| 1 |
<?php |
| 2 |
// phpcs:disable Yoast.NamingConventions.NamespaceName.TooLong -- Needed in the folder structure. |
| 3 |
namespace Yoast\WP\SEO\Schema_Aggregator\Application\Enhancement; |
| 4 |
|
| 5 |
use Exception; |
| 6 |
use Yoast\WP\SEO\Models\Indexable; |
| 7 |
use Yoast\WP\SEO\Schema_Aggregator\Domain\Enhancement\Schema_Enhancement_Interface; |
| 8 |
use Yoast\WP\SEO\Schema_Aggregator\Domain\Schema_Piece; |
| 9 |
use Yoast\WP\SEO\Schema_Aggregator\Infrastructure\Enhancement\Article_Config; |
| 10 |
|
| 11 |
/** |
| 12 |
* The Article schema pieces enhancer. |
| 13 |
*/ |
| 14 |
class Article_Schema_Enhancer extends Abstract_Schema_Enhancer implements Schema_Enhancement_Interface { |
| 15 |
|
| 16 |
/** |
| 17 |
* The config. |
| 18 |
* |
| 19 |
* @var Article_Config |
| 20 |
*/ |
| 21 |
private $config; |
| 22 |
|
| 23 |
/** |
| 24 |
* Sets the Article_Config instance. |
| 25 |
* |
| 26 |
* @required |
| 27 |
* |
| 28 |
* @param Article_Config $config The Article_Config instance. |
| 29 |
* |
| 30 |
* @return void |
| 31 |
*/ |
| 32 |
public function set_article_config( Article_Config $config ) { |
| 33 |
$this->config = $config; |
| 34 |
} |
| 35 |
|
| 36 |
/** |
| 37 |
* Enhances specific Article schema pieces. |
| 38 |
* |
| 39 |
* @param Schema_Piece $schema_piece The schema piece to enhance. |
| 40 |
* @param Indexable $indexable The indexable object that is the source of the schema piece. |
| 41 |
* |
| 42 |
* @return Schema_Piece The enhanced schema piece. |
| 43 |
*/ |
| 44 |
public function enhance( Schema_Piece $schema_piece, Indexable $indexable ): Schema_Piece { |
| 45 |
$schema_data = $schema_piece->get_data(); |
| 46 |
if ( ! isset( $schema_data['@type'] ) ) { |
| 47 |
return $schema_piece; |
| 48 |
} |
| 49 |
if ( |
| 50 |
\in_array( |
| 51 |
$schema_data['@type'], |
| 52 |
[ |
| 53 |
'Article', |
| 54 |
'NewsArticle', |
| 55 |
'BlogPosting', |
| 56 |
], |
| 57 |
true, |
| 58 |
) ) { |
| 59 |
$schema_data = $this->enhance_schema_piece( $schema_data, $indexable ); |
| 60 |
} |
| 61 |
|
| 62 |
if ( |
| 63 |
\is_array( $schema_data['@type'] ) && \in_array( 'Article', $schema_data['@type'], true ) ) { |
| 64 |
$schema_data = $this->enhance_schema_piece( $schema_data, $indexable ); |
| 65 |
} |
| 66 |
|
| 67 |
return new Schema_Piece( $schema_data, $schema_piece->get_type() ); |
| 68 |
} |
| 69 |
|
| 70 |
/** |
| 71 |
* Enhance a single schema piece |
| 72 |
* |
| 73 |
* @param array<string> $schema_data The schema data to enhance. |
| 74 |
* @param Indexable $indexable The indexable object that is the source of the schema piece. |
| 75 |
* |
| 76 |
* @return array<string> The enhanced schema data. |
| 77 |
*/ |
| 78 |
protected function enhance_schema_piece( array $schema_data, Indexable $indexable ): array { |
| 79 |
try { |
| 80 |
$has_excerpt = false; |
| 81 |
|
| 82 |
if ( $this->config->is_enhancement_enabled( 'use_excerpt' ) ) { |
| 83 |
$excerpt = $this->get_excerpt( $indexable->object_id ); |
| 84 |
$has_excerpt = ! empty( $excerpt ); |
| 85 |
|
| 86 |
if ( $has_excerpt && ! isset( $schema_data['description'] ) ) { |
| 87 |
$schema_data['description'] = $excerpt; |
| 88 |
} |
| 89 |
} |
| 90 |
|
| 91 |
if ( $this->config->is_enhancement_enabled( 'article_body' ) && ! isset( $schema_data['articleBody'] ) ) { |
| 92 |
if ( $this->config->should_include_article_body( $has_excerpt ) ) { |
| 93 |
$article_body = $this->get_article_body( $indexable->object_id ); |
| 94 |
if ( ! empty( $article_body ) ) { |
| 95 |
$schema_data['articleBody'] = $article_body; |
| 96 |
} |
| 97 |
} |
| 98 |
} |
| 99 |
|
| 100 |
if ( $this->config->is_enhancement_enabled( 'keywords' ) ) { |
| 101 |
$keywords = $this->get_article_keywords( $indexable->object_id ); |
| 102 |
if ( ! empty( $keywords ) ) { |
| 103 |
$existing = (array) ( $schema_data['keywords'] ?? [] ); |
| 104 |
$schema_data['keywords'] = \implode( ', ', \array_unique( \array_merge( $existing, $keywords ) ) ); |
| 105 |
} |
| 106 |
} |
| 107 |
|
| 108 |
return $schema_data; |
| 109 |
