insideout10 / wordlift-plugin

src/includes/class-wordlift-post-to-jsonld-converter.php

Indentation +394 added lines, -394 removed lines

@@ -15,418 +15,418 @@
  */
 class Wordlift_Post_To_Jsonld_Converter extends Wordlift_Abstract_Post_To_Jsonld_Converter {
 
-	/**
-	 * A {@link Wordlift_Configuration_Service} instance.
-	 *
-	 * @since  3.10.0
-	 * @access private
-	 * @var \Wordlift_Configuration_Service $configuration_service A {@link Wordlift_Configuration_Service} instance.
-	 */
-	private $configuration_service;
-
-	/**
-	 * A {@link Wordlift_Log_Service} instance.
-	 *
-	 * @since  3.10.0
-	 * @access private
-	 * @var Wordlift_Log_Service $log A {@link Wordlift_Log_Service} instance.
-	 */
-	private $log;
-
-	/**
-	 * Wordlift_Post_To_Jsonld_Converter constructor.
-	 *
-	 * @param \Wordlift_Entity_Type_Service $entity_type_service A {@link Wordlift_Entity_Type_Service} instance.
-	 * @param \Wordlift_Entity_Service $entity_service A {@link Wordlift_Entity_Service} instance.
-	 * @param \Wordlift_User_Service $user_service A {@link Wordlift_User_Service} instance.
-	 * @param \Wordlift_Attachment_Service $attachment_service A {@link Wordlift_Attachment_Service} instance.
-	 * @param \Wordlift_Configuration_Service $configuration_service A {@link Wordlift_Configuration_Service} instance.
-	 *
-	 * @since 3.10.0
-	 *
-	 */
-	public function __construct( $entity_type_service, $entity_service, $user_service, $attachment_service, $configuration_service ) {
-		parent::__construct( $entity_type_service, $entity_service, $user_service, $attachment_service );
-
-		$this->configuration_service = $configuration_service;
-
-		// Set a reference to the logger.
-		$this->log = Wordlift_Log_Service::get_logger( 'Wordlift_Post_To_Jsonld_Converter' );
-	}
-
-	/**
-	 * Convert the provided {@link WP_Post} to a JSON-LD array. Any entity reference
-	 * found while processing the post is set in the $references array.
-	 *
-	 * @param int $post_id The post id.
-	 * @param array $references An array of entity references.
-	 * @param array $references_infos
-	 *
-	 * @return array A JSON-LD array.
-	 * @since 3.10.0
-	 */
-	public function convert( $post_id, &$references = array(), &$references_infos = array() ) {
-
-		// Get the post instance.
-		if ( null === $post = get_post( $post_id ) ) {
-			// Post not found.
-			return null;
-		}
-
-		// Get the base JSON-LD and the list of entities referenced by this entity.
-		$jsonld = parent::convert( $post_id, $references, $references_infos );
-
-		// Set WebPage by default.
-		if ( empty( $jsonld['@type'] ) ) {
-			$jsonld['@type'] = 'WebPage';
-		}
-
-		// Get the entity name.
-		$jsonld['headline'] = $post->post_title;
-
-		// Set the published and modified dates.
-		/*
+    /**
+     * A {@link Wordlift_Configuration_Service} instance.
+     *
+     * @since  3.10.0
+     * @access private
+     * @var \Wordlift_Configuration_Service $configuration_service A {@link Wordlift_Configuration_Service} instance.
+     */
+    private $configuration_service;
+
+    /**
+     * A {@link Wordlift_Log_Service} instance.
+     *
+     * @since  3.10.0
+     * @access private
+     * @var Wordlift_Log_Service $log A {@link Wordlift_Log_Service} instance.
+     */
+    private $log;
+
+    /**
+     * Wordlift_Post_To_Jsonld_Converter constructor.
+     *
+     * @param \Wordlift_Entity_Type_Service $entity_type_service A {@link Wordlift_Entity_Type_Service} instance.
+     * @param \Wordlift_Entity_Service $entity_service A {@link Wordlift_Entity_Service} instance.
+     * @param \Wordlift_User_Service $user_service A {@link Wordlift_User_Service} instance.
+     * @param \Wordlift_Attachment_Service $attachment_service A {@link Wordlift_Attachment_Service} instance.
+     * @param \Wordlift_Configuration_Service $configuration_service A {@link Wordlift_Configuration_Service} instance.
+     *
+     * @since 3.10.0
+     *
+     */
+    public function __construct( $entity_type_service, $entity_service, $user_service, $attachment_service, $configuration_service ) {
+        parent::__construct( $entity_type_service, $entity_service, $user_service, $attachment_service );
+
+        $this->configuration_service = $configuration_service;
+
+        // Set a reference to the logger.
+        $this->log = Wordlift_Log_Service::get_logger( 'Wordlift_Post_To_Jsonld_Converter' );
+    }
+
+    /**
+     * Convert the provided {@link WP_Post} to a JSON-LD array. Any entity reference
+     * found while processing the post is set in the $references array.
+     *
+     * @param int $post_id The post id.
+     * @param array $references An array of entity references.
+     * @param array $references_infos
+     *
+     * @return array A JSON-LD array.
+     * @since 3.10.0
+     */
+    public function convert( $post_id, &$references = array(), &$references_infos = array() ) {
+
+        // Get the post instance.
+        if ( null === $post = get_post( $post_id ) ) {
+            // Post not found.
+            return null;
+        }
+
+        // Get the base JSON-LD and the list of entities referenced by this entity.
+        $jsonld = parent::convert( $post_id, $references, $references_infos );
+
+        // Set WebPage by default.
+        if ( empty( $jsonld['@type'] ) ) {
+            $jsonld['@type'] = 'WebPage';
+        }
+
+        // Get the entity name.
+        $jsonld['headline'] = $post->post_title;
+
+        // Set the published and modified dates.
+        /*
 		 * Set the `datePublished` and `dateModified` using the local timezone.
 		 *
 		 * @see https://github.com/insideout10/wordlift-plugin/issues/887
 		 *
 		 * @since 3.20.0
 		 */
-		try {
-			$default_timezone = date_default_timezone_get();
-			date_default_timezone_set( get_option( 'timezone_string' ) );
-			$jsonld['datePublished'] = get_post_time( 'Y-m-d\TH:i:sP', false, $post );
-			$jsonld['dateModified']  = get_post_modified_time( 'Y-m-d\TH:i:sP', false, $post );
-			date_default_timezone_set( $default_timezone );
-		} catch ( Exception $e ) {
-			$jsonld['datePublished'] = get_post_time( 'Y-m-d\TH:i', true, $post, false );
-			$jsonld['dateModified']  = get_post_modified_time( 'Y-m-d\TH:i', true, $post, false );
-		}
-
-		// Get the word count for the post.
-		/*
+        try {
+            $default_timezone = date_default_timezone_get();
+            date_default_timezone_set( get_option( 'timezone_string' ) );
+            $jsonld['datePublished'] = get_post_time( 'Y-m-d\TH:i:sP', false, $post );
+            $jsonld['dateModified']  = get_post_modified_time( 'Y-m-d\TH:i:sP', false, $post );
+            date_default_timezone_set( $default_timezone );
+        } catch ( Exception $e ) {
+            $jsonld['datePublished'] = get_post_time( 'Y-m-d\TH:i', true, $post, false );
+            $jsonld['dateModified']  = get_post_modified_time( 'Y-m-d\TH:i', true, $post, false );
+        }
+
+        // Get the word count for the post.
+        /*
 		 * Do not display the `wordCount` on a `WebPage`.
 		 *
 		 * @see https://github.com/insideout10/wordlift-plugin/issues/888
 		 *
 		 * @since 3.20.0
 		 */
-		if ( ! empty( $jsonld['@type'] ) && 'WebPage' !== $jsonld['@type'] ) {
-			$post_adapter        = new Wordlift_Post_Adapter( $post_id );
-			$jsonld['wordCount'] = $post_adapter->word_count();
-		}
+        if ( ! empty( $jsonld['@type'] ) && 'WebPage' !== $jsonld['@type'] ) {
+            $post_adapter        = new Wordlift_Post_Adapter( $post_id );
+            $jsonld['wordCount'] = $post_adapter->word_count();
+        }
 
