wp-includes/canonical.php

   1 <?php
   2 /**
   3  * Canonical API to handle WordPress Redirecting
   4  *
   5  * Based on "Permalink Redirect" from Scott Yang and "Enforce www. Preference"
   6  * by Mark Jaquith
   7  *
   8  * @package WordPress
   9  * @since 2.3.0
  10  */
  11
  12 /**
  13  * Redirects incoming links to the proper URL based on the site url.
  14  *
  15  * Search engines consider www.somedomain.com and somedomain.com to be two
  16  * different URLs when they both go to the same location. This SEO enhancement
  17  * prevents penality for duplicate content by redirecting all incoming links to
  18  * one or the other.
  19  *
  20  * Prevents redirection for feeds, trackbacks, searches, comment popup, and
  21  * admin URLs. Does not redirect on IIS, page/post previews, and on form data.
  22  *
  23  * Will also attempt to find the correct link when a user enters a URL that does
  24  * not exist based on exact WordPress query. Will instead try to parse the URL
  25  * or query in an attempt to figure the correct page to go to.
  26  *
  27  * @since 2.3.0
  28  * @uses $wp_rewrite
  29  * @uses $is_IIS
  30  *
  31  * @param string $requested_url Optional. The URL that was requested, used to
  32  *              figure if redirect is needed.
  33  * @param bool $do_redirect Optional. Redirect to the new URL.
  34  * @return null|false|string Null, if redirect not needed. False, if redirect
  35  *              not needed or the string of the URL
  36  */
  37 function redirect_canonical( $requested_url = null, $do_redirect = true ) {
  38         global $wp_rewrite, $is_IIS, $wp_query, $wpdb;
  39
  40         if ( is_trackback() || is_search() || is_comments_popup() || is_admin() || !empty($_POST) || is_preview() || is_robots() || $is_IIS )
  41                 return;
  42
  43         if ( !$requested_url ) {
  44                 // build the URL in the address bar
  45                 $requested_url  = is_ssl() ? 'https://' : 'http://';
  46                 $requested_url .= $_SERVER['HTTP_HOST'];
  47                 $requested_url .= $_SERVER['REQUEST_URI'];
  48         }
  49
  50         $original = @parse_url($requested_url);
  51         if ( false === $original )
  52                 return;
  53
  54         // Some PHP setups turn requests for / into /index.php in REQUEST_URI
  55         // See: http://trac.wordpress.org/ticket/5017
  56         // See: http://trac.wordpress.org/ticket/7173
  57         // Disabled, for now:
  58         // $original['path'] = preg_replace('|/index\.php$|', '/', $original['path']);
  59
  60         $redirect = $original;
  61         $redirect_url = false;
  62
  63         // Notice fixing
  64         if ( !isset($redirect['path']) )
  65                 $redirect['path'] = '';
  66         if ( !isset($redirect['query']) )
  67                 $redirect['query'] = '';
  68
  69         if ( is_singular() && 1 > $wp_query->post_count && ($id = get_query_var('p')) ) {
  70
  71                 $vars = $wpdb->get_results( $wpdb->prepare("SELECT post_type, post_parent FROM $wpdb->posts WHERE ID = %d", $id) );
  72
  73                 if ( isset($vars[0]) && $vars = $vars[0] ) {
  74                         if ( 'revision' == $vars->post_type && $vars->post_parent > 0 )
  75                                 $id = $vars->post_parent;
  76
  77                         if ( $redirect_url = get_permalink($id) )
  78                                 $redirect['query'] = remove_query_arg(array('p', 'page_id', 'attachment_id', 'post_type'), $redirect['query']);
  79                 }
  80         }
  81
  82         // These tests give us a WP-generated permalink
  83         if ( is_404() ) {
  84
  85                 // Redirect ?page_id, ?p=, ?attachment_id= to their respective url's
  86                 $id = max( get_query_var('p'), get_query_var('page_id'), get_query_var('attachment_id') );
  87                 if ( $id && $redirect_post = get_post($id) ) {
  88                         $post_type_obj = get_post_type_object($redirect_post->post_type);
  89                         if ( $post_type_obj->public ) {
  90                                 $redirect_url = get_permalink($redirect_post);
  91                                 $redirect['query'] = remove_query_arg(array('p', 'page_id', 'attachment_id', 'post_type'), $redirect['query']);
  92                         }
  93                 }
  94
  95                 if ( ! $redirect_url )
  96                         $redirect_url = redirect_guess_404_permalink();
