/wp-content/plugins/wordpress-seo/inc/sitemaps/class-post-type-sitemap-provider.php
PHP | 641 lines | 301 code | 128 blank | 212 comment | 48 complexity | 6a4519005405420c64050d82c5bf6477 MD5 | raw file
1<?php 2/** 3 * WPSEO plugin file. 4 * 5 * @package WPSEO\XML_Sitemaps 6 */ 7 8/** 9 * Sitemap provider for author archives. 10 */ 11class WPSEO_Post_Type_Sitemap_Provider implements WPSEO_Sitemap_Provider { 12 13 /** @var string $home_url Holds the home_url() value. */ 14 protected static $home_url; 15 16 /** @var WPSEO_Sitemap_Image_Parser $image_parser Holds image parser instance. */ 17 protected static $image_parser; 18 19 /** @var object $classifier Holds instance of classifier for a link. */ 20 protected static $classifier; 21 22 /** @var int $page_on_front_id Static front page ID. */ 23 protected static $page_on_front_id; 24 25 /** @var int $page_for_posts_id Posts page ID. */ 26 protected static $page_for_posts_id; 27 28 /** 29 * Set up object properties for data reuse. 30 */ 31 public function __construct() { 32 add_filter( 'save_post', array( $this, 'save_post' ) ); 33 } 34 35 /** 36 * Get all the options 37 * 38 * @deprecated 7.0 39 */ 40 protected function get_options() { 41 _deprecated_function( __METHOD__, 'WPSEO 7.0', 'WPSEO_Options::get' ); 42 } 43 44 /** 45 * Get front page ID 46 * 47 * @return int 48 */ 49 protected function get_page_on_front_id() { 50 if ( ! isset( self::$page_on_front_id ) ) { 51 self::$page_on_front_id = (int) get_option( 'page_on_front' ); 52 } 53 54 return self::$page_on_front_id; 55 } 56 57 /** 58 * Get page for posts ID 59 * 60 * @return int 61 */ 62 protected function get_page_for_posts_id() { 63 if ( ! isset( self::$page_for_posts_id ) ) { 64 self::$page_for_posts_id = (int) get_option( 'page_for_posts' ); 65 } 66 67 return self::$page_for_posts_id; 68 } 69 70 /** 71 * Get the Image Parser 72 * 73 * @return WPSEO_Sitemap_Image_Parser 74 */ 75 protected function get_image_parser() { 76 if ( ! isset( self::$image_parser ) ) { 77 self::$image_parser = new WPSEO_Sitemap_Image_Parser(); 78 } 79 80 return self::$image_parser; 81 } 82 83 /** 84 * Get the Classifier for a link 85 * 86 * @return WPSEO_Link_Type_Classifier 87 */ 88 protected function get_classifier() { 89 if ( ! isset( self::$classifier ) ) { 90 self::$classifier = new WPSEO_Link_Type_Classifier( $this->get_home_url() ); 91 } 92 93 return self::$classifier; 94 } 95 96 /** 97 * Get Home URL 98 * 99 * This has been moved from the constructor because wp_rewrite is not available on plugins_loaded in multisite. 100 * It will now be requested on need and not on initialization. 101 * 102 * @return string 103 */ 104 protected function get_home_url() { 105 if ( ! isset( self::$home_url ) ) { 106 self::$home_url = WPSEO_Utils::home_url(); 107 } 108 109 return self::$home_url; 110 } 111 112 /** 113 * Check if provider supports given item type. 114 * 115 * @param string $type Type string to check for. 116 * 117 * @return boolean 118 */ 119 public function handles_type( $type ) { 120 121 return post_type_exists( $type ); 122 } 123 124 /** 125 * @param int $max_entries Entries per sitemap. 126 * 127 * @return array 128 */ 129 public function get_index_links( $max_entries ) { 130 131 global $wpdb; 132 133 // Consider using WPSEO_Post_Type::get_accessible_post_types() to filter out any `no-index` post-types. 