| 1 |
<?php |
| 2 |
/** |
| 3 |
* WPSEO plugin file. |
| 4 |
* |
| 5 |
* @package WPSEO\XML_Sitemaps |
| 6 |
*/ |
| 7 |
|
| 8 |
/** |
| 9 |
* Parses images from the given post. |
| 10 |
*/ |
| 11 |
class WPSEO_Sitemap_Image_Parser { |
| 12 |
|
| 13 |
/** |
| 14 |
* Holds the home_url() value to speed up loops. |
| 15 |
* |
| 16 |
* @var string |
| 17 |
*/ |
| 18 |
protected $home_url = ''; |
| 19 |
|
| 20 |
/** |
| 21 |
* Holds site URL hostname. |
| 22 |
* |
| 23 |
* @var string |
| 24 |
*/ |
| 25 |
protected $host = ''; |
| 26 |
|
| 27 |
/** |
| 28 |
* Holds site URL protocol. |
| 29 |
* |
| 30 |
* @var string |
| 31 |
*/ |
| 32 |
protected $scheme = 'http'; |
| 33 |
|
| 34 |
/** |
| 35 |
* Cached set of attachments for multiple posts. |
| 36 |
* |
| 37 |
* @var array |
| 38 |
*/ |
| 39 |
protected $attachments = []; |
| 40 |
|
| 41 |
/** |
| 42 |
* Holds blog charset value for use in DOM parsing. |
| 43 |
* |
| 44 |
* @var string |
| 45 |
*/ |
| 46 |
protected $charset = 'UTF-8'; |
| 47 |
|
| 48 |
/** |
| 49 |
* Set up URL properties for reuse. |
| 50 |
*/ |
| 51 |
public function __construct() { |
| 52 |
|
| 53 |
$this->home_url = home_url(); |
| 54 |
$parsed_home = wp_parse_url( $this->home_url ); |
| 55 |
|
| 56 |
if ( ! empty( $parsed_home['host'] ) ) { |
| 57 |
$this->host = str_replace( 'www.', '', $parsed_home['host'] ); |
| 58 |
} |
| 59 |
|
| 60 |
if ( ! empty( $parsed_home['scheme'] ) ) { |
| 61 |
$this->scheme = $parsed_home['scheme']; |
| 62 |
} |
| 63 |
|
| 64 |
$this->charset = esc_attr( get_bloginfo( 'charset' ) ); |
| 65 |
} |
| 66 |
|
| 67 |
/** |
| 68 |
* Get set of image data sets for the given post. |
| 69 |
* |
| 70 |
* @param object $post Post object to get images for. |
| 71 |
* |
| 72 |
* @return array |
| 73 |
*/ |
| 74 |
public function get_images( $post ) { |
| 75 |
|
| 76 |
$images = []; |
| 77 |
|
| 78 |
if ( ! is_object( $post ) ) { |
| 79 |
return $images; |
| 80 |
} |
| 81 |
|
| 82 |
$thumbnail_id = get_post_thumbnail_id( $post->ID ); |
| 83 |
|
| 84 |
if ( $thumbnail_id ) { |
| 85 |
|
| 86 |
$src = $this->get_absolute_url( $this->image_url( $thumbnail_id ) ); |
| 87 |
$alt = WPSEO_Image_Utils::get_alt_tag( $thumbnail_id ); |
| 88 |
$title = get_post_field( 'post_title', $thumbnail_id ); |
| 89 |
$images[] = $this->get_image_item( $post, $src, $title, $alt ); |
| 90 |
} |
| 91 |
|
| 92 |
/** |
| 93 |
* Filter: 'wpseo_sitemap_content_before_parse_html_images' - Filters the post content |
| 94 |
* before it is parsed for images. |
| 95 |
* |
| 96 |
* @param string $content The raw/unprocessed post content. |
| 97 |
*/ |
| 98 |
$content = apply_filters( 'wpseo_sitemap_content_before_parse_html_images', $post->post_content ); |
| 99 |
|
| 100 |
$unfiltered_images = $this->parse_html_images( $content ); |
| 101 |
|
| 102 |
foreach ( $unfiltered_images as $image ) { |
| 103 |
$images[] = $this->get_image_item( $post, $image['src'], $image['title'], $image['alt'] ); |
| 104 |
} |
| 105 |
|
| 106 |
foreach ( $this->parse_galleries( $content, $post->ID ) as $attachment ) { |
| 107 |
|
| 108 |
$src = $this->get_absolute_url( $this->image_url( $attachment->ID ) ); |
| 109 |
$alt = WPSEO_Image_Utils::get_alt_tag( $attachment->ID ); |