  97
  98         } elseif ( is_object($wp_rewrite) && $wp_rewrite->using_permalinks() ) {
  99                 // rewriting of old ?p=X, ?m=2004, ?m=200401, ?m=20040101
 100                 if ( is_attachment() && !empty($_GET['attachment_id']) && ! $redirect_url ) {
 101                         if ( $redirect_url = get_attachment_link(get_query_var('attachment_id')) )
 102                                 $redirect['query'] = remove_query_arg('attachment_id', $redirect['query']);
 103                 } elseif ( is_single() && !empty($_GET['p']) && ! $redirect_url ) {
 104                         if ( $redirect_url = get_permalink(get_query_var('p')) )
 105                                 $redirect['query'] = remove_query_arg(array('p', 'post_type'), $redirect['query']);
 106                 } elseif ( is_single() && !empty($_GET['name'])  && ! $redirect_url ) {
 107                         if ( $redirect_url = get_permalink( $wp_query->get_queried_object_id() ) )
 108                                 $redirect['query'] = remove_query_arg('name', $redirect['query']);
 109                 } elseif ( is_page() && !empty($_GET['page_id']) && ! $redirect_url ) {
 110                         if ( $redirect_url = get_permalink(get_query_var('page_id')) )
 111                                 $redirect['query'] = remove_query_arg('page_id', $redirect['query']);
 112                 } elseif ( is_page() && !is_feed() && isset($wp_query->queried_object) && 'page' == get_option('show_on_front') && $wp_query->queried_object->ID == get_option('page_on_front')  && ! $redirect_url ) {
 113                         $redirect_url = home_url('/');
 114                 } elseif ( is_home() && !empty($_GET['page_id']) && 'page' == get_option('show_on_front') && get_query_var('page_id') == get_option('page_for_posts')  && ! $redirect_url ) {
 115                         if ( $redirect_url = get_permalink(get_option('page_for_posts')) )
 116                                 $redirect['query'] = remove_query_arg('page_id', $redirect['query']);
 117                 } elseif ( !empty($_GET['m']) && ( is_year() || is_month() || is_day() ) ) {
 118                         $m = get_query_var('m');
 119                         switch ( strlen($m) ) {
 120                                 case 4: // Yearly
 121                                         $redirect_url = get_year_link($m);
 122                                         break;
 123                                 case 6: // Monthly
 124                                         $redirect_url = get_month_link( substr($m, 0, 4), substr($m, 4, 2) );
 125                                         break;
 126                                 case 8: // Daily
 127                                         $redirect_url = get_day_link(substr($m, 0, 4), substr($m, 4, 2), substr($m, 6, 2));
 128                                         break;
 129                         }
 130                         if ( $redirect_url )
 131                                 $redirect['query'] = remove_query_arg('m', $redirect['query']);
 132                 // now moving on to non ?m=X year/month/day links
 133                 } elseif ( is_day() && get_query_var('year') && get_query_var('monthnum') && !empty($_GET['day']) ) {
 134                         if ( $redirect_url = get_day_link(get_query_var('year'), get_query_var('monthnum'), get_query_var('day')) )
 135                                 $redirect['query'] = remove_query_arg(array('year', 'monthnum', 'day'), $redirect['query']);
 136                 } elseif ( is_month() && get_query_var('year') && !empty($_GET['monthnum']) ) {
 137                         if ( $redirect_url = get_month_link(get_query_var('year'), get_query_var('monthnum')) )
 138                                 $redirect['query'] = remove_query_arg(array('year', 'monthnum'), $redirect['query']);
 139                 } elseif ( is_year() && !empty($_GET['year']) ) {
 140                         if ( $redirect_url = get_year_link(get_query_var('year')) )
 141                                 $redirect['query'] = remove_query_arg('year', $redirect['query']);
 142                 } elseif ( is_author() && !empty($_GET['author']) && preg_match( '|^[0-9]+$|', $_GET['author'] ) ) {
 143                         $author = get_userdata(get_query_var('author'));
 144                         if ( false !== $author && $redirect_url = get_author_posts_url($author->ID, $author->user_nicename) )
