| 1 |
<?php |
| 2 |
|
| 3 |
namespace Yoast\WP\SEO\Builders; |
| 4 |
|
| 5 |
use Yoast\WP\SEO\Helpers\Image_Helper; |
| 6 |
use Yoast\WP\SEO\Helpers\Post_Helper; |
| 7 |
use Yoast\WP\SEO\Helpers\Url_Helper; |
| 8 |
use Yoast\WP\SEO\Models\Indexable; |
| 9 |
use Yoast\WP\SEO\Models\SEO_Links; |
| 10 |
use Yoast\WP\SEO\Repositories\Indexable_Repository; |
| 11 |
use Yoast\WP\SEO\Repositories\SEO_Links_Repository; |
| 12 |
|
| 13 |
/** |
| 14 |
* Indexable link builder. |
| 15 |
*/ |
| 16 |
class Indexable_Link_Builder { |
| 17 |
|
| 18 |
/** |
| 19 |
* The SEO links repository. |
| 20 |
* |
| 21 |
* @var SEO_Links_Repository |
| 22 |
*/ |
| 23 |
protected $seo_links_repository; |
| 24 |
|
| 25 |
/** |
| 26 |
* The url helper. |
| 27 |
* |
| 28 |
* @var Url_Helper |
| 29 |
*/ |
| 30 |
protected $url_helper; |
| 31 |
|
| 32 |
/** |
| 33 |
* The image helper. |
| 34 |
* |
| 35 |
* @var Image_Helper |
| 36 |
*/ |
| 37 |
protected $image_helper; |
| 38 |
|
| 39 |
/** |
| 40 |
* The post helper. |
| 41 |
* |
| 42 |
* @var Post_Helper |
| 43 |
*/ |
| 44 |
protected $post_helper; |
| 45 |
|
| 46 |
/** |
| 47 |
* The indexable repository. |
| 48 |
* |
| 49 |
* @var Indexable_Repository |
| 50 |
*/ |
| 51 |
protected $indexable_repository; |
| 52 |
|
| 53 |
/** |
| 54 |
* Indexable_Link_Builder constructor. |
| 55 |
* |
| 56 |
* @param SEO_Links_Repository $seo_links_repository The SEO links repository. |
| 57 |
* @param Url_Helper $url_helper The URL helper. |
| 58 |
* @param Post_Helper $post_helper The post helper. |
| 59 |
*/ |
| 60 |
public function __construct( |
| 61 |
SEO_Links_Repository $seo_links_repository, |
| 62 |
Url_Helper $url_helper, |
| 63 |
Post_Helper $post_helper |
| 64 |
) { |
| 65 |
$this->seo_links_repository = $seo_links_repository; |
| 66 |
$this->url_helper = $url_helper; |
| 67 |
$this->post_helper = $post_helper; |
| 68 |
} |
| 69 |
|
| 70 |
/** |
| 71 |
* Sets the indexable repository. |
| 72 |
* |
| 73 |
* @required |
| 74 |
* |
| 75 |
* @param Indexable_Repository $indexable_repository The indexable repository. |
| 76 |
* @param Image_Helper $image_helper The image helper. |
| 77 |
* |
| 78 |
* @return void |
| 79 |
*/ |
| 80 |
public function set_dependencies( |
| 81 |
Indexable_Repository $indexable_repository, |
| 82 |
Image_Helper $image_helper |
| 83 |
) { |
| 84 |
$this->indexable_repository = $indexable_repository; |
| 85 |
$this->image_helper = $image_helper; |
| 86 |
} |
| 87 |
|
| 88 |
/** |
| 89 |
* Builds the links for a post. |
| 90 |
* |
| 91 |
* @param Indexable $indexable The indexable. |
| 92 |
* @param string $content The content. Expected to be unfiltered. |
| 93 |
* |
| 94 |
* @return SEO_Links[] The created SEO links. |
| 95 |
*/ |
| 96 |
public function build( $indexable, $content ) { |
| 97 |
global $post; |
| 98 |
if ( $indexable->object_type === 'post' ) { |
| 99 |
$post_backup = $post; |
| 100 |
// phpcs:ignore WordPress.WP.GlobalVariablesOverride.Prohibited -- To setup the post we need to do this explicitly. |
| 101 |
$post = $this->post_helper->get_post( $indexable->object_id ); |
| 102 |
\setup_postdata( $post ); |
| 103 |
$content = \apply_filters( 'the_content', $content ); |
| 104 |
\wp_reset_postdata(); |
| 105 |
// phpcs:ignore WordPress.WP.GlobalVariablesOverride.Prohibited -- To setup the post we need to do this explicitly. |
| 106 |
$post = $post_backup; |
| 107 |
} |
| 108 |
|
| 109 |
$content = \str_replace( ']]>', ']]>', $content ); |
| 110 |
