Issue #3337321: Provide a Href to Media action

html_processors.services.yml

@@ -11,3 +11,6 @@ services:
   html_processors.media_generator:
     class: Drupal\html_processors\Service\MediaGenerator
     arguments: ['@file_system', '@entity_type.manager', '@token', '@current_user', '@file.repository', '@logger.channel.html_processors', '@http_client']
+  html_processors.href_to_media_replacer:
+    class: Drupal\html_processors\Service\HrefToMediaReplacer
+    arguments: ['@html_processors.media_generator']

modules/html_processors_tampers/src/Plugin/Tamper/HrefToMedia.php

@@ -6,6 +6,7 @@ use Drupal\Core\Form\FormStateInterface;
 use Drupal\tamper\TamperableItemInterface;
 use Drupal\tamper\TamperBase;
 use Drupal\html_processors\Service\MediaGenerator;
+use Drupal\html_processors\Service\HrefToMediaReplacer;
 use Symfony\Component\DependencyInjection\ContainerInterface;
 use Drupal\Core\Plugin\ContainerFactoryPluginInterface;
 
@@ -31,6 +32,13 @@ class HrefToMedia extends TamperBase implements ContainerFactoryPluginInterface
    */
   protected $mediaGenerator;
 
+  /**
+   * The href to media replacer.
+   *
+   * @var \Drupal\html_processors\Service\HrefToMediaReplacer
+   */
+  protected $hrefToMediaReplacer;
+
   /**
    * Constructs a new HrefToMedia instance.
    *
@@ -45,10 +53,13 @@ class HrefToMedia extends TamperBase implements ContainerFactoryPluginInterface
    *   The plugin implementation definition.
    * @param \Drupal\html_processors\Service\MediaGenerator $media_generator
    *   The media generator service.
+   * @param \Drupal\html_processors\Service\HrefToMediaReplacer $href_to_media_replacer
+   *   The href to media replacer.
    */
-  public function __construct(array $configuration, $plugin_id, $plugin_definition, MediaGenerator $media_generator) {
+  public function __construct(array $configuration, $plugin_id, $plugin_definition, MediaGenerator $media_generator, HrefToMediaReplacer $href_to_media_replacer) {
     parent::__construct($configuration, $plugin_id, $plugin_definition, $configuration['source_definition']);
     $this->mediaGenerator = $media_generator;
+    $this->hrefToMediaReplacer = $href_to_media_replacer;
   }
 
   /**
@@ -62,7 +73,8 @@ class HrefToMedia extends TamperBase implements ContainerFactoryPluginInterface
       $configuration,
       $plugin_id,
       $plugin_definition,
-      $container->get('html_processors.media_generator')
+      $container->get('html_processors.media_generator'),
+      $container->get('html_processors.href_to_media_replacer')
     );
   }
 
@@ -133,60 +145,16 @@ class HrefToMedia extends TamperBase implements ContainerFactoryPluginInterface
   /**
    * {@inheritdoc}
    *
-   * @throws \Drupal\Core\File\Exception\FileException
    * @throws \Drupal\Core\Entity\EntityStorageException
+   * @throws \Drupal\Core\File\Exception\FileException
    * @throws \InvalidArgumentException
+   * @throws \Drupal\Component\Plugin\Exception\PluginException
    * @throws \Drupal\Component\Plugin\Exception\InvalidPluginDefinitionException
    * @throws \Drupal\Component\Plugin\Exception\PluginNotFoundException
+   * @throws \GuzzleHttp\Exception\GuzzleException
    */
   public function tamper($data, TamperableItemInterface $item = NULL) {
-    $remote_url = $this->getSetting(self::REMOTE_SITE_URL);
-    $only_relative = $this->getSetting(self::ONLY_RELATIVE);
-    // Load the HTML as a DOM document to traverse through.
-    $source = new \DOMDocument('5.0', 'UTF-8');
-    // Remove comments.
-    $data = preg_replace('/<!--(.|\s)*?-->/', '', $data);
-    // Replace &nbsp; to space.
-    $data = str_replace("&nbsp;", " ", $data);
-    try {
-      @$source->loadHTML(mb_convert_encoding($data, 'HTML-ENTITIES', 'UTF-8'));
-    }
-    catch (\Exception $th) {
-      // Skipping malformed HTML.
-    }
-    $source->encoding = 'utf-8';
-
-    $links = $source->getElementsByTagName('a');
-    // Iterate each link.
-    foreach ($links as $link) {
-      $href = $link->getAttribute('href');
-      if ($href) {
-        $remote_file_path = $href;
-        // Skip when only relative is enable.
-        if ($only_relative && str_starts_with($href, 'http')) {
-          continue;
-        }
-        // Prepend source site URL if relative path.
-        if (!str_starts_with($href, 'http')) {
-          $remote_file_path = $remote_url . $href;
-        }
-        /** @var \Drupal\media\MediaInterface $media */
-        $media = $this->mediaGenerator->generateFromRemoteFile($remote_file_path, $this->configuration);
-        if ($media) {
-          /** @var \Drupal\media\MediaTypeInterface $media_type */
-          $media_type = $media->bundle->entity;
-          $source_field = $media->getSource()->getSourceFieldDefinition($media_type)->getName();
-          /** @var \Drupal\file\FileInterface $file */
-          $file = $media->get($source_field)->entity;
-          // Updating link href.
-          $link->setAttribute('href', $file->createFileUrl());
-        }
-      }
-    }
-    $result = preg_replace('/<!DOCTYPE .*>/', '', $source->saveHTML());
-    $trim_off_front = strlen('<html><body>') + 1;
-    $trim_off_end = (strrpos($result, '</body></html>')) - strlen($result);
-    $result = substr($result, $trim_off_front, $trim_off_end);
+    $result = $this->hrefToMediaReplacer->process($data, $this->configuration);
     return $result;
   }
 