 145                                 $redirect['query'] = remove_query_arg('author', $redirect['query']);
 146                 } elseif ( is_category() || is_tag() || is_tax() ) { // Terms (Tags/categories)
 147
 148                         $term_count = 0;
 149                         foreach ( $wp_query->tax_query->queries as $tax_query )
 150                                 $term_count += count( $tax_query['terms'] );
 151
 152                         $obj = $wp_query->get_queried_object();
 153                         if ( $term_count <= 1 && !empty($obj->term_id) && ( $tax_url = get_term_link((int)$obj->term_id, $obj->taxonomy) ) && !is_wp_error($tax_url) && !empty($redirect['query']) ) {
 154                                 if ( !empty($redirect['query']) ) {
 155                                         if ( is_category() ) {
 156                                                 $redirect['query'] = remove_query_arg( array( 'category_name', 'category', 'cat'), $redirect['query']);
 157                                         } elseif ( is_tag() ) {
 158                                                 $redirect['query'] = remove_query_arg( array( 'tag', 'tag_id'), $redirect['query']);
 159                                         } elseif ( is_tax() ) { // Custom taxonomies will have a custom query var, remove those too:
 160                                                 $tax = get_taxonomy( $obj->taxonomy );
 161                                                 if ( false !== $tax->query_var)
 162                                                         $redirect['query'] = remove_query_arg($tax->query_var, $redirect['query']);
 163                                                 else
 164                                                         $redirect['query'] = remove_query_arg( array( 'term', 'taxonomy'), $redirect['query']);
 165                                         }
 166                                 }
 167                                 $tax_url = parse_url($tax_url);
 168                                 if ( ! empty($tax_url['query']) ) { // Custom taxonomies may only be accessable via ?taxonomy=..&term=..
 169                                         parse_str($tax_url['query'], $query_vars);
 170                                         $redirect['query'] = add_query_arg($query_vars, $redirect['query']);
 171                                 } else { // Taxonomy is accessable via a "pretty-URL"
 172                                         $redirect['path'] = $tax_url['path'];
 173                                 }
 174                         }
 175                 } elseif ( is_single() && strpos($wp_rewrite->permalink_structure, '%category%') !== false ) {
 176                         $category = get_category_by_path(get_query_var('category_name'));
 177                         $post_terms = wp_get_object_terms($wp_query->get_queried_object_id(), 'category', array('fields' => 'tt_ids'));
 178                         if ( (!$category || is_wp_error($category)) || ( !is_wp_error($post_terms) && !empty($post_terms) && !in_array($category->term_taxonomy_id, $post_terms) ) )
 179                                 $redirect_url = get_permalink($wp_query->get_queried_object_id());
 180                 }
 181
 182                 // Post Paging
 183                 if ( is_singular() && get_query_var('page') && $redirect_url ) {
 184                         $redirect_url = trailingslashit( $redirect_url ) . user_trailingslashit( get_query_var( 'page' ), 'single_paged' );
 185                         $redirect['query'] = remove_query_arg( 'page', $redirect['query'] );
 186                 }
 187
 188                 // paging and feeds
 189                 if ( get_query_var('paged') || is_feed() || get_query_var('cpage') ) {
 190                         while ( preg_match( "#/$wp_rewrite->pagination_base/?[0-9]+?(/+)?$#", $redirect['path'] ) || preg_match( '#/(comments/?)?(feed|rss|rdf|atom|rss2)(/+)?$#', $redirect['path'] ) || preg_match( '#/comment-page-[0-9]+(/+)?$#', $redirect['path'] ) ) {
 191                                 // Strip off paging and feed
 192                                 $redirect['path'] = preg_replace("#/$wp_rewrite->pagination_base/?[0-9]+?(/+)?$#", '/', $redirect['path']); // strip off any existing paging
 193                                 $redirect['path'] = preg_replace('#/(comments/?)?(feed|rss2?|rdf|atom)(/+|$)#', '/', $redirect['path']); // strip off feed endings
 194                                 $redirect['path'] = preg_replace('#/comment-page-[0-9]+?(/+)?$#', '/', $redirect['path']); // strip off any existing comment paging
