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  * @author Scott Yang
   9  * @author Mark Jaquith
  10  * @package WordPress
  11  * @since 2.3.0
  12  */
  13
  14 /**
  15  * Redirects incoming links to the proper URL based on the site url.
  16  *
  17  * Search engines consider www.somedomain.com and somedomain.com to be two
  18  * different URLs when they both go to the same location. This SEO enhancement
  19  * prevents penality for duplicate content by redirecting all incoming links to
  20  * one or the other.
  21  *
  22  * Prevents redirection for feeds, trackbacks, searches, comment popup, and
  23  * admin URLs. Does not redirect on IIS, page/post previews, and on form data.
  24  *
  25  * Will also attempt to find the correct link when a user enters a URL that does
  26  * not exist based on exact WordPress query. Will instead try to parse the URL
  27  * or query in an attempt to figure the correct page to go to.
  28  *
  29  * @since 2.3.0
  30  * @uses $wp_rewrite
  31  * @uses $is_IIS
  32  *
  33  * @param string $requested_url Optional. The URL that was requested, used to
  34  *              figure if redirect is needed.
  35  * @param bool $do_redirect Optional. Redirect to the new URL.
  36  * @return null|false|string Null, if redirect not needed. False, if redirect
  37  *              not needed or the string of the URL
  38  */
  39 function redirect_canonical($requested_url=null, $do_redirect=true) {
  40         global $wp_rewrite, $is_IIS, $wp_query, $wpdb;
  41
  42         if ( is_trackback() || is_search() || is_comments_popup() || is_admin() || $is_IIS || ( isset($_POST) && count($_POST) ) || is_preview() || is_robots() )
  43                 return;
  44
  45         if ( !$requested_url ) {
  46                 // build the URL in the address bar
  47                 $requested_url  = ( !empty($_SERVER['HTTPS'] ) && strtolower($_SERVER['HTTPS']) == 'on' ) ? 'https://' : 'http://';
  48                 $requested_url .= $_SERVER['HTTP_HOST'];
  49                 $requested_url .= $_SERVER['REQUEST_URI'];
  50         }
  51
  52         $original = @parse_url($requested_url);
  53         if ( false === $original )
  54                 return;
  55
  56         // Some PHP setups turn requests for / into /index.php in REQUEST_URI
  57         // See: http://trac.wordpress.org/ticket/5017
  58         // See: http://trac.wordpress.org/ticket/7173
  59         // Disabled, for now:
  60         // $original['path'] = preg_replace('|/index\.php$|', '/', $original['path']);
  61
  62         $redirect = $original;
  63         $redirect_url = false;
  64
  65         // Notice fixing
  66         if ( !isset($redirect['path']) )  $redirect['path'] = '';
  67         if ( !isset($redirect['query']) ) $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'), $redirect['query']);
  79                 }
  80         }
  81
  82         // These tests give us a WP-generated permalink
  83         if ( is_404() ) {
  84                 $redirect_url = redirect_guess_404_permalink();
  85         } elseif ( is_object($wp_rewrite) && $wp_rewrite->using_permalinks() ) {
  86                 // rewriting of old ?p=X, ?m=2004, ?m=200401, ?m=20040101
  87                 if ( is_single() && !empty($_GET['p']) && ! $redirect_url ) {
  88                         if ( $redirect_url = get_permalink(get_query_var('p')) )
  89                                 $redirect['query'] = remove_query_arg('p', $redirect['query']);
  90                         if ( get_query_var( 'page' ) ) {
  91                                 $redirect_url = trailingslashit( $redirect_url ) . user_trailingslashit( get_query_var( 'page' ), 'single_paged' );
  92                                 $redirect['query'] = remove_query_arg( 'page', $redirect['query'] );
  93                         }
  94                 } elseif ( is_page() && !empty($_GET['page_id']) && ! $redirect_url ) {
  95                         if ( $redirect_url = get_permalink(get_query_var('page_id')) )
  96                                 $redirect['query'] = remove_query_arg('page_id', $redirect['query']);
  97                 } elseif ( !empty($_GET['m']) && ( is_year() || is_month() || is_day() ) ) {
  98                         $m = get_query_var('m');
  99                         switch ( strlen($m) ) {
 100                                 case 4: // Yearly
 101                                         $redirect_url = get_year_link($m);
 102                                         break;
 103                                 case 6: // Monthly
 104                                         $redirect_url = get_month_link( substr($m, 0, 4), substr($m, 4, 2) );
