wp-410 / wp-410.php

wp-410.php in HTTP 410 (Gone) responses 0.7.1, at wp-410.php

355 lines 13.5 KB

Raw Download Zip

1	<?php
2	/*
3	Plugin Name: 410 for WordPress
4	Plugin URI: http://rayofsolaris.net/code/410-for-wordpress
5	Description: Sends HTTP 410 (Gone) responses to requests for pages that no longer exist on your blog.
6	Version: 0.7.1
7	Author: Samir Shah
8	Author URI: http://rayofsolaris.net/
9	License: GPL2
10	*/
11
12	if( ! defined( 'ABSPATH' ) )
13	exit;
14
15	class WP_410 {
16	const db_version = 5;
17	private $permalinks;
18	private $table;
19
20	function __construct() {
21	$this->permalinks = (bool) get_option('permalink_structure');
22	$this->table = $GLOBALS['wpdb']->prefix . '410_links';
23
24	add_action( 'plugins_loaded', array( $this, 'upgrade_check' ) );
25
26	if( is_admin() )
27	add_action( 'admin_menu', array( $this, 'settings_menu' ) );
28	else
29	add_action( 'template_redirect', array( $this, 'check_for_410' ) );
30
31	// these could theoretically happen both with/without is_admin()
32	add_action('delete_post', array( $this, 'note_deleted_post' ) );
33	add_action('wp_insert_post', array( $this, 'note_inserted_post' ));
34
35	}
36
37	private function install_table() {
38	// remember, two spaces after PRIMARY KEY otherwise WP borks
39	$sql = "CREATE TABLE $this->table (
40	gone_id MEDIUMINT unsigned NOT NULL AUTO_INCREMENT,
41	gone_key VARCHAR(512) NOT NULL,
42	gone_regex VARCHAR(512) NOT NULL,
43	is_404 SMALLINT(1) unsigned NOT NULL DEFAULT 0,
44	PRIMARY KEY (gone_id),
45	KEY is_404 (is_404)
46	);";
47
48	require_once(ABSPATH . 'wp-admin/includes/upgrade.php');
49	dbDelta($sql);
50	}
51
52	private function get_links(){
53	global $wpdb;
54	return $wpdb->get_results( "SELECT gone_key, gone_regex FROM $this->table WHERE is_404 = 0", OBJECT_K ); // indexed by gone_key
55	}
56
57	private function get_404s(){
58	global $wpdb;
59	$links = $wpdb->get_results( "SELECT gone_key, gone_regex FROM $this->table WHERE is_404 = 1", OBJECT_K ); // indexed by gone_key
60	// limit how many 404s we store
61	if( count( $links ) > 50 ) {
62	$n = count( $links ) - 50;
63	$wpdb->query( "DELETE FROM `$this->table` WHERE is_404 = 1 ORDER BY gone_id LIMIT $n" );
64	return $this->get_404s();
65	}
66	return $links;
67	}
68
69	private function add_link( $key, $is_404 = false ){ // just supply the link
70	global $wpdb;
71	// build regex
72	$parts = preg_split( '/(\*)/', $key, -1, PREG_SPLIT_DELIM_CAPTURE \| PREG_SPLIT_NO_EMPTY );
73	foreach( $parts as &$part ) if( '*' != $part )
74	$part = preg_quote( $part, '\|' );
75	$parts = str_replace( '', '.', $parts );
76	$regex = '\|^' . implode( '', $parts ) . '$\|i';
77
78	// avoid duplicates - messy but MySQL doesn't allow url-length unique keys
79	if( $wpdb->get_var( $wpdb->prepare( "SELECT COUNT(*) FROM `$this->table` WHERE `gone_key` = %s", $key ) ) )
80	return 0;
81
82	return $wpdb->insert( $this->table, array( 'gone_key' => $key, 'gone_regex' => $regex, 'is_404' => intval( $is_404 ) ) );
83	}
84
85	private function convert_404( $key ) {
86	global $wpdb;
87	return $wpdb->query( $wpdb->prepare( "UPDATE `$this->table` SET is_404 = 0 WHERE `gone_key` = %s LIMIT 1", $key ) );
88	}
89
90	private function remove_link( $key ){
91	global $wpdb;
92	return $wpdb->query( $wpdb->prepare( "DELETE FROM $this->table WHERE gone_key = %s", array( $key ) ) );
93	}
94
95	function upgrade_check() {
96	$options_version = get_option( 'wp_410_options_version', 0 );
97
98	if( $options_version == self::db_version ) // nothing to do
99	return;