| 110 |
|
| 111 |
$images[] = $this->get_image_item( $post, $src, $attachment->post_title, $alt ); |
| 112 |
} |
| 113 |
|
| 114 |
if ( $post->post_type === 'attachment' && wp_attachment_is_image( $post ) ) { |
| 115 |
|
| 116 |
$src = $this->get_absolute_url( $this->image_url( $post->ID ) ); |
| 117 |
$alt = WPSEO_Image_Utils::get_alt_tag( $post->ID ); |
| 118 |
|
| 119 |
$images[] = $this->get_image_item( $post, $src, $post->post_title, $alt ); |
| 120 |
} |
| 121 |
|
| 122 |
foreach ( $images as $key => $image ) { |
| 123 |
|
| 124 |
if ( empty( $image['src'] ) ) { |
| 125 |
unset( $images[ $key ] ); |
| 126 |
} |
| 127 |
} |
| 128 |
|
| 129 |
/** |
| 130 |
* Filter images to be included for the post in XML sitemap. |
| 131 |
* |
| 132 |
* @param array $images Array of image items. |
| 133 |
* @param int $post_id ID of the post. |
| 134 |
*/ |
| 135 |
$images = apply_filters( 'wpseo_sitemap_urlimages', $images, $post->ID ); |
| 136 |
|
| 137 |
return $images; |
| 138 |
} |
| 139 |
|
| 140 |
/** |
| 141 |
* Get the images in the term description. |
| 142 |
* |
| 143 |
* @param object $term Term to get images from description for. |
| 144 |
* |
| 145 |
* @return array |
| 146 |
*/ |
| 147 |
public function get_term_images( $term ) { |
| 148 |
|
| 149 |
$images = $this->parse_html_images( $term->description ); |
| 150 |
|
| 151 |
foreach ( $this->parse_galleries( $term->description ) as $attachment ) { |
| 152 |
|
| 153 |
$images[] = [ |
| 154 |
'src' => $this->get_absolute_url( $this->image_url( $attachment->ID ) ), |
| 155 |
'title' => $attachment->post_title, |
| 156 |
'alt' => WPSEO_Image_Utils::get_alt_tag( $attachment->ID ), |
| 157 |
]; |
| 158 |
} |
| 159 |
|
| 160 |
return $images; |
| 161 |
} |
| 162 |
|
| 163 |
/** |
| 164 |
* Parse `<img />` tags in content. |
| 165 |
* |
| 166 |
* @param string $content Content string to parse. |
| 167 |
* |
| 168 |
* @return array |
| 169 |
*/ |
| 170 |
private function parse_html_images( $content ) { |
| 171 |
|
| 172 |
$images = []; |
| 173 |
|
| 174 |
if ( ! class_exists( 'DOMDocument' ) ) { |
| 175 |
return $images; |
| 176 |
} |
| 177 |
|
| 178 |
if ( empty( $content ) ) { |
| 179 |
return $images; |
| 180 |
} |
| 181 |
|
| 182 |
// Prevent DOMDocument from bubbling warnings about invalid HTML. |
| 183 |
libxml_use_internal_errors( true ); |
| 184 |
|
| 185 |
$post_dom = new DOMDocument(); |
| 186 |
$post_dom->loadHTML( '<?xml encoding="' . $this->charset . '">' . $content ); |
| 187 |
|
| 188 |
// Clear the errors, so they don't get kept in memory. |
| 189 |
libxml_clear_errors(); |
| 190 |
|
| 191 |
/** @var DOMElement $img */ |
| 192 |
foreach ( $post_dom->getElementsByTagName( 'img' ) as $img ) { |
| 193 |
|
| 194 |
$src = $img->getAttribute( 'src' ); |
| 195 |
|
| 196 |
if ( empty( $src ) ) { |
| 197 |
continue; |
| 198 |
} |
| 199 |
|
| 200 |
$class = $img->getAttribute( 'class' ); |
| 201 |
|
| 202 |
if ( // This detects WP-inserted images, which we need to upsize. R. |
| 203 |
! empty( $class ) |
| 204 |
&& ( strpos( $class, 'size-full' ) === false ) |
| 205 |
&& preg_match( '|wp-image-(?P<id>\d+)|', $class, $matches ) |
| 206 |
&& get_post_status( $matches['id'] ) |
| 207 |
) { |
| 208 |