134 $post_types = WPSEO_Post_Type::get_accessible_post_types(); 135 $post_types = array_filter( $post_types, array( $this, 'is_valid_post_type' ) ); 136 $last_modified_times = WPSEO_Sitemaps::get_last_modified_gmt( $post_types, true ); 137 $index = array(); 138 139 foreach ( $post_types as $post_type ) { 140 141 $total_count = $this->get_post_type_count( $post_type ); 142 143 if ( $total_count === 0 ) { 144 continue; 145 } 146 147 $max_pages = 1; 148 149 if ( $total_count > $max_entries ) { 150 $max_pages = (int) ceil( $total_count / $max_entries ); 151 } 152 153 $all_dates = array(); 154 155 if ( $max_pages > 1 ) { 156 157 $sql = " 158 SELECT post_modified_gmt 159 FROM ( SELECT @rownum:=0 ) init 160 JOIN {$wpdb->posts} USE INDEX( type_status_date ) 161 WHERE post_status IN ( 'publish', 'inherit' ) 162 AND post_type = %s 163 AND ( @rownum:=@rownum+1 ) %% %d = 0 164 ORDER BY post_modified_gmt ASC 165 "; 166 167 $all_dates = $wpdb->get_col( $wpdb->prepare( $sql, $post_type, $max_entries ) ); 168 } 169 170 for ( $page_counter = 0; $page_counter < $max_pages; $page_counter++ ) { 171 172 $current_page = ( $max_pages > 1 ) ? ( $page_counter + 1 ) : ''; 173 $date = false; 174 175 if ( empty( $current_page ) || $current_page === $max_pages ) { 176 177 if ( ! empty( $last_modified_times[ $post_type ] ) ) { 178 $date = $last_modified_times[ $post_type ]; 179 } 180 } 181 else { 182 $date = $all_dates[ $page_counter ]; 183 } 184 185 $index[] = array( 186 'loc' => WPSEO_Sitemaps_Router::get_base_url( $post_type . '-sitemap' . $current_page . '.xml' ), 187 'lastmod' => $date, 188 ); 189 } 190 } 191 192 return $index; 193 } 194 195 /** 196 * Get set of sitemap link data. 197 * 198 * @param string $type Sitemap type. 199 * @param int $max_entries Entries per sitemap. 200 * @param int $current_page Current page of the sitemap. 201 * 202 * @return array 203 */ 204 public function get_sitemap_links( $type, $max_entries, $current_page ) { 205 206 $links = array(); 207 $post_type = $type; 208 209 if ( ! $this->is_valid_post_type( $post_type ) ) { 210 return $links; 211 } 212 213 $steps = min( 100, $max_entries ); 214 $offset = ( $current_page > 1 ) ? ( ( $current_page - 1 ) * $max_entries ) : 0; 215 $total = ( $offset + $max_entries ); 216 217 $typecount = $this->get_post_type_count( $post_type ); 218 219 if ( $total > $typecount ) { 220 $total = $typecount; 221 } 222 223 if ( $current_page === 1 ) { 224 $links = array_merge( $links, $this->get_first_links( $post_type ) ); 225 } 226 227 if ( $typecount === 0 ) { 228 229 return $links; 230 } 231 232 $stacked_urls = array(); 233 $posts_to_exclude = $this->get_excluded_posts(); 234 235 while ( $total > $offset ) { 236 237 $posts = $this->get_posts( $post_type, $steps, $offset ); 238 239 $offset += $steps; 240 241 if ( empty( $posts ) ) { 242 continue; 243 } 244 245 foreach ( $posts as $post ) { 246 247 if ( in_array( $post->ID, $posts_to_exclude, true ) ) { 248 continue; 249 } 250 251 if ( WPSEO_Meta::get_value( 'meta-robots-noindex', $post->ID ) === '1' ) { 252 continue; 253 } 254 255 $url = $this->get_url( $post ); 256 257 if ( ! isset( $url['loc'] ) ) { 258 continue; 259 } 260 261 /** 262 * Filter URL entry before it gets added to the sitemap. 263 * 264 * @param array $url Array of URL parts. 