[ Index ]

PHP Cross Reference of WordPress 3.0 beta 1

[ Index ]     [ Variables ]     [ Functions ]     [ Classes ]     [ Constants ]     [ Statistics ]

title

Body

[close]

/wp-includes/ -> canonical.php (source)

   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() || $is_IIS || ( isset($_POST) && count($_POST) ) || is_preview() || is_robots() )
  41          return;
  42  
  43      if ( !$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'), $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) && $wp_rewrite->using_permalinks() ) {
  86          // rewriting of old ?p=X, ?m=2004, ?m=200401, ?m=20040101
  87          if ( is_attachment() && !empty($_GET['attachment_id']) && ! $redirect_url ) {
  88              if ( $redirect_url = get_attachment_link(get_query_var('attachment_id')) )
  89                  $redirect['query'] = remove_query_arg('attachment_id', $redirect['query']);
  90          } elseif ( is_single() && !empty($_GET['p']) && ! $redirect_url ) {
  91              if ( $redirect_url = get_permalink(get_query_var('p')) )
  92                  $redirect['query'] = remove_query_arg('p', $redirect['query']);
  93              if ( get_query_var( 'page' ) ) {
  94                  $redirect_url = trailingslashit( $redirect_url ) . user_trailingslashit( get_query_var( 'page' ), 'single_paged' );
  95                  $redirect['query'] = remove_query_arg( 'page', $redirect['query'] );
  96              }
  97          } elseif ( is_single() && !empty($_GET['name'])  && ! $redirect_url ) {
  98              if ( $redirect_url = get_permalink( $wp_query->get_queried_object_id() ) )
  99                  $redirect['query'] = remove_query_arg('name', $redirect['query']);
 100          } elseif ( is_page() && !empty($_GET['page_id']) && ! $redirect_url ) {
 101              if ( $redirect_url = get_permalink(get_query_var('page_id')) )
 102                  $redirect['query'] = remove_query_arg('page_id', $redirect['query']);
 103          } elseif ( is_page() && isset($wp_query->queried_object) && 'page' == get_option('show_on_front') && $wp_query->queried_object->ID == get_option('page_on_front')  && ! $redirect_url ) {
 104              $redirect_url = home_url('/');
 105          } elseif ( is_home() && !empty($_GET['page_id']) && 'page' == get_option('show_on_front') && get_query_var('page_id') == get_option('page_for_posts')  && ! $redirect_url ) {
 106              if ( $redirect_url = get_permalink(get_option('page_for_posts')) )
 107                  $redirect['query'] = remove_query_arg('page_id', $redirect['query']);
 108          } elseif ( !empty($_GET['m']) && ( is_year() || is_month() || is_day() ) ) {
 109              $m = get_query_var('m');
 110              switch ( strlen($m) ) {
 111                  case 4: // Yearly
 112                      $redirect_url = get_year_link($m);
 113                      break;
 114                  case 6: // Monthly
 115                      $redirect_url = get_month_link( substr($m, 0, 4), substr($m, 4, 2) );
 116                      break;
 117                  case 8: // Daily
 118                      $redirect_url = get_day_link(substr($m, 0, 4), substr($m, 4, 2), substr($m, 6, 2));
 119                      break;
 120              }
 121              if ( $redirect_url )
 122                  $redirect['query'] = remove_query_arg('m', $redirect['query']);
 123          // now moving on to non ?m=X year/month/day links
 124          } elseif ( is_day() && get_query_var('year') && get_query_var('monthnum') && !empty($_GET['day']) ) {
 125              if ( $redirect_url = get_day_link(get_query_var('year'), get_query_var('monthnum'), get_query_var('day')) )
 126                  $redirect['query'] = remove_query_arg(array('year', 'monthnum', 'day'), $redirect['query']);
 127          } elseif ( is_month() && get_query_var('year') && !empty($_GET['monthnum']) ) {
 128              if ( $redirect_url = get_month_link(get_query_var('year'), get_query_var('monthnum')) )
 129                  $redirect['query'] = remove_query_arg(array('year', 'monthnum'), $redirect['query']);
 130          } elseif ( is_year() && !empty($_GET['year']) ) {
 131              if ( $redirect_url = get_year_link(get_query_var('year')) )
 132                  $redirect['query'] = remove_query_arg('year', $redirect['query']);
 133          } elseif ( is_author() && !empty($_GET['author']) && preg_match( '|^[0-9]+$|', $_GET['author'] ) ) {
 134              $author = get_userdata(get_query_var('author'));
