+
+  /**
+   * The generator ID.
+   *
+   * @var string
+   */
+  public $id;
+
+  /**
+   * The human-readable name of the generator.
+   *
+   * @ingroup plugin_translatable
+   *
+   * @var \Drupal\Core\Annotation\Translation
+   */
+  public $title;
+
+  /**
+   * A short description of the generator.
+   *
+   * @ingroup plugin_translatable
+   *
+   * @var \Drupal\Core\Annotation\Translation
+   */
+  public $description;
+
+  /**
+   * An integer to determine the weight of this generator relative to others.
+   *
+   * @var int
+   */
+  public $weight = 0;
+
+  /**
+   * Whether the generator is enabled by default.
+   *
+   * @var bool
+   */
+  public $enabled = TRUE;
+
+  /**
+   * Default generator settings.
+   *
+   * @var array
+   */
+  public $settings = [
+    'list' => TRUE,
+    'default' => FALSE,
+  ];
+}

src/Annotation/UrlGenerator.php

@@ -61,6 +61,7 @@ class UrlGenerator extends Plugin {
    * @var array
    */
   public $settings = [
+    'default_sitemap_generator' => 'default',
     'instantiate_for_each_data_set' => FALSE,
   ];
 }

src/Batch.php

@@ -26,11 +26,16 @@ class Batch {
    */
   protected $batchSettings;
 
+  /**
+   * @var array
+   */
+  protected $batchMeta;
+
   const BATCH_TITLE = 'Generating XML sitemap';
   const BATCH_INIT_MESSAGE = 'Initializing batch...';
   const BATCH_ERROR_MESSAGE = 'An error has occurred. This may result in an incomplete XML sitemap.';
   const BATCH_PROGRESS_MESSAGE = 'Processing @current out of @total link types.';
-  const REGENERATION_FINISHED_MESSAGE = 'The <a href="@url" target="_blank">XML sitemap</a> has been regenerated.';
+  const REGENERATION_FINISHED_MESSAGE = 'The XML sitemaps have been regenerated.';
   const REGENERATION_FINISHED_ERROR_MESSAGE = 'The sitemap generation finished with an error.';
 
   /**
@@ -54,11 +59,24 @@ class Batch {
     $this->batchSettings = $batch_settings;
   }
 
+  /**
+   * @param array $batch_meta
+   */
+  public function setBatchMeta(array $batch_meta) {
+    $this->batchMeta = $batch_meta;
+  }
+
   /**
    * Starts the batch process depending on where it was requested from.
    */
   public function start() {
-    switch ($this->batchSettings['from']) {
+
+    // Update total operation count for each operation.
+    foreach ($this->batch['operations'] as $i => $operation) {
+      $this->batch['operations'][$i][1][3]['operations_count'] = count($this->batch['operations']);
+    }
+
+    switch ($this->batchMeta['from']) {
 
       case 'form':
         // Start batch process.
@@ -98,8 +116,7 @@ class Batch {
           $operation[1][] = &$context;
           call_user_func_array($operation[0], $operation[1]);
         }
-        $this->finishGeneration(TRUE, !empty($context['results']) ? $context['results'] : [], []);
-        return TRUE;
+        return $this->finishGeneration(TRUE, !empty($context['results']) ? $context['results'] : [], []);
     }
     return FALSE;
   }
@@ -107,29 +124,32 @@ class Batch {
   /**
    * Adds an operation to the batch.
    *
-   * @param $plugin_id
+   * @param string $url_generator_id$data_sets
    * @param array|null $data_sets
    */
-  public function addOperation($plugin_id, $data_sets = NULL) {
-    $this->batch['operations'][] = [
-      __CLASS__ . '::generate', [$plugin_id, $data_sets, $this->batchSettings],
+  public function addOperation($url_generator_id, $data_sets = NULL) {
+    $operation_no = count($this->batch['operations']) + 1;
+    $this->batch['operations'][$operation_no] = [
+      __CLASS__ . '::generate', [$url_generator_id, $data_sets, $this->batchSettings, $this->batchMeta + ['current_operation_no' => $operation_no]],
     ];
   }
 