265 * @param string $type URL type. 266 * @param object $post Data object for the URL. 267 */ 268 $url = apply_filters( 'wpseo_sitemap_entry', $url, 'post', $post ); 269 270 if ( empty( $url ) ) { 271 continue; 272 } 273 274 if ( $post->ID === $this->get_page_for_posts_id() || $post->ID === $this->get_page_on_front_id() ) { 275 276 array_unshift( $links, $url ); 277 continue; 278 } 279 $links[] = $url; 280 } 281 282 unset( $post, $url ); 283 } 284 285 return $links; 286 } 287 288 /** 289 * Check for relevant post type before invalidation. 290 * 291 * @param int $post_id Post ID to possibly invalidate for. 292 */ 293 public function save_post( $post_id ) { 294 295 if ( $this->is_valid_post_type( get_post_type( $post_id ) ) ) { 296 WPSEO_Sitemaps_Cache::invalidate_post( $post_id ); 297 } 298 } 299 300 /** 301 * Check if post type should be present in sitemaps. 302 * 303 * @param string $post_type Post type string to check for. 304 * 305 * @return bool 306 */ 307 public function is_valid_post_type( $post_type ) { 308 if ( ! WPSEO_Post_Type::is_post_type_indexable( $post_type ) ) { 309 return false; 310 } 311 312 /** 313 * Filter decision if post type is excluded from the XML sitemap. 314 * 315 * @param bool $exclude Default false. 316 * @param string $post_type Post type name. 317 */ 318 if ( apply_filters( 'wpseo_sitemap_exclude_post_type', false, $post_type ) ) { 319 return false; 320 } 321 322 return true; 323 } 324 325 /** 326 * Retrieves a list with the excluded post ids. 327 * 328 * @return array Array with post ids to exclude. 329 */ 330 protected function get_excluded_posts() { 331 /** 332 * Filter: 'wpseo_exclude_from_sitemap_by_post_ids' - Allow extending and modifying the posts to exclude. 333 * 334 * @api array $posts_to_exclude The posts to exclude. 335 */ 336 $excluded_posts_ids = apply_filters( 'wpseo_exclude_from_sitemap_by_post_ids', array() ); 337 if ( ! is_array( $excluded_posts_ids ) || $excluded_posts_ids === array() ) { 338 return array(); 339 } 340 341 return array_map( 'intval', $excluded_posts_ids ); 342 } 343 344 /** 345 * Get count of posts for post type. 346 * 347 * @param string $post_type Post type to retrieve count for. 348 * 349 * @return int 350 */ 351 protected function get_post_type_count( $post_type ) { 352 353 global $wpdb; 354 355 /** 356 * Filter JOIN query part for type count of post type. 357 * 358 * @param string $join SQL part, defaults to empty string. 359 * @param string $post_type Post type name. 360 */ 361 $join_filter = apply_filters( 'wpseo_typecount_join', '', $post_type ); 362 363 /** 364 * Filter WHERE query part for type count of post type. 365 * 366 * @param string $where SQL part, defaults to empty string. 367 * @param string $post_type Post type name. 368 */ 369 $where_filter = apply_filters( 'wpseo_typecount_where', '', $post_type ); 370 371 $where = $this->get_sql_where_clause( $post_type ); 372 373 $sql = " 374 SELECT COUNT({$wpdb->posts}.ID) 375 FROM {$wpdb->posts} 376 {$join_filter} 377 {$where} 378 {$where_filter} 379 "; 380 381 return (int) $wpdb->get_var( $sql ); 382 } 383 384 /** 385 * Produces set of links to prepend at start of first sitemap page. 386 * 387 * @param string $post_type Post type to produce links for. 388 * 389 * @return array 390 */ 391 protected function get_first_links( $post_type ) { 392 393 $links = array(); 394 395 $needs_archive = true; 396 397 if ( ! $this->get_page_on_front_id() && ( $post_type === 'post' || $post_type === 'page' ) ) { 398 399 $links[] = array( 400 'loc' => $this->get_home_url(), 401 402 // Deprecated, kept for backwards data compat. R. 403 'chf' => 'daily', 404 'pri' => 1, 405 ); 406 407 $needs_archive = false; 408 } 409 elseif ( $this->get_page_on_front_id() && $post_type === 'post' && $this->get_page_for_posts_id() ) { 410 411 $page_for_posts_url = get_permalink( $this->get_page_for_posts_id() ); 412 413 $links[] = array( 414 'loc' => $page_for_posts_url, 415 416 // Deprecated, kept for backwards data compat. R. 417 'chf' => 'daily', 418 'pri' => 1, 419 ); 420 421 $needs_archive = false; 422 } 423 424 if ( ! $needs_archive ) { 425 return $links; 426 } 427 428 $archive_url = $this->get_post_type_archive_link( $post_type ); 429 430 /** 431 * Filter the URL Yoast SEO uses in the XML sitemap for this post type archive. 432 * 433 * @param string $archive_url The URL of this archive 434 * @param string $post_type The post type this archive is for. 435 */ 436 $archive_url = apply_filters( 'wpseo_sitemap_post_type_archive_link', $archive_url, $post_type ); 437 438 if ( $archive_url ) { 439 /** 440 * Filter the priority of the URL Yoast SEO uses in the XML sitemap. 441 * 442 * @param float $priority The priority for this URL, ranging from 0 to 1 443 * @param string $post_type The post type this archive is for. 444 */ 445 $links[] = array( 446 'loc' => $archive_url, 447 'mod' => WPSEO_Sitemaps::get_last_modified_gmt( $post_type ), 448 449 // Deprecated, kept for backwards data compat. R. 450 'chf' => 'daily', 451 'pri' => 1, 452 ); 453 } 454 455 return $links; 456 } 457 458 /** 459 * Get URL for a post type archive. 460 * 461 * @since 5.3 462 * 463 * @param string $post_type Post type. 464 * 465 * @return string|bool URL or false if it should be excluded. 466 */ 467 protected function get_post_type_archive_link( $post_type ) { 468 469 if ( WPSEO_Options::get( 'noindex-ptarchive-' . $post_type, false ) ) { 470 return false; 471 } 472 473 // Post archive should be excluded if it isn't front page or posts page. 474 if ( $post_type === 'post' && get_option( 'show_on_front' ) !== 'posts' && ! $this->get_page_for_posts_id() ) { 475 return false; 476 } 477 478 $archive_url = get_post_type_archive_link( $post_type ); 479 480 return $archive_url; 481 } 482 483 /** 484 * Retrieve set of posts with optimized query routine. 485 * 486 * @param string $post_type Post type to retrieve. 487 * @param int $count Count of posts to retrieve. 488 * @param int $offset Starting offset. 489 * 490 * @return object[] 491 */ 492 protected function get_posts( $post_type, $count, $offset ) { 493 494 global $wpdb; 495 496 static $filters = array(); 497 498 if ( ! isset( $filters[ $post_type ] ) ) { 499 // Make sure you're wpdb->preparing everything you throw into this!! 500 $filters[ $post_type ] = array( 501 /** 502 * Filter JOIN query part for the post type. 503 * 504 * @param string $join SQL part, defaults to false. 505 * @param string $post_type Post type name. 506 */ 507 'join' => apply_filters( 'wpseo_posts_join', false, $post_type ), 508 509 /** 510 * Filter Where query part for the post type. 511 * 512 * @param string $where SQL part, defaults to false. 513 * @param string $post_type Post type name. 514 */ 515 'where' => apply_filters( 'wpseo_posts_where', false, $post_type ), 516 ); 517 } 518 519 $join_filter = $filters[ $post_type ]['join']; 520 $where_filter = $filters[ $post_type ]['where']; 521 $where = $this->get_sql_where_clause( $post_type ); 522 523 // Optimized query per this thread: http://wordpress.org/support/topic/plugin-wordpress-seo-by-yoast-performance-suggestion. 524 // Also see http://explainextended.com/2009/10/23/mysql-order-by-limit-performance-late-row-lookups/. 