$src = $this->image_url( $matches['id'] ); |
| 209 |
} |
| 210 |
|
| 211 |
$src = $this->get_absolute_url( $src ); |
| 212 |
|
| 213 |
if ( strpos( $src, $this->host ) === false ) { |
| 214 |
continue; |
| 215 |
} |
| 216 |
|
| 217 |
if ( $src !== esc_url( $src ) ) { |
| 218 |
continue; |
| 219 |
} |
| 220 |
|
| 221 |
$images[] = [ |
| 222 |
'src' => $src, |
| 223 |
'title' => $img->getAttribute( 'title' ), |
| 224 |
'alt' => $img->getAttribute( 'alt' ), |
| 225 |
]; |
| 226 |
} |
| 227 |
|
| 228 |
return $images; |
| 229 |
} |
| 230 |
|
| 231 |
/** |
| 232 |
* Parse gallery shortcodes in a given content. |
| 233 |
* |
| 234 |
* @param string $content Content string. |
| 235 |
* @param int $post_id Optional. ID of post being parsed. |
| 236 |
* |
| 237 |
* @return array Set of attachment objects. |
| 238 |
*/ |
| 239 |
protected function parse_galleries( $content, $post_id = 0 ) { |
| 240 |
|
| 241 |
$attachments = []; |
| 242 |
$galleries = $this->get_content_galleries( $content ); |
| 243 |
|
| 244 |
foreach ( $galleries as $gallery ) { |
| 245 |
|
| 246 |
$id = $post_id; |
| 247 |
|
| 248 |
if ( ! empty( $gallery['id'] ) ) { |
| 249 |
$id = intval( $gallery['id'] ); |
| 250 |
} |
| 251 |
|
| 252 |
// Forked from core gallery_shortcode() to have exact same logic. R. |
| 253 |
if ( ! empty( $gallery['ids'] ) ) { |
| 254 |
$gallery['include'] = $gallery['ids']; |
| 255 |
} |
| 256 |
|
| 257 |
$gallery_attachments = $this->get_gallery_attachments( $id, $gallery ); |
| 258 |
|
| 259 |
$attachments = array_merge( $attachments, $gallery_attachments ); |
| 260 |
} |
| 261 |
|
| 262 |
return array_unique( $attachments, SORT_REGULAR ); |
| 263 |
} |
| 264 |
|
| 265 |
/** |
| 266 |
* Retrieves galleries from the passed content. |
| 267 |
* |
| 268 |
* Forked from core to skip executing shortcodes for performance. |
| 269 |
* |
| 270 |
* @param string $content Content to parse for shortcodes. |
| 271 |
* |
| 272 |
* @return array A list of arrays, each containing gallery data. |
| 273 |
*/ |
| 274 |
protected function get_content_galleries( $content ) { |
| 275 |
|
| 276 |
$galleries = []; |
| 277 |
|
| 278 |
if ( ! preg_match_all( '/' . get_shortcode_regex( [ 'gallery' ] ) . '/s', $content, $matches, PREG_SET_ORDER ) ) { |
| 279 |
return $galleries; |
| 280 |
} |
| 281 |
|
| 282 |
foreach ( $matches as $shortcode ) { |
| 283 |
|
| 284 |
$attributes = shortcode_parse_atts( $shortcode[3] ); |
| 285 |
|
| 286 |
if ( $attributes === '' ) { // Valid shortcode without any attributes. R. |
| 287 |
$attributes = []; |
| 288 |
} |
| 289 |
|
| 290 |
$galleries[] = $attributes; |
| 291 |
} |
| 292 |
|
| 293 |
return $galleries; |
| 294 |
} |
| 295 |
|
| 296 |
/** |
| 297 |
* Get image item array with filters applied. |
| 298 |
* |
| 299 |
* @param WP_Post $post Post object for the context. |
| 300 |
* @param string $src Image URL. |
| 301 |
* @param string $title Optional image title. |
| 302 |
* @param string $alt Optional image alt text. |
| 303 |
* |
| 304 |
* @return array |
| 305 |
*/ |
| 306 |
protected function get_image_item( $post, $src, $title = '', $alt = '' ) { |
| 307 |
|
| 308 |
$image = []; |
| 309 |
|
| 310 |
/** |
| 311 |
* Filter image URL to be included in XML sitemap for the post. |
