1<?php
  2/**
  3 * WordPress Rewrite API
  4 *
  5 * @package WordPress
  6 * @subpackage Rewrite
  7 */
  8
  9/**
 10 * Endpoint mask that matches nothing.
 11 *
 12 * @since 2.1.0
 13 */
 14define( 'EP_NONE', 0 );
 15
 16/**
 17 * Endpoint mask that matches post permalinks.
 18 *
 19 * @since 2.1.0
 20 */
 21define( 'EP_PERMALINK', 1 );
 22
 23/**
 24 * Endpoint mask that matches attachment permalinks.
 25 *
 26 * @since 2.1.0
 27 */
 28define( 'EP_ATTACHMENT', 2 );
 29
 30/**
 31 * Endpoint mask that matches any date archives.
 32 *
 33 * @since 2.1.0
 34 */
 35define( 'EP_DATE', 4 );
 36
 37/**
 38 * Endpoint mask that matches yearly archives.
 39 *
 40 * @since 2.1.0
 41 */
 42define( 'EP_YEAR', 8 );
 43
 44/**
 45 * Endpoint mask that matches monthly archives.
 46 *
 47 * @since 2.1.0
 48 */
 49define( 'EP_MONTH', 16 );
 50
 51/**
 52 * Endpoint mask that matches daily archives.
 53 *
 54 * @since 2.1.0
 55 */
 56define( 'EP_DAY', 32 );
 57
 58/**
 59 * Endpoint mask that matches the site root.
 60 *
 61 * @since 2.1.0
 62 */
 63define( 'EP_ROOT', 64 );
 64
 65/**
 66 * Endpoint mask that matches comment feeds.
 67 *
 68 * @since 2.1.0
 69 */
 70define( 'EP_COMMENTS', 128 );
 71
 72/**
 73 * Endpoint mask that matches searches.
 74 *
 75 * Note that this only matches a search at a "pretty" URL such as
 76 * `/search/my-search-term`, not `?s=my-search-term`.
 77 *
 78 * @since 2.1.0
 79 */
 80define( 'EP_SEARCH', 256 );
 81
 82/**
 83 * Endpoint mask that matches category archives.
 84 *
 85 * @since 2.1.0
 86 */
 87define( 'EP_CATEGORIES', 512 );
 88
 89/**
 90 * Endpoint mask that matches tag archives.
 91 *
 92 * @since 2.3.0
 93 */
 94define( 'EP_TAGS', 1024 );
 95
 96/**
 97 * Endpoint mask that matches author archives.
 98 *
 99 * @since 2.1.0
100 */
101define( 'EP_AUTHORS', 2048 );
102
103/**
104 * Endpoint mask that matches pages.
105 *
106 * @since 2.1.0
107 */
108define( 'EP_PAGES', 4096 );
109
110/**
111 * Endpoint mask that matches all archive views.
112 *
113 * @since 3.7.0
114 */
115define( 'EP_ALL_ARCHIVES', EP_DATE | EP_YEAR | EP_MONTH | EP_DAY | EP_CATEGORIES | EP_TAGS | EP_AUTHORS );
116
117/**
118 * Endpoint mask that matches everything.
119 *
120 * @since 2.1.0
121 */
122define( 'EP_ALL', EP_PERMALINK | EP_ATTACHMENT | EP_ROOT | EP_COMMENTS | EP_SEARCH | EP_PAGES | EP_ALL_ARCHIVES );
123
124/**
125 * Adds a rewrite rule that transforms a URL structure to a set of query vars.
126 *
127 * Any value in the $after parameter that isn't 'bottom' will result in the rule
128 * being placed at the top of the rewrite rules.
129 *
130 * @since 2.1.0
131 * @since 4.4.0 Array support was added to the `$query` parameter.
132 *
133 * @global WP_Rewrite $wp_rewrite WordPress rewrite component.
134 *
135 * @param string       $regex Regular expression to match request against.
136 * @param string|array $query The corresponding query vars for this rewrite rule.
137 * @param string       $after Optional. Priority of the new rule. Accepts 'top'
138 *                            or 'bottom'. Default 'bottom'.
