404-solution / includes / PermalinkCache.php

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

306 lines 10.7 KB

Raw Download Zip

1	<?php
2
3
4	if (!defined('ABSPATH')) {
5	exit;
6	}
7
8	/* Functions in this class should only be for plugging into WordPress listeners (filters, actions, etc). */
9
10	class ABJ_404_Solution_PermalinkCache {
11
12	/** The name of the hook to use in WordPress. */
13	const UPDATE_PERMALINK_CACHE_HOOK = 'abj404_updatePermalinkCacheAction';
14
15	/** The maximum number of times in a row to run the hook. */
16	const MAX_EXECUTIONS = 15;
17
18	/** @var self\|null */
19	private static $instance = null;
20
21	/** @var ABJ_404_Solution_ContentRepository */
22	private $contentRepository;
23
24	/** @var mixed */
25	private $statsRepository;
26
27	/** @var ABJ_404_Solution_Logging */
28	private $logger;
29
30	/** @var ABJ_404_Solution_PluginLogic */
31	private $logic;
32
33	/**
34	* Constructor with dependency injection.
35	*
36	* @param ABJ_404_Solution_ContentRepository\|null $contentRepository Content repository
37	* @param ABJ_404_Solution_Logging\|null $logging Logging service
38	* @param ABJ_404_Solution_PluginLogic\|null $pluginLogic Business logic service
39	* @param ABJ_404_Solution_StatsRepository\|null $statsRepository Stats repository
40	*/
41	public function __construct($contentRepository = null, $logging = null, $pluginLogic = null, $statsRepository = null) {
42	// Use injected dependencies or fall back to getInstance() for backward compatibility
43	$this->contentRepository = $contentRepository !== null ? $contentRepository : abj_service('content_repository');
44	$this->logger = $logging !== null ? $logging : abj_service('logging');
45	$this->logic = $pluginLogic !== null ? $pluginLogic : abj_service('plugin_logic');
46	$this->statsRepository = $statsRepository !== null ? $statsRepository :
47	(is_object($contentRepository) && method_exists($contentRepository, 'getPostsNeedingContentKeywords') ? $contentRepository : abj_service('stats_repository'));
48	}
49
50	/** @return self */
51	public static function getInstance(): self {
52	if (self::$instance == null) {
53	self::$instance = new ABJ_404_Solution_PermalinkCache();
54	}
55
56	return self::$instance;
57	}
58
59	/** @return void */
60	static function init(): void {
61	$me = abj_service('permalink_cache');
62
63	add_action('updated_option', array($me, 'permalinkStructureChanged'), 10, 2);
64	}
65
66	/** If the permalink structure changes then truncate the cache table and update some values.
67	* @global type $abj404logging
68	* @param string $var1
69	* @param string $newStructure
70	*/
71	/**
72	* @param string $var1
73	* @param string $newStructure
74	* @return void
75	*/
76	function permalinkStructureChanged($var1, $newStructure): void {
77	if ($var1 != 'permalink_structure') {
78	return;
79	}
80
81	// we need to truncate the permlink cache since the structure changed
82
83	$this->logger->debugMessage(__CLASS__ . "/" . __FUNCTION__ .
84	": Truncating and updating permalink cache because the permalink structure changed to " .
85	$newStructure);
86
87	$this->contentRepository->truncatePermalinkCacheTable();
88
89	// let's take this opportunity to update some of the values in the cache table.
90	$this->updatePermalinkCache(1);
91	}
92
93	/**
94	* @param int $maxExecutionTime
95	* @param int $executionCount
96	* @return int
97	* @throws Exception
98	*/
99	function updatePermalinkCache($maxExecutionTime, $executionCount = 1) {
100	// check to see if we need to upgrade the database.
101	// we must pass "true" here to avoid an infinite loop when updating the database.
102	$this->logic->getOptions(true);
103
104	// insert the new rows.
105	$results = $this->contentRepository->updatePermalinkCache();
106	$rowsInserted = (is_array($results) && isset($results['rows_affected']) && is_int($results['rows_affected'])) ? $results['rows_affected'] : 0;
107
108	// Invalidate coverage ratio if rows were inserted (new permalinks may lack N-grams)
109	if ($rowsInserted > 0) {
110	abj_service('ngram_filter')->invalidateCoverageCaches();
111	}
112
113	// now we have to update the the pages that have parents to include the parent
114	// part of the URL.
115	// wherever the post_parent != 0, prepend the parent ID URL onto the current URL
116	// and update the post_parent to be the parent ID of the parent.
117	$this->contentRepository->updatePermalinkCacheParentPages();
118
119	$this->populateContentKeywords();
120
121	$this->checkPermalinkCacheStaleness();
122
123	return $rowsInserted;
124	}
125
126	/** @return void */
127	private function checkPermalinkCacheStaleness(): void {
128	$cacheCount = $this->contentRepository->getPermalinkCacheCount();
129	if ($cacheCount > 0) {
130	return;
131	}
132	$postCount = function_exists('wp_count_posts') ? (int) (wp_count_posts('post')->publish ?? 0) : 0;
133	$pageCount = function_exists('wp_count_posts') ? (int) (wp_count_posts('page')->publish ?? 0) : 0;
134	if ($postCount + $pageCount === 0) {
135	return;
136	}
137	$message = function_exists('__')
138	? __('Permalink cache appears empty after rebuild — suggestions may be degraded. Try rebuilding again or check available disk space.', '404-solution')