 135              if ( false !== $author && $redirect_url = get_author_posts_url($author->ID, $author->user_nicename) )
 136                  $redirect['query'] = remove_query_arg('author', $redirect['query']);
 137          } elseif ( is_category() || is_tag() || is_tax() ) { // Terms (Tags/categories)
 138  
 139              $term_count = 0;
 140              foreach ( array('category__in', 'category__not_in', 'category__and', 'post__in', 'post__not_in',
 141              'tag__in', 'tag__not_in', 'tag__and', 'tag_slug__in', 'tag_slug__and') as $key )
 142                  $term_count += count($wp_query->query_vars[$key]);
 143  
 144              $obj = $wp_query->get_queried_object();
 145  
 146              if ( $term_count <= 1 && !empty($obj->term_id) && ( $tax_url = get_term_link((int)$obj->term_id, $obj->taxonomy) ) && !is_wp_error($tax_url) ) {
 147  
 148                  if ( is_category() ) {
 149                      $redirect['query'] = remove_query_arg( array( 'category_name', 'category', 'cat'), $redirect['query']);
 150                  } elseif ( is_tag() ) {
 151                      $redirect['query'] = remove_query_arg( array( 'tag', 'tag_id'), $redirect['query']);
 152                  } elseif ( is_tax() ) { // Custom taxonomies will have a custom query var, remove those too:
 153                      $tax = get_taxonomy( $obj->taxonomy );
 154                      if ( false !== $tax->query_var)
 155                          $redirect['query'] = remove_query_arg($tax->query_var, $redirect['query']);
 156                      else
 157                          $redirect['query'] = remove_query_arg( array( 'term', 'taxonomy'), $redirect['query']);
 158                  }
 159  
 160                  $tax_url = parse_url($tax_url);
 161                  if ( ! empty($tax_url['query']) ) { // Custom taxonomies may only be accessable via ?taxonomy=..&term=..
 162                      parse_str($tax_url['query'], $query_vars);
 163                      $redirect['query'] = add_query_arg($query_vars, $redirect['query']);
 164                  } else { // Taxonomy is accessable via a "pretty-URL"
 165                      $redirect['path'] = $tax_url['path'];
 166                  }
 167  
 168              }
 169          } elseif ( is_single() && strpos($wp_rewrite->permalink_structure, '%category%') !== false ) {
 170              $category = get_term_by('slug', get_query_var('category_name'), 'category');
 171              $post_terms = wp_get_object_terms($wp_query->get_queried_object_id(), 'category');
 172              if ( (!$category || is_wp_error($category)) || ( !is_wp_error($post_terms) && !empty($post_terms) && !in_array($category, $post_terms) ) )
 173                  $redirect_url = get_permalink($wp_query->get_queried_object_id());
 174          }
 175  
 176          // paging and feeds
 177          if ( get_query_var('paged') || is_feed() || get_query_var('cpage') ) {
 178              if ( !$redirect_url )
 179                  $redirect_url = $requested_url;
 180              $paged_redirect = @parse_url($redirect_url);
 181              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'] ) ) {
 182                  // Strip off paging and feed
 183                  $paged_redirect['path'] = preg_replace('#/page/?[0-9]+?(/+)?$#', '/', $paged_redirect['path']); // strip off any existing paging
 184                  $paged_redirect['path'] = preg_replace('#/(comments/?)?(feed|rss2?|rdf|atom)(/+|$)#', '/', $paged_redirect['path']); // strip off feed endings
 185                  $paged_redirect['path'] = preg_replace('#/comment-page-[0-9]+?(/+)?$#', '/', $paged_redirect['path']); // strip off any existing comment paging
 186              }
 187  
 188              $addl_path = '';
 189              if ( is_feed() ) {
 190                  $addl_path = !empty( $addl_path ) ? trailingslashit($addl_path) : '';
 191                  if ( get_query_var( 'withcomments' ) )
 192                      $addl_path .= 'comments/';
 193                  $addl_path .= user_trailingslashit( 'feed/' . ( ( 'rss2' ==  get_query_var('feed') || 'feed' == get_query_var('feed') ) ? '' : get_query_var('feed') ), 'feed' );
 194                  $redirect['query'] = remove_query_arg( 'feed', $redirect['query'] );
 195              }
 196  
 197              if ( get_query_var('paged') > 0 ) {
 198                  $paged = get_query_var('paged');
 199                  $redirect['query'] = remove_query_arg( 'paged', $redirect['query'] );
 200                  if ( !is_feed() ) {