139 */
140function add_rewrite_rule( $regex, $query, $after = 'bottom' ) {
141	global $wp_rewrite;
142
143	$wp_rewrite->add_rule( $regex, $query, $after );
144}
145
146/**
147 * Adds a new rewrite tag (like %postname%).
148 *
149 * The `$query` parameter is optional. If it is omitted you must ensure that you call
150 * this on, or before, the {@see 'init'} hook. This is because `$query` defaults to
151 * `$tag=`, and for this to work a new query var has to be added.
152 *
153 * @since 2.1.0
154 *
155 * @global WP_Rewrite $wp_rewrite WordPress rewrite component.
156 * @global WP         $wp         Current WordPress environment instance.
157 *
158 * @param string $tag   Name of the new rewrite tag.
159 * @param string $regex Regular expression to substitute the tag for in rewrite rules.
160 * @param string $query Optional. String to append to the rewritten query. Must end in '='. Default empty.
161 */
162function add_rewrite_tag( $tag, $regex, $query = '' ) {
163	// Validate the tag's name.
164	if ( strlen( $tag ) < 3 || '%' !== $tag[0] || '%' !== $tag[ strlen( $tag ) - 1 ] ) {
165		return;
166	}
167
168	global $wp_rewrite, $wp;
169
170	if ( empty( $query ) ) {
171		$qv = trim( $tag, '%' );
172		$wp->add_query_var( $qv );
173		$query = $qv . '=';
174	}
175
176	$wp_rewrite->add_rewrite_tag( $tag, $regex, $query );
177}
178
179/**
180 * Removes an existing rewrite tag (like %postname%).
181 *
182 * @since 4.5.0
183 *
184 * @global WP_Rewrite $wp_rewrite WordPress rewrite component.
185 *
186 * @param string $tag Name of the rewrite tag.
187 */
188function remove_rewrite_tag( $tag ) {
189	global $wp_rewrite;
190	$wp_rewrite->remove_rewrite_tag( $tag );
191}
192
193/**
194 * Adds a permalink structure.
195 *
196 * @since 3.0.0
197 *
198 * @see WP_Rewrite::add_permastruct()
199 * @global WP_Rewrite $wp_rewrite WordPress rewrite component.
200 *
201 * @param string $name   Name for permalink structure.
202 * @param string $struct Permalink structure.
203 * @param array  $args   Optional. Arguments for building the rules from the permalink structure,
204 *                       see WP_Rewrite::add_permastruct() for full details. Default empty array.
205 */
206function add_permastruct( $name, $struct, $args = array() ) {
207	global $wp_rewrite;
208
209	// Back-compat for the old parameters: $with_front and $ep_mask.
210	if ( ! is_array( $args ) ) {
211		$args = array( 'with_front' => $args );
212	}
213
214	if ( func_num_args() === 4 ) {
215		$args['ep_mask'] = func_get_arg( 3 );
216	}
217
218	$wp_rewrite->add_permastruct( $name, $struct, $args );
219}
220
221/**
222 * Removes a permalink structure.
223 *
224 * Can only be used to remove permastructs that were added using add_permastruct().
225 * Built-in permastructs cannot be removed.
226 *
227 * @since 4.5.0
228 *
229 * @see WP_Rewrite::remove_permastruct()
230 * @global WP_Rewrite $wp_rewrite WordPress rewrite component.
231 *
232 * @param string $name Name for permalink structure.
233 */
234function remove_permastruct( $name ) {
235	global $wp_rewrite;
236
237	$wp_rewrite->remove_permastruct( $name );
238}
239
240/**
241 * Adds a new feed type like /atom1/.
242 *
243 * @since 2.1.0
244 *
245 * @global WP_Rewrite $wp_rewrite WordPress rewrite component.
246 *
247 * @param string   $feedname Feed name. Should not start with '_'.
248 * @param callable $callback Callback to run on feed display.
249 * @return string Feed action name.