100
101	// last db change was in version 5
102	if( $options_version < 5 )
103	$this->install_table();
104
105	if( $options_version < 3 ) {
106	$old_links = get_option( 'wp_410_links_list', array() );
107	$new_links = array(); // just a simple array of links
108
109	if( 0 == $options_version ) // links were stored just as links
110	$new_links = array_map( 'rawurldecode', $old_links );
111	elseif( 1 == $options_version ) // links were stored as array( link => regex ), We only need the link
112	$new_links = array_map( 'rawurldecode', array_keys( $old_links ) );
113	else // moved to using the database in db_version 3
114	$new_links = array_keys( $old_links );
115
116	foreach( $new_links as $link )
117	$this->add_link( $link );
118
119	delete_option( 'wp_410_links_list' ); // remove old option
120	}
121
122	update_option( 'wp_410_options_version', self::db_version );
123	}
124
125	function settings_menu() {
126	add_submenu_page('plugins.php', '410 for WordPress', '410 for WordPress', 'manage_options', 'wp_410_settings', array( $this, 'settings_page') );
127	}
128
129	function settings_page() {
130	$links = $this->get_links();
131	$logged_404s = $this->get_404s();
132	$links_to_add = array();
133
134	if ( isset( $_POST['submit'] ) ) {
135	$failed_to_add = array();
136	if( !empty( $_POST['links_to_add'] ) ) {
137	foreach( preg_split( '/(\r?\n)+/', $_POST['links_to_add'], -1, PREG_SPLIT_NO_EMPTY ) as $link ) {
138	if( $this->is_valid_url( $link ) ) {
139	$this->add_link( $link );
140	$links[$link] = $link; // the value doesn't matter, we don't use it here
141	}
142	else {
143	$failed_to_add[] = '<code>' . htmlspecialchars( $link ) . '</code>';
144	}
145	}
146	}
147
148	if( !empty( $_POST['add_404s'] ) ) {
149	foreach( $_POST['add_404s'] as $link ) {
150	$this->convert_404( $link );
151	$links[$link] = $link; // the value doesn't matter, we don't use it here
152	unset( $logged_404s[$link] );
153	}
154	}
155
156	if( !empty( $_POST['old_links_to_remove'] ) ) {
157	foreach( $_POST['old_links_to_remove'] as $key ) {
158	if( isset( $links[$key] ) ) {
159	$this->remove_link( $key );
160	unset( $links[$key] );
161	}
162	}
163	}
164
165	echo '<div id="message" class="updated fade"><p>Options updated.</p></div>';
166
167	if( $failed_to_add )
168	echo '<div class="error"><p>The following entries could not be recognised as URLs that your WordPress site handles, and were not added to the list. This can be because the domain name and path does not match that of your WordPress site, or because pretty permalinks are disabled.</p><p>- ' . implode( '<br> - ', $failed_to_add ) .'</p></div>';
169	}
170
171	ksort( $links );
172	ksort( $logged_404s );
173	?>
174	<style>
175	.indent {padding-left: 2em}
176	.wp_gone_deleted label {text-decoration: line-through}
177	label.invalid {color: red}
178	p.invalid { background: #FFEBE8; border: 1px solid red; border-radius: 5px; padding: 5px 10px}
179	#wp_410_table th {font-weight: bold}
180	#wp_410_table tr {border-top: 1px solid #EEE}
181	</style>
182	<?php screen_icon(); ?>
183	<div class="wrap">
184	<h2>410 for WordPress</h2>
185	<?php
186	if( WP_CACHE )
187	echo '<div class="error"><p><strong>It seems that a caching/performance plugin is active on this site. Please note that with most caching plugins, WordPress does not handle requests for pages that are not found on your site (your web server handles them directly), and so this plugin cannot intercept them to issue a 410 response.</strong></div>';
188	?>