modules/html_processors_tampers/src/Plugin/Tamper/HtmlToGutenberg.php

@@ -83,11 +83,8 @@ class HtmlToGutenberg extends TamperBase implements ContainerFactoryPluginInterf
    * @throws \Drupal\Component\Plugin\Exception\PluginException
    */
   public function tamper($data, TamperableItemInterface $item = NULL) {
-    $result = "";
     $processors_config = $this->configuration['gutenberg_processors'];
-    if (!empty($data)) {
     $result = $this->htmlGutenbergParser->parse($data, $processors_config);
-    }
     return $result;
   }
 

src/Plugin/Action/HrefToMedia.php

+<?php
+
+namespace Drupal\html_processors\Plugin\Action;
+
+use Drupal\Core\Action\ConfigurableActionBase;
+use Drupal\Core\Session\AccountInterface;
+use Drupal\Core\Form\FormStateInterface;
+use Drupal\html_processors\Service\MediaGenerator;
+use Drupal\html_processors\Service\HrefToMediaReplacer;
+use Symfony\Component\DependencyInjection\ContainerInterface;
+use Drupal\Core\Plugin\ContainerFactoryPluginInterface;
+
+/**
+ * Provides a HREF to Media action.
+ *
+ * @Action(
+ *   id = "html_processors_href_to_media",
+ *   label = @Translation("Create media items from body content Href attributes"),
+ *   type = "node",
+ *   category = @Translation("HtmlProcessors")
+ * )
+ */
+class HrefToMedia extends ConfigurableActionBase implements ContainerFactoryPluginInterface {
+
+  const REMOTE_SITE_URL = 'remote_url';
+  const ONLY_RELATIVE = 'only_relative';
+
+  /**
+   * The media generator service.
+   *
+   * @var \Drupal\html_processors\Service\MediaGenerator
+   */
+  protected $mediaGenerator;
+
+  /**
+   * The href to media replacer.
+   *
+   * @var \Drupal\html_processors\Service\HrefToMediaReplacer
+   */
+  protected $hrefToMediaReplacer;
+
+  /**
+   * Constructs a new HrefToMedia instance.
+   *
+   * @param array $configuration
+   *   The plugin configuration, i.e. an array with configuration values keyed
+   *   by configuration option name. The special key 'context' may be used to
+   *   initialize the defined contexts by setting it to an array of context
+   *   values keyed by context names.
+   * @param string $plugin_id
+   *   The plugin_id for the plugin instance.
+   * @param mixed $plugin_definition
+   *   The plugin implementation definition.
+   * @param \Drupal\html_processors\Service\MediaGenerator $media_generator
+   *   The media generator service.
+   * @param \Drupal\html_processors\Service\HrefToMediaReplacer $href_to_media_replacer
+   *   The href to media replacer.
+   */
+  public function __construct(array $configuration, $plugin_id, $plugin_definition, MediaGenerator $media_generator, HrefToMediaReplacer $href_to_media_replacer) {
+    parent::__construct($configuration, $plugin_id, $plugin_definition);
+    $this->mediaGenerator = $media_generator;
+    $this->hrefToMediaReplacer = $href_to_media_replacer;
+  }
+
+  /**
+   * {@inheritdoc}
+   *
+   * @throws \Symfony\Component\DependencyInjection\Exception\ServiceCircularReferenceException
+   * @throws \Symfony\Component\DependencyInjection\Exception\ServiceNotFoundException
+   */
+  public static function create(ContainerInterface $container, array $configuration, $plugin_id, $plugin_definition) {
+    return new static(
+      $configuration,
+      $plugin_id,
+      $plugin_definition,
+      $container->get('html_processors.media_generator'),
+      $container->get('html_processors.href_to_media_replacer')
+    );
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function defaultConfiguration() {
+    $config = parent::defaultConfiguration();