 105                                         break;
 106                                 case 8: // Daily
 107                                         $redirect_url = get_day_link(substr($m, 0, 4), substr($m, 4, 2), substr($m, 6, 2));
 108                                         break;
 109                         }
 110                         if ( $redirect_url )
 111                                 $redirect['query'] = remove_query_arg('m', $redirect['query']);
 112                 // now moving on to non ?m=X year/month/day links
 113                 } elseif ( is_day() && get_query_var('year') && get_query_var('monthnum') && !empty($_GET['day']) ) {
 114                         if ( $redirect_url = get_day_link(get_query_var('year'), get_query_var('monthnum'), get_query_var('day')) )
 115                                 $redirect['query'] = remove_query_arg(array('year', 'monthnum', 'day'), $redirect['query']);
 116                 } elseif ( is_month() && get_query_var('year') && !empty($_GET['monthnum']) ) {
 117                         if ( $redirect_url = get_month_link(get_query_var('year'), get_query_var('monthnum')) )
 118                                 $redirect['query'] = remove_query_arg(array('year', 'monthnum'), $redirect['query']);
 119                 } elseif ( is_year() && !empty($_GET['year']) ) {
 120                         if ( $redirect_url = get_year_link(get_query_var('year')) )
 121                                 $redirect['query'] = remove_query_arg('year', $redirect['query']);
 122                 } elseif ( is_category() && !empty($_GET['cat']) && preg_match( '|^[0-9]+$|', $_GET['cat'] ) ) {
 123                         if ( $redirect_url = get_category_link(get_query_var('cat')) )
 124                                 $redirect['query'] = remove_query_arg('cat', $redirect['query']);
 125                 } elseif ( is_author() && !empty($_GET['author']) ) {
 126                         $author = get_userdata(get_query_var('author'));
 127                         if ( false !== $author && $redirect_url = get_author_posts_url($author->ID, $author->user_nicename) )
 128                                 $redirect['query'] = remove_query_arg('author', $redirect['author']);
 129                 }
 130
 131         // paging and feeds
 132                 if ( get_query_var('paged') || is_feed() || get_query_var('cpage') ) {
 133                         if ( !$redirect_url )
 134                                 $redirect_url = $requested_url;
 135                         $paged_redirect = @parse_url($redirect_url);
 136                         while ( preg_match( '#/page/[0-9]+?(/+)?$#', $paged_redirect['path'] ) || preg_match( '#/(comments/?)?(feed|rss|rdf|atom|rss2)(/+)?$#', $paged_redirect['path'] ) || preg_match( '#/comment-page-[0-9]+(/+)?$#', $paged_redirect['path'] ) ) {
 137                                 // Strip off paging and feed
 138                                 $paged_redirect['path'] = preg_replace('#/page/[0-9]+?(/+)?$#', '/', $paged_redirect['path']); // strip off any existing paging
 139                                 $paged_redirect['path'] = preg_replace('#/(comments/?)?(feed|rss2?|rdf|atom)(/+|$)#', '/', $paged_redirect['path']); // strip off feed endings
 140                                 $paged_redirect['path'] = preg_replace('#/comment-page-[0-9]+?(/+)?$#', '/', $paged_redirect['path']); // strip off any existing comment paging
 141                         }
 142
 143                         $addl_path = '';
 144                         if ( is_feed() ) {
 145                                 $addl_path = !empty( $addl_path ) ? trailingslashit($addl_path) : '';
 146                                 if ( get_query_var( 'withcomments' ) )
 147                                         $addl_path .= 'comments/';
 148                                 $addl_path .= user_trailingslashit( 'feed/' . ( ( 'rss2' ==  get_query_var('feed') || 'feed' == get_query_var('feed') ) ? '' : get_query_var('feed') ), 'feed' );
 149                                 $redirect['query'] = remove_query_arg( 'feed', $redirect['query'] );
 150                         }
 151
 152                         if ( get_query_var('paged') > 0 ) {
 153                                 $paged = get_query_var('paged');
 154                                 $redirect['query'] = remove_query_arg( 'paged', $redirect['query'] );
 155                                 if ( !is_feed() ) {
 156                                         if ( $paged > 1 && !is_single() ) {
 157                                                 $addl_path = ( !empty( $addl_path ) ? trailingslashit($addl_path) : '' ) . user_trailingslashit("page/$paged", 'paged');