-		/*
+        /*
 		 * Add keywords, articleSection, commentCount and inLanguage properties to `Article` JSON-LD
 		 *
 		 * @see https://github.com/insideout10/wordlift-plugin/issues/1140
 		 *
 		 * @since 3.27.2
 		 */
-		if ( ! empty( $jsonld['@type'] ) && 'Article' === $jsonld['@type'] ) {
-			$post_adapter    = new Wordlift_Post_Adapter( $post_id );
-			$keywords        = $post_adapter->keywords();
-			$article_section = $post_adapter->article_section();
-			$comment_count   = $post_adapter->comment_count();
-			$locale          = $post_adapter->locale();
-
-			if ( isset( $keywords ) ) {
-				$jsonld['keywords'] = $keywords;
-			}
-			if ( isset( $article_section ) ) {
-				$jsonld['articleSection'] = $article_section;
-			}
-			$jsonld['commentCount'] = $comment_count;
-			$jsonld['inLanguage']   = $locale;
-			//$post_adapter->add_mentions($post_id, $references );
-		}
-
-		// Set the publisher.
-		$this->set_publisher( $jsonld );
-
-		// Process the references if any.
-		if ( 0 < count( $references ) ) {
-
-			// Prepare the `about` and `mentions` array.
-			$about = $mentions = array();
-
-			// If the entity is in the title, then it should be an `about`.
-			foreach ( $references as $reference ) {
-
-				// Get the entity labels.
-				$labels = $this->entity_service->get_labels( $reference );
-
-				// Get the entity URI.
-				$item = array(
-					'@id' => $this->entity_service->get_uri( $reference ),
-				);
-
-				$escaped_labels = array_map( function ( $value ) {
-					return preg_quote( $value, '/' );
-				}, $labels );
-
-				// Check if the labels match any part of the title.
-				$matches = 1 === preg_match( '/' . implode( '|', $escaped_labels ) . '/', $post->post_title );
-
-				// If the title matches, assign the entity to the about, otherwise to the mentions.
-				if ( $matches ) {
-					$about[] = $item;
-				} else {
-					$mentions[] = $item;
-				}
-			}
-
-			// If we have abouts, assign them to the JSON-LD.
-			if ( 0 < count( $about ) ) {
-				$jsonld['about'] = $about;
-			}
-
-			// If we have mentions, assign them to the JSON-LD.
-			if ( 0 < count( $mentions ) ) {
-				$jsonld['mentions'] = $mentions;
-			}
-		}
-
-		// Finally set the author.
-		$jsonld['author'] = $this->get_author( $post->post_author, $references );
-
-		/**
-		 * Call the `wl_post_jsonld_array` filter. This filter allows 3rd parties to also modify the references.
-		 *
-		 * @param array $value {
-		 *
-		 * @type array $jsonld The JSON-LD structure.
-		 * @type int[] $references An array of post IDs.
-		 * }
-		 * @since 3.25.0
-		 *
-		 * @see https://www.geeklab.info/2010/04/wordpress-pass-variables-by-reference-with-apply_filter/
-		 *
-		 * @api
-		 */
-		$ret_val    = apply_filters( 'wl_post_jsonld_array', array(
-			'jsonld'     => $jsonld,
-			'references' => $references,
-		), $post_id );
-		$jsonld     = $ret_val['jsonld'];
-		$references = $ret_val['references'];
-
-		/**
-		 * Call the `wl_post_jsonld` filter.
-		 *
-		 * @param array $jsonld The JSON-LD structure.
-		 * @param int $post_id The {@link WP_Post} `id`.
-		 * @param array $references The array of referenced entities.
-		 *
-		 * @since 3.14.0
-		 *
-		 * @api
-		 */
-		return apply_filters( 'wl_post_jsonld', $jsonld, $post_id, $references );
-	}
-
-	/**
-	 * Get the author's JSON-LD fragment.
-	 *
-	 * The JSON-LD fragment is generated using the {@link WP_User}'s data or
-	 * the referenced entity if configured for the {@link WP_User}.
-	 *
-	 * @param int $author_id The author {@link WP_User}'s `id`.
-	 * @param array $references An array of referenced entities.
-	 *
-	 * @return string|array A JSON-LD structure.
-	 * @since 3.14.0
-	 *
-	 */
-	private function get_author( $author_id, &$references ) {
-
-		// Get the entity bound to this user.
-		$entity_id = $this->user_service->get_entity( $author_id );
-
-		// If there's no entity bound return a simple author structure.
-		if ( empty( $entity_id ) || 'publish' !== get_post_status( $entity_id ) ) {
-
-			$author     = get_the_author_meta( 'display_name', $author_id );
-			$author_uri = $this->user_service->get_uri( $author_id );
-
-			return array(
-				'@type' => 'Person',
-				'@id'   => $author_uri,
-				'name'  => $author,
-			);
-		}
-
-		// Add the author to the references.
-		$author_uri   = $this->entity_service->get_uri( $entity_id );
-		$references[] = $entity_id;
-
-		// Return the JSON-LD for the referenced entity.
-		return array(
-			'@id' => $author_uri,
-		);
-	}
-
-	/**
-	 * Enrich the provided params array with publisher data, if available.
-	 *
-	 * @param array $params The parameters array.
-	 *
-	 * @since 3.10.0
-	 *
-	 */
-	protected function set_publisher( &$params ) {
-
-		// If the publisher id isn't set don't do anything.
-		if ( null === $publisher_id = $this->configuration_service->get_publisher_id() ) {
-			return;
-		}
-
-		// Get the post instance.
-		if ( null === $post = get_post( $publisher_id ) ) {
-			// Publisher not found.
-			return;
-		}
-
-		// Get the item id.
-		$id = $this->entity_service->get_uri( $publisher_id );
-
-		// Get the type.
-		$type = $this->entity_type_service->get( $publisher_id );
-
-		// Get the name.
-		$name = $post->post_title;
-
-		// Set the publisher data.
-		$params['publisher'] = array(
-			'@type' => $this->relative_to_context( $type['uri'] ),
-			'@id'   => $id,
-			'name'  => $name,
-		);
-
-		// Add the sameAs values associated with the publisher.
-		$storage_factory = Wordlift_Storage_Factory::get_instance();
-		$sameas          = $storage_factory->post_meta( Wordlift_Schema_Service::FIELD_SAME_AS )->get( $publisher_id );
-		if ( ! empty( $sameas ) ) {
-			$params['publisher']['sameAs'] = $sameas;
-		}
-
-		// Set the logo, only for http://schema.org/Organization as Person doesn't
-		// support the logo property.
-		//
-		// See http://schema.org/logo.
-		if ( 'http://schema.org/Organization' !== $type['uri'] ) {
-			return;
-		}
-
-		// Get the publisher logo.
-		$publisher_logo = $this->get_publisher_logo( $post->ID );
-
-		// Bail out if the publisher logo isn't set.
-		if ( false === $publisher_logo ) {
-			return;
-		}
-
-		// Copy over some useful properties.
-		//
-		// See https://developers.google.com/search/docs/data-types/articles.
-		$params['publisher']['logo']['@type'] = 'ImageObject';
-		$params['publisher']['logo']['url']   = $publisher_logo['url'];
-
-		// If you specify a "width" or "height" value you should leave out
-		// 'px'. For example: "width":"4608px" should be "width":"4608".
-		//
-		// See https://github.com/insideout10/wordlift-plugin/issues/451.
-		$params['publisher']['logo']['width']  = $publisher_logo['width'];
-		$params['publisher']['logo']['height'] = $publisher_logo['height'];
-
-	}
-
-	/**
-	 * Get the publisher logo structure.
-	 *
-	 * The function returns false when the publisher logo cannot be determined, i.e.:
-	 *  - the post has no featured image.
-	 *  - the featured image has no file.
-	 *  - a wp_image_editor instance cannot be instantiated on the original file or on the publisher logo file.
-	 *
-	 * @param int $post_id The post id.
-	 *
-	 * @return array|false Returns an array with the `url`, `width` and `height` for the publisher logo or false in case
-	 *  of errors.
-	 * @since 3.19.2
-	 * @see https://github.com/insideout10/wordlift-plugin/issues/823 related issue.
-	 *
-	 */
-	private function get_publisher_logo( $post_id ) {
-
-		// Get the featured image for the post.
-		$thumbnail_id = get_post_thumbnail_id( $post_id );
-
-		// Bail out if thumbnail not available.
-		if ( empty( $thumbnail_id ) || 0 === $thumbnail_id ) {
-			$this->log->info( "Featured image not set for post $post_id." );
-
-			return false;
-		}
-
-		// Get the uploads base URL.
-		$uploads_dir = wp_upload_dir();
-
-		// Get the attachment metadata.
-		$metadata = wp_get_attachment_metadata( $thumbnail_id );
-
-		// Bail out if the file isn't set.
-		if ( ! isset( $metadata['file'] ) ) {
-			$this->log->warn( "Featured image file not found for post $post_id." );
-
-			return false;
-		}
-
-		// Retrieve the relative filename, e.g. "2018/05/logo_publisher.png"
-		$path = $uploads_dir['basedir'] . DIRECTORY_SEPARATOR . $metadata['file'];
-
-		// Bail out if the file isn't found.
-		if ( ! file_exists( $path ) ) {
-			$this->log->warn( "Featured image file $path doesn't exist for post $post_id." );
-
-			return false;
-		}
-
-		// Try to get the image editor and bail out if the editor cannot be instantiated.
-		$original_file_editor = wp_get_image_editor( $path );
-		if ( is_wp_error( $original_file_editor ) ) {
-			$this->log->warn( "Cannot instantiate WP Image Editor on file $path for post $post_id." );
-
-			return false;
-		}
-
-		// Generate the publisher logo filename, we cannot use the `width` and `height` because we're scaling
-		// and we don't actually know the end values.
-		$publisher_logo_path = $original_file_editor->generate_filename( '-publisher-logo' );
-
-		// If the file doesn't exist yet, create it.
-		if ( ! file_exists( $publisher_logo_path ) ) {
-			$original_file_editor->resize( 600, 60 );
-			$original_file_editor->save( $publisher_logo_path );
-		}
-
-		// Try to get the image editor and bail out if the editor cannot be instantiated.
-		$publisher_logo_editor = wp_get_image_editor( $publisher_logo_path );
-		if ( is_wp_error( $publisher_logo_editor ) ) {
-			$this->log->warn( "Cannot instantiate WP Image Editor on file $publisher_logo_path for post $post_id." );
-
-			return false;
-		}
-
-		// Get the actual size.
-		$size = $publisher_logo_editor->get_size();
-
-		// Finally return the array with data.
-		return array(
-			'url'    => $uploads_dir['baseurl'] . substr( $publisher_logo_path, strlen( $uploads_dir['basedir'] ) ),
-			'width'  => $size['width'],
-			'height' => $size['height'],
-		);
-	}
+        if ( ! empty( $jsonld['@type'] ) && 'Article' === $jsonld['@type'] ) {
+            $post_adapter    = new Wordlift_Post_Adapter( $post_id );
+            $keywords        = $post_adapter->keywords();
+            $article_section = $post_adapter->article_section();
+            $comment_count   = $post_adapter->comment_count();
+            $locale          = $post_adapter->locale();
+
+            if ( isset( $keywords ) ) {
+                $jsonld['keywords'] = $keywords;
+            }
+            if ( isset( $article_section ) ) {
+                $jsonld['articleSection'] = $article_section;
+            }
+            $jsonld['commentCount'] = $comment_count;
+            $jsonld['inLanguage']   = $locale;