+    $config[self::REMOTE_SITE_URL] = '';
+    $config[self::ONLY_RELATIVE] = FALSE;
+    return $config;
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function buildConfigurationForm(array $form, FormStateInterface $form_state) {
+    // Remote site URL.
+    $form[self::REMOTE_SITE_URL] = [
+      '#type' => 'url',
+      '#title' => $this->t('Remote site URL'),
+      '#default_value' => $this->configuration[self::REMOTE_SITE_URL],
+      '#description' => $this->t('The remote site URL without trailing "/"'),
+      '#required' => TRUE,
+    ];
+    // Only relative paths.
+    $form[self::ONLY_RELATIVE] = [
+      '#type' => 'checkbox',
+      '#title' => $this->t('Only process relative paths'),
+      '#default_value' => $this->configuration[self::ONLY_RELATIVE],
+      '#description' => $this->t('If checked the absolute paths will be skipped.'),
+    ];
+    // Media types paths.
+    $form['media_paths'] = [
+      '#type' => 'details',
+      '#title' => $this->t('Media Paths'),
+      '#open' => TRUE,
+    ];
+    foreach ($this->mediaGenerator->getMediaSettingsMap() as $media_type_id => $settings) {
+      $form['media_paths'][$media_type_id] = [
+        '#type' => 'textfield',
+        '#title' => $this->t('Path to store files of type: @media_type', [
+          '@media_type' => $settings['label'],
+        ]),
+        '#default_value' => $this->configuration[$media_type_id] ?? '',
+        '#description' => $this->t('Leave empty to use media type file directory.'),
+        '#field_prefix' => 'public://',
+        '#attributes' => [
+          'placeholder' => $settings['file_directory'] ?? '',
+        ],
+      ];
+    }
+    return $form;
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function submitConfigurationForm(array &$form, FormStateInterface $form_state) {
+    $media_paths = $form_state->getValue('media_paths') ?? [];
+    $this->setConfiguration([
+      self::REMOTE_SITE_URL => $form_state->getValue(self::REMOTE_SITE_URL),
+      self::ONLY_RELATIVE => $form_state->getValue(self::ONLY_RELATIVE),
+    ] + $media_paths);
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function access($node, AccountInterface $account = NULL, $return_as_object = FALSE) {
+    /** @var \Drupal\node\NodeInterface $node */
+    $access = $node->access('update', $account, TRUE)
+      ->andIf($node->title->access('edit', $account, TRUE));
+    return $return_as_object ? $access : $access->isAllowed();
+  }
+
+  /**
+   * {@inheritdoc}
+   *
+   * @throws \Drupal\Core\Entity\EntityStorageException
+   * @throws \InvalidArgumentException
+   * @throws \Drupal\Component\Plugin\Exception\PluginException
+   * @throws \Drupal\Core\File\Exception\FileException
+   * @throws \GuzzleHttp\Exception\GuzzleException
+   */
+  public function execute($node = NULL) {
+    /** @var \Drupal\node\NodeInterface $node */
+    $body = $node->body->value;
+    $processed_body = $this->hrefToMediaReplacer->process($body, $this->configuration);
+    $node->set('body', $processed_body)->save();
+  }
+
+}

src/Plugin/HtmlGutenbergProcessor/Image.php

@@ -123,6 +123,7 @@ class Image extends HtmlGutenbergProcessorBase implements ContainerFactoryPlugin
    * @throws \Drupal\Component\Plugin\Exception\InvalidPluginDefinitionException
    * @throws \Drupal\Component\Plugin\Exception\PluginNotFoundException
    * @throws \DOMException
+   * @throws \GuzzleHttp\Exception\GuzzleException
    */
   public function process(\DOMDocument &$source) {
     $tag = $this->pluginDefinition['tag'];

