404-solution / includes / InternalLinkScanner.php

InternalLinkScanner.php in 404 Solution 4.2.0, at includes/InternalLinkScanner.php

238 lines 8.3 KB

Raw Download Zip

1	<?php
2
3	if (!defined('ABSPATH')) {
4	exit;
5	}
6
7	/**
8	* Proactive Internal Link Scanner.
9	*
10	* Scans published post/page content for broken internal links — links whose
11	* relative path matches a URL captured as a 404 in the plugin's redirect table.
12	*
13	* No HTTP requests are made: this is entirely database + post_content inspection.
14	*/
15	class ABJ_404_Solution_InternalLinkScanner {
16
17	/** Transient key used to cache scan results. */
18	const TRANSIENT_KEY = 'abj404_broken_links_scan';
19
20	/** Cache TTL in seconds (6 hours). */
21	const CACHE_TTL = 21600;
22
23	/** Maximum posts to inspect in a single batch to guard large sites. */
24	const BATCH_LIMIT = 1000;
25
26	/**
27	* Scan all published posts/pages for broken internal links.
28	*
29	* Broken = URL appears in the captured-404 list (status = ABJ404_STATUS_CAPTURED,
30	* disabled = 0). External links are ignored.
31	*
32	* @return array<int, array{post_id: int, post_title: string, broken_url: string, hit_count: int}>
33	*/
34	public function scanForBrokenLinks(): array {
35	// 1. Collect captured 404 URLs from the redirects table.
36	$capturedUrls = $this->getCapturedUrlSet();
37	if (empty($capturedUrls)) {
38	return array();
39	}
40
41	// 2. Get published posts/pages in batches (guard against very large sites).
42	$siteHomeUrl = function_exists('home_url') ? (string)home_url() : '';
43	$results = array();
44	$offset = 0;
45
46	do {
47	$posts = $this->fetchPublishedPosts($offset, self::BATCH_LIMIT);
48	if (empty($posts)) {
49	break;
50	}
51
52	foreach ($posts as $post) {
53	// Boundary normalizer: WP_Post shape-probing lives in the VO.
54	$ref = ABJ_404_Solution_PostRef::fromWpPost($post);
55	if ($ref === null) {
56	continue;
57	}
58	$content = $ref->getContent();
59	$postId = $ref->getId();
60	$postTitle = $ref->getTitle();
61
62	if ($content === '') {
63	continue;
64	}
65
66	// Extract all href values.
67	preg_match_all('/href=["\']([^"\']+)["\']/', $content, $matches);
68	$hrefs = $matches[1];
69
70	foreach ($hrefs as $href) {
71	$href = trim($href);
72	if ($href === '') {
73	continue;
74	}
75
76	// Normalize: strip the home_url prefix to get a relative path.
77	$relative = $this->normalizeToRelative($href, $siteHomeUrl);
78
79	// Skip purely external links (those we could not strip to a relative path).
80	if ($relative === null) {
81	continue;
82	}
83
84	// Check if this relative path is a captured 404.
85	if (!isset($capturedUrls[$relative])) {
86	continue;
87	}
88
89	$results[] = array(
90	'post_id' => $postId,
91	'post_title' => $postTitle,
92	'broken_url' => $relative,
93	'hit_count' => $capturedUrls[$relative],
94	);
95	}
96	}
97
98	$offset += count($posts);
99	} while (count($posts) >= self::BATCH_LIMIT);
100
101	return $results;
102	}
103
104	/**
105	* Run a scan and cache results in a transient with a 6-hour TTL.
106	* Called nightly by the maintenance cron.
107	* @return void
108	*/
109	public function runNightlyScan(): void {
110	$results = $this->scanForBrokenLinks();
111	set_transient(self::TRANSIENT_KEY, $results, self::CACHE_TTL);
112	}
113
114	/**
115	* Get cached scan results.
116	*
117	* @return array<int, array{post_id: int, post_title: string, broken_url: string, hit_count: int}>\|false
118	* Cached results array, or false if no cache exists.
119	*/
120	public function getCachedResults() {
121	$result = get_transient(self::TRANSIENT_KEY);
122	return is_array($result) ? $result : false;
123	}
124
125	/**
126	* Build a map of captured-404 URLs to their hit counts from the redirects table.
127	*