 195                         }
 196
 197                         $addl_path = '';
 198                         if ( is_feed() && in_array( get_query_var('feed'), $wp_rewrite->feeds ) ) {
 199                                 $addl_path = !empty( $addl_path ) ? trailingslashit($addl_path) : '';
 200                                 if ( get_query_var( 'withcomments' ) )
 201                                         $addl_path .= 'comments/';
 202                                 $addl_path .= user_trailingslashit( 'feed/' . ( ( get_default_feed() ==  get_query_var('feed') || 'feed' == get_query_var('feed') ) ? '' : get_query_var('feed') ), 'feed' );
 203                                 $redirect['query'] = remove_query_arg( 'feed', $redirect['query'] );
 204                         }
 205
 206                         if ( get_query_var('paged') > 0 ) {
 207                                 $paged = get_query_var('paged');
 208                                 $redirect['query'] = remove_query_arg( 'paged', $redirect['query'] );
 209                                 if ( !is_feed() ) {
 210                                         if ( $paged > 1 && !is_single() ) {
 211                                                 $addl_path = ( !empty( $addl_path ) ? trailingslashit($addl_path) : '' ) . user_trailingslashit("$wp_rewrite->pagination_base/$paged", 'paged');
 212                                         } elseif ( !is_single() ) {
 213                                                 $addl_path = !empty( $addl_path ) ? trailingslashit($addl_path) : '';
 214                                         }
 215                                 } elseif ( $paged > 1 ) {
 216                                         $redirect['query'] = add_query_arg( 'paged', $paged, $redirect['query'] );
 217                                 }
 218                         }
 219
 220                         if ( get_option('page_comments') && ( ( 'newest' == get_option('default_comments_page') && get_query_var('cpage') > 0 ) || ( 'newest' != get_option('default_comments_page') && get_query_var('cpage') > 1 ) ) ) {
 221                                 $addl_path = ( !empty( $addl_path ) ? trailingslashit($addl_path) : '' ) . user_trailingslashit( 'comment-page-' . get_query_var('cpage'), 'commentpaged' );
 222                                 $redirect['query'] = remove_query_arg( 'cpage', $redirect['query'] );
 223                         }
 224
 225                         $redirect['path'] = user_trailingslashit( preg_replace('|/index.php/?$|', '/', $redirect['path']) ); // strip off trailing /index.php/
 226                         if ( !empty( $addl_path ) && $wp_rewrite->using_index_permalinks() && strpos($redirect['path'], '/index.php/') === false )
 227                                 $redirect['path'] = trailingslashit($redirect['path']) . 'index.php/';
 228                         if ( !empty( $addl_path ) )
 229                                 $redirect['path'] = trailingslashit($redirect['path']) . $addl_path;
 230                         $redirect_url = $redirect['scheme'] . '://' . $redirect['host'] . $redirect['path'];
 231                 }
 232         }
 233
 234         // tack on any additional query vars
 235         $redirect['query'] = preg_replace( '#^\??&*?#', '', $redirect['query'] );
 236         if ( $redirect_url && !empty($redirect['query']) ) {
 237                 parse_str( $redirect['query'], $_parsed_query );
 238                 $redirect = @parse_url($redirect_url);
 239
 240                 if ( ! empty( $_parsed_query['name'] ) && ! empty( $redirect['query'] ) ) {
 241                         parse_str( $redirect['query'], $_parsed_redirect_query );
 242
 243                         if ( empty( $_parsed_redirect_query['name'] ) )
 244                                 unset( $_parsed_query['name'] );
 245                 }
 246
 247                 $redirect_url = add_query_arg( $_parsed_query, $redirect_url );
 248         }
 249
 250         if ( $redirect_url )
 251                 $redirect = @parse_url($redirect_url);
 252
 253         // www.example.com vs example.com
 254         $user_home = @parse_url(home_url());
 255         if ( !empty($user_home['host']) )
 256                 $redirect['host'] = $user_home['host'];
 257         if ( empty($user_home['path']) )
 258                 $user_home['path'] = '/';
 259
 260         // Handle ports
 261         if ( !empty($user_home['port']) )