$links = $this->gather_links( $content ); |
| 111 |
$images = $this->gather_images( $content ); |
| 112 |
|
| 113 |
if ( empty( $links ) && empty( $images ) ) { |
| 114 |
$indexable->link_count = 0; |
| 115 |
$this->update_related_indexables( $indexable, [] ); |
| 116 |
|
| 117 |
return []; |
| 118 |
} |
| 119 |
|
| 120 |
$links = $this->create_links( $indexable, $links, $images ); |
| 121 |
|
| 122 |
$this->update_related_indexables( $indexable, $links ); |
| 123 |
|
| 124 |
$indexable->link_count = $this->get_internal_link_count( $links ); |
| 125 |
|
| 126 |
return $links; |
| 127 |
} |
| 128 |
|
| 129 |
/** |
| 130 |
* Deletes all SEO links for an indexable. |
| 131 |
* |
| 132 |
* @param Indexable $indexable The indexable. |
| 133 |
* |
| 134 |
* @return void |
| 135 |
*/ |
| 136 |
public function delete( $indexable ) { |
| 137 |
$links = ( $this->seo_links_repository->find_all_by_indexable_id( $indexable->id ) ); |
| 138 |
$this->seo_links_repository->delete_all_by_indexable_id( $indexable->id ); |
| 139 |
|
| 140 |
$linked_indexable_ids = []; |
| 141 |
foreach ( $links as $link ) { |
| 142 |
$linked_indexable_ids[] = $link->target_indexable_id; |
| 143 |
} |
| 144 |
|
| 145 |
$this->update_incoming_links_for_related_indexables( $linked_indexable_ids ); |
| 146 |
} |
| 147 |
|
| 148 |
/** |
| 149 |
* Gathers all links from content. |
| 150 |
* |
| 151 |
* @param string $content The content. |
| 152 |
* |
| 153 |
* @return string[] An array of urls. |
| 154 |
*/ |
| 155 |
protected function gather_links( $content ) { |
| 156 |
if ( \strpos( $content, 'href' ) === false ) { |
| 157 |
// Nothing to do. |
| 158 |
return []; |
| 159 |
} |
| 160 |
|
| 161 |
$links = []; |
| 162 |
$regexp = '<a\s[^>]*href=("??)([^" >]*?)\1[^>]*>'; |
| 163 |
// Used modifiers iU to match case insensitive and make greedy quantifiers lazy. |
| 164 |
if ( \preg_match_all( "/$regexp/iU", $content, $matches, \PREG_SET_ORDER ) ) { |
| 165 |
foreach ( $matches as $match ) { |
| 166 |
$links[] = \trim( $match[2], "'" ); |
| 167 |
} |
| 168 |
} |
| 169 |
|
| 170 |
return $links; |
| 171 |
} |
| 172 |
|
| 173 |
/** |
| 174 |
* Gathers all images from content. |
| 175 |
* |
| 176 |
* @param string $content The content. |
| 177 |
* |
| 178 |
* @return string[] An array of urls. |
| 179 |
*/ |
| 180 |
protected function gather_images( $content ) { |
| 181 |
if ( \strpos( $content, 'src' ) === false ) { |
| 182 |
// Nothing to do. |
| 183 |
return []; |
| 184 |
} |
| 185 |
|
| 186 |
$images = []; |
| 187 |
$regexp = '<img\s[^>]*src=("??)([^" >]*?)\\1[^>]*>'; |
| 188 |
// Used modifiers iU to match case insensitive and make greedy quantifiers lazy. |
| 189 |
if ( \preg_match_all( "/$regexp/iU", $content, $matches, \PREG_SET_ORDER ) ) { |
| 190 |
foreach ( $matches as $match ) { |
| 191 |
$images[] = \trim( $match[2], "'" ); |
| 192 |
} |
| 193 |
} |
| 194 |
|
| 195 |
return $images; |
| 196 |
} |
| 197 |
|
| 198 |
/** |
| 199 |
* Creates link models from lists of URLs and image sources. |
| 200 |
* |
| 201 |
* @param Indexable $indexable The indexable. |
| 202 |
* @param string[] $links The link URLs. |
| 203 |
* @param string[] $images The image sources. |
| 204 |
* |
| 205 |
* @return SEO_Links[] The link models. |
| 206 |
*/ |
| 207 |
protected function create_links( $indexable, $links, $images ) { |
| 208 |
$home_url = \wp_parse_url( \home_url() ); |
| 209 |
$current_url = \wp_parse_url( $indexable->permalink ); |
| 210 |
$links = \array_map( |