+            //$post_adapter->add_mentions($post_id, $references );
+        }
+
+        // Set the publisher.
+        $this->set_publisher( $jsonld );
+
+        // Process the references if any.
+        if ( 0 < count( $references ) ) {
+
+            // Prepare the `about` and `mentions` array.
+            $about = $mentions = array();
+
+            // If the entity is in the title, then it should be an `about`.
+            foreach ( $references as $reference ) {
+
+                // Get the entity labels.
+                $labels = $this->entity_service->get_labels( $reference );
+
+                // Get the entity URI.
+                $item = array(
+                    '@id' => $this->entity_service->get_uri( $reference ),
+                );
+
+                $escaped_labels = array_map( function ( $value ) {
+                    return preg_quote( $value, '/' );
+                }, $labels );
+
+                // Check if the labels match any part of the title.
+                $matches = 1 === preg_match( '/' . implode( '|', $escaped_labels ) . '/', $post->post_title );
+
+                // If the title matches, assign the entity to the about, otherwise to the mentions.
+                if ( $matches ) {
+                    $about[] = $item;
+                } else {
+                    $mentions[] = $item;
+                }
+            }
+
+            // If we have abouts, assign them to the JSON-LD.
+            if ( 0 < count( $about ) ) {
+                $jsonld['about'] = $about;
+            }
+
+            // If we have mentions, assign them to the JSON-LD.
+            if ( 0 < count( $mentions ) ) {
+                $jsonld['mentions'] = $mentions;
+            }
+        }
+
+        // Finally set the author.
+        $jsonld['author'] = $this->get_author( $post->post_author, $references );
+
+        /**
+         * Call the `wl_post_jsonld_array` filter. This filter allows 3rd parties to also modify the references.
+         *
+         * @param array $value {
+         *
+         * @type array $jsonld The JSON-LD structure.
+         * @type int[] $references An array of post IDs.
+         * }
+         * @since 3.25.0
+         *
+         * @see https://www.geeklab.info/2010/04/wordpress-pass-variables-by-reference-with-apply_filter/
+         *
+         * @api
+         */
+        $ret_val    = apply_filters( 'wl_post_jsonld_array', array(
+            'jsonld'     => $jsonld,
+            'references' => $references,
+        ), $post_id );
+        $jsonld     = $ret_val['jsonld'];
+        $references = $ret_val['references'];
+
+        /**
+         * Call the `wl_post_jsonld` filter.
+         *
+         * @param array $jsonld The JSON-LD structure.
+         * @param int $post_id The {@link WP_Post} `id`.
+         * @param array $references The array of referenced entities.
+         *
+         * @since 3.14.0
+         *
+         * @api
+         */
+        return apply_filters( 'wl_post_jsonld', $jsonld, $post_id, $references );
+    }
+
+    /**
+     * Get the author's JSON-LD fragment.
+     *
+     * The JSON-LD fragment is generated using the {@link WP_User}'s data or
+     * the referenced entity if configured for the {@link WP_User}.
+     *
+     * @param int $author_id The author {@link WP_User}'s `id`.
+     * @param array $references An array of referenced entities.
+     *
+     * @return string|array A JSON-LD structure.
+     * @since 3.14.0
+     *
+     */
+    private function get_author( $author_id, &$references ) {
+
+        // Get the entity bound to this user.
+        $entity_id = $this->user_service->get_entity( $author_id );
+
+        // If there's no entity bound return a simple author structure.
+        if ( empty( $entity_id ) || 'publish' !== get_post_status( $entity_id ) ) {
+
+            $author     = get_the_author_meta( 'display_name', $author_id );
+            $author_uri = $this->user_service->get_uri( $author_id );
+
+            return array(
+                '@type' => 'Person',
+                '@id'   => $author_uri,
+                'name'  => $author,
+            );
+        }
+
+        // Add the author to the references.
+        $author_uri   = $this->entity_service->get_uri( $entity_id );
+        $references[] = $entity_id;
+
+        // Return the JSON-LD for the referenced entity.
+        return array(
+            '@id' => $author_uri,
+        );
+    }
+
+    /**
+     * Enrich the provided params array with publisher data, if available.
+     *
+     * @param array $params The parameters array.
+     *
+     * @since 3.10.0
+     *
+     */
+    protected function set_publisher( &$params ) {
+
+        // If the publisher id isn't set don't do anything.
+        if ( null === $publisher_id = $this->configuration_service->get_publisher_id() ) {
+            return;
+        }
+
+        // Get the post instance.
+        if ( null === $post = get_post( $publisher_id ) ) {
+            // Publisher not found.
+            return;
+        }
+
+        // Get the item id.
+        $id = $this->entity_service->get_uri( $publisher_id );
+
+        // Get the type.
+        $type = $this->entity_type_service->get( $publisher_id );
+
+        // Get the name.
+        $name = $post->post_title;
+
+        // Set the publisher data.
+        $params['publisher'] = array(
+            '@type' => $this->relative_to_context( $type['uri'] ),
+            '@id'   => $id,
+            'name'  => $name,
+        );
+
+        // Add the sameAs values associated with the publisher.
+        $storage_factory = Wordlift_Storage_Factory::get_instance();
+        $sameas          = $storage_factory->post_meta( Wordlift_Schema_Service::FIELD_SAME_AS )->get( $publisher_id );
+        if ( ! empty( $sameas ) ) {
+            $params['publisher']['sameAs'] = $sameas;
+        }
+
+        // Set the logo, only for http://schema.org/Organization as Person doesn't
+        // support the logo property.
+        //
+        // See http://schema.org/logo.
+        if ( 'http://schema.org/Organization' !== $type['uri'] ) {
+            return;
+        }
+
+        // Get the publisher logo.
+        $publisher_logo = $this->get_publisher_logo( $post->ID );
+
+        // Bail out if the publisher logo isn't set.
+        if ( false === $publisher_logo ) {
+            return;
+        }
+
+        // Copy over some useful properties.
+        //
+        // See https://developers.google.com/search/docs/data-types/articles.
+        $params['publisher']['logo']['@type'] = 'ImageObject';
+        $params['publisher']['logo']['url']   = $publisher_logo['url'];
+
+        // If you specify a "width" or "height" value you should leave out
+        // 'px'. For example: "width":"4608px" should be "width":"4608".
+        //
+        // See https://github.com/insideout10/wordlift-plugin/issues/451.
+        $params['publisher']['logo']['width']  = $publisher_logo['width'];
+        $params['publisher']['logo']['height'] = $publisher_logo['height'];
+
+    }
+
+    /**
+     * Get the publisher logo structure.
+     *
+     * The function returns false when the publisher logo cannot be determined, i.e.:
+     *  - the post has no featured image.
+     *  - the featured image has no file.
+     *  - a wp_image_editor instance cannot be instantiated on the original file or on the publisher logo file.
+     *
+     * @param int $post_id The post id.
+     *
+     * @return array|false Returns an array with the `url`, `width` and `height` for the publisher logo or false in case
+     *  of errors.
+     * @since 3.19.2
+     * @see https://github.com/insideout10/wordlift-plugin/issues/823 related issue.
+     *
+     */
+    private function get_publisher_logo( $post_id ) {
+
+        // Get the featured image for the post.
+        $thumbnail_id = get_post_thumbnail_id( $post_id );
+
+        // Bail out if thumbnail not available.
+        if ( empty( $thumbnail_id ) || 0 === $thumbnail_id ) {
+            $this->log->info( "Featured image not set for post $post_id." );
+
+            return false;
+        }
+
+        // Get the uploads base URL.
+        $uploads_dir = wp_upload_dir();
+
+        // Get the attachment metadata.
+        $metadata = wp_get_attachment_metadata( $thumbnail_id );
+
+        // Bail out if the file isn't set.
+        if ( ! isset( $metadata['file'] ) ) {
+            $this->log->warn( "Featured image file not found for post $post_id." );
+
+            return false;
+        }
+
+        // Retrieve the relative filename, e.g. "2018/05/logo_publisher.png"
+        $path = $uploads_dir['basedir'] . DIRECTORY_SEPARATOR . $metadata['file'];
+
+        // Bail out if the file isn't found.
+        if ( ! file_exists( $path ) ) {
+            $this->log->warn( "Featured image file $path doesn't exist for post $post_id." );
+
+            return false;
+        }
+
+        // Try to get the image editor and bail out if the editor cannot be instantiated.
+        $original_file_editor = wp_get_image_editor( $path );
+        if ( is_wp_error( $original_file_editor ) ) {
+            $this->log->warn( "Cannot instantiate WP Image Editor on file $path for post $post_id." );
+
+            return false;
+        }
+
+        // Generate the publisher logo filename, we cannot use the `width` and `height` because we're scaling
+        // and we don't actually know the end values.
+        $publisher_logo_path = $original_file_editor->generate_filename( '-publisher-logo' );
+
+        // If the file doesn't exist yet, create it.
+        if ( ! file_exists( $publisher_logo_path ) ) {
+            $original_file_editor->resize( 600, 60 );
+            $original_file_editor->save( $publisher_logo_path );
+        }
+
+        // Try to get the image editor and bail out if the editor cannot be instantiated.
+        $publisher_logo_editor = wp_get_image_editor( $publisher_logo_path );
+        if ( is_wp_error( $publisher_logo_editor ) ) {
+            $this->log->warn( "Cannot instantiate WP Image Editor on file $publisher_logo_path for post $post_id." );
+
+            return false;
+        }
+
+        // Get the actual size.
+        $size = $publisher_logo_editor->get_size();
+
+        // Finally return the array with data.
+        return array(
+            'url'    => $uploads_dir['baseurl'] . substr( $publisher_logo_path, strlen( $uploads_dir['basedir'] ) ),
+            'width'  => $size['width'],
+            'height' => $size['height'],
+        );
+    }
 