139	: 'Permalink cache appears empty after rebuild — suggestions may be degraded. Try rebuilding again or check available disk space.';
140	if (function_exists('set_transient')) {
141	// allow-cache-empty: notice payload is constructed locally and intentionally persisted as-is.
142	set_transient('abj404_plugin_db_notice', array(
143	'type' => 'stale_permalink_cache',
144	'message' => $message,
145	'timestamp' => time(),
146	), 86400);
147	}
148	}
149
150	/**
151	* @param int $executionCount
152	* @return void
153	*/
154	function scheduleToRunAgain(int $executionCount): void {
155	$maxExecutionTime = (int)ini_get('max_execution_time') - 5;
156	$maxExecutionTime = max($maxExecutionTime, 25);
157
158	wp_schedule_single_event(1, ABJ_404_Solution_PermalinkCache::UPDATE_PERMALINK_CACHE_HOOK,
159	array($maxExecutionTime, $executionCount));
160	}
161
162	/** Maximum unique keywords to store per post. */
163	const MAX_CONTENT_KEYWORDS = 30;
164
165	/** Minimum word length to keep during keyword extraction. */
166	const MIN_KEYWORD_LENGTH = 3;
167
168	/**
169	* Populate content_keywords for permalink cache rows that have NULL.
170	*
171	* Reads post_content, strips HTML/shortcodes, filters stop words,
172	* keeps top keywords by frequency. Runs in batches to stay within
173	* PHP time limits.
174	*
175	* @param int $batchSize Maximum posts to process per call.
176	* @return int Number of rows updated.
177	*/
178	function populateContentKeywords(int $batchSize = 500): int {
179	$rows = $this->getPostsNeedingContentKeywords($batchSize);
180
181	if (empty($rows)) {
182	return 0;
183	}
184
185	$idToKeywords = array();
186	foreach ($rows as $row) {
187	if (!is_object($row)) {
188	continue;
189	}
190	$id = isset($row->id) ? (int)$row->id : 0;
191	if ($id <= 0) {
192	continue;
193	}
194	$content = isset($row->post_content) && is_string($row->post_content) ? $row->post_content : '';
195	$idToKeywords[$id] = self::extractContentKeywords($content);
196	}
197
198	if (empty($idToKeywords)) {
199	return 0;
200	}
201
202	$this->bulkUpdateContentKeywords($idToKeywords);
203
204	return count($idToKeywords);
205	}
206
207	/**
208	* @param int $batchSize
209	* @return array<int, mixed>
210	*/
211	private function getPostsNeedingContentKeywords(int $batchSize): array {
212	if (!is_object($this->statsRepository) \|\| !method_exists($this->statsRepository, 'getPostsNeedingContentKeywords')) {
213	return array();
214	}
215	$rows = call_user_func(array($this->statsRepository, 'getPostsNeedingContentKeywords'), $batchSize);
216	return is_array($rows) ? $rows : array();
217	}
218
219	/**
220	* @param array<int, string> $idToKeywords
221	* @return void
222	*/
223	private function bulkUpdateContentKeywords(array $idToKeywords): void {
224	if (!is_object($this->statsRepository) \|\| !method_exists($this->statsRepository, 'bulkUpdateContentKeywords')) {
225	return;
226	}
227	call_user_func(array($this->statsRepository, 'bulkUpdateContentKeywords'), $idToKeywords);
228	}
229
230	/**
231	* Extract significant keywords from HTML post content.
232	*
233	* 1. Strip shortcodes ([shortcode attr=val]...[/shortcode] and [self-closing])
234	* 2. Strip HTML tags
235	* 3. Decode HTML entities
236	* 4. Split on whitespace, lowercase, strip non-alpha
237	* 5. Filter: length < MIN_KEYWORD_LENGTH, stop words
238	* 6. Count frequency, take top MAX_CONTENT_KEYWORDS unique words
239	* 7. Return space-joined string
240	*
241	* @param string $htmlContent Raw post_content (may contain HTML and shortcodes).
242	* @return string Space-separated lowercase keywords.
243	*/
244	public static function extractContentKeywords(string $htmlContent): string {
245	if (trim($htmlContent) === '') {
246	return '';
247	}
248
249	// Strip shortcodes: [tag attr="val"]content[/tag] and [self-closing /]
250	$text = preg_replace('/\[\/?\w+[^\]]*\]/', '', $htmlContent);
251	if (!is_string($text)) {
252	$text = $htmlContent;
253	}
254
255	// Strip HTML tags
256	$text = strip_tags($text);
257
258	// Decode HTML entities
259	$text = html_entity_decode($text, ENT_QUOTES \| ENT_HTML5, 'UTF-8');
260
261	// Lowercase
262	$text = strtolower($text);
263
264	// Replace non-alpha characters with spaces (keeps Unicode letters via \p{L})
265	$text = preg_replace('/[^\p{L}]+/u', ' ', $text);
266	if (!is_string($text)) {
267	return '';
268	}
269
270	// Split on whitespace
271	$words = preg_split('/\s+/', trim($text));
272	if (!is_array($words)) {
273	return '';
274	}
275
276	$stopLookup = array_flip(ABJ_404_Solution_ContentMatchingEngine::$stopWords);
277	$freq = [];
278
279	foreach ($words as $word) {
280	if (!is_string($word) \|\| strlen($word) < self::MIN_KEYWORD_LENGTH) {
281	continue;
282	}
283	if (isset($stopLookup[$word])) {
284	continue;
285	}
286	if (!isset($freq[$word])) {
287	$freq[$word] = 0;
288	}
289	$freq[$word]++;
290	}
291
292	if (empty($freq)) {
293	return '';
294	}
295
296	// Sort by frequency descending
297	arsort($freq);
298
299	// Take top N unique words
300	$top = array_slice(array_keys($freq), 0, self::MAX_CONTENT_KEYWORDS);
301
302	return implode(' ', $top);
303	}
304
305	}
306