525 $sql = " 526 SELECT l.ID, post_title, post_content, post_name, post_parent, post_author, post_modified_gmt, post_date, post_date_gmt 527 FROM ( 528 SELECT {$wpdb->posts}.ID 529 FROM {$wpdb->posts} 530 {$join_filter} 531 {$where} 532 {$where_filter} 533 ORDER BY {$wpdb->posts}.post_modified ASC LIMIT %d OFFSET %d 534 ) 535 o JOIN {$wpdb->posts} l ON l.ID = o.ID 536 "; 537 538 $posts = $wpdb->get_results( $wpdb->prepare( $sql, $count, $offset ) ); 539 540 $post_ids = array(); 541 542 foreach ( $posts as $post ) { 543 $post->post_type = $post_type; 544 $post->post_status = 'publish'; 545 $post->filter = 'sample'; 546 $post->ID = (int) $post->ID; 547 $post->post_parent = (int) $post->post_parent; 548 $post->post_author = (int) $post->post_author; 549 $post_ids[] = $post->ID; 550 } 551 552 update_meta_cache( 'post', $post_ids ); 553 554 return $posts; 555 } 556 557 /** 558 * @param string $post_type Post type slug. 559 * 560 * @return string 561 */ 562 protected function get_sql_where_clause( $post_type ) { 563 564 global $wpdb; 565 566 $join = ''; 567 $status = "{$wpdb->posts}.post_status = 'publish'"; 568 569 // Based on WP_Query->get_posts(). R. 570 if ( 'attachment' === $post_type ) { 571 $join = " LEFT JOIN {$wpdb->posts} AS p2 ON ({$wpdb->posts}.post_parent = p2.ID) "; 572 $status = "p2.post_status = 'publish' AND p2.post_password = ''"; 573 } 574 575 $where_clause = " 576 {$join} 577 WHERE {$status} 578 AND {$wpdb->posts}.post_type = %s 579 AND {$wpdb->posts}.post_password = '' 580 AND {$wpdb->posts}.post_date != '0000-00-00 00:00:00' 581 "; 582 583 return $wpdb->prepare( $where_clause, $post_type ); 584 } 585 586 /** 587 * Produce array of URL parts for given post object. 588 * 589 * @param object $post Post object to get URL parts for. 590 * 591 * @return array|bool 592 */ 593 protected function get_url( $post ) { 594 595 $url = array(); 596 597 /** 598 * Filter the URL Yoast SEO uses in the XML sitemap. 599 * 600 * Note that only absolute local URLs are allowed as the check after this removes external URLs. 601 * 602 * @param string $url URL to use in the XML sitemap 603 * @param object $post Post object for the URL. 604 */ 605 $url['loc'] = apply_filters( 'wpseo_xml_sitemap_post_url', get_permalink( $post ), $post ); 606 607 /** 608 * Do not include external URLs. 609 * 610 * @see https://wordpress.org/plugins/page-links-to/ can rewrite permalinks to external URLs. 611 */ 612 if ( $this->get_classifier()->classify( $url['loc'] ) === WPSEO_Link::TYPE_EXTERNAL ) { 613 return false; 614 } 615 616 $modified = max( $post->post_modified_gmt, $post->post_date_gmt ); 617 618 if ( $modified !== '0000-00-00 00:00:00' ) { 619 $url['mod'] = $modified; 620 } 621 622 $url['chf'] = 'daily'; // Deprecated, kept for backwards data compat. R. 623 624 $canonical = WPSEO_Meta::get_value( 'canonical', $post->ID ); 625 626 if ( $canonical !== '' && $canonical !== $url['loc'] ) { 627 /* 628 * Let's assume that if a canonical is set for this page and it's different from 629 * the URL of this post, that page is either already in the XML sitemap OR is on 630 * an external site, either way, we shouldn't include it here. 631 */ 632 return false; 633 } 634 unset( $canonical ); 635 636 $url['pri'] = 1; // Deprecated, kept for backwards data compat. R. 637 $url['images'] = $this->get_image_parser()->get_images( $post ); 638 639 return $url; 640 } 641}