| 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_NotFoundResponseService|null */ |
| 47 |
private $notFoundResponse; |
| 48 |
|
| 49 |
/** @var ABJ_404_Solution_ContentRepository */ |
| 50 |
private $contentRepository; |
| 51 |
|
| 52 |
/** @var ABJ_404_Solution_Logging */ |
| 53 |
private $logger; |
| 54 |
|
| 55 |
/** @var ABJ_404_Solution_SpellURLMatcher */ |
| 56 |
private $urlMatcher; |
| 57 |
|
| 58 |
/** @var ABJ_404_Solution_SpellLevenshteinEngine */ |
| 59 |
private $levenshteinEngine; |
| 60 |
|
| 61 |
/** @var ABJ_404_Solution_SpellCandidateFilter */ |
| 62 |
private $candidateFilter; |
| 63 |
|
| 64 |
/** @var ABJ_404_Solution_SpellPostListeners */ |
| 65 |
private $postListeners; |
| 66 |
|
| 67 |
/** @var ABJ_404_Solution_SpellSuggestionShortcodeDetector */ |
| 68 |
private $shortcodeDetector; |
| 69 |
|
| 70 |
/** @var ABJ_404_Solution_SuggestionPublisher */ |
| 71 |
private $suggestionPublisher; |
| 72 |
|
| 73 |
/** |
| 74 |
* @param ABJ_404_Solution_SpellCheckerDependencies|null $deps |
| 75 |
*/ |
| 76 |
public function __construct(?ABJ_404_Solution_SpellCheckerDependencies $deps = null) { |
| 77 |
$deps = $deps ?? new ABJ_404_Solution_SpellCheckerDependencies(); |
| 78 |
$contentRepository = $deps->contentRepository; |
| 79 |
$this->f = $deps->functions !== null ? $deps->functions : abj_service('functions'); |
| 80 |
$this->logic = $deps->pluginLogic !== null ? $deps->pluginLogic : abj_service('plugin_logic'); |
| 81 |
$resolvedNotFoundResponse = function_exists('abj_service_optional') |
| 82 |
? abj_service_optional('not_found_response') : null; |
| 83 |
$this->notFoundResponse = $resolvedNotFoundResponse instanceof ABJ_404_Solution_NotFoundResponseService |
| 84 |
? $resolvedNotFoundResponse : null; |
| 85 |
$this->contentRepository = $contentRepository !== null ? $contentRepository : abj_service('content_repository'); |
| 86 |
$this->logger = $deps->logging !== null ? $deps->logging : abj_service('logging'); |
| 87 |
$permalinkCacheResolved = $deps->permalinkCache !== null ? $deps->permalinkCache : abj_service('permalink_cache'); |
| 88 |
$ngramFilterResolved = $deps->ngramFilter !== null ? $deps->ngramFilter : abj_service('ngram_filter'); |
| 89 |
$viewReadServiceResolved = $deps->viewReadService !== null ? $deps->viewReadService : |
| 90 |
(is_object($contentRepository) && method_exists($contentRepository, 'getRedirectsWithRegEx') ? $contentRepository : abj_service('view_read_service')); |
| 91 |
|
| 92 |
$options = abj_service('options_repository')->getOptions(true); |
| 93 |
$custom404PageIDRaw = |
| 94 |
(is_array($options) && isset($options['dest404page']) ? |
| 95 |
$options['dest404page'] : null); |
| 96 |
$custom404PageID = is_string($custom404PageIDRaw) ? $custom404PageIDRaw : (is_int($custom404PageIDRaw) ? (string)$custom404PageIDRaw : null); |
| 97 |
$custom404PageIDResolved = null; |
| 98 |
if ($this->notFoundResponse instanceof ABJ_404_Solution_NotFoundResponseService |
| 99 |
&& $this->notFoundResponse->thereIsAUserSpecified404Page($custom404PageID)) { |
| 100 |
$custom404PageIDResolved = $custom404PageID; |
| 101 |
} |
| 102 |
|
| 103 |
$this->urlMatcher = new ABJ_404_Solution_SpellURLMatcher( |
| 104 |
$this->f, $this->logger, $this->contentRepository, |