   /**
    * Batch callback function which generates URLs.
    *
-   * @param $plugin_id
+   * @param string $url_generator_id
    * @param array|null $data_sets
    * @param array $batch_settings
+   * @param array $batch_meta
    * @param $context
    *
    * @see https://api.drupal.org/api/drupal/core!includes!form.inc/group/batch/8
    */
-  public static function generate($plugin_id, $data_sets, array $batch_settings, &$context) {
+  public static function generate($url_generator_id, $data_sets, array $batch_settings, array $batch_meta, &$context) {
     \Drupal::service('plugin.manager.simple_sitemap.url_generator')
-      ->createInstance($plugin_id)
+      ->createInstance($url_generator_id)
       ->setContext($context)
+      ->setBatchMeta($batch_meta)
       ->setBatchSettings($batch_settings)
       ->generate($data_sets);
   }
@@ -141,19 +161,14 @@ class Batch {
    * @param $results
    * @param $operations
    *
+   * @return bool
+   *
    * @see https://api.drupal.org/api/drupal/core!includes!form.inc/group/batch/8
    */
   public static function finishGeneration($success, $results, $operations) {
     if ($success) {
-      $remove_sitemap = empty($results['chunk_count']);
-      if (!empty($results['generate']) || $remove_sitemap) {
-        \Drupal::service('simple_sitemap.sitemap_generator')
-          ->setSettings(['excluded_languages' => \Drupal::service('simple_sitemap.generator')->getSetting('excluded_languages', [])])
-          ->generateSitemap(!empty($results['generate']) ? $results['generate'] : [], $remove_sitemap);
-      }
       Cache::invalidateTags(['simple_sitemap']);
-      \Drupal::service('simple_sitemap.logger')->m(self::REGENERATION_FINISHED_MESSAGE,
-        ['@url' => $GLOBALS['base_url'] . '/sitemap.xml'])
+      \Drupal::service('simple_sitemap.logger')->m(self::REGENERATION_FINISHED_MESSAGE)
 //        ['@url' => $this->sitemapGenerator->getCustomBaseUrl() . '/sitemap.xml']) //todo: Use actual base URL for message.
         ->display('status')
         ->log('info');
@@ -163,6 +178,8 @@ class Batch {
         ->display('error', 'administer sitemap settings')
         ->log('error');
     }
+
+    return $success;
   }
 
 }

src/Commands/SimplesitemapCommands.php

@@ -25,7 +25,7 @@ class SimplesitemapCommands extends DrushCommands {
   }
 
   /**
-   * Regenerate the XML sitemap according to the module settings.
+   * Regenerate the XML sitemaps according to the module settings.
    *
    * @command simple-sitemap:generate
    * @validate-module-enabled simple_sitemap

src/Controller/SimplesitemapController.php

@@ -8,6 +8,7 @@ use Symfony\Component\HttpFoundation\Response;
 use Drupal\Core\Cache\CacheableResponse;
 use Symfony\Component\HttpKernel\Exception\NotFoundHttpException;
 use Drupal\simple_sitemap\Simplesitemap;
+use Drupal\simple_sitemap\Plugin\simple_sitemap\SitemapGenerator\SitemapGeneratorBase;
 use Drupal\Core\PageCache\ResponsePolicy\KillSwitch;
 
 /**
@@ -47,26 +48,30 @@ class SimplesitemapController extends ControllerBase {
   }
 
   /**
-   * Returns the whole sitemap, a requested sitemap chunk, or the sitemap index file.
+   * Returns the whole sitemap of a certain type, its requested sitemap chunk,
+   * or its sitemap index file.
    * Caches the response in case of expected output, prevents caching otherwise.
    *
-   * @param int $chunk_id
-   *   Optional ID of the sitemap chunk. If none provided, the first chunk or
-   *   the sitemap index is fetched.
+   * @param string $type
+   *  Optional name of sitemap type. This corresponds to the plugin ID
+   *  of the sitemap generator plugin to be responsible for the generation.
+   *
+   * @param int $delta
+   *  Optional delta of the sitemap chunk. If none provided, the first chunk or
+   *  the sitemap index is fetched.
    *
    * @throws NotFoundHttpException
    *
    * @return object
-   *   Returns an XML response.
+   *  Returns an XML response.
    */
-  public function getSitemap($chunk_id = NULL) {
-    $output = $this->generator->getSitemap($chunk_id);
+  public function getSitemap($type = SitemapGeneratorBase::DEFAULT_SITEMAP_TYPE, $delta = NULL) {
+    $output = $this->generator->getSitemap($type, $delta);
     if (!$output) {
       $this->cacheKillSwitch->trigger();
       throw new NotFoundHttpException();
     }
 