 }

Spacing +74 added lines, -74 removed lines

@@ -45,13 +45,13 @@ 
 	 * @since 3.10.0
 	 *
 	 */
-	public function __construct( $entity_type_service, $entity_service, $user_service, $attachment_service, $configuration_service ) {
-		parent::__construct( $entity_type_service, $entity_service, $user_service, $attachment_service );
+	public function __construct($entity_type_service, $entity_service, $user_service, $attachment_service, $configuration_service) {
+		parent::__construct($entity_type_service, $entity_service, $user_service, $attachment_service);
 
 		$this->configuration_service = $configuration_service;
 
 		// Set a reference to the logger.
-		$this->log = Wordlift_Log_Service::get_logger( 'Wordlift_Post_To_Jsonld_Converter' );
+		$this->log = Wordlift_Log_Service::get_logger('Wordlift_Post_To_Jsonld_Converter');
 	}
 
 	/**
@@ -65,19 +65,19 @@ 
 	 * @return array A JSON-LD array.
 	 * @since 3.10.0
 	 */
-	public function convert( $post_id, &$references = array(), &$references_infos = array() ) {
+	public function convert($post_id, &$references = array(), &$references_infos = array()) {
 
 		// Get the post instance.
-		if ( null === $post = get_post( $post_id ) ) {
+		if (null === $post = get_post($post_id)) {
 			// Post not found.
 			return null;
 		}
 
 		// Get the base JSON-LD and the list of entities referenced by this entity.
-		$jsonld = parent::convert( $post_id, $references, $references_infos );
+		$jsonld = parent::convert($post_id, $references, $references_infos);
 
 		// Set WebPage by default.
-		if ( empty( $jsonld['@type'] ) ) {
+		if (empty($jsonld['@type'])) {
 			$jsonld['@type'] = 'WebPage';
 		}
 
@@ -94,13 +94,13 @@ 
 		 */
 		try {
 			$default_timezone = date_default_timezone_get();
-			date_default_timezone_set( get_option( 'timezone_string' ) );
-			$jsonld['datePublished'] = get_post_time( 'Y-m-d\TH:i:sP', false, $post );
-			$jsonld['dateModified']  = get_post_modified_time( 'Y-m-d\TH:i:sP', false, $post );
-			date_default_timezone_set( $default_timezone );
-		} catch ( Exception $e ) {
-			$jsonld['datePublished'] = get_post_time( 'Y-m-d\TH:i', true, $post, false );
-			$jsonld['dateModified']  = get_post_modified_time( 'Y-m-d\TH:i', true, $post, false );
+			date_default_timezone_set(get_option('timezone_string'));
+			$jsonld['datePublished'] = get_post_time('Y-m-d\TH:i:sP', false, $post);
+			$jsonld['dateModified']  = get_post_modified_time('Y-m-d\TH:i:sP', false, $post);
+			date_default_timezone_set($default_timezone);
+		} catch (Exception $e) {
+			$jsonld['datePublished'] = get_post_time('Y-m-d\TH:i', true, $post, false);
+			$jsonld['dateModified']  = get_post_modified_time('Y-m-d\TH:i', true, $post, false);
 		}
 
 		// Get the word count for the post.
@@ -111,8 +111,8 @@ 
 		 *
 		 * @since 3.20.0
 		 */
-		if ( ! empty( $jsonld['@type'] ) && 'WebPage' !== $jsonld['@type'] ) {
-			$post_adapter        = new Wordlift_Post_Adapter( $post_id );
+		if ( ! empty($jsonld['@type']) && 'WebPage' !== $jsonld['@type']) {
+			$post_adapter        = new Wordlift_Post_Adapter($post_id);
 			$jsonld['wordCount'] = $post_adapter->word_count();
 		}
 
@@ -123,17 +123,17 @@ 
 		 *
 		 * @since 3.27.2
 		 */
-		if ( ! empty( $jsonld['@type'] ) && 'Article' === $jsonld['@type'] ) {
-			$post_adapter    = new Wordlift_Post_Adapter( $post_id );
+		if ( ! empty($jsonld['@type']) && 'Article' === $jsonld['@type']) {
+			$post_adapter    = new Wordlift_Post_Adapter($post_id);
 			$keywords        = $post_adapter->keywords();
 			$article_section = $post_adapter->article_section();
 			$comment_count   = $post_adapter->comment_count();
 			$locale          = $post_adapter->locale();
 
-			if ( isset( $keywords ) ) {
+			if (isset($keywords)) {
 				$jsonld['keywords'] = $keywords;
 			}
-			if ( isset( $article_section ) ) {
+			if (isset($article_section)) {
 				$jsonld['articleSection'] = $article_section;
 			}
 			$jsonld['commentCount'] = $comment_count;
@@ -142,34 +142,34 @@ 
 		}
 