189	<p>This plugin will issue a HTTP 410 response to articles that no longer exist on your blog. When an article is deleted, it records this and issues a 410 response when pages unique to that article are requested. This HTTP code informs robots visiting your site that the requested page has been permanently removed, and that they should stop trying to access it.</p>
190	<p><strong>Note: this plugin will only return a 410 response if WordPress has not found something valid to display for the requested URL.</strong></p>
191	<form action="" method="post" id="wp-gone-captcha-settings">
192	<table id="wp_410_table" class="form-table"><tbody>
193	<tr><th>Obsolete URLs</th><td>
194	<?php
195	if( empty( $links ) ) {
196	echo '<p>There are currently no obsolete URLs in this list. You can add some manually below.</p>';
197	}
198	else {
199	echo '<p>The following URLs are currently considered obsolete. To remove items from the list, check the corresponding box.</p>';
200	echo '<ul id="wp_gone_old_links" class="indent">';
201
202	$invalid_links_exist = false;
203	foreach( array_keys( $links ) as $k ) {
204	$valid = $this->is_valid_url( $k );
205
206	if( ! $valid )
207	$invalid_links_exist = true;
208
209	$k = htmlspecialchars( $k );
210	$class = $valid ? '' : 'class="invalid"';
211
212	echo "<li><input type='checkbox' name='old_links_to_remove[]' value='$k' id='old_link_$k' /> <label for='old_link_$k' $class><code>$k</code></label></li>";
213	}
214	echo '</ul>';
215
216	if( $invalid_links_exist )
217	echo '<p class="invalid">Warning: WordPress is not able to issue 410 responses for the URLs marked in red above. This is because those URLs are not handled by your WordPress installation. This can be because the domain name and path does not match that of your WordPress site, or because pretty permalinks are disabled.</p>';
218	}
219	?>
220	<p><small>If you create or update an article whose URL matches one of those above, that URL will automatically be removed from this list.</small></p>
221	</td></tr>
222	<tr><th>Recent 404 errors</th><td>
223
224	<?php
225	if( empty( $logged_404s ) ) {
226	echo '<p>This plugin will automatically log recent 404 (Page Not Found) errors on your site, and show them here, so that you can easily add them to the list above. There are currently no 404 errors reported.</p>';
227	}
228	else {
229	echo '<p>Below are recent 404 (Page Not Found) errors that have occurred on your site. If you want to issue a 410 response for any of these requests, check the corresponding box.</p>';
230	echo '<ul id="wp_gone_404s" class="indent">';
231	foreach( array_keys( $logged_404s ) as $k ) {
232	$k = htmlspecialchars( $k );
233	echo "<li><input type='checkbox' name='add_404s[]' value='$k' id='404_$k' /> <label for='404_$k'><code>$k</code></label></li>";
234	}
235	echo '</ul>';
236	}
237	?>
238	</td></tr>
239	<tr><th>Manually add URLs</th><td>
240	<p>You can manually add items to the list by entering them below. Please enter one <strong>fully qualified</strong> URL per line.</p>
241	<p>Use <code></code> as a wildcard character. So <code>http://www.example.com//music/</code> will match all URLs ending in <code>/music/</code>.</p>
242	<textarea name="links_to_add" rows="8" cols="80"></textarea>
243	</td></tr>
244	<tr><th>410 reponse message</th><td>
245	<p>By default, the plugin issues the following plain-text message as part of the 410 response: <code>Sorry, the page you requested has been permanently removed.</code></p>
246	<?php
247	if( locate_template( '410.php' ) )