| 105 |
$viewReadServiceResolved, $custom404PageIDResolved |
| 106 |
); |
| 107 |
|
| 108 |
$this->postListeners = new ABJ_404_Solution_SpellPostListeners( |
| 109 |
$this->f, $this->logger, $this->contentRepository, |
| 110 |
$permalinkCacheResolved, $ngramFilterResolved |
| 111 |
); |
| 112 |
|
| 113 |
$this->levenshteinEngine = new ABJ_404_Solution_SpellLevenshteinEngine( |
| 114 |
new ABJ_404_Solution_SpellLevenshteinEngineDependencies( |
| 115 |
$this->f, $this->logic, $this->logger, $this->contentRepository, |
| 116 |
$ngramFilterResolved, $this->urlMatcher, $this->separatingCharacters |
| 117 |
) |
| 118 |
); |
| 119 |
|
| 120 |
$this->candidateFilter = new ABJ_404_Solution_SpellCandidateFilter( |
| 121 |
$this->f, $this->logic, $this->logger, $this->contentRepository, |
| 122 |
$this->urlMatcher, $this->levenshteinEngine, $this->postListeners, |
| 123 |
$custom404PageIDResolved, $this->separatingCharacters, $this->separatingCharactersForImages |
| 124 |
); |
| 125 |
|
| 126 |
$this->shortcodeDetector = new ABJ_404_Solution_SpellSuggestionShortcodeDetector( |
| 127 |
$this->notFoundResponse |
| 128 |
); |
| 129 |
|
| 130 |
$this->suggestionPublisher = new ABJ_404_Solution_SuggestionPublisher($this->logger); |
| 131 |
} |
| 132 |
|
| 133 |
/** |
| 134 |
* Test seam (M105): install the cached singleton; pass null to clear it. |
| 135 |
* @param self|null $instance |
| 136 |
* @return void |
| 137 |
*/ |
| 138 |
public static function setInstance($instance) { |
| 139 |
self::$instance = $instance; |
| 140 |
} |
| 141 |
|
| 142 |
/** |
| 143 |
* Return the already-built singleton without resolving the container or |
| 144 |
* building a new one, so the `spell_checker` factory can honor a |
| 145 |
* test-installed override. Mirrors PluginLogic / Logging peekInstance(). |
| 146 |
* @return self|null |
| 147 |
*/ |
| 148 |
public static function peekInstance(): ?self { |
| 149 |
return self::$instance; |
| 150 |
} |
| 151 |
|
| 152 |
public static function getInstance(): self { |
| 153 |
if (self::$instance !== null) { |
| 154 |
return self::$instance; |
| 155 |
} |
| 156 |
|
| 157 |
if (class_exists('ABJ_404_Solution_ServiceContainer')) { |
| 158 |
$resolved = ABJ_404_Solution_ServiceContainer::safeGet('spell_checker'); |
| 159 |
if ($resolved instanceof self) { |
| 160 |
self::$instance = $resolved; |
| 161 |
return self::$instance; |
| 162 |
} |
| 163 |
} |
| 164 |
|
| 165 |
self::$instance = new ABJ_404_Solution_SpellChecker(); |
| 166 |
|
| 167 |
return self::$instance; |
| 168 |
} |
| 169 |
|
| 170 |
public function enablePerformanceCounters(bool $enable = true): void { |
| 171 |
$this->levenshteinEngine->enablePerformanceCounters($enable); |
| 172 |
} |
| 173 |
|
| 174 |
public function setSkipNgramGate4(bool $skip = true): void { |
| 175 |
$this->levenshteinEngine->setSkipNgramGate4($skip); |
| 176 |
} |
| 177 |
|
| 178 |
public function resetPerformanceCounters(): void { |
| 179 |
$this->levenshteinEngine->resetPerformanceCounters(); |
| 180 |
} |
| 181 |
|
| 182 |
/** |
| 183 |
* @return array{levenshtein_calls: int, pages_considered: int, efficiency_percent: float} |
| 184 |
*/ |
| 185 |
public function getPerformanceCounters(): array { |
| 186 |