 		// Set the publisher.
-		$this->set_publisher( $jsonld );
+		$this->set_publisher($jsonld);
 
 		// Process the references if any.
-		if ( 0 < count( $references ) ) {
+		if (0 < count($references)) {
 
 			// Prepare the `about` and `mentions` array.
 			$about = $mentions = array();
 
 			// If the entity is in the title, then it should be an `about`.
-			foreach ( $references as $reference ) {
+			foreach ($references as $reference) {
 
 				// Get the entity labels.
-				$labels = $this->entity_service->get_labels( $reference );
+				$labels = $this->entity_service->get_labels($reference);
 
 				// Get the entity URI.
 				$item = array(
-					'@id' => $this->entity_service->get_uri( $reference ),
+					'@id' => $this->entity_service->get_uri($reference),
 				);
 
-				$escaped_labels = array_map( function ( $value ) {
-					return preg_quote( $value, '/' );
-				}, $labels );
+				$escaped_labels = array_map(function($value) {
+					return preg_quote($value, '/');
+				}, $labels);
 
 				// Check if the labels match any part of the title.
-				$matches = 1 === preg_match( '/' . implode( '|', $escaped_labels ) . '/', $post->post_title );
+				$matches = 1 === preg_match('/'.implode('|', $escaped_labels).'/', $post->post_title);
 
 				// If the title matches, assign the entity to the about, otherwise to the mentions.
-				if ( $matches ) {
+				if ($matches) {
 					$about[] = $item;
 				} else {
 					$mentions[] = $item;
@@ -177,18 +177,18 @@ 
 			}
 
 			// If we have abouts, assign them to the JSON-LD.
-			if ( 0 < count( $about ) ) {
+			if (0 < count($about)) {
 				$jsonld['about'] = $about;
 			}
 
 			// If we have mentions, assign them to the JSON-LD.
-			if ( 0 < count( $mentions ) ) {
+			if (0 < count($mentions)) {
 				$jsonld['mentions'] = $mentions;
 			}
 		}
 
 		// Finally set the author.
-		$jsonld['author'] = $this->get_author( $post->post_author, $references );
+		$jsonld['author'] = $this->get_author($post->post_author, $references);
 
 		/**
 		 * Call the `wl_post_jsonld_array` filter. This filter allows 3rd parties to also modify the references.
@@ -204,10 +204,10 @@ 
 		 *
 		 * @api
 		 */
-		$ret_val    = apply_filters( 'wl_post_jsonld_array', array(
+		$ret_val = apply_filters('wl_post_jsonld_array', array(
 			'jsonld'     => $jsonld,
 			'references' => $references,
-		), $post_id );
+		), $post_id);
 		$jsonld     = $ret_val['jsonld'];
 		$references = $ret_val['references'];
 
@@ -222,7 +222,7 @@ 
 		 *
 		 * @api
 		 */
-		return apply_filters( 'wl_post_jsonld', $jsonld, $post_id, $references );
+		return apply_filters('wl_post_jsonld', $jsonld, $post_id, $references);
 	}
 
 	/**
@@ -238,16 +238,16 @@ 
 	 * @since 3.14.0
 	 *
 	 */
-	private function get_author( $author_id, &$references ) {
+	private function get_author($author_id, &$references) {
 
 		// Get the entity bound to this user.
-		$entity_id = $this->user_service->get_entity( $author_id );
+		$entity_id = $this->user_service->get_entity($author_id);
 
 		// If there's no entity bound return a simple author structure.
-		if ( empty( $entity_id ) || 'publish' !== get_post_status( $entity_id ) ) {
+		if (empty($entity_id) || 'publish' !== get_post_status($entity_id)) {
 
-			$author     = get_the_author_meta( 'display_name', $author_id );
-			$author_uri = $this->user_service->get_uri( $author_id );
+			$author     = get_the_author_meta('display_name', $author_id);
+			$author_uri = $this->user_service->get_uri($author_id);
 
 			return array(
 				'@type' => 'Person',
@@ -257,7 +257,7 @@ 
 		}
 
 		// Add the author to the references.
-		$author_uri   = $this->entity_service->get_uri( $entity_id );
+		$author_uri   = $this->entity_service->get_uri($entity_id);
 		$references[] = $entity_id;
 
 		// Return the JSON-LD for the referenced entity.
@@ -274,39 +274,39 @@ 
 	 * @since 3.10.0
 	 *
 	 */
-	protected function set_publisher( &$params ) {
+	protected function set_publisher(&$params) {
 
 		// If the publisher id isn't set don't do anything.
-		if ( null === $publisher_id = $this->configuration_service->get_publisher_id() ) {
+		if (null === $publisher_id = $this->configuration_service->get_publisher_id()) {
 			return;
 		}
 
 		// Get the post instance.
-		if ( null === $post = get_post( $publisher_id ) ) {
+		if (null === $post = get_post($publisher_id)) {
 			// Publisher not found.
 			return;
 		}
 
 		// Get the item id.
-		$id = $this->entity_service->get_uri( $publisher_id );
+		$id = $this->entity_service->get_uri($publisher_id);
 
 		// Get the type.
-		$type = $this->entity_type_service->get( $publisher_id );
+		$type = $this->entity_type_service->get($publisher_id);
 
 		// Get the name.
 		$name = $post->post_title;
 
 		// Set the publisher data.
 		$params['publisher'] = array(
-			'@type' => $this->relative_to_context( $type['uri'] ),
+			'@type' => $this->relative_to_context($type['uri']),
 			'@id'   => $id,
 			'name'  => $name,
 		);
 
 		// Add the sameAs values associated with the publisher.
 		$storage_factory = Wordlift_Storage_Factory::get_instance();
-		$sameas          = $storage_factory->post_meta( Wordlift_Schema_Service::FIELD_SAME_AS )->get( $publisher_id );
-		if ( ! empty( $sameas ) ) {
+		$sameas          = $storage_factory->post_meta(Wordlift_Schema_Service::FIELD_SAME_AS)->get($publisher_id);
+		if ( ! empty($sameas)) {
 			$params['publisher']['sameAs'] = $sameas;
 		}
 
@@ -314,15 +314,15 @@ 
 		// support the logo property.
 		//
 		// See http://schema.org/logo.
-		if ( 'http://schema.org/Organization' !== $type['uri'] ) {
+		if ('http://schema.org/Organization' !== $type['uri']) {
 			return;
 		}
 
 		// Get the publisher logo.
-		$publisher_logo = $this->get_publisher_logo( $post->ID );
+		$publisher_logo = $this->get_publisher_logo($post->ID);
 
 		// Bail out if the publisher logo isn't set.
-		if ( false === $publisher_logo ) {
+		if (false === $publisher_logo) {
 			return;
 		}
 
@@ -357,14 +357,14 @@ 
 	 * @see https://github.com/insideout10/wordlift-plugin/issues/823 related issue.
 	 *
 	 */
-	private function get_publisher_logo( $post_id ) {
+	private function get_publisher_logo($post_id) {
 
 		// Get the featured image for the post.
-		$thumbnail_id = get_post_thumbnail_id( $post_id );
+		$thumbnail_id = get_post_thumbnail_id($post_id);
 
 		// Bail out if thumbnail not available.
-		if ( empty( $thumbnail_id ) || 0 === $thumbnail_id ) {
-			$this->log->info( "Featured image not set for post $post_id." );
+		if (empty($thumbnail_id) || 0 === $thumbnail_id) {
+			$this->log->info("Featured image not set for post $post_id.");
 
 			return false;
 		}
@@ -373,47 +373,47 @@ 
 		$uploads_dir = wp_upload_dir();
 
 		// Get the attachment metadata.
-		$metadata = wp_get_attachment_metadata( $thumbnail_id );
+		$metadata = wp_get_attachment_metadata($thumbnail_id);
 
 		// Bail out if the file isn't set.
-		if ( ! isset( $metadata['file'] ) ) {
-			$this->log->warn( "Featured image file not found for post $post_id." );
+		if ( ! isset($metadata['file'])) {
+			$this->log->warn("Featured image file not found for post $post_id.");
 
 			return false;
 		}
 
 		// Retrieve the relative filename, e.g. "2018/05/logo_publisher.png"
-		$path = $uploads_dir['basedir'] . DIRECTORY_SEPARATOR . $metadata['file'];
+		$path = $uploads_dir['basedir'].DIRECTORY_SEPARATOR.$metadata['file'];
 