 201                      if ( $paged > 1 && !is_single() ) {
 202                          $addl_path = ( !empty( $addl_path ) ? trailingslashit($addl_path) : '' ) . user_trailingslashit("page/$paged", 'paged');
 203                      } elseif ( !is_single() ) {
 204                          $addl_path = !empty( $addl_path ) ? trailingslashit($addl_path) : '';
 205                      }
 206                  } elseif ( $paged > 1 ) {
 207                      $redirect['query'] = add_query_arg( 'paged', $paged, $redirect['query'] );
 208                  }
 209              }
 210  
 211              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 ) ) ) {
 212                  $addl_path = ( !empty( $addl_path ) ? trailingslashit($addl_path) : '' ) . user_trailingslashit( 'comment-page-' . get_query_var('cpage'), 'commentpaged' );
 213                  $redirect['query'] = remove_query_arg( 'cpage', $redirect['query'] );
 214              }
 215  
 216              $paged_redirect['path'] = user_trailingslashit( preg_replace('|/index.php/?$|', '/', $paged_redirect['path']) ); // strip off trailing /index.php/
 217              if ( !empty( $addl_path ) && $wp_rewrite->using_index_permalinks() && strpos($paged_redirect['path'], '/index.php/') === false )
 218                  $paged_redirect['path'] = trailingslashit($paged_redirect['path']) . 'index.php/';
 219              if ( !empty( $addl_path ) )
 220                  $paged_redirect['path'] = trailingslashit($paged_redirect['path']) . $addl_path;
 221              $redirect_url = $paged_redirect['scheme'] . '://' . $paged_redirect['host'] . $paged_redirect['path'];
 222              $redirect['path'] = $paged_redirect['path'];
 223          }
 224      }
 225  
 226      // tack on any additional query vars
 227      $redirect['query'] = preg_replace( '#^\??&*?#', '', $redirect['query'] );
 228      if ( $redirect_url && !empty($redirect['query']) ) {
 229          if ( strpos($redirect_url, '?') !== false )
 230              $redirect_url .= '&';
 231          else
 232              $redirect_url .= '?';
 233          $redirect_url .= $redirect['query'];
 234      }
 235  
 236      if ( $redirect_url )
 237          $redirect = @parse_url($redirect_url);
 238  
 239      // www.example.com vs example.com
 240      $user_home = @parse_url(home_url());
 241      if ( !empty($user_home['host']) )
 242          $redirect['host'] = $user_home['host'];
 243      if ( empty($user_home['path']) )
 244          $user_home['path'] = '/';
 245  
 246      // Handle ports
 247      if ( !empty($user_home['port']) )
 248          $redirect['port'] = $user_home['port'];
 249      else
 250          unset($redirect['port']);
 251  
 252      // trailing /index.php
 253      $redirect['path'] = preg_replace('|/index.php/*?$|', '/', $redirect['path']);
 254  
 255      // Remove trailing spaces from the path
 256      $redirect['path'] = preg_replace( '#(%20| )+$#', '', $redirect['path'] );
 257  
 258      if ( !empty( $redirect['query'] ) ) {
 259          // Remove trailing spaces from certain terminating query string args
 260          $redirect['query'] = preg_replace( '#((p|page_id|cat|tag)=[^&]*?)(%20| )+$#', '$1', $redirect['query'] );
 261  
 262          // Clean up empty query strings
 263          $redirect['query'] = trim(preg_replace( '#(^|&)(p|page_id|cat|tag)=?(&|$)#', '&', $redirect['query']), '&');
 264  
 265          // Remove redundant leading ampersands
 266          $redirect['query'] = preg_replace( '#^\??&*?#', '', $redirect['query'] );
 267      }
 268  
 269      // strip /index.php/ when we're not using PATHINFO permalinks
 270      if ( !$wp_rewrite->using_index_permalinks() )
 271          $redirect['path'] = str_replace('/index.php/', '/', $redirect['path']);
 272  
 273      // trailing slashes
 274      if ( is_object($wp_rewrite) && $wp_rewrite->using_permalinks() && !is_404() && (!is_front_page() || ( is_front_page() && (get_query_var('paged') > 1) ) ) ) {
 275          $user_ts_type = '';
 276          if ( get_query_var('paged') > 0 ) {
 277              $user_ts_type = 'paged';
 278          } else {
 279              foreach ( array('single', 'category', 'page', 'day', 'month', 'year', 'home') as $type ) {
 280                  $func = 'is_' . $type;
 281                  if ( call_user_func($func) ) {
 282                      $user_ts_type = $type;
 283                      break;
 284                  }
 285              }
 286          }
 287          $redirect['path'] = user_trailingslashit($redirect['path'], $user_ts_type);