| 211 |
function( $link ) use ( $home_url, $indexable ) { |
| 212 |
return $this->create_internal_link( $link, $home_url, $indexable ); |
| 213 |
}, |
| 214 |
$links |
| 215 |
); |
| 216 |
// Filter out links to the same page with a fragment or query. |
| 217 |
$links = \array_filter( |
| 218 |
$links, |
| 219 |
function ( $link ) use ( $current_url ) { |
| 220 |
return $this->filter_link( $link, $current_url ); |
| 221 |
} |
| 222 |
); |
| 223 |
|
| 224 |
$images = \array_map( |
| 225 |
function( $link ) use ( $home_url, $indexable ) { |
| 226 |
return $this->create_internal_link( $link, $home_url, $indexable, true ); |
| 227 |
}, |
| 228 |
$images |
| 229 |
); |
| 230 |
return \array_merge( $links, $images ); |
| 231 |
} |
| 232 |
|
| 233 |
/** |
| 234 |
* Get the post ID based on the link's type and its target's permalink. |
| 235 |
* |
| 236 |
* @param string $type The type of link (either SEO_Links::TYPE_INTERNAL or SEO_Links::TYPE_INTERNAL_IMAGE). |
| 237 |
* @param string $permalink The permalink of the link's target. |
| 238 |
* |
| 239 |
* @return int The post ID. |
| 240 |
*/ |
| 241 |
protected function get_post_id( $type, $permalink ) { |
| 242 |
if ( $type === SEO_Links::TYPE_INTERNAL ) { |
| 243 |
return \url_to_postid( $permalink ); |
| 244 |
} |
| 245 |
|
| 246 |
return $this->image_helper->get_attachment_by_url( $permalink ); |
| 247 |
} |
| 248 |
|
| 249 |
/** |
| 250 |
* Creates an internal link. |
| 251 |
* |
| 252 |
* @param string $url The url of the link. |
| 253 |
* @param array $home_url The home url, as parsed by wp_parse_url. |
| 254 |
* @param Indexable $indexable The indexable of the post containing the link. |
| 255 |
* @param bool $is_image Whether or not the link is an image. |
| 256 |
* |
| 257 |
* @return SEO_Links The created link. |
| 258 |
*/ |
| 259 |
protected function create_internal_link( $url, $home_url, $indexable, $is_image = false ) { |
| 260 |
$parsed_url = \wp_parse_url( $url ); |
| 261 |
$link_type = $this->url_helper->get_link_type( $parsed_url, $home_url, $is_image ); |
| 262 |
|
| 263 |
/** |
| 264 |
* ORM representing a link in the SEO Links table. |
| 265 |
* |
| 266 |
* @var SEO_Links $model |
| 267 |
*/ |
| 268 |
$model = $this->seo_links_repository->query()->create( |
| 269 |
[ |
| 270 |
'url' => $url, |
| 271 |
'type' => $link_type, |
| 272 |
'indexable_id' => $indexable->id, |
| 273 |
'post_id' => $indexable->object_id, |
| 274 |
] |
| 275 |
); |
| 276 |
|
| 277 |
$model->parsed_url = $parsed_url; |
| 278 |
|
| 279 |
if ( $model->type === SEO_Links::TYPE_INTERNAL || $model->type === SEO_Links::TYPE_INTERNAL_IMAGE ) { |
| 280 |
$permalink = $this->get_permalink( $url, $home_url ); |
| 281 |
if ( $this->url_helper->is_relative( $permalink ) ) { |
| 282 |
// Make sure we're checking against the absolute URL, and add a trailing slash if the site has a trailing slash in its permalink settings. |
| 283 |
$permalink = $this->url_helper->ensure_absolute_url( \user_trailingslashit( $permalink ) ); |
| 284 |
} |
| 285 |
$target = $this->indexable_repository->find_by_permalink( $permalink ); |
| 286 |
|
| 287 |
if ( ! $target ) { |
| 288 |
// If target indexable cannot be found, create one based on the post's post ID. |
| 289 |
$post_id = $this->get_post_id( $model->type, $permalink ); |
| 290 |
if ( $post_id && $post_id !== 0 ) { |
| 291 |
$target = $this->indexable_repository->find_by_id_and_type( $post_id, 'post' ); |
| 292 |
} |
| 293 |
} |
| 294 |