250 */
251function add_feed( $feedname, $callback ) {
252	global $wp_rewrite;
253
254	if ( ! in_array( $feedname, $wp_rewrite->feeds, true ) ) {
255		$wp_rewrite->feeds[] = $feedname;
256	}
257
258	$hook = 'do_feed_' . $feedname;
259
260	// Remove default function hook.
261	remove_action( $hook, $hook );
262
263	add_action( $hook, $callback, 10, 2 );
264
265	return $hook;
266}
267
268/**
269 * Removes rewrite rules and then recreate rewrite rules.
270 *
271 * @since 3.0.0
272 *
273 * @global WP_Rewrite $wp_rewrite WordPress rewrite component.
274 *
275 * @param bool $hard Whether to update .htaccess (hard flush) or just update
276 *                   rewrite_rules option (soft flush). Default is true (hard).
277 */
278function flush_rewrite_rules( $hard = true ) {
279	global $wp_rewrite;
280
281	if ( is_callable( array( $wp_rewrite, 'flush_rules' ) ) ) {
282		$wp_rewrite->flush_rules( $hard );
283	}
284}
285
286/**
287 * Adds an endpoint, like /trackback/.
288 *
289 * Adding an endpoint creates extra rewrite rules for each of the matching
290 * places specified by the provided bitmask. For example:
291 *
292 *     add_rewrite_endpoint( 'json', EP_PERMALINK | EP_PAGES );
293 *
294 * will add a new rewrite rule ending with "json(/(.*))?/?$" for every permastruct
295 * that describes a permalink (post) or page. This is rewritten to "json=$match"
296 * where $match is the part of the URL matched by the endpoint regex (e.g. "foo" in
297 * "[permalink]/json/foo/").
298 *
299 * A new query var with the same name as the endpoint will also be created.
300 *
301 * When specifying $places ensure that you are using the EP_* constants (or a
302 * combination of them using the bitwise OR operator) as their values are not
303 * guaranteed to remain static (especially `EP_ALL`).
304 *
305 * Be sure to flush the rewrite rules - see flush_rewrite_rules() - when your plugin gets
306 * activated and deactivated.
307 *
308 * @since 2.1.0
309 * @since 4.3.0 Added support for skipping query var registration by passing `false` to `$query_var`.
310 *
311 * @global WP_Rewrite $wp_rewrite WordPress rewrite component.
312 *
313 * @param string      $name      Name of the endpoint.
314 * @param int         $places    Endpoint mask describing the places the endpoint should be added.
315 *                               Accepts a mask of:
316 *                               - `EP_ALL`
317 *                               - `EP_NONE`
318 *                               - `EP_ALL_ARCHIVES`
319 *                               - `EP_ATTACHMENT`
320 *                               - `EP_AUTHORS`
321 *                               - `EP_CATEGORIES`
322 *                               - `EP_COMMENTS`
323 *                               - `EP_DATE`
324 *                               - `EP_DAY`
325 *                               - `EP_MONTH`
326 *                               - `EP_PAGES`
327 *                               - `EP_PERMALINK`
328 *                               - `EP_ROOT`
329 *                               - `EP_SEARCH`
330 *                               - `EP_TAGS`
331 *                               - `EP_YEAR`
332 * @param string|bool $query_var Name of the corresponding query variable. Pass `false` to skip registering a query_var
333 *                               for this endpoint. Defaults to the value of `$name`.
334 */
335function add_rewrite_endpoint( $name, $places, $query_var = true ) {
336	global $wp_rewrite;
337	$wp_rewrite->add_endpoint( $name, $places, $query_var );
338}
339
340/**
341 * Filters the URL base for taxonomies.
342 *
343 * To remove any manually prepended /index.php/.
344 *
345 * @access private
346 * @since 2.6.0
347 *
348 * @param string $base The taxonomy base that we're going to filter
349 * @return string
350 */
351function _wp_filter_taxonomy_base( $base ) {
352	if ( ! empty( $base ) ) {
353		$base = preg_replace( '|^/index\.php/|', '', $base );
354		$base = trim( $base, '/' );
355	}
356	return $base;
357}
358
359
360/**
361 * Resolves numeric slugs that collide with date permalinks.