 262                 $redirect['port'] = $user_home['port'];
 263         else
 264                 unset($redirect['port']);
 265
 266         // trailing /index.php
 267         $redirect['path'] = preg_replace('|/index.php/*?$|', '/', $redirect['path']);
 268
 269         // Remove trailing spaces from the path
 270         $redirect['path'] = preg_replace( '#(%20| )+$#', '', $redirect['path'] );
 271
 272         if ( !empty( $redirect['query'] ) ) {
 273                 // Remove trailing spaces from certain terminating query string args
 274                 $redirect['query'] = preg_replace( '#((p|page_id|cat|tag)=[^&]*?)(%20| )+$#', '$1', $redirect['query'] );
 275
 276                 // Clean up empty query strings
 277                 $redirect['query'] = trim(preg_replace( '#(^|&)(p|page_id|cat|tag)=?(&|$)#', '&', $redirect['query']), '&');
 278
 279                 // Remove redundant leading ampersands
 280                 $redirect['query'] = preg_replace( '#^\??&*?#', '', $redirect['query'] );
 281         }
 282
 283         // strip /index.php/ when we're not using PATHINFO permalinks
 284         if ( !$wp_rewrite->using_index_permalinks() )
 285                 $redirect['path'] = str_replace('/index.php/', '/', $redirect['path']);
 286
 287         // trailing slashes
 288         if ( is_object($wp_rewrite) && $wp_rewrite->using_permalinks() && !is_404() && (!is_front_page() || ( is_front_page() && (get_query_var('paged') > 1) ) ) ) {
 289                 $user_ts_type = '';
 290                 if ( get_query_var('paged') > 0 ) {
 291                         $user_ts_type = 'paged';
 292                 } else {
 293                         foreach ( array('single', 'category', 'page', 'day', 'month', 'year', 'home') as $type ) {
 294                                 $func = 'is_' . $type;
 295                                 if ( call_user_func($func) ) {
 296                                         $user_ts_type = $type;
 297                                         break;
 298                                 }
 299                         }
 300                 }
 301                 $redirect['path'] = user_trailingslashit($redirect['path'], $user_ts_type);
 302         } elseif ( is_front_page() ) {
 303                 $redirect['path'] = trailingslashit($redirect['path']);
 304         }
 305
 306         // Strip multiple slashes out of the URL
 307         if ( strpos($redirect['path'], '//') > -1 )
 308                 $redirect['path'] = preg_replace('|/+|', '/', $redirect['path']);
 309
 310         // Always trailing slash the Front Page URL
 311         if ( trailingslashit( $redirect['path'] ) == trailingslashit( $user_home['path'] ) )
 312                 $redirect['path'] = trailingslashit($redirect['path']);
 313
 314         // Ignore differences in host capitalization, as this can lead to infinite redirects
 315         // Only redirect no-www <=> yes-www
 316         if ( strtolower($original['host']) == strtolower($redirect['host']) ||
 317                 ( strtolower($original['host']) != 'www.' . strtolower($redirect['host']) && 'www.' . strtolower($original['host']) != strtolower($redirect['host']) ) )
 318                 $redirect['host'] = $original['host'];
 319
 320         $compare_original = array($original['host'], $original['path']);
 321
 322         if ( !empty( $original['port'] ) )
 323                 $compare_original[] = $original['port'];
 324
 325         if ( !empty( $original['query'] ) )
 326                 $compare_original[] = $original['query'];
 327
 328         $compare_redirect = array($redirect['host'], $redirect['path']);
 329
 330         if ( !empty( $redirect['port'] ) )
 331                 $compare_redirect[] = $redirect['port'];
 332
 333         if ( !empty( $redirect['query'] ) )
 334                 $compare_redirect[] = $redirect['query'];
 335
 336         if ( $compare_original !== $compare_redirect ) {
 337                 $redirect_url = $redirect['scheme'] . '://' . $redirect['host'];
 338                 if ( !empty($redirect['port']) )
 339                         $redirect_url .= ':' . $redirect['port'];
 340                 $redirect_url .= $redirect['path'];
 341                 if ( !empty($redirect['query']) )
 342                         $redirect_url .= '?' . $redirect['query'];
 343         }
 344
 345         if ( !$redirect_url || $redirect_url == $requested_url )
 346                 return false;
 347
 348         // Hex encoded octets are case-insensitive.
