404-solution / includes / SpellChecker.php

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

429 lines 14.6 KB

Raw Download Zip

1	<?php
2
3
4	if (!defined('ABSPATH')) {
5	exit;
6	}
7
8	/* Finds similar pages.
9	* Finds search suggestions. */
10
11	class ABJ_404_Solution_SpellChecker {
12
13	/** @var array<int, string> */
14	private array $separatingCharacters = array("-","_",".","~",'%20');
15
16	/** Same as above except without the period (.) because of the extension in the file name.
17	* @var array<int, string> */
18	private array $separatingCharactersForImages = array("-","_","~",'%20');
19
20	const MAX_DIST = 2083;
21
22	const MAX_LIKELY_DISTANCE = 300;
23
24	const NGRAM_PREFILTER_THRESHOLD = 0.3;
25
26	const NGRAM_PREFILTER_MAX_CANDIDATES = 500;
27
28	const NGRAM_MIN_CACHE_ENTRIES = 50;
29
30	const NGRAM_SECONDARY_THRESHOLD = 0.4;
31
32	const NGRAM_SECONDARY_MAX_CANDIDATES = 100;
33
34	const NGRAM_MIN_COVERAGE_RATIO = 0.8;
35
36	const NGRAM_SECONDARY_MIN_CANDIDATES = 50;
37
38	private static ?self $instance = null;
39
40	/** @var ABJ_404_Solution_Functions */
41	private $f;
42
43	/** @var ABJ_404_Solution_PluginLogic */
44	private $logic;
45
46	/** @var ABJ_404_Solution_ContentRepository */
47	private $contentRepository;
48
49	/** @var ABJ_404_Solution_Logging */
50	private $logger;
51
52	/** @var ABJ_404_Solution_SpellURLMatcher */
53	private $urlMatcher;
54
55	/** @var ABJ_404_Solution_SpellLevenshteinEngine */
56	private $levenshteinEngine;
57
58	/** @var ABJ_404_Solution_SpellCandidateFilter */
59	private $candidateFilter;
60
61	/** @var ABJ_404_Solution_SpellPostListeners */
62	private $postListeners;
63
64	/**
65	* @param ABJ_404_Solution_Functions\|null $functions
66	* @param ABJ_404_Solution_PluginLogic\|null $pluginLogic
67	* @param ABJ_404_Solution_ContentRepository\|null $contentRepository
68	* @param ABJ_404_Solution_Logging\|null $logging
69	* @param ABJ_404_Solution_PermalinkCache\|null $permalinkCache
70	* @param ABJ_404_Solution_NGramFilter\|null $ngramFilter
71	* @param ABJ_404_Solution_ViewReadService\|null $viewReadService
72	*/
73	public function __construct($functions = null, $pluginLogic = null, $contentRepository = null, $logging = null, $permalinkCache = null, $ngramFilter = null, $viewReadService = null) {
74	$this->f = $functions !== null ? $functions : abj_service('functions');
75	$this->logic = $pluginLogic !== null ? $pluginLogic : abj_service('plugin_logic');
76	$this->contentRepository = $contentRepository !== null ? $contentRepository : abj_service('content_repository');
77	$this->logger = $logging !== null ? $logging : abj_service('logging');
78	$permalinkCacheResolved = $permalinkCache !== null ? $permalinkCache : abj_service('permalink_cache');
79	$ngramFilterResolved = $ngramFilter !== null ? $ngramFilter : abj_service('ngram_filter');
80	$viewReadServiceResolved = $viewReadService !== null ? $viewReadService :
81	(is_object($contentRepository) && method_exists($contentRepository, 'getRedirectsWithRegEx') ? $contentRepository : abj_service('view_read_service'));
82
83	$options = $this->logic->getOptions();
84	$custom404PageIDRaw =
85	(is_array($options) && isset($options['dest404page']) ?