| 312 |
* |
| 313 |
* @param string $src Image URL. |
| 314 |
* @param object $post Post object. |
| 315 |
*/ |
| 316 |
$image['src'] = apply_filters( 'wpseo_xml_sitemap_img_src', $src, $post ); |
| 317 |
|
| 318 |
if ( ! empty( $title ) ) { |
| 319 |
$image['title'] = $title; |
| 320 |
} |
| 321 |
|
| 322 |
if ( ! empty( $alt ) ) { |
| 323 |
$image['alt'] = $alt; |
| 324 |
} |
| 325 |
|
| 326 |
/** |
| 327 |
* Filter image data to be included in XML sitemap for the post. |
| 328 |
* |
| 329 |
* @param array $image { |
| 330 |
* Array of image data. |
| 331 |
* |
| 332 |
* @type string $src Image URL. |
| 333 |
* @type string $title Image title attribute (optional). |
| 334 |
* @type string $alt Image alt attribute (optional). |
| 335 |
* } |
| 336 |
* |
| 337 |
* @param object $post Post object. |
| 338 |
*/ |
| 339 |
return apply_filters( 'wpseo_xml_sitemap_img', $image, $post ); |
| 340 |
} |
| 341 |
|
| 342 |
/** |
| 343 |
* Get attached image URL with filters applied. Adapted from core for speed. |
| 344 |
* |
| 345 |
* @param int $post_id ID of the post. |
| 346 |
* |
| 347 |
* @return string |
| 348 |
*/ |
| 349 |
private function image_url( $post_id ) { |
| 350 |
|
| 351 |
static $uploads; |
| 352 |
|
| 353 |
if ( empty( $uploads ) ) { |
| 354 |
$uploads = wp_upload_dir(); |
| 355 |
} |
| 356 |
|
| 357 |
if ( $uploads['error'] !== false ) { |
| 358 |
return ''; |
| 359 |
} |
| 360 |
|
| 361 |
$file = get_post_meta( $post_id, '_wp_attached_file', true ); |
| 362 |
|
| 363 |
if ( empty( $file ) ) { |
| 364 |
return ''; |
| 365 |
} |
| 366 |
|
| 367 |
// Check that the upload base exists in the file location. |
| 368 |
if ( strpos( $file, $uploads['basedir'] ) === 0 ) { |
| 369 |
$src = str_replace( $uploads['basedir'], $uploads['baseurl'], $file ); |
| 370 |
} |
| 371 |
elseif ( strpos( $file, 'wp-content/uploads' ) !== false ) { |
| 372 |
$src = $uploads['baseurl'] . substr( $file, ( strpos( $file, 'wp-content/uploads' ) + 18 ) ); |
| 373 |
} |
| 374 |
else { |
| 375 |
// It's a newly uploaded file, therefore $file is relative to the baseurl. |
| 376 |
$src = $uploads['baseurl'] . '/' . $file; |
| 377 |
} |
| 378 |
|
| 379 |
return apply_filters( 'wp_get_attachment_url', $src, $post_id ); |
| 380 |
} |
| 381 |
|
| 382 |
/** |
| 383 |
* Make absolute URL for domain or protocol-relative one. |
| 384 |
* |
| 385 |
* @param string $src URL to process. |
| 386 |
* |
| 387 |
* @return string |
| 388 |
*/ |
| 389 |
protected function get_absolute_url( $src ) { |
| 390 |
|
| 391 |
if ( empty( $src ) || ! is_string( $src ) ) { |
| 392 |
return $src; |
| 393 |
} |
| 394 |
|
| 395 |
if ( YoastSEO()->helpers->url->is_relative( $src ) === true ) { |
| 396 |
|
| 397 |
if ( $src[0] !== '/' ) { |
| 398 |
return $src; |
| 399 |
} |
| 400 |
|
| 401 |
// The URL is relative, we'll have to make it absolute. |
| 402 |
return $this->home_url . $src; |
| 403 |
} |
| 404 |
|
| 405 |
if ( strpos( $src, 'http' ) !== 0 ) { |
| 406 |
// Protocol relative URL, we add the scheme as the standard requires a protocol. |
| 407 |
return $this->scheme . ':' . $src; |
| 408 |
} |
| 409 |
|
| 410 |
return $src; |
| 411 |
} |
| 412 |
|
| 413 |
/** |
| 414 |