362 *
363 * Permalinks of posts with numeric slugs can sometimes look to WP_Query::parse_query()
364 * like a date archive, as when your permalink structure is `/%year%/%postname%/` and
365 * a post with post_name '05' has the URL `/2015/05/`.
366 *
367 * This function detects conflicts of this type and resolves them in favor of the
368 * post permalink.
369 *
370 * Note that, since 4.3.0, wp_unique_post_slug() prevents the creation of post slugs
371 * that would result in a date archive conflict. The resolution performed in this
372 * function is primarily for legacy content, as well as cases when the admin has changed
373 * the site's permalink structure in a way that introduces URL conflicts.
374 *
375 * @since 4.3.0
376 *
377 * @param array $query_vars Optional. Query variables for setting up the loop, as determined in
378 *                          WP::parse_request(). Default empty array.
379 * @return array Returns the original array of query vars, with date/post conflicts resolved.
380 */
381function wp_resolve_numeric_slug_conflicts( $query_vars = array() ) {
382	if ( ! isset( $query_vars['year'] ) && ! isset( $query_vars['monthnum'] ) && ! isset( $query_vars['day'] ) ) {
383		return $query_vars;
384	}
385
386	// Identify the 'postname' position in the permastruct array.
387	$permastructs   = array_values( array_filter( explode( '/', get_option( 'permalink_structure' ) ) ) );
388	$postname_index = array_search( '%postname%', $permastructs, true );
389
390	if ( false === $postname_index ) {
391		return $query_vars;
392	}
393
394	/*
395	 * A numeric slug could be confused with a year, month, or day, depending on position. To account for
396	 * the possibility of post pagination (eg 2015/2 for the second page of a post called '2015'), our
397	 * `is_*` checks are generous: check for year-slug clashes when `is_year` *or* `is_month`, and check
398	 * for month-slug clashes when `is_month` *or* `is_day`.
399	 */
400	$compare = '';
401	if ( 0 === $postname_index && ( isset( $query_vars['year'] ) || isset( $query_vars['monthnum'] ) ) ) {
402		$compare = 'year';
403	} elseif ( $postname_index && '%year%' === $permastructs[ $postname_index - 1 ] && ( isset( $query_vars['monthnum'] ) || isset( $query_vars['day'] ) ) ) {
404		$compare = 'monthnum';
405	} elseif ( $postname_index && '%monthnum%' === $permastructs[ $postname_index - 1 ] && isset( $query_vars['day'] ) ) {
406		$compare = 'day';
407	}
408
409	if ( ! $compare ) {
410		return $query_vars;
411	}
412
413	// This is the potentially clashing slug.
414	$value = '';
415	if ( array_key_exists( $compare, $query_vars ) ) {
416		$value = $query_vars[ $compare ];
417	}
418
419	$post = get_page_by_path( $value, OBJECT, 'post' );
420	if ( ! ( $post instanceof WP_Post ) ) {
421		return $query_vars;
422	}
423
424	// If the date of the post doesn't match the date specified in the URL, resolve to the date archive.
425	if ( preg_match( '/^([0-9]{4})\-([0-9]{2})/', $post->post_date, $matches ) && isset( $query_vars['year'] ) && ( 'monthnum' === $compare || 'day' === $compare ) ) {
426		// $matches[1] is the year the post was published.
427		if ( (int) $query_vars['year'] !== (int) $matches[1] ) {
428			return $query_vars;
429		}
430
431		// $matches[2] is the month the post was published.
432		if ( 'day' === $compare && isset( $query_vars['monthnum'] ) && (int) $query_vars['monthnum'] !== (int) $matches[2] ) {
433			return $query_vars;
434		}
435	}
436
437	/*
438	 * If the located post contains nextpage pagination, then the URL chunk following postname may be
439	 * intended as the page number. Verify that it's a valid page before resolving to it.