} catch ( Exception $e ) { |
| 110 |
return $schema_data; |
| 111 |
} |
| 112 |
} |
| 113 |
|
| 114 |
/** |
| 115 |
* Get article keywords |
| 116 |
* |
| 117 |
* Extracts post tags and optionally categories as keywords. |
| 118 |
* |
| 119 |
* @param int $post_id Post ID. |
| 120 |
* |
| 121 |
* @return array<string> Array of keyword strings. |
| 122 |
*/ |
| 123 |
private function get_article_keywords( int $post_id ): array { |
| 124 |
try { |
| 125 |
$keywords = []; |
| 126 |
|
| 127 |
if ( $this->config->get_config_value( 'categories_as_keywords', false ) ) { |
| 128 |
$categories = \get_the_category( $post_id ); |
| 129 |
if ( \is_array( $categories ) && ! empty( $categories ) ) { |
| 130 |
foreach ( $categories as $category ) { |
| 131 |
if ( isset( $category->name ) && $category->name !== 'Uncategorized' ) { |
| 132 |
$keywords[] = $category->name; |
| 133 |
} |
| 134 |
} |
| 135 |
} |
| 136 |
} |
| 137 |
|
| 138 |
return \array_unique( $keywords ); |
| 139 |
} catch ( Exception $e ) { |
| 140 |
return []; |
| 141 |
} |
| 142 |
} |
| 143 |
|
| 144 |
/** |
| 145 |
* Get article excerpt for description field |
| 146 |
* |
| 147 |
* Retrieves post excerpt with robust validation (no empty/whitespace-only). |
| 148 |
* Falls back to auto-generated excerpt from content unless prefer_manual is enabled. |
| 149 |
* |
| 150 |
* @param int $post_id Post ID. |
| 151 |
* |
| 152 |
* @return string|null Excerpt or null if unavailable/invalid. |
| 153 |
*/ |
| 154 |
private function get_excerpt( int $post_id ): ?string { |
| 155 |
try { |
| 156 |
$excerpt = \get_post_field( 'post_excerpt', $post_id ); |
| 157 |
if ( \is_wp_error( $excerpt ) ) { |
| 158 |
$excerpt = ''; |
| 159 |
} |
| 160 |
|
| 161 |
if ( empty( $excerpt ) || \trim( $excerpt ) === '' ) { |
| 162 |
if ( $this->config->get_config_value( 'excerpt_prefer_manual', false ) ) { |
| 163 |
return null; |
| 164 |
} |
| 165 |
|
| 166 |
$content = \get_post_field( 'post_content', $post_id ); |
| 167 |
|
| 168 |
if ( \is_wp_error( $content ) || empty( $content ) ) { |
| 169 |
return null; |
| 170 |
} |
| 171 |
$excerpt = \wp_trim_excerpt( $content, $post_id ); |
| 172 |
|
| 173 |
if ( empty( $excerpt ) || \trim( $excerpt ) === '' ) { |
| 174 |
return null; |
| 175 |
} |
| 176 |
} |
| 177 |
$excerpt = \wp_strip_all_tags( $excerpt ); |
| 178 |
|
| 179 |
// Apply max length if configured. |
| 180 |
$max_length = $this->config->get_config_value( 'excerpt_max_length', 0 ); |
| 181 |
$excerpt = $this->trim_content_to_max_length( $max_length, $excerpt ); |
| 182 |
|
| 183 |
$excerpt = \trim( $excerpt ); |
| 184 |
|
| 185 |
return ( $excerpt !== '' ) ? $excerpt : null; |
| 186 |
} catch ( Exception $e ) { |
| 187 |
return null; |
| 188 |
} |
| 189 |
} |
| 190 |
|
| 191 |
/** |
| 192 |
* Get article body (full post content) |
| 193 |
* |
| 194 |
* Extracts full post content with optional HTML and shortcode stripping. |
| 195 |
* Respects max_length configuration if set. |
| 196 |
* |
| 197 |
* @param int $post_id Post ID. |
| 198 |
* |
| 199 |
* @return string|null Article body or null if unavailable. |
| 200 |
*/ |
| 201 |
private function get_article_body( int $post_id ): ?string { |
| 202 |
try { |
| 203 |
$content = \get_post_field( 'post_content', $post_id ); |
| 204 |
|
| 205 |
if ( \is_wp_error( $content ) || empty( $content ) ) { |
| 206 |
return null; |
| 207 |
} |
| 208 |
|
| 209 |
if ( $this->config->get_config_value( 'strip_shortcodes_from_body', true ) ) { |
| 210 |
$content = \strip_shortcodes( $content ); |
| 211 |
} |
| 212 |
|
| 213 |
if ( $this->config->get_config_value( 'strip_html_from_body', true ) ) { |
| 214 |
$content = \wp_strip_all_tags( $content ); |
| 215 |
} |
| 216 |
|
| 217 |
$max_length = $this->config->get_config_value( 'article_body_max_length', Article_Config::DEFAULT_MAX_ARTICLE_BODY_LENGTH ); |
| 218 |
|
| 219 |
return $this->trim_content_to_max_length( $max_length, $content ); |
| 220 |
} catch ( Exception $e ) { |
| 221 |
return null; |
| 222 |
} |
| 223 |
} |
| 224 |
} |
| 225 |
|