86	$options['dest404page'] : null);
87	$custom404PageID = is_string($custom404PageIDRaw) ? $custom404PageIDRaw : (is_int($custom404PageIDRaw) ? (string)$custom404PageIDRaw : null);
88	$custom404PageIDResolved = null;
89	if ($this->logic->thereIsAUserSpecified404Page($custom404PageID)) {
90	$custom404PageIDResolved = $custom404PageID;
91	}
92
93	$this->urlMatcher = new ABJ_404_Solution_SpellURLMatcher(
94	$this->f, $this->logic, $this->logger, $this->contentRepository,
95	$viewReadServiceResolved, $custom404PageIDResolved
96	);
97
98	$this->postListeners = new ABJ_404_Solution_SpellPostListeners(
99	$this->f, $this->logic, $this->logger, $this->contentRepository,
100	$permalinkCacheResolved, $ngramFilterResolved
101	);
102
103	$this->levenshteinEngine = new ABJ_404_Solution_SpellLevenshteinEngine(
104	$this->f, $this->logic, $this->logger, $this->contentRepository,
105	$ngramFilterResolved, $this->urlMatcher, $this->separatingCharacters
106	);
107
108	$this->candidateFilter = new ABJ_404_Solution_SpellCandidateFilter(
109	$this->f, $this->logic, $this->logger, $this->contentRepository,
110	$this->urlMatcher, $this->levenshteinEngine, $this->postListeners,
111	$custom404PageIDResolved, $this->separatingCharacters, $this->separatingCharactersForImages
112	);
113	}
114
115	public static function resetForTests(): void {
116	self::$instance = null;
117	}
118
119	public static function getInstance(): self {
120	if (self::$instance !== null) {
121	return self::$instance;
122	}
123
124	if (class_exists('ABJ_404_Solution_ServiceContainer')) {
125	$resolved = ABJ_404_Solution_ServiceContainer::safeGet('spell_checker');
126	if ($resolved instanceof self) {
127	self::$instance = $resolved;
128	return self::$instance;
129	}
130	}
131
132	self::$instance = new ABJ_404_Solution_SpellChecker();
133
134	return self::$instance;
135	}
136
137	public function enablePerformanceCounters(bool $enable = true): void {
138	$this->levenshteinEngine->enablePerformanceCounters($enable);
139	}
140
141	public function setSkipNgramGate4(bool $skip = true): void {
142	$this->levenshteinEngine->setSkipNgramGate4($skip);
143	}
144
145	public function resetPerformanceCounters(): void {
146	$this->levenshteinEngine->resetPerformanceCounters();
147	}
148
149	/**
150	* @return array{levenshtein_calls: int, pages_considered: int, efficiency_percent: float}
151	*/
152	public function getPerformanceCounters(): array {
153	return $this->levenshteinEngine->getPerformanceCounters();
154	}
155
156	/** @return array<string, mixed>\|null */
157	function getPermalinkUsingRegEx(string $requestedURL, $options = null) {
158	return $this->urlMatcher->getPermalinkUsingRegEx($requestedURL, $options);
159	}
160
161	/** @return array<string, mixed>\|null */
162	function getPermalinkUsingSlug(string $requestedURL) {
163	return $this->urlMatcher->getPermalinkUsingSlug($requestedURL);
164	}
165
166	function requestIsForAnImage(string $requestedURL): bool {
167	return $this->urlMatcher->requestIsForAnImage($requestedURL);
168	}
169
170	/** @return array<int, array<string, mixed>> */
171	function getOnlyIDandTermID(array $rowsAsObject): array {
172	return $this->urlMatcher->getOnlyIDandTermID($rowsAsObject);
173	}
174
175	/** @return array<int\|string, mixed> */
176	function getFromPermalinkCache(string $requestedURL): array {
177	return $this->urlMatcher->getFromPermalinkCache($requestedURL);
178	}
179
180	/**
181	* @return string\|null
182	* @throws Exception
183	*/
184	function getPermalink($id, $rowType) {
185	return $this->urlMatcher->getPermalink($id, $rowType);
186	}
187
188	function getLastURLPart($url) {
189	return $this->urlMatcher->getLastURLPart($url);
190	}
191
192	/** @return array<int, mixed> */
193	function findMatchingPosts(string $requestedURLRaw, string $includeCats = '1', string $includeTags = '1') {
194	return $this->candidateFilter->findMatchingPosts($requestedURLRaw, $includeCats, $includeTags);
195	}
196
197	/** @return array<string, string> */
198	function removeExcludedPages(array $options, array $permalinks): array {
199	return $this->candidateFilter->removeExcludedPages($options, $permalinks);
200	}
201
202	/** @return array<string, string> */
203	function removeExcludedPagesWithRegex(array $options, array $permalinks, int $maxCacheCount): array {
204	return $this->candidateFilter->removeExcludedPagesWithRegex($options, $permalinks, $maxCacheCount);
205	}
206
207	/** @return array<string, string> */
208	function matchOnCats(array $permalinks, string $requestedURLCleaned, string $fullURLspacesCleaned, string $rowType): array {