return $this->levenshteinEngine->getPerformanceCounters(); |
| 187 |
} |
| 188 |
|
| 189 |
/** @return array<string, mixed>|null */ |
| 190 |
function getPermalinkUsingRegEx(string $requestedURL, $options = null) { |
| 191 |
return $this->urlMatcher->getPermalinkUsingRegEx($requestedURL, $options); |
| 192 |
} |
| 193 |
|
| 194 |
/** @return array<string, mixed>|null */ |
| 195 |
function getPermalinkUsingSlug(string $requestedURL) { |
| 196 |
return $this->urlMatcher->getPermalinkUsingSlug($requestedURL); |
| 197 |
} |
| 198 |
|
| 199 |
function requestIsForAnImage(string $requestedURL): bool { |
| 200 |
return $this->urlMatcher->requestIsForAnImage($requestedURL); |
| 201 |
} |
| 202 |
|
| 203 |
/** @return array<int, array<string, mixed>> */ |
| 204 |
function getOnlyIDandTermID(array $rowsAsObject): array { |
| 205 |
return $this->urlMatcher->getOnlyIDandTermID($rowsAsObject); |
| 206 |
} |
| 207 |
|
| 208 |
/** @return array<int|string, mixed> */ |
| 209 |
function getFromPermalinkCache(string $requestedURL): array { |
| 210 |
return $this->urlMatcher->getFromPermalinkCache($requestedURL); |
| 211 |
} |
| 212 |
|
| 213 |
/** |
| 214 |
* @return string|null |
| 215 |
* @throws Exception |
| 216 |
*/ |
| 217 |
function getPermalink($id, $rowType) { |
| 218 |
return $this->urlMatcher->getPermalink($id, $rowType); |
| 219 |
} |
| 220 |
|
| 221 |
function getLastURLPart($url) { |
| 222 |
return $this->urlMatcher->getLastURLPart($url); |
| 223 |
} |
| 224 |
|
| 225 |
/** @return array<int, mixed> */ |
| 226 |
function findMatchingPosts(string $requestedURLRaw, string $includeCats = '1', string $includeTags = '1') { |
| 227 |
return $this->candidateFilter->findMatchingPosts($requestedURLRaw, $includeCats, $includeTags); |
| 228 |
} |
| 229 |
|
| 230 |
/** @return array<string, string> */ |
| 231 |
function removeExcludedPages(array $options, array $permalinks): array { |
| 232 |
return $this->candidateFilter->removeExcludedPages($options, $permalinks); |
| 233 |
} |
| 234 |
|
| 235 |
/** @return array<string, string> */ |
| 236 |
function removeExcludedPagesWithRegex(array $options, array $permalinks, int $maxCacheCount): array { |
| 237 |
return $this->candidateFilter->removeExcludedPagesWithRegex($options, $permalinks, $maxCacheCount); |
| 238 |
} |
| 239 |
|
| 240 |
/** @return array<string, string> */ |
| 241 |
function matchOnCats(array $permalinks, string $requestedURLCleaned, string $fullURLspacesCleaned, string $rowType): array { |
| 242 |
return $this->candidateFilter->matchOnCats($permalinks, $requestedURLCleaned, $fullURLspacesCleaned, $rowType); |
| 243 |
} |
| 244 |
|
| 245 |
/** @return array<string, string> */ |
| 246 |
function matchOnTags(array $permalinks, string $requestedURLCleaned, string $fullURLspacesCleaned, string $rowType): array { |
| 247 |
return $this->candidateFilter->matchOnTags($permalinks, $requestedURLCleaned, $fullURLspacesCleaned, $rowType); |
| 248 |
} |
| 249 |
|
| 250 |
/** @return array<string, string> */ |
| 251 |
function matchOnPosts(array $permalinks, string $requestedURLRaw, string $requestedURLCleaned, string $fullURLspacesCleaned, string $rowType): array { |
| 252 |
return $this->candidateFilter->matchOnPosts($permalinks, $requestedURLRaw, $requestedURLCleaned, $fullURLspacesCleaned, $rowType); |