 158                                         } elseif ( !is_single() ) {
 159                                                 $addl_path = ( !empty( $addl_path ) ? trailingslashit($addl_path) : '' ) . user_trailingslashit($paged_redirect['path'], 'paged');
 160                                         }
 161                                 } elseif ( $paged > 1 ) {
 162                                         $redirect['query'] = add_query_arg( 'paged', $paged, $redirect['query'] );
 163                                 }
 164                         }
 165
 166                         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 ) ) ) {
 167                                 $addl_path = ( !empty( $addl_path ) ? trailingslashit($addl_path) : '' ) . user_trailingslashit( 'comment-page-' . get_query_var('cpage'), 'commentpaged' );
 168                                 $redirect['query'] = remove_query_arg( 'cpage', $redirect['query'] );
 169                         }
 170
 171                         $paged_redirect['path'] = user_trailingslashit( preg_replace('|/index.php/?$|', '/', $paged_redirect['path']) ); // strip off trailing /index.php/
 172                         if ( !empty( $addl_path ) && $wp_rewrite->using_index_permalinks() && strpos($paged_redirect['path'], '/index.php/') === false )
 173                                 $paged_redirect['path'] = trailingslashit($paged_redirect['path']) . 'index.php/';
 174                         if ( !empty( $addl_path ) )
 175                                 $paged_redirect['path'] = trailingslashit($paged_redirect['path']) . $addl_path;
 176                         $redirect_url = $paged_redirect['scheme'] . '://' . $paged_redirect['host'] . $paged_redirect['path'];
 177                         $redirect['path'] = $paged_redirect['path'];
 178                 }
 179         }
 180
 181         // tack on any additional query vars
 182         $redirect['query'] = preg_replace( '#^\??&*?#', '', $redirect['query'] );
 183         if ( $redirect_url && !empty($redirect['query']) ) {
 184                 if ( strpos($redirect_url, '?') !== false )
 185                         $redirect_url .= '&';
 186                 else
 187                         $redirect_url .= '?';
 188                 $redirect_url .= $redirect['query'];
 189         }
 190
 191         if ( $redirect_url )
 192                 $redirect = @parse_url($redirect_url);
 193
 194         // www.example.com vs example.com
 195         $user_home = @parse_url(get_option('home'));
 196         if ( !empty($user_home['host']) )
 197                 $redirect['host'] = $user_home['host'];
 198         if ( empty($user_home['path']) )
 199                 $user_home['path'] = '/';
 200
 201         // Handle ports
 202         if ( !empty($user_home['port']) )
 203                 $redirect['port'] = $user_home['port'];
 204         else
 205                 unset($redirect['port']);
 206
 207         // trailing /index.php
 208         $redirect['path'] = preg_replace('|/index.php/*?$|', '/', $redirect['path']);
 209
 210         // Remove trailing spaces from the path
 211         $redirect['path'] = preg_replace( '#(%20| )+$#', '', $redirect['path'] );
 212
 213         if ( !empty( $redirect['query'] ) ) {
 214                 // Remove trailing spaces from certain terminating query string args
 215                 $redirect['query'] = preg_replace( '#((p|page_id|cat|tag)=[^&]*?)(%20| )+$#', '$1', $redirect['query'] );
 216
 217                 // Clean up empty query strings
 218                 $redirect['query'] = trim(preg_replace( '#(^|&)(p|page_id|cat|tag)=?(&|$)#', '&', $redirect['query']), '&');
 219
 220                 // Remove redundant leading ampersands
 221                 $redirect['query'] = preg_replace( '#^\??&*?#', '', $redirect['query'] );
 222         }
 223
 224         // strip /index.php/ when we're not using PATHINFO permalinks
 225         if ( !$wp_rewrite->using_index_permalinks() )
 226                 $redirect['path'] = str_replace('/index.php/', '/', $redirect['path']);
 227
 228         // trailing slashes
 229         if ( is_object($wp_rewrite) && $wp_rewrite->using_permalinks() && !is_404() && (!is_front_page() || ( is_front_page() && (get_query_var('paged') > 1) ) ) ) {
 230                 $user_ts_type = '';
 231                 if ( get_query_var('paged') > 0 ) {
 232                         $user_ts_type = 'paged';
 233                 } else {
 234                         foreach ( array('single', 'category', 'page', 'day', 'month', 'year', 'home') as $type ) {
 235                                 $func = 'is_' . $type;
 236                                 if ( call_user_func($func) ) {