* Returns the attachments for a gallery. |
| 415 |
* |
| 416 |
* @param int $id The post ID. |
| 417 |
* @param array $gallery The gallery config. |
| 418 |
* |
| 419 |
* @return array The selected attachments. |
| 420 |
*/ |
| 421 |
protected function get_gallery_attachments( $id, $gallery ) { |
| 422 |
|
| 423 |
// When there are attachments to include. |
| 424 |
if ( ! empty( $gallery['include'] ) ) { |
| 425 |
return $this->get_gallery_attachments_for_included( $gallery['include'] ); |
| 426 |
} |
| 427 |
|
| 428 |
// When $id is empty, just return empty array. |
| 429 |
if ( empty( $id ) ) { |
| 430 |
return []; |
| 431 |
} |
| 432 |
|
| 433 |
return $this->get_gallery_attachments_for_parent( $id, $gallery ); |
| 434 |
} |
| 435 |
|
| 436 |
/** |
| 437 |
* Returns the attachments for the given ID. |
| 438 |
* |
| 439 |
* @param int $id The post ID. |
| 440 |
* @param array $gallery The gallery config. |
| 441 |
* |
| 442 |
* @return array The selected attachments. |
| 443 |
*/ |
| 444 |
protected function get_gallery_attachments_for_parent( $id, $gallery ) { |
| 445 |
$query = [ |
| 446 |
'posts_per_page' => -1, |
| 447 |
'post_parent' => $id, |
| 448 |
]; |
| 449 |
|
| 450 |
// When there are posts that should be excluded from result set. |
| 451 |
if ( ! empty( $gallery['exclude'] ) ) { |
| 452 |
$query['post__not_in'] = wp_parse_id_list( $gallery['exclude'] ); |
| 453 |
} |
| 454 |
|
| 455 |
return $this->get_attachments( $query ); |
| 456 |
} |
| 457 |
|
| 458 |
/** |
| 459 |
* Returns an array with attachments for the post IDs that will be included. |
| 460 |
* |
| 461 |
* @param array $included_ids Array with IDs to include. |
| 462 |
* |
| 463 |
* @return array The found attachments. |
| 464 |
*/ |
| 465 |
protected function get_gallery_attachments_for_included( $included_ids ) { |
| 466 |
$ids_to_include = wp_parse_id_list( $included_ids ); |
| 467 |
$attachments = $this->get_attachments( |
| 468 |
[ |
| 469 |
'posts_per_page' => count( $ids_to_include ), |
| 470 |
'post__in' => $ids_to_include, |
| 471 |
] |
| 472 |
); |
| 473 |
|
| 474 |
$gallery_attachments = []; |
| 475 |
foreach ( $attachments as $key => $val ) { |
| 476 |
$gallery_attachments[ $val->ID ] = $val; |
| 477 |
} |
| 478 |
|
| 479 |
return $gallery_attachments; |
| 480 |
} |
| 481 |
|
| 482 |
/** |
| 483 |
* Returns the attachments. |
| 484 |
* |
| 485 |
* @param array $args Array with query args. |
| 486 |
* |
| 487 |
* @return array The found attachments. |
| 488 |
*/ |
| 489 |
protected function get_attachments( $args ) { |
| 490 |
$default_args = [ |
| 491 |
'post_status' => 'inherit', |
| 492 |
'post_type' => 'attachment', |
| 493 |
'post_mime_type' => 'image', |
| 494 |
|
| 495 |
// Defaults taken from function get_posts. |
| 496 |
'orderby' => 'date', |
| 497 |
'order' => 'DESC', |
| 498 |
'meta_key' => '', |
| 499 |
'meta_value' => '', |
| 500 |
'suppress_filters' => true, |
| 501 |
'ignore_sticky_posts' => true, |
| 502 |
'no_found_rows' => true, |
| 503 |
]; |
| 504 |
|
| 505 |
$args = wp_parse_args( $args, $default_args ); |
| 506 |
|
| 507 |
$get_attachments = new WP_Query(); |
| 508 |
return $get_attachments->query( $args ); |
| 509 |
} |
| 510 |
} |
| 511 |
|