209	return $this->candidateFilter->matchOnCats($permalinks, $requestedURLCleaned, $fullURLspacesCleaned, $rowType);
210	}
211
212	/** @return array<string, string> */
213	function matchOnTags(array $permalinks, string $requestedURLCleaned, string $fullURLspacesCleaned, string $rowType): array {
214	return $this->candidateFilter->matchOnTags($permalinks, $requestedURLCleaned, $fullURLspacesCleaned, $rowType);
215	}
216
217	/** @return array<string, string> */
218	function matchOnPosts(array $permalinks, string $requestedURLRaw, string $requestedURLCleaned, string $fullURLspacesCleaned, string $rowType): array {
219	return $this->candidateFilter->matchOnPosts($permalinks, $requestedURLRaw, $requestedURLCleaned, $fullURLspacesCleaned, $rowType);
220	}
221
222	/** @return array<int\|string, mixed> */
223	function getLikelyMatchIDs(string $requestedURLCleaned, string $fullURLspaces, string $rowType, ?array $rows = null) {
224	return $this->levenshteinEngine->getLikelyMatchIDs($requestedURLCleaned, $fullURLspaces, $rowType, $rows);
225	}
226
227	function getMaxAcceptableDistance(array $maxDistances, int $onlyNeedThisManyPages): int {
228	return $this->levenshteinEngine->getMaxAcceptableDistance($maxDistances, $onlyNeedThisManyPages);
229	}
230
231	function customLevenshtein($str1, $str2) {
232	return $this->levenshteinEngine->customLevenshtein($str1, $str2);
233	}
234
235	function save_postListener($post_id, $post = null, $update = null): void {
236	// @hook-lifecycle: opt-out - delegated SpellPostListeners::save_postListener owns request-level dedup.
237	$this->postListeners->save_postListener($post_id, $post, $update);
238	}
239
240	function delete_postListener($post_id, $post = null): void {
241	$this->postListeners->delete_postListener($post_id, $post);
242	}
243
244	function savePostHandler($post_id, $post, $update, $saveOrDelete): void {
245	$this->postListeners->savePostHandler($post_id, $post, $update, $saveOrDelete);
246	}
247
248	function permalinkStructureChanged($var1, $newStructure): void {
249	$this->postListeners->permalinkStructureChanged($var1, $newStructure);
250	}
251
252	function initializePublishedPostsProvider(): void {
253	$this->postListeners->initializePublishedPostsProvider();
254	}
255
256	/**
257	* @return array<int, mixed>
258	*/
259	public function findSuggestionsForURLUsingSmartCache($requestedURL, $includeCats = '1', $includeTags = true) {
260	$includeTagsStr = $includeTags ? '1' : '0';
261	return $this->findMatchingPosts($requestedURL, $includeCats, $includeTagsStr);
262	}
263
264	static function init(): void {
265	$me = abj_service('spell_checker');
266
267	add_action('updated_option', array($me,'permalinkStructureChanged'), 10, 2);
268	add_action('save_post', array($me,'save_postListener'), 10, 3);
269	add_action('delete_post', array($me,'delete_postListener'), 10, 2);
270	}
271
272	/**
273	* @return array<string, mixed>\|null
274	*/
275	function getPermalinkUsingSpelling(string $requestedURL, ?string $fullRequestedURL = null, $optionsOverride = null) {
276	$abj404spellChecker = abj_service('spell_checker');
277
278	$options = is_array($optionsOverride) ? $optionsOverride : $this->logic->getOptions();
279
280	if (@$options['auto_redirects'] == '1') {
281	$autoCats = isset($options['auto_cats']) && is_string($options['auto_cats']) ? $options['auto_cats'] : '1';
282	$autoTags = isset($options['auto_tags']) && is_string($options['auto_tags']) ? $options['auto_tags'] : '1';
283	$permalinksPacket = $abj404spellChecker->findMatchingPosts($requestedURL,
284	$autoCats, $autoTags);
285
286	$permalinks = $permalinksPacket[0];
287	$rowType = $permalinksPacket[1];
288
289	$minScore = $options['auto_score'];
290
291	if (!is_array($permalinks) \|\| empty($permalinks)) {
292	return null;
293	}
294	$linkScore = reset($permalinks);
295	$idAndType = key($permalinks);
296	$idAndTypeStr = is_string($idAndType) ? $idAndType : (string)$idAndType;
297	$linkScoreInt = is_scalar($linkScore) ? (int)$linkScore : 0;
298	$permalink = ABJ_404_Solution_Functions::permalinkInfoToArray($idAndTypeStr, $linkScoreInt,
299	is_string($rowType) ? $rowType : null, $options);
300
301	if ($permalink['score'] >= $minScore) {
302	$redirectType = $permalink['type'];
303	if (('' . $redirectType != ABJ404_TYPE_404_DISPLAYED) && ('' . $redirectType != ABJ404_TYPE_HOME)) {
304	return $permalink;
305
306	} else {
307	$permalinkJson = json_encode($permalink);
308	$this->logger->errorMessage("Unhandled permalink type: " .