248	echo '<p><strong>A template file <code>410.php</code> has been detected in your theme directory. This file will be used to display 410 responses.</strong> To revert back to the default message, remove the file from your theme directory.</p>';
249	else
250	echo '<p>If you would like to use your own template instead, simply place a file called <code>410.php</code> in your theme directory, containing your template. Have a look at your theme\'s <code>404.php</code> template to see what it should look like.</p>';
251	?>
252	</td></tr>
253	</tbody></table>
254	<p class="submit"><input class="button-primary" type="submit" name="submit" value="Update settings" /></p>
255	</form>
256	</div>
257	<script>
258	jQuery(document).ready(function($){
259	$('#wp_gone_old_links input').change( function(){
260	$(this).parent().toggleClass('wp_gone_deleted', $(this).is(':checked'));
261	});
262	$("#wp-gone-captcha-settings input").change( function(){
263	$("#message").slideUp('slow');
264	});
265	});
266	</script>
267	<?php
268	}
269
270	private function is_valid_url ( $link ) {
271	// Determine whether WP will handle a request for this URL
272	$home = home_url();
273	if( strpos( $link, $home ) !== 0 )
274	return false;
275
276	if( !$this->permalinks ) {
277	$req = preg_replace( '^\|' . preg_quote( $home, '\|' ) . '/?\|' , '', $link );
278	if( strlen( $req ) && $req[0] != '?' ) // this is a pretty permalink, but pretty permalinks are disabled
279	return false;
280	}
281
282	return true;
283	}
284
285	function note_deleted_post( $id ) {
286	// NOTE: function wp_delete_post calls the delete_post action twice, so we store deleted IDs in an array to avoid repeating
287	// but still allow repeated use of the function for batch deletes
288	static $noted = array();
289	$id = (int) $id;
290
291	if( in_array( $id, $noted ) )
292	return;
293	$noted[] = $id;
294
295	$post = get_post( $id );
296
297	if( 'revision' == $post->post_type )
298	return;
299
300	$link = rawurldecode( get_permalink( $id ) );
301	if( $this->permalinks )
302	$link .= '*'; // catch everything like /page/ and /feed/
303
304	$this->add_link( $link );
305	}
306
307	function note_inserted_post( $id ) {
308	$post = get_post( $id );
309
310	if( 'revision' == $post->post_type \|\| 'draft' == $post->post_status )
311	return;
312
313	// Check our list of URLs against the new/updated post's permalink, and if they match, scratch it from our list
314	$created_links = array();
315
316	$created_links[] = rawurldecode( get_permalink( $id ) );
317	$created_links[] = get_post_comments_feed_link( $id ); // back compat
318
319	if( $this->permalinks )
320	$created_links[0] .= '*';
321
322	$list = get_option( 'wp_410_links_list', array() );
323
324	foreach( $created_links as $link )
325	$this->remove_link( $link );
326	}
327
328	function check_for_410() {
329	// Don't mess if Wordpress has found something to display
330	if( !is_404() )
331	return;
332
333	$links = $this->get_links();
334	$req = ( is_ssl() ? 'https://' : 'http://' ) . $_SERVER['HTTP_HOST'] . $_SERVER['REQUEST_URI'];
335	$req = rawurldecode( $req );
336
337	foreach( $links as $link ) {
338	if( @preg_match( $link->gone_regex, $req ) ) {
339	status_header( 410 );
340	do_action( 'wp_410_response' ); // you can use this to customise the response message
341
342	if( ! locate_template( '410.php', true ) )
343	echo 'Sorry, the page you requested has been permanently removed.';
344
345	exit;
346	}
347	}
348
349	// no hit, log 404
350	$this->add_link( $req, true );
351	}
352	}
353
354	new WP_410();
355