128	* @return array<string, int> Keyed by relative URL, value = hit count estimate.
129	*/
130	private function getCapturedUrlSet(): array {
131	global $wpdb;
132
133	$capturedStatus = defined('ABJ404_STATUS_CAPTURED') ? intval(ABJ404_STATUS_CAPTURED) : 3;
134
135	// Use the plugin's DatabaseCore if available, otherwise fall back to strtolower prefix.
136	$dbCore = null;
137	if (class_exists('ABJ_404_Solution_DatabaseCore')) {
138	$dbCore = abj_service('db_core');
139	$redirectsTable = $dbCore->doTableNameReplacements('{wp_abj404_redirects}');
140	} else {
141	$redirectsTable = strtolower($wpdb->prefix) . 'abj404_redirects';
142	}
143
144	$sql = "SELECT `url` FROM `{$redirectsTable}` WHERE `status` = %d AND `disabled` = 0";
145
146	// Route through queryAndGetResults() so this nightly-cron scan inherits
147	// the centralized timeout, retry, and corrupted-table recovery instead
148	// of failing silently or hanging. Only fall back to the raw $wpdb path
149	// when the DatabaseCore class is unavailable (e.g. integration test bootstraps).
150	if ($dbCore !== null) {
151	$result = $dbCore->queryAndGetResults($sql, array(
152	'query_params' => array($capturedStatus),
153	));
154	if (!empty($result['timed_out']) \|\|
155	(isset($result['last_error']) && $result['last_error'] != '')) {
156	return array();
157	}
158	$rows = is_array($result['rows'] ?? null) ? $result['rows'] : array();
159	} else {
160	$prepared = $wpdb->prepare($sql, $capturedStatus);
161	// DAO-bypass-approved: Test-environment fallback — primary path goes through queryAndGetResults at :149
162	$rows = $wpdb->get_results($prepared, ARRAY_A);
163	if (!is_array($rows)) {
164	return array();
165	}
166	}
167
168	$urlSet = array();
169	foreach ($rows as $row) {
170	if (!is_array($row) \|\| !isset($row['url'])) {
171	continue;
172	}
173	$url = (string)$row['url'];
174	if ($url !== '') {
175	// Use 0 as a placeholder hit count; real hit counts would need a join.
176	$urlSet[$url] = 0;
177	}
178	}
179
180	return $urlSet;
181	}
182
183	/**
184	* Fetch a batch of published posts and pages.
185	*
186	* @param int $offset
187	* @param int $limit
188	* @return array<int, object>
189	*/
190	private function fetchPublishedPosts(int $offset, int $limit): array {
191	if (!function_exists('get_posts')) {
192	return array();
193	}
194
195	$posts = get_posts(array(
196	'post_type' => array('post', 'page'),
197	'post_status' => 'publish',
198	'posts_per_page' => $limit,
199	'offset' => $offset,
200	'fields' => 'all',
201	// Suppress filters so we get raw post_content.
202	'suppress_filters' => true,
203	));
204
205	return $posts;
206	}
207
208	/**
209	* Normalize a URL to a relative path by stripping the home_url prefix.
210	*
211	* Returns null when the URL is purely external (not on this site).
212	*
213	* @param string $href The href value extracted from post content.
214	* @param string $siteHome The result of home_url() for this site.
215	* @return string\|null Relative path (e.g. "/old-page/") or null for external links.
216	*/
217	private function normalizeToRelative(string $href, string $siteHome): ?string {
218	// Already a relative path.
219	if (strpos($href, '/') === 0 && strpos($href, '//') !== 0) {
220	return $href;
221	}
222
223	// Absolute URL: strip the home_url prefix if it matches this site.
224	if ($siteHome !== '' && strpos($href, $siteHome) === 0) {
225	$relative = substr($href, strlen($siteHome));
226	return ($relative === '') ? '/' : $relative;
227	}
228
229	// Fragment-only or javascript: — not a real internal link.
230	if (strpos($href, '#') === 0 \|\| strpos($href, 'javascript:') === 0 \|\| strpos($href, 'mailto:') === 0) {
231	return null;
232	}
233
234	// Anything else is an external URL.
235	return null;
236	}
237	}
238