 288      } elseif ( is_front_page() ) {
 289          $redirect['path'] = trailingslashit($redirect['path']);
 290      }
 291  
 292      // Strip multiple slashes out of the URL
 293      if ( strpos($redirect['path'], '//') > -1 )
 294          $redirect['path'] = preg_replace('|/+|', '/', $redirect['path']);
 295  
 296      // Always trailing slash the Front Page URL
 297      if ( trailingslashit( $redirect['path'] ) == trailingslashit( $user_home['path'] ) )
 298          $redirect['path'] = trailingslashit($redirect['path']);
 299  
 300      // Ignore differences in host capitalization, as this can lead to infinite redirects
 301      // Only redirect no-www <=> yes-www
 302      if ( strtolower($original['host']) == strtolower($redirect['host']) ||
 303          ( strtolower($original['host']) != 'www.' . strtolower($redirect['host']) && 'www.' . strtolower($original['host']) != strtolower($redirect['host']) ) )
 304          $redirect['host'] = $original['host'];
 305  
 306      $compare_original = array($original['host'], $original['path']);
 307  
 308      if ( !empty( $original['port'] ) )
 309          $compare_original[] = $original['port'];
 310  
 311      if ( !empty( $original['query'] ) )
 312          $compare_original[] = $original['query'];
 313  
 314      $compare_redirect = array($redirect['host'], $redirect['path']);
 315  
 316      if ( !empty( $redirect['port'] ) )
 317          $compare_redirect[] = $redirect['port'];
 318  
 319      if ( !empty( $redirect['query'] ) )
 320          $compare_redirect[] = $redirect['query'];
 321  
 322      if ( $compare_original !== $compare_redirect ) {
 323          $redirect_url = $redirect['scheme'] . '://' . $redirect['host'];
 324          if ( !empty($redirect['port']) )
 325              $redirect_url .= ':' . $redirect['port'];
 326          $redirect_url .= $redirect['path'];
 327          if ( !empty($redirect['query']) )
 328              $redirect_url .= '?' . $redirect['query'];
 329      }
 330  
 331      if ( !$redirect_url || $redirect_url == $requested_url )
 332          return false;
 333  
 334      // Note that you can use the "redirect_canonical" filter to cancel a canonical redirect for whatever reason by returning FALSE
 335      $redirect_url = apply_filters('redirect_canonical', $redirect_url, $requested_url);
 336  
 337      if ( !$redirect_url || $redirect_url == $requested_url ) // yes, again -- in case the filter aborted the request
 338          return false;
 339  
 340      if ( $do_redirect ) {
 341          // protect against chained redirects
 342          if ( !redirect_canonical($redirect_url, false) ) {
 343              wp_redirect($redirect_url, 301);
 344              exit();
 345          } else {
 346              // Debug
 347              // die("1: $redirect_url<br />2: " . redirect_canonical( $redirect_url, false ) );
 348              return false;
 349          }
 350      } else {
 351          return $redirect_url;
 352      }
 353  }
 354  
 355  /**
 356   * Attempts to guess correct post based on query vars.
 357   *
 358   * @since 2.3.0
 359   * @uses $wpdb
 360   *
 361   * @return bool|string Returns False, if it can't find post, returns correct
 362   *        location on success.
 363   */
 364  function redirect_guess_404_permalink() {
 365      global $wpdb;
 366  
 367      if ( !get_query_var('name') )
 368          return false;
 369  
 370      $where = $wpdb->prepare("post_name LIKE %s", get_query_var('name') . '%');
 371  
 372      // if any of post_type, year, monthnum, or day are set, use them to refine the query
 373      if ( get_query_var('post_type') )
 374          $where .= $wpdb->prepare(" AND post_type = %s", get_query_var('post_type'));
 375      if ( get_query_var('year') )
 376          $where .= $wpdb->prepare(" AND YEAR(post_date) = %d", get_query_var('year'));
 377      if ( get_query_var('monthnum') )
 378          $where .= $wpdb->prepare(" AND MONTH(post_date) = %d", get_query_var('monthnum'));
 379      if ( get_query_var('day') )
 380          $where .= $wpdb->prepare(" AND DAYOFMONTH(post_date) = %d", get_query_var('day'));
 381  
 382      $post_id = $wpdb->get_var("SELECT ID FROM $wpdb->posts WHERE $where AND post_status = 'publish'");
 383      if ( !$post_id )
 384          return false;
 385      return get_permalink($post_id);
 386  }
 387  
 388  add_action('template_redirect', 'redirect_canonical');
 389  
 390  ?>


Generated: Mon Apr 5 14:26:09 2010 Cross-referenced by PHPXref 0.7