-    // Display sitemap with correct XML header.
     $response = new CacheableResponse($output, Response::HTTP_OK, [
       'content-type' => 'application/xml',
       'X-Robots-Tag' => 'noindex', // Do not index the sitemap itself.

src/Form/SimplesitemapSettingsForm.php

@@ -27,13 +27,13 @@ class SimplesitemapSettingsForm extends SimplesitemapFormBase {
 
     $form['simple_sitemap_settings']['regenerate'] = [
       '#type' => 'fieldset',
-      '#title' => $this->t('Regenerate sitemap'),
-      '#markup' => '<p>' . $this->t('This will regenerate the XML sitemap immediately.') . '</p>',
+      '#title' => $this->t('Regenerate sitemaps'),
+      '#markup' => '<p>' . $this->t('This will regenerate the XML sitemaps immediately.') . '</p>',
     ];
 
     $form['simple_sitemap_settings']['regenerate']['regenerate_submit'] = [
       '#type' => 'submit',
-      '#value' => $this->t('Regenerate sitemap'),
+      '#value' => $this->t('Regenerate sitemaps'),
       '#submit' => ['::generateSitemap'],
       // Skip form-level validator.
       '#validate' => [],
@@ -46,8 +46,8 @@ class SimplesitemapSettingsForm extends SimplesitemapFormBase {
 
     $form['simple_sitemap_settings']['settings']['cron_generate'] = [
       '#type' => 'checkbox',
-      '#title' => $this->t('Regenerate the sitemap during cron runs'),
-      '#description' => $this->t('Uncheck this if you intend to only regenerate the sitemap manually or via drush.'),
+      '#title' => $this->t('Regenerate the sitemaps during cron runs'),
+      '#description' => $this->t('Uncheck this if you intend to only regenerate the sitemaps manually or via drush.'),
       '#default_value' => $this->generator->getSetting('cron_generate', TRUE),
     ];
 

src/SitemapGenerator.php → src/Plugin/simple_sitemap/SitemapGenerator/DefaultSitemapGenerator.php

 <?php
 
-namespace Drupal\simple_sitemap;
+namespace Drupal\simple_sitemap\Plugin\simple_sitemap\SitemapGenerator;
 
+use Symfony\Component\DependencyInjection\ContainerInterface;
 use Drupal\Core\Database\Connection;
 use Drupal\Core\Extension\ModuleHandler;
 use Drupal\Core\Language\LanguageManagerInterface;
 use Drupal\Component\Datetime\Time;
 
 /**
- * Class SitemapGenerator
- * @package Drupal\simple_sitemap
+ * Class DefaultSitemapGenerator
+ * @package Drupal\simple_sitemap\Plugin\simple_sitemap\SitemapGenerator
+ *
+ * @SitemapGenerator(
+ *   id = "default",
+ *   title = @Translation("Default sitemap generator"),
+ *   description = @Translation("Generates a standard conform hreflang sitemap of your content."),
+ *   weight = 0,
+ *   settings = {
+ *     "default" = true,
+ *   },
+ * )
  */
-class SitemapGenerator {
+class DefaultSitemapGenerator extends SitemapGeneratorBase {
 
-  const XML_VERSION = '1.0';
-  const ENCODING = 'UTF-8';
-  const XMLNS = 'http://www.sitemaps.org/schemas/sitemap/0.9';
   const XMLNS_XHTML = 'http://www.w3.org/1999/xhtml';
-  const GENERATED_BY = 'Generated by the Simple XML sitemap Drupal module: https://drupal.org/project/simple_sitemap.';
-  const FIRST_CHUNK_INDEX = 1;
   const XMLNS_IMAGE = 'http://www.google.com/schemas/sitemap-image/1.1';
 
-  /**
-   * @var \Drupal\simple_sitemap\EntityHelper
-   */
-  protected $entityHelper;
-
-  /**
-   * @var \Drupal\Core\Database\Connection
-   */
-