src/Service/HrefToMediaReplacer.php

+<?php
+
+namespace Drupal\html_processors\Service;
+
+use Drupal\html_processors\Service\MediaGenerator;
+
+/**
+ * Service to parse HTML to Gutenberg.
+ */
+class HrefToMediaReplacer {
+
+  /**
+   * The media generator service.
+   *
+   * @var \Drupal\html_processors\Service\MediaGenerator
+   */
+  protected $mediaGenerator;
+
+  /**
+   * Constructs a HrefToMediaReplacer object.
+   *
+   * @param \Drupal\html_processors\Service\MediaGenerator $media_generator
+   *   The media generator service.
+   */
+  public function __construct(MediaGenerator $media_generator) {
+    $this->mediaGenerator = $media_generator;
+  }
+
+  /**
+   * Process HTML to generate media items and replace href attributes.
+   *
+   * @param string $data
+   *   The original HTML.
+   * @param array $config
+   *   The config to process media items.
+   *
+   * @return string
+   *   The processed HTML.
+   *
+   * @throws \Drupal\Component\Plugin\Exception\PluginException
+   * @throws \Drupal\Core\Entity\EntityStorageException
+   * @throws \Drupal\Core\File\Exception\FileException
+   * @throws \InvalidArgumentException
+   * @throws \GuzzleHttp\Exception\GuzzleException
+   */
+  public function process($data, array $config) {
+    $remote_url = $config['remote_url'];
+    $only_relative = $config['only_relative'];
+    // Load the HTML as a DOM document to traverse through.
+    $source = new \DOMDocument('5.0', 'UTF-8');
+    try {
+      @$source->loadHTML(mb_convert_encoding($data, 'HTML-ENTITIES', 'UTF-8'));
+    }
+    catch (\Exception $th) {
+      // Skipping malformed HTML.
+    }
+    $source->encoding = 'utf-8';
+
+    $links = $source->getElementsByTagName('a');
+    // Iterate each link.
+    foreach ($links as $link) {
+      $href = $link->getAttribute('href');
+      if ($href) {
+        $remote_file_path = $href;
+        // Skip when only relative is enable.
+        if ($only_relative && str_starts_with($href, 'http')) {
+          continue;
+        }
+        // Prepend source site URL if relative path.
+        if (!str_starts_with($href, 'http')) {
+          $remote_file_path = $remote_url . $href;
+        }
+        /** @var \Drupal\media\MediaInterface $media */
+        $media = $this->mediaGenerator->generateFromRemoteFile($remote_file_path, $config);
+        if ($media) {
+          /** @var \Drupal\media\MediaTypeInterface $media_type */
+          $media_type = $media->bundle->entity;
+          $source_field = $media->getSource()->getSourceFieldDefinition($media_type)->getName();
+          /** @var \Drupal\file\FileInterface $file */
+          $file = $media->get($source_field)->entity;
+          // Updating link href.
+          $link->setAttribute('href', $file->createFileUrl());
+        }
+      }
+    }
+    $result = preg_replace('/<!DOCTYPE .*>/', '', $source->saveHTML());
+    $result = preg_replace('/<html><body>/', '', $result);
+    $result = preg_replace('/<\/body><\/html>/', '', $result);
+    return $result;
+  }
+
+}

src/Service/MediaGenerator.php

@@ -136,12 +136,13 @@ class MediaGenerator {
    * @throws \InvalidArgumentException
    * @throws \Drupal\Component\Plugin\Exception\InvalidPluginDefinitionException
    * @throws \Drupal\Component\Plugin\Exception\PluginNotFoundException
+   * @throws \GuzzleHttp\Exception\GuzzleException
    */
   public function generateFromRemoteFile($remote_file_path, array $paths_overrides = []) {
     $generated_media = NULL;
     $ext = pathinfo($remote_file_path, PATHINFO_EXTENSION);
     // Check if file extension.
-    if (strlen($ext) === 3) {
+    if (strlen($ext) >= 3 && strlen($ext) <= 4) {
       $media_type_settings = $this->getMediaTypeSettingsFromExt($ext);
       if ($media_type_settings) {
         $media_type_id = $media_type_settings['id'];
@@ -265,6 +266,9 @@ class MediaGenerator {
    *   - If it succeeds a \Drupal\file\FileInterface
    *     object which describes the file.
    *   - If it fails, FALSE.
+   *
+   * @throws \GuzzleHttp\Exception\GuzzleException
+   * @throws \Drupal\Core\Entity\EntityStorageException
    */
   protected function retrieveFile($url, $destination, $replace = FileSystemInterface::EXISTS_RENAME) {
     $parsed_url = parse_url($url);