309	wp_kses_post(is_string($permalinkJson) ? $permalinkJson : '{}'));
310	return null;
311	}
312	}
313
314	if ($fullRequestedURL !== null) {
315	$this->cacheComputedSuggestionsForShortcode($fullRequestedURL, $permalinksPacket);
316	}
317	}
318
319	return null;
320	}
321
322	private function cacheComputedSuggestionsForShortcode(string $fullRequestedURL, array $permalinksPacket): void {
323	$normalizedURL = $this->f->normalizeURLForCacheKey($fullRequestedURL);
324
325	$urlKey = md5($normalizedURL);
326	$transientKey = 'abj404_suggest_' . $urlKey;
327
328	$existing = get_transient($transientKey);
329	if ($existing !== false) {
330	return;
331	}
332
333	// allow-cache-empty: factory-built typed array; SuggestionTransient::completeArray
334	// always returns a non-empty associative array with at minimum a 'status' key.
335	set_transient(
336	$transientKey,
337	ABJ_404_Solution_SuggestionTransient::completeArray(
338	$normalizedURL,
339	$permalinksPacket,
340	time(),
341	''
342	),
343	300
344	); // 5 minute TTL
345
346	$this->logger->debugMessage("Cached spell-check suggestions for shortcode: " .
347	esc_html($normalizedURL));
348	}
349
350	public function triggerAsyncSuggestionComputation($requestedURL) {
351	$f = abj_service('functions');
352
353	$normalizedURL = $f->normalizeURLForCacheKey($requestedURL);
354
355	$urlKey = md5($normalizedURL);
356	$transientKey = 'abj404_suggest_' . $urlKey;
357
358	$existing = ABJ_404_Solution_SuggestionTransient::fromRaw(get_transient($transientKey));
359	if ($existing !== null) {
360	$this->logger->debugMessage("Async suggestions: skipping, transient already exists for " .
361	esc_html($normalizedURL) . " (status: " . esc_html($existing->getStatus()) . ")");
362	return false;
363	}
364
365	$token = wp_generate_password(32, false);
366
367	// allow-cache-empty: factory-built typed array; keep the TTL at 120
368	// seconds so slow hosts can start before the polling UI gives up.
369	set_transient(
370	$transientKey,
371	ABJ_404_Solution_SuggestionTransient::pendingArray(
372	$normalizedURL,
373	$token,
374	0,
375	time()
376	),
377	120
378	); // 2 minute TTL
379
380	$this->logger->debugMessage("Async suggestions: triggering background computation for " .
381	esc_html($normalizedURL));
382
383	$response = wp_remote_post(admin_url('admin-ajax.php'), array(
384	'blocking' => false,
385	'timeout' => 5,
386	'sslverify' => apply_filters('https_local_ssl_verify', false),
387	'body' => array(
388	'action' => 'abj404_compute_suggestions',
389	'url' => $normalizedURL,
390	'token' => $token
391	)
392	));
393
394	if (is_wp_error($response)) {
395	$this->logger->debugMessage("Async suggestions: dispatch failed for " .
396	esc_html($normalizedURL) . " - " . $response->get_error_message());
397	delete_transient($transientKey);
398	return false;
399	}
400
401	return true;
402	}
403
404	public function does404PageHaveSuggestionsShortcode() {
405	$options = $this->logic->getOptions();
406	$dest404pageRaw = isset($options['dest404page']) ? $options['dest404page'] : null;
407	$dest404page = is_string($dest404pageRaw) ? $dest404pageRaw : null;
408
409	if (!$this->logic->thereIsAUserSpecified404Page($dest404page)) {
410	return false;
411	}
412
413	$parts = explode('\|', $dest404page ?? '');
414	$page404Id = isset($parts[0]) ? intval($parts[0]) : 0;
415
416	if ($page404Id <= 0) {
417	return false;
418	}
419
420	$page = get_post($page404Id);
421	if (!$page) {
422	return false;
423	}
424
425	return has_shortcode($page->post_content, ABJ404_SHORTCODE_NAME);
426	}
427
428	}
429