 		// Bail out if the file isn't found.
-		if ( ! file_exists( $path ) ) {
-			$this->log->warn( "Featured image file $path doesn't exist for post $post_id." );
+		if ( ! file_exists($path)) {
+			$this->log->warn("Featured image file $path doesn't exist for post $post_id.");
 
 			return false;
 		}
 
 		// Try to get the image editor and bail out if the editor cannot be instantiated.
-		$original_file_editor = wp_get_image_editor( $path );
-		if ( is_wp_error( $original_file_editor ) ) {
-			$this->log->warn( "Cannot instantiate WP Image Editor on file $path for post $post_id." );
+		$original_file_editor = wp_get_image_editor($path);
+		if (is_wp_error($original_file_editor)) {
+			$this->log->warn("Cannot instantiate WP Image Editor on file $path for post $post_id.");
 
 			return false;
 		}
 
 		// Generate the publisher logo filename, we cannot use the `width` and `height` because we're scaling
 		// and we don't actually know the end values.
-		$publisher_logo_path = $original_file_editor->generate_filename( '-publisher-logo' );
+		$publisher_logo_path = $original_file_editor->generate_filename('-publisher-logo');
 
 		// If the file doesn't exist yet, create it.
-		if ( ! file_exists( $publisher_logo_path ) ) {
-			$original_file_editor->resize( 600, 60 );
-			$original_file_editor->save( $publisher_logo_path );
+		if ( ! file_exists($publisher_logo_path)) {
+			$original_file_editor->resize(600, 60);
+			$original_file_editor->save($publisher_logo_path);
 		}
 
 		// Try to get the image editor and bail out if the editor cannot be instantiated.
-		$publisher_logo_editor = wp_get_image_editor( $publisher_logo_path );
-		if ( is_wp_error( $publisher_logo_editor ) ) {
-			$this->log->warn( "Cannot instantiate WP Image Editor on file $publisher_logo_path for post $post_id." );
+		$publisher_logo_editor = wp_get_image_editor($publisher_logo_path);
+		if (is_wp_error($publisher_logo_editor)) {
+			$this->log->warn("Cannot instantiate WP Image Editor on file $publisher_logo_path for post $post_id.");
 
 			return false;
 		}
@@ -423,7 +423,7 @@ 
 
 		// Finally return the array with data.
 		return array(
-			'url'    => $uploads_dir['baseurl'] . substr( $publisher_logo_path, strlen( $uploads_dir['basedir'] ) ),
+			'url'    => $uploads_dir['baseurl'].substr($publisher_logo_path, strlen($uploads_dir['basedir'])),
 			'width'  => $size['width'],
 			'height' => $size['height'],
 		);