440	 */
441	$maybe_page = '';
442	if ( 'year' === $compare && isset( $query_vars['monthnum'] ) ) {
443		$maybe_page = $query_vars['monthnum'];
444	} elseif ( 'monthnum' === $compare && isset( $query_vars['day'] ) ) {
445		$maybe_page = $query_vars['day'];
446	}
447	// Bug found in #11694 - 'page' was returning '/4'.
448	$maybe_page = (int) trim( $maybe_page, '/' );
449
450	$post_page_count = substr_count( $post->post_content, '<!--nextpage-->' ) + 1;
451
452	// If the post doesn't have multiple pages, but a 'page' candidate is found, resolve to the date archive.
453	if ( 1 === $post_page_count && $maybe_page ) {
454		return $query_vars;
455	}
456
457	// If the post has multiple pages and the 'page' number isn't valid, resolve to the date archive.
458	if ( $post_page_count > 1 && $maybe_page > $post_page_count ) {
459		return $query_vars;
460	}
461
462	// If we've gotten to this point, we have a slug/date clash. First, adjust for nextpage.
463	$query_vars['page'] = $maybe_page;
464
465	// Next, unset autodetected date-related query vars.
466	unset( $query_vars['year'] );
467	unset( $query_vars['monthnum'] );
468	unset( $query_vars['day'] );
469
470	// Then, set the identified post.
471	$query_vars['name'] = $post->post_name;
472
473	// Finally, return the modified query vars.
474	return $query_vars;
475}
476
477/**
478 * Examines a URL and try to determine the post ID it represents.
479 *
480 * Checks are supposedly from the hosted site blog.
481 *
482 * @since 1.0.0
483 *
484 * @global WP_Rewrite $wp_rewrite WordPress rewrite component.
485 * @global WP         $wp         Current WordPress environment instance.
486 *
487 * @param string $url Permalink to check.
488 * @return int Post ID, or 0 on failure.
489 */
490function url_to_postid( $url ) {
491	global $wp_rewrite;
492
493	/**
494	 * Filters the URL to derive the post ID from.
495	 *
496	 * @since 2.2.0
497	 *
498	 * @param string $url The URL to derive the post ID from.
499	 */
500	$url = apply_filters( 'url_to_postid', $url );
501
502	$url_host = parse_url( $url, PHP_URL_HOST );
503
504	if ( is_string( $url_host ) ) {
505		$url_host = str_replace( 'www.', '', $url_host );
506	} else {
507		$url_host = '';
508	}
509
510	$home_url_host = parse_url( home_url(), PHP_URL_HOST );
511
512	if ( is_string( $home_url_host ) ) {
513		$home_url_host = str_replace( 'www.', '', $home_url_host );
514	} else {
515		$home_url_host = '';
516	}
517
518	// Bail early if the URL does not belong to this site.
519	if ( $url_host && $url_host !== $home_url_host ) {
520		return 0;
521	}
522
523	// First, check to see if there is a 'p=N' or 'page_id=N' to match against.
524	if ( preg_match( '#[?&](p|page_id|attachment_id)=(\d+)#', $url, $values ) ) {
525		$id = absint( $values[2] );
526		if ( $id ) {
527			return $id;
528		}
529	}
530
531	// Get rid of the #anchor.
532	$url_split = explode( '#', $url );
533	$url       = $url_split[0];
534
535	// Get rid of URL ?query=string.
536	$url_split = explode( '?', $url );
537	$url       = $url_split[0];
538
539	// Set the correct URL scheme.
540	$scheme = parse_url( home_url(), PHP_URL_SCHEME );
541	$url    = set_url_scheme( $url, $scheme );
542
543	// Add 'www.' if it is absent and should be there.
544	if ( str_contains( home_url(), '://www.' ) && ! str_contains( $url, '://www.' ) ) {
545		$url = str_replace( '://', '://www.', $url );
546	}
547
548	// Strip 'www.' if it is present and shouldn't be.