 349         if ( false !== strpos($requested_url, '%') ) {
 350                 if ( !function_exists('lowercase_octets') ) {
 351                         function lowercase_octets($matches) {
 352                                 return strtolower( $matches[0] );
 353                         }
 354                 }
 355                 $requested_url = preg_replace_callback('|%[a-fA-F0-9][a-fA-F0-9]|', 'lowercase_octets', $requested_url);
 356         }
 357
 358         // Note that you can use the "redirect_canonical" filter to cancel a canonical redirect for whatever reason by returning FALSE
 359         $redirect_url = apply_filters('redirect_canonical', $redirect_url, $requested_url);
 360
 361         if ( !$redirect_url || $redirect_url == $requested_url ) // yes, again -- in case the filter aborted the request
 362                 return false;
 363
 364         if ( $do_redirect ) {
 365                 // protect against chained redirects
 366                 if ( !redirect_canonical($redirect_url, false) ) {
 367                         wp_redirect($redirect_url, 301);
 368                         exit();
 369                 } else {
 370                         // Debug
 371                         // die("1: $redirect_url<br />2: " . redirect_canonical( $redirect_url, false ) );
 372                         return false;
 373                 }
 374         } else {
 375                 return $redirect_url;
 376         }
 377 }
 378
 379 /**
 380  * Attempts to guess correct post based on query vars.
 381  *
 382  * @since 2.3.0
 383  * @uses $wpdb
 384  *
 385  * @return bool|string Returns False, if it can't find post, returns correct
 386  *              location on success.
 387  */
 388 function redirect_guess_404_permalink() {
 389         global $wpdb;
 390
 391         if ( !get_query_var('name') )
 392                 return false;
 393
 394         $where = $wpdb->prepare("post_name LIKE %s", like_escape( get_query_var('name') ) . '%');
 395
 396         // if any of post_type, year, monthnum, or day are set, use them to refine the query
 397         if ( get_query_var('post_type') )
 398                 $where .= $wpdb->prepare(" AND post_type = %s", get_query_var('post_type'));
 399         if ( get_query_var('year') )
 400                 $where .= $wpdb->prepare(" AND YEAR(post_date) = %d", get_query_var('year'));
 401         if ( get_query_var('monthnum') )
 402                 $where .= $wpdb->prepare(" AND MONTH(post_date) = %d", get_query_var('monthnum'));
 403         if ( get_query_var('day') )
 404                 $where .= $wpdb->prepare(" AND DAYOFMONTH(post_date) = %d", get_query_var('day'));
 405
 406         $post_id = $wpdb->get_var("SELECT ID FROM $wpdb->posts WHERE $where AND post_status = 'publish'");
 407         if ( !$post_id )
 408                 return false;
 409         return get_permalink($post_id);
 410 }
 411
 412 add_action('template_redirect', 'redirect_canonical');
 413
 414 ?>