src/includes/class-wordlift-post-adapter.php

Indentation +248 added lines, -248 removed lines

@@ -16,258 +16,258 @@
  */
 class Wordlift_Post_Adapter {
 
-	/**
-	 * The post id to which the adopter relates.
-	 *
-	 * @since 3.14.0
-	 *
-	 * @var integer $post_id .
-	 */
-	private $post_id;
-
-	const TYPE_ENTITY_LINK = 0;
-	const TYPE_TERM_LINK = 1;
-
-	/**
-	 * Create the {@link Wordlift_Post_Adatpter} instance.
-	 *
-	 * @param integer $post_id the post ID of the post the adopter relates to.
-	 *
-	 * @since 3.14.0
-	 *
-	 */
-	public function __construct( $post_id ) {
-
-		$this->post_id = $post_id;
-
-	}
-
-	/**
-	 * Get the word count of the post content.
-	 *
-	 * The count is calculated over the post content after stripping shortcodes and html tags.
-	 *
-	 * @return integer the number of words in the content after stripping shortcodes and html tags..
-	 * @since 3.14.0
-	 *
-	 */
-	public function word_count() {
-
-		$post = get_post( $this->post_id );
-
-		/*
+    /**
+     * The post id to which the adopter relates.
+     *
+     * @since 3.14.0
+     *
+     * @var integer $post_id .
+     */
+    private $post_id;
+
+    const TYPE_ENTITY_LINK = 0;
+    const TYPE_TERM_LINK = 1;
+
+    /**
+     * Create the {@link Wordlift_Post_Adatpter} instance.
+     *
+     * @param integer $post_id the post ID of the post the adopter relates to.
+     *
+     * @since 3.14.0
+     *
+     */
+    public function __construct( $post_id ) {
+
+        $this->post_id = $post_id;
+
+    }
+
+    /**
+     * Get the word count of the post content.
+     *
+     * The count is calculated over the post content after stripping shortcodes and html tags.
+     *
+     * @return integer the number of words in the content after stripping shortcodes and html tags..
+     * @since 3.14.0
+     *
+     */
+    public function word_count() {
+
+        $post = get_post( $this->post_id );
+
+        /*
 		 * Apply the `wl_post_content` filter, in case 3rd parties want to change the post content, e.g.
 		 * because the content is written elsewhere.
 		 *
 		 * @since 3.20.0
 		 */
-		$post_content = apply_filters( 'wl_post_content', $post->post_content, $post );
-
-		return self::str_word_count_utf8( strip_tags( strip_shortcodes( $post_content ) ) );
-	}
-
-	/**
-	 * Count words in the string, taking into account UTF-8 characters.
-	 *
-	 * @see https://github.com/insideout10/wordlift-plugin/issues/884
-	 *
-	 * @since 3.20.0
-	 *
-	 * @param string $str The target string.
-	 *
-	 * @return int The number of words.
-	 */
-	private static function str_word_count_utf8( $str ) {
-
-		return count( preg_split( '~[^\p{L}\p{N}\']+~u', $str ) );
-	}
-
-	/**
-	 * Get the {@link WP_Post} permalink allowing 3rd parties to alter the URL.
-	 *
-	 * @param int $post_id Post ID.
-	 *
-	 * @return string The post permalink.
-	 * @since 3.20.0
-	 *
-	 */
-	public static function get_production_permalink( $post_id ) {
-
-		/**
-		 * WordPress 4.4 doesn't support meta queries for terms, therefore we only support post permalinks here.
-		 *
-		 * Later on for WordPress 4.5+ we look for terms bound to the entity and we use the term link instead of the
-		 * post permalink if we find them.
-		 */
-		global $wp_version;
-		if ( version_compare( $wp_version, '4.5', '<' ) ) {
-			$permalink = get_permalink( $post_id );
-
-			return apply_filters( 'wl_production_permalink', $permalink, $post_id, self::TYPE_ENTITY_LINK, null );
-		}
-
-		/**
-		 * The `wl_production_permalink` filter allows to change the permalink, this is useful in contexts
-		 * when the production environment is copied over from a staging environment with staging
-		 * URLs.
-		 *
-		 * @param string $permalink_url The default permalink.
-		 * @param int $post_id The post id.
-		 *
-		 * @since 3.23.0 we check whether the entity is bound to a term and, in that case, we link to the term.
-		 * @since 3.20.0
-		 *
-		 * @see https://github.com/insideout10/wordlift-plugin/issues/850
-		 *
-		 */
-
-		// Get all the URIs for the entity, i.e. itemid and sameAs.
-		$uris = array_merge(
-			(array) Wordlift_Entity_Service::get_instance()->get_uri( $post_id ),
-			get_post_meta( $post_id, Wordlift_Schema_Service::FIELD_SAME_AS )
-		);
-
-		// Only try to link a term if `WL_ENABLE_TERM_LINKING` is enabled.
-		$terms = array();
-		if ( defined( 'WL_ENABLE_TERM_LINKING' ) && WL_ENABLE_TERM_LINKING ) {
-			// Try to find one term matching the entity.
-			$terms = get_terms( array(
-				'number'                 => 1,
-				'hide_empty'             => false,
-				'update_term_meta_cache' => false,
-				'meta_key'               => '_wl_entity_id',
-				'meta_value'             => $uris,
-				'meta_compare'           => 'IN',
-			) );
-		}
-
-		// If found use the term link, otherwise the permalink.
-		if ( 1 === count( $terms ) ) {
-			$term      = current( $terms );
-			$permalink = get_term_link( $term );
-			$type      = self::TYPE_TERM_LINK;
-		} else {
-			$term      = null;
-			$permalink = get_permalink( $post_id );
-			$type      = self::TYPE_ENTITY_LINK;
-		}
-
-		/**
-		 * Apply the `wl_production_permalink` filter.
-		 *
-		 * @param string $permalink The permalink.
-		 * @param int $post_id The post id.
-		 * @param int $type The permalink type: 0 = entity permalink, 1 = term link.
-		 * @param WP_Term $term The term if type is term link, otherwise null.
-		 *
-		 * @since 3.23.0 add the permalink type and term parameters.
-		 *
-		 */
-		return apply_filters( 'wl_production_permalink', $permalink, $post_id, $type, $term );
-	}
-
-	/**
-	 * Get comma separated tags to be used as keywords
-	 *
-	 * @return string|void Comma separated tags
-	 *
-	 * @since 3.27.2
-	 */
-	public function keywords() {
-		$tags = get_the_tags( $this->post_id );
-
-		if ( empty( $tags ) ) {
-			return;
-		}
-
-		return implode( ',', array_map( function ( $tag ) {
-			return $tag->name;
-		}, $tags ) );
-	}
-
-	/**
-	 * Get comma separated categories to be used as article section
-	 *
-	 * @return string|void Comma separated categories
-	 *
-	 * @since 3.27.2
-	 */
-	public function article_section() {
-		$categories = get_the_category( $this->post_id );
-
-		if ( empty( $categories ) ) {
-			return;
-		}
-
-		return implode( ',', array_map( function ( $tag ) {
-			return $tag->cat_name;
-		}, $categories ) );
-	}
-
-	/**
-	 * Get comment count
-	 *
-	 * @return string|void Comment count
-	 *
-	 * @since 3.27.2
-	 */
-	public function comment_count() {
-		return get_comments_number( $this->post_id );
-	}
-
-	/**
-	 * Get Language
-	 * Try WPML, Polylang for post specific languages, else fallback on get_locale()
-	 *
-	 * @return string|void Language code (locale)
-	 *
-	 * @since 3.27.2
-	 */
-	public function locale() {
-		// WPML handling
-		if ( function_exists( 'wpml_get_language_information' ) ) {
-			$post_language = wpml_get_language_information( $this->post_id );
-
-			return $post_language['locale'];
-		}
-
-		// Polylang handling
-		if ( function_exists( 'pll_get_post_language' ) ) {
-			return pll_get_post_language( $this->post_id, 'locale' );
-		}
-
-		return get_locale();
-
-	}
-
-	/**
-	 * Add mentions to an article.
-	 * To add mentions we just push the referenced entities to mentions.
-	 *
-	 * @param $post_id int
-	 * @param $references int[]
-	 */
-	public function add_mentions( $post_id, &$references ) {
-		// Create empty array for references.
-		if ( ! is_array( $references ) ) {
-			$references = array();
-		}
-
-		$tags = get_the_tags( $post_id );
-		// Loop through the tags and push it to references.
-		foreach ( $tags as $tag ) {
-			/**
-			 * @var $tag WP_Term
-			 */
-			$entity_uris = get_term_meta( $tag->term_id, '_wl_entity_id' );
-			foreach ( $entity_uris as $uri ) {
-				$referenced_entity = Wordlift_Entity_Uri_Service::get_instance()->get_entity( $uri );
-				if ( $referenced_entity instanceof WP_Post ) {
-					// push the referenced entities to references.
-					$references[] = $referenced_entity->ID;
-				}
-			}
-		}
-	}
+        $post_content = apply_filters( 'wl_post_content', $post->post_content, $post );
+
+        return self::str_word_count_utf8( strip_tags( strip_shortcodes( $post_content ) ) );
+    }
+
+    /**
+     * Count words in the string, taking into account UTF-8 characters.
+     *
+     * @see https://github.com/insideout10/wordlift-plugin/issues/884
+     *
+     * @since 3.20.0
+     *
+     * @param string $str The target string.
+     *
+     * @return int The number of words.
+     */
+    private static function str_word_count_utf8( $str ) {
+
+        return count( preg_split( '~[^\p{L}\p{N}\']+~u', $str ) );
+    }
+
+    /**
+     * Get the {@link WP_Post} permalink allowing 3rd parties to alter the URL.
+     *
+     * @param int $post_id Post ID.
+     *
+     * @return string The post permalink.
+     * @since 3.20.0
+     *
+     */
+    public static function get_production_permalink( $post_id ) {
+
+        /**
+         * WordPress 4.4 doesn't support meta queries for terms, therefore we only support post permalinks here.
+         *
+         * Later on for WordPress 4.5+ we look for terms bound to the entity and we use the term link instead of the
+         * post permalink if we find them.
+         */
+        global $wp_version;
+        if ( version_compare( $wp_version, '4.5', '<' ) ) {
+            $permalink = get_permalink( $post_id );
+
+            return apply_filters( 'wl_production_permalink', $permalink, $post_id, self::TYPE_ENTITY_LINK, null );
+        }
+
+        /**
+         * The `wl_production_permalink` filter allows to change the permalink, this is useful in contexts
+         * when the production environment is copied over from a staging environment with staging
+         * URLs.
+         *
+         * @param string $permalink_url The default permalink.
+         * @param int $post_id The post id.
+         *
+         * @since 3.23.0 we check whether the entity is bound to a term and, in that case, we link to the term.
+         * @since 3.20.0
+         *
+         * @see https://github.com/insideout10/wordlift-plugin/issues/850
+         *
+         */
+
+        // Get all the URIs for the entity, i.e. itemid and sameAs.
+        $uris = array_merge(
+            (array) Wordlift_Entity_Service::get_instance()->get_uri( $post_id ),
+            get_post_meta( $post_id, Wordlift_Schema_Service::FIELD_SAME_AS )
+        );
+
+        // Only try to link a term if `WL_ENABLE_TERM_LINKING` is enabled.
+        $terms = array();
+        if ( defined( 'WL_ENABLE_TERM_LINKING' ) && WL_ENABLE_TERM_LINKING ) {
+            // Try to find one term matching the entity.
+            $terms = get_terms( array(
+                'number'                 => 1,
+                'hide_empty'             => false,
+                'update_term_meta_cache' => false,
+                'meta_key'               => '_wl_entity_id',
+                'meta_value'             => $uris,
+                'meta_compare'           => 'IN',
+            ) );
+        }
+
+        // If found use the term link, otherwise the permalink.
+        if ( 1 === count( $terms ) ) {
+            $term      = current( $terms );
+            $permalink = get_term_link( $term );
+            $type      = self::TYPE_TERM_LINK;
+        } else {
+            $term      = null;
+            $permalink = get_permalink( $post_id );
+            $type      = self::TYPE_ENTITY_LINK;
+        }
+
+        /**
+         * Apply the `wl_production_permalink` filter.
+         *
+         * @param string $permalink The permalink.
+         * @param int $post_id The post id.
+         * @param int $type The permalink type: 0 = entity permalink, 1 = term link.
+         * @param WP_Term $term The term if type is term link, otherwise null.
+         *
+         * @since 3.23.0 add the permalink type and term parameters.
+         *
+         */
+        return apply_filters( 'wl_production_permalink', $permalink, $post_id, $type, $term );
+    }
+
+    /**
+     * Get comma separated tags to be used as keywords
+     *
+     * @return string|void Comma separated tags
+     *
+     * @since 3.27.2
+     */
+    public function keywords() {
+        $tags = get_the_tags( $this->post_id );
+
+        if ( empty( $tags ) ) {
+            return;
+        }
+
+        return implode( ',', array_map( function ( $tag ) {
+            return $tag->name;
+        }, $tags ) );
+    }
+
+    /**
+     * Get comma separated categories to be used as article section
+     *
+     * @return string|void Comma separated categories
+     *
+     * @since 3.27.2
+     */
+    public function article_section() {
+        $categories = get_the_category( $this->post_id );
+
+        if ( empty( $categories ) ) {
+            return;
+        }
+
+        return implode( ',', array_map( function ( $tag ) {
+            return $tag->cat_name;
+        }, $categories ) );
+    }
+
+    /**
+     * Get comment count
+     *
+     * @return string|void Comment count
+     *
+     * @since 3.27.2
+     */
+    public function comment_count() {
+        return get_comments_number( $this->post_id );
+    }
+
+    /**
+     * Get Language
+     * Try WPML, Polylang for post specific languages, else fallback on get_locale()
+     *
+     * @return string|void Language code (locale)
+     *
+     * @since 3.27.2
+     */
+    public function locale() {
+        // WPML handling
+        if ( function_exists( 'wpml_get_language_information' ) ) {
+            $post_language = wpml_get_language_information( $this->post_id );
+
+            return $post_language['locale'];
+        }
+
+        // Polylang handling
+        if ( function_exists( 'pll_get_post_language' ) ) {
+            return pll_get_post_language( $this->post_id, 'locale' );
+        }
+
+        return get_locale();
+
+    }
+
+    /**
+     * Add mentions to an article.
+     * To add mentions we just push the referenced entities to mentions.
+     *
+     * @param $post_id int
+     * @param $references int[]
+     */
+    public function add_mentions( $post_id, &$references ) {
+        // Create empty array for references.
+        if ( ! is_array( $references ) ) {
+            $references = array();
+        }
+
+        $tags = get_the_tags( $post_id );
+        // Loop through the tags and push it to references.
+        foreach ( $tags as $tag ) {
+            /**
+             * @var $tag WP_Term
+             */
+            $entity_uris = get_term_meta( $tag->term_id, '_wl_entity_id' );
+            foreach ( $entity_uris as $uri ) {
+                $referenced_entity = Wordlift_Entity_Uri_Service::get_instance()->get_entity( $uri );
+                if ( $referenced_entity instanceof WP_Post ) {
+                    // push the referenced entities to references.
+                    $references[] = $referenced_entity->ID;
+                }
+            }
+        }
+    }
 