|
| 295 |
if ( ! $target ) { |
| 296 |
return $model; |
| 297 |
} |
| 298 |
|
| 299 |
$model->target_indexable_id = $target->id; |
| 300 |
if ( $target->object_type === 'post' ) { |
| 301 |
$model->target_post_id = $target->object_id; |
| 302 |
} |
| 303 |
} |
| 304 |
|
| 305 |
if ( $is_image && $model->target_post_id ) { |
| 306 |
$file = \get_attached_file( $model->target_post_id ); |
| 307 |
if ( $file ) { |
| 308 |
list( , $width, $height ) = \wp_get_attachment_image_src( $model->target_post_id, 'full' ); |
| 309 |
|
| 310 |
$model->width = $width; |
| 311 |
$model->height = $height; |
| 312 |
$model->size = \filesize( $file ); |
| 313 |
} |
| 314 |
else { |
| 315 |
$model->width = 0; |
| 316 |
$model->height = 0; |
| 317 |
$model->size = 0; |
| 318 |
} |
| 319 |
} |
| 320 |
|
| 321 |
if ( $model->target_indexable_id ) { |
| 322 |
$model->language = $target->language; |
| 323 |
$model->region = $target->region; |
| 324 |
} |
| 325 |
|
| 326 |
return $model; |
| 327 |
} |
| 328 |
|
| 329 |
/** |
| 330 |
* Filters out links that point to the same page with a fragment or query. |
| 331 |
* |
| 332 |
* @param SEO_Links $link The link. |
| 333 |
* @param array $current_url The url of the page the link is on, as parsed by wp_parse_url. |
| 334 |
* |
| 335 |
* @return bool Whether or not the link should be filtered. |
| 336 |
*/ |
| 337 |
protected function filter_link( SEO_Links $link, $current_url ) { |
| 338 |
$url = $link->parsed_url; |
| 339 |
|
| 340 |
// Always keep external links. |
| 341 |
if ( $link->type === SEO_Links::TYPE_EXTERNAL ) { |
| 342 |
return true; |
| 343 |
} |
| 344 |
|
| 345 |
// Always keep links with an empty path or pointing to other pages. |
| 346 |
if ( isset( $url['path'] ) ) { |
| 347 |
return empty( $url['path'] ) || $url['path'] !== $current_url['path']; |
| 348 |
} |
| 349 |
|
| 350 |
// Only keep links to the current page without a fragment or query. |
| 351 |
return ( ! isset( $url['fragment'] ) && ! isset( $url['query'] ) ); |
| 352 |
} |
| 353 |
|
| 354 |
/** |
| 355 |
* Updates the link counts for related indexables. |
| 356 |
* |
| 357 |
* @param Indexable $indexable The indexable. |
| 358 |
* @param SEO_Links[] $links The link models. |
| 359 |
* |
| 360 |
* @return void |
| 361 |
*/ |
| 362 |
protected function update_related_indexables( $indexable, $links ) { |
| 363 |
// Old links were only stored by post id, so remove all old seo links for this post that have no indexable id. |
| 364 |
// This can be removed if we ever fully clear all seo links. |
| 365 |
if ( $indexable->object_type === 'post' ) { |
| 366 |
$this->seo_links_repository->delete_all_by_post_id_where_indexable_id_null( $indexable->object_id ); |
| 367 |
} |
| 368 |
|
| 369 |
$updated_indexable_ids = []; |
| 370 |
$old_links = $this->seo_links_repository->find_all_by_indexable_id( $indexable->id ); |
| 371 |
|
| 372 |
$links_to_remove = $this->links_diff( $old_links, $links ); |
| 373 |
$links_to_add = $this->links_diff( $links, $old_links ); |
| 374 |
|
| 375 |
if ( ! empty( $links_to_remove ) ) { |
| 376 |
$this->seo_links_repository->delete_many_by_id( \wp_list_pluck( $links_to_remove, 'id' ) ); |
| 377 |
} |
| 378 |
|
| 379 |
if ( ! empty( $links_to_add ) ) { |
| 380 |
$this->seo_links_repository->insert_many( $links_to_add ); |
| 381 |
} |
| 382 |
|
| 383 |
foreach ( $links_to_add as $link ) { |
| 384 |
if ( $link->target_indexable_id ) { |
| 385 |
$updated_indexable_ids[] = $link->target_indexable_id; |