 237                                         $user_ts_type = $type;
 238                                         break;
 239                                 }
 240                         }
 241                 }
 242                 $redirect['path'] = user_trailingslashit($redirect['path'], $user_ts_type);
 243         } elseif ( is_front_page() ) {
 244                 $redirect['path'] = trailingslashit($redirect['path']);
 245         }
 246
 247         // Always trailing slash the Front Page URL
 248         if ( trailingslashit( $redirect['path'] ) == trailingslashit( $user_home['path'] ) )
 249                 $redirect['path'] = trailingslashit($redirect['path']);
 250
 251         // Ignore differences in host capitalization, as this can lead to infinite redirects
 252         // Only redirect no-www <=> yes-www
 253         if ( strtolower($original['host']) == strtolower($redirect['host']) ||
 254                 ( strtolower($original['host']) != 'www.' . strtolower($redirect['host']) && 'www.' . strtolower($original['host']) != strtolower($redirect['host']) ) )
 255                 $redirect['host'] = $original['host'];
 256
 257         $compare_original = array($original['host'], $original['path']);
 258
 259         if ( !empty( $original['port'] ) )
 260                 $compare_original[] = $original['port'];
 261
 262         if ( !empty( $original['query'] ) )
 263                 $compare_original[] = $original['query'];
 264
 265         $compare_redirect = array($redirect['host'], $redirect['path']);
 266
 267         if ( !empty( $redirect['port'] ) )
 268                 $compare_redirect[] = $redirect['port'];
 269
 270         if ( !empty( $redirect['query'] ) )
 271                 $compare_redirect[] = $redirect['query'];
 272
 273         if ( $compare_original !== $compare_redirect ) {
 274                 $redirect_url = $redirect['scheme'] . '://' . $redirect['host'];
 275                 if ( !empty($redirect['port']) )
 276                         $redirect_url .= ':' . $redirect['port'];
 277                 $redirect_url .= $redirect['path'];
 278                 if ( !empty($redirect['query']) )
 279                         $redirect_url .= '?' . $redirect['query'];
 280         }
 281
 282         if ( !$redirect_url || $redirect_url == $requested_url )
 283                 return false;
 284
 285         // Note that you can use the "redirect_canonical" filter to cancel a canonical redirect for whatever reason by returning FALSE
 286         $redirect_url = apply_filters('redirect_canonical', $redirect_url, $requested_url);
 287
 288         if ( !$redirect_url || $redirect_url == $requested_url ) // yes, again -- in case the filter aborted the request
 289                 return false;
 290
 291         if ( $do_redirect ) {
 292                 // protect against chained redirects
 293                 if ( !redirect_canonical($redirect_url, false) ) {
 294                         wp_redirect($redirect_url, 301);
 295                         exit();
 296                 } else {
 297                         // Debug
 298                         // die("1: $redirect_url<br />2: " . redirect_canonical( $redirect_url, false ) );
 299                         return false;
 300                 }
 301         } else {
 302                 return $redirect_url;
 303         }
 304 }
 305
 306 /**
 307  * Attempts to guess correct post based on query vars.
 308  *
 309  * @since 2.3.0
 310  * @uses $wpdb
 311  *
 312  * @return bool|string Returns False, if it can't find post, returns correct
 313  *              location on success.
 314  */
 315 function redirect_guess_404_permalink() {
 316         global $wpdb;
 317
 318         if ( !get_query_var('name') )
 319                 return false;
 320
 321         $where = $wpdb->prepare("post_name LIKE %s", get_query_var('name') . '%');
 322
 323         // if any of year, monthnum, or day are set, use them to refine the query
 324         if ( get_query_var('year') )
 325                 $where .= $wpdb->prepare(" AND YEAR(post_date) = %d", get_query_var('year'));
 326         if ( get_query_var('monthnum') )
 327                 $where .= $wpdb->prepare(" AND MONTH(post_date) = %d", get_query_var('monthnum'));
 328         if ( get_query_var('day') )
 329                 $where .= $wpdb->prepare(" AND DAYOFMONTH(post_date) = %d", get_query_var('day'));
 330
 331         $post_id = $wpdb->get_var("SELECT ID FROM $wpdb->posts WHERE $where AND post_status = 'publish'");
 332         if ( !$post_id )
 333                 return false;
 334         return get_permalink($post_id);
 335 }
 336
 337 add_action('template_redirect', 'redirect_canonical');
 338
 339 ?>