549	if ( ! str_contains( home_url(), '://www.' ) ) {
550		$url = str_replace( '://www.', '://', $url );
551	}
552
553	if ( trim( $url, '/' ) === home_url() && 'page' === get_option( 'show_on_front' ) ) {
554		$page_on_front = get_option( 'page_on_front' );
555
556		if ( $page_on_front && get_post( $page_on_front ) instanceof WP_Post ) {
557			return (int) $page_on_front;
558		}
559	}
560
561	// Check to see if we are using rewrite rules.
562	$rewrite = $wp_rewrite->wp_rewrite_rules();
563
564	// Not using rewrite rules, and 'p=N' and 'page_id=N' methods failed, so we're out of options.
565	if ( empty( $rewrite ) ) {
566		return 0;
567	}
568
569	// Strip 'index.php/' if we're not using path info permalinks.
570	if ( ! $wp_rewrite->using_index_permalinks() ) {
571		$url = str_replace( $wp_rewrite->index . '/', '', $url );
572	}
573
574	if ( str_contains( trailingslashit( $url ), home_url( '/' ) ) ) {
575		// Chop off http://domain.com/[path].
576		$url = str_replace( home_url(), '', $url );
577	} else {
578		// Chop off /path/to/blog.
579		$home_path = parse_url( home_url( '/' ) );
580		$home_path = isset( $home_path['path'] ) ? $home_path['path'] : '';
581		$url       = preg_replace( sprintf( '#^%s#', preg_quote( $home_path ) ), '', trailingslashit( $url ) );
582	}
583
584	// Trim leading and lagging slashes.
585	$url = trim( $url, '/' );
586
587	$request              = $url;
588	$post_type_query_vars = array();
589
590	foreach ( get_post_types( array(), 'objects' ) as $post_type => $t ) {
591		if ( ! empty( $t->query_var ) ) {
592			$post_type_query_vars[ $t->query_var ] = $post_type;
593		}
594	}
595
596	// Look for matches.
597	$request_match = $request;
598	foreach ( (array) $rewrite as $match => $query ) {
599
600		/*
601		 * If the requesting file is the anchor of the match,
602		 * prepend it to the path info.
603		 */
604		if ( ! empty( $url ) && ( $url !== $request ) && str_starts_with( $match, $url ) ) {
605			$request_match = $url . '/' . $request;
606		}
607
608		if ( preg_match( "#^$match#", $request_match, $matches ) ) {
609
610			if ( $wp_rewrite->use_verbose_page_rules && preg_match( '/pagename=\$matches\[([0-9]+)\]/', $query, $varmatch ) ) {
611				// This is a verbose page match, let's check to be sure about it.
612				$page = get_page_by_path( $matches[ $varmatch[1] ] );
613				if ( ! $page ) {
614					continue;
615				}
616
617				$post_status_obj = get_post_status_object( $page->post_status );
618				if ( ! $post_status_obj->public && ! $post_status_obj->protected
619					&& ! $post_status_obj->private && $post_status_obj->exclude_from_search ) {
620					continue;
621				}
622			}
623
624			/*
625			 * Got a match.
626			 * Trim the query of everything up to the '?'.
627			 */
628			$query = preg_replace( '!^.+\?!', '', $query );
629
630			// Substitute the substring matches into the query.
631			$query = addslashes( WP_MatchesMapRegex::apply( $query, $matches ) );
632
633			// Filter out non-public query vars.
634			global $wp;
635			parse_str( $query, $query_vars );
636			$query = array();
637			foreach ( (array) $query_vars as $key => $value ) {
638				if ( in_array( (string) $key, $wp->public_query_vars, true ) ) {
639					$query[ $key ] = $value;
640					if ( isset( $post_type_query_vars[ $key ] ) ) {
641						$query['post_type'] = $post_type_query_vars[ $key ];
642						$query['name']      = $value;
643					}
644				}
645			}
646
647			// Resolve conflicts between posts with numeric slugs and date archive queries.
648			$query = wp_resolve_numeric_slug_conflicts( $query );
649
650			// Do the query.
651			$query = new WP_Query( $query );
652			if ( ! empty( $query->posts ) && $query->is_singular ) {
653				return $query->post->ID;
654			} else {
655				return 0;
656			}
657		}
658	}
659	return 0;
660}
661