 }

Spacing +41 added lines, -41 removed lines

@@ -36,7 +36,7 @@ 
 	 * @since 3.14.0
 	 *
 	 */
-	public function __construct( $post_id ) {
+	public function __construct($post_id) {
 
 		$this->post_id = $post_id;
 
@@ -53,7 +53,7 @@ 
 	 */
 	public function word_count() {
 
-		$post = get_post( $this->post_id );
+		$post = get_post($this->post_id);
 
 		/*
 		 * Apply the `wl_post_content` filter, in case 3rd parties want to change the post content, e.g.
@@ -61,9 +61,9 @@ 
 		 *
 		 * @since 3.20.0
 		 */
-		$post_content = apply_filters( 'wl_post_content', $post->post_content, $post );
+		$post_content = apply_filters('wl_post_content', $post->post_content, $post);
 
-		return self::str_word_count_utf8( strip_tags( strip_shortcodes( $post_content ) ) );
+		return self::str_word_count_utf8(strip_tags(strip_shortcodes($post_content)));
 	}
 
 	/**
@@ -77,9 +77,9 @@ 
 	 *
 	 * @return int The number of words.
 	 */
-	private static function str_word_count_utf8( $str ) {
+	private static function str_word_count_utf8($str) {
 
-		return count( preg_split( '~[^\p{L}\p{N}\']+~u', $str ) );
+		return count(preg_split('~[^\p{L}\p{N}\']+~u', $str));
 	}
 
 	/**
@@ -91,7 +91,7 @@ 
 	 * @since 3.20.0
 	 *
 	 */
-	public static function get_production_permalink( $post_id ) {
+	public static function get_production_permalink($post_id) {
 
 		/**
 		 * WordPress 4.4 doesn't support meta queries for terms, therefore we only support post permalinks here.
@@ -100,10 +100,10 @@ 
 		 * post permalink if we find them.
 		 */
 		global $wp_version;
-		if ( version_compare( $wp_version, '4.5', '<' ) ) {
-			$permalink = get_permalink( $post_id );
+		if (version_compare($wp_version, '4.5', '<')) {
+			$permalink = get_permalink($post_id);
 
-			return apply_filters( 'wl_production_permalink', $permalink, $post_id, self::TYPE_ENTITY_LINK, null );
+			return apply_filters('wl_production_permalink', $permalink, $post_id, self::TYPE_ENTITY_LINK, null);
 		}
 
 		/**
@@ -123,32 +123,32 @@ 
 
 		// Get all the URIs for the entity, i.e. itemid and sameAs.
 		$uris = array_merge(
-			(array) Wordlift_Entity_Service::get_instance()->get_uri( $post_id ),
-			get_post_meta( $post_id, Wordlift_Schema_Service::FIELD_SAME_AS )
+			(array) Wordlift_Entity_Service::get_instance()->get_uri($post_id),
+			get_post_meta($post_id, Wordlift_Schema_Service::FIELD_SAME_AS)
 		);
 
 		// Only try to link a term if `WL_ENABLE_TERM_LINKING` is enabled.
 		$terms = array();
-		if ( defined( 'WL_ENABLE_TERM_LINKING' ) && WL_ENABLE_TERM_LINKING ) {
+		if (defined('WL_ENABLE_TERM_LINKING') && WL_ENABLE_TERM_LINKING) {
 			// Try to find one term matching the entity.
-			$terms = get_terms( array(
+			$terms = get_terms(array(
 				'number'                 => 1,
 				'hide_empty'             => false,
 				'update_term_meta_cache' => false,
 				'meta_key'               => '_wl_entity_id',
 				'meta_value'             => $uris,
 				'meta_compare'           => 'IN',
-			) );
+			));
 		}
 
 		// If found use the term link, otherwise the permalink.
-		if ( 1 === count( $terms ) ) {
-			$term      = current( $terms );
-			$permalink = get_term_link( $term );
+		if (1 === count($terms)) {
+			$term      = current($terms);
+			$permalink = get_term_link($term);
 			$type      = self::TYPE_TERM_LINK;
 		} else {
 			$term      = null;
-			$permalink = get_permalink( $post_id );
+			$permalink = get_permalink($post_id);
 			$type      = self::TYPE_ENTITY_LINK;
 		}
 
@@ -163,7 +163,7 @@ 
 		 * @since 3.23.0 add the permalink type and term parameters.
 		 *
 		 */
-		return apply_filters( 'wl_production_permalink', $permalink, $post_id, $type, $term );
+		return apply_filters('wl_production_permalink', $permalink, $post_id, $type, $term);
 	}
 
 	/**
@@ -174,15 +174,15 @@ 
 	 * @since 3.27.2
 	 */
 	public function keywords() {
-		$tags = get_the_tags( $this->post_id );
+		$tags = get_the_tags($this->post_id);
 
-		if ( empty( $tags ) ) {
+		if (empty($tags)) {
 			return;
 		}
 
-		return implode( ',', array_map( function ( $tag ) {
+		return implode(',', array_map(function($tag) {
 			return $tag->name;
-		}, $tags ) );
+		}, $tags));
 	}
 
 	/**
@@ -193,15 +193,15 @@ 
 	 * @since 3.27.2
 	 */
 	public function article_section() {
-		$categories = get_the_category( $this->post_id );
+		$categories = get_the_category($this->post_id);
 
-		if ( empty( $categories ) ) {
+		if (empty($categories)) {
 			return;
 		}
 
-		return implode( ',', array_map( function ( $tag ) {
+		return implode(',', array_map(function($tag) {
 			return $tag->cat_name;
-		}, $categories ) );
+		}, $categories));
 	}
 
 	/**
@@ -212,7 +212,7 @@ 
 	 * @since 3.27.2
 	 */
 	public function comment_count() {
-		return get_comments_number( $this->post_id );
+		return get_comments_number($this->post_id);
 	}
 
 	/**
@@ -225,15 +225,15 @@ 
 	 */
 	public function locale() {
 		// WPML handling
-		if ( function_exists( 'wpml_get_language_information' ) ) {
-			$post_language = wpml_get_language_information( $this->post_id );
+		if (function_exists('wpml_get_language_information')) {
+			$post_language = wpml_get_language_information($this->post_id);
 
 			return $post_language['locale'];
 		}
 
 		// Polylang handling
-		if ( function_exists( 'pll_get_post_language' ) ) {
-			return pll_get_post_language( $this->post_id, 'locale' );
+		if (function_exists('pll_get_post_language')) {
+			return pll_get_post_language($this->post_id, 'locale');
 		}
 
 		return get_locale();
@@ -247,22 +247,22 @@ 
 	 * @param $post_id int
 	 * @param $references int[]
 	 */
-	public function add_mentions( $post_id, &$references ) {
+	public function add_mentions($post_id, &$references) {
 		// Create empty array for references.
-		if ( ! is_array( $references ) ) {
+		if ( ! is_array($references)) {
 			$references = array();
 		}
 
-		$tags = get_the_tags( $post_id );
+		$tags = get_the_tags($post_id);
 		// Loop through the tags and push it to references.
-		foreach ( $tags as $tag ) {
+		foreach ($tags as $tag) {
 			/**
 			 * @var $tag WP_Term
 			 */
-			$entity_uris = get_term_meta( $tag->term_id, '_wl_entity_id' );
-			foreach ( $entity_uris as $uri ) {
-				$referenced_entity = Wordlift_Entity_Uri_Service::get_instance()->get_entity( $uri );
-				if ( $referenced_entity instanceof WP_Post ) {
+			$entity_uris = get_term_meta($tag->term_id, '_wl_entity_id');
+			foreach ($entity_uris as $uri) {
+				$referenced_entity = Wordlift_Entity_Uri_Service::get_instance()->get_entity($uri);
+				if ($referenced_entity instanceof WP_Post) {
 					// push the referenced entities to references.
 					$references[] = $referenced_entity->ID;
 				}