| 386 |
} |
| 387 |
} |
| 388 |
foreach ( $links_to_remove as $link ) { |
| 389 |
$updated_indexable_ids[] = $link->target_indexable_id; |
| 390 |
} |
| 391 |
|
| 392 |
$this->update_incoming_links_for_related_indexables( $updated_indexable_ids ); |
| 393 |
} |
| 394 |
|
| 395 |
/** |
| 396 |
* Creates a diff between two arrays of SEO links, based on urls. |
| 397 |
* |
| 398 |
* @param SEO_Links[] $links_a The array to compare. |
| 399 |
* @param SEO_Links[] $links_b The array to compare against. |
| 400 |
* |
| 401 |
* @return SEO_Links[] Links that are in $links_a, but not in $links_b. |
| 402 |
*/ |
| 403 |
protected function links_diff( $links_a, $links_b ) { |
| 404 |
return \array_udiff( |
| 405 |
$links_a, |
| 406 |
$links_b, |
| 407 |
static function( SEO_Links $link_a, SEO_Links $link_b ) { |
| 408 |
return \strcmp( $link_a->url, $link_b->url ); |
| 409 |
} |
| 410 |
); |
| 411 |
} |
| 412 |
|
| 413 |
/** |
| 414 |
* Returns the number of internal links in an array of link models. |
| 415 |
* |
| 416 |
* @param SEO_Links[] $links The link models. |
| 417 |
* |
| 418 |
* @return int The number of internal links. |
| 419 |
*/ |
| 420 |
protected function get_internal_link_count( $links ) { |
| 421 |
$internal_link_count = 0; |
| 422 |
|
| 423 |
foreach ( $links as $link ) { |
| 424 |
if ( $link->type === SEO_Links::TYPE_INTERNAL ) { |
| 425 |
++$internal_link_count; |
| 426 |
} |
| 427 |
} |
| 428 |
|
| 429 |
return $internal_link_count; |
| 430 |
} |
| 431 |
|
| 432 |
/** |
| 433 |
* Returns a cleaned permalink for a given link. |
| 434 |
* |
| 435 |
* @param string $link The raw URL. |
| 436 |
* @param array $home_url The home URL, as parsed by wp_parse_url. |
| 437 |
* |
| 438 |
* @return string The cleaned permalink. |
| 439 |
*/ |
| 440 |
protected function get_permalink( $link, $home_url ) { |
| 441 |
// Get rid of the #anchor. |
| 442 |
$url_split = \explode( '#', $link ); |
| 443 |
$link = $url_split[0]; |
| 444 |
|
| 445 |
// Get rid of URL ?query=string. |
| 446 |
$url_split = \explode( '?', $link ); |
| 447 |
$link = $url_split[0]; |
| 448 |
|
| 449 |
// Set the correct URL scheme. |
| 450 |
$link = \set_url_scheme( $link, $home_url['scheme'] ); |
| 451 |
|
| 452 |
// Add 'www.' if it is absent and should be there. |
| 453 |
if ( \strpos( $home_url['host'], 'www.' ) === 0 && \strpos( $link, '://www.' ) === false ) { |
| 454 |
$link = \str_replace( '://', '://www.', $link ); |
| 455 |
} |
| 456 |
|
| 457 |
// Strip 'www.' if it is present and shouldn't be. |
| 458 |
if ( \strpos( $home_url['host'], 'www.' ) !== 0 ) { |
| 459 |
$link = \str_replace( '://www.', '://', $link ); |
| 460 |
} |
| 461 |
|
| 462 |
return $link; |
| 463 |
} |
| 464 |
|
| 465 |
/** |
| 466 |
* Updates incoming link counts for related indexables. |
| 467 |
* |
| 468 |
* @param int[] $related_indexable_ids The IDs of all related indexables. |
| 469 |
* |
| 470 |
* @return void |
| 471 |
*/ |
| 472 |
protected function update_incoming_links_for_related_indexables( $related_indexable_ids ) { |
| 473 |
if ( empty( $related_indexable_ids ) ) { |
| 474 |
return; |
| 475 |
} |
| 476 |
|
| 477 |
$counts = $this->seo_links_repository->get_incoming_link_counts_for_indexable_ids( $related_indexable_ids ); |
| 478 |
foreach ( $counts as $count ) { |
| 479 |
$this->indexable_repository->update_incoming_link_count( $count['target_indexable_id'], $count['incoming'] ); |
| 480 |
} |
| 481 |
} |
| 482 |
} |
| 483 |
|