| 253 |
} |
| 254 |
|
| 255 |
/** @return array<int|string, mixed> */ |
| 256 |
function getLikelyMatchIDs(string $requestedURLCleaned, string $fullURLspaces, string $rowType, ?array $rows = null) { |
| 257 |
return $this->levenshteinEngine->getLikelyMatchIDs($requestedURLCleaned, $fullURLspaces, $rowType, $rows); |
| 258 |
} |
| 259 |
|
| 260 |
function customLevenshtein($str1, $str2) { |
| 261 |
return $this->levenshteinEngine->customLevenshtein($str1, $str2); |
| 262 |
} |
| 263 |
|
| 264 |
function save_postListener($post_id, $post = null, $update = null): void { |
| 265 |
// @hook-lifecycle: opt-out - delegated SpellPostListeners::save_postListener owns request-level dedup. |
| 266 |
$this->postListeners->save_postListener($post_id, $post, $update); |
| 267 |
} |
| 268 |
|
| 269 |
function delete_postListener($post_id, $post = null): void { |
| 270 |
$this->postListeners->delete_postListener($post_id, $post); |
| 271 |
} |
| 272 |
|
| 273 |
function term_changedListener(int $term_id, int $tt_id = 0, string $taxonomy = ''): void { |
| 274 |
$this->postListeners->term_changedListener($term_id, $tt_id, $taxonomy); |
| 275 |
} |
| 276 |
|
| 277 |
function savePostHandler($post_id, $post, $update, $saveOrDelete): void { |
| 278 |
$this->postListeners->savePostHandler($post_id, $post, $update, $saveOrDelete); |
| 279 |
} |
| 280 |
|
| 281 |
function permalinkStructureChanged($var1, $newStructure): void { |
| 282 |
$this->postListeners->permalinkStructureChanged($var1, $newStructure); |
| 283 |
} |
| 284 |
|
| 285 |
function initializePublishedPostsProvider(): void { |
| 286 |
$this->postListeners->initializePublishedPostsProvider(); |
| 287 |
} |
| 288 |
|
| 289 |
/** |
| 290 |
* @return array<int, mixed> |
| 291 |
*/ |
| 292 |
public function findSuggestionsForURLUsingSmartCache($requestedURL, $includeCats = '1', $includeTags = true) { |
| 293 |
$includeTagsStr = $includeTags ? '1' : '0'; |
| 294 |
return $this->findMatchingPosts($requestedURL, $includeCats, $includeTagsStr); |
| 295 |
} |
| 296 |
|
| 297 |
static function init(): void { |
| 298 |
$me = abj_service('spell_checker'); |
| 299 |
|
| 300 |
add_action('updated_option', array($me,'permalinkStructureChanged'), 10, 2); |
| 301 |
add_action('save_post', array($me,'save_postListener'), 10, 3); |
| 302 |
add_action('delete_post', array($me,'delete_postListener'), 10, 2); |
| 303 |
// A category/tag create/rename/delete changes spelling-match results, so |
| 304 |
// invalidate the spelling cache (including memoized no-match entries). |
| 305 |
add_action('created_term', array($me,'term_changedListener'), 10, 3); |
| 306 |
add_action('edited_term', array($me,'term_changedListener'), 10, 3); |
| 307 |
add_action('delete_term', array($me,'term_changedListener'), 10, 3); |
| 308 |
} |
| 309 |
|
| 310 |
/** |
| 311 |
* True only for a cached confirmed no-match: a [permalinks, rowType] packet |
| 312 |
* whose permalink list is an empty array. A cache miss (empty array()), a |
| 313 |
* positive result, and any malformed payload all return false so the caller |
| 314 |
* recomputes -- the negative short-circuit can never emit a destination. |
| 315 |
* |
| 316 |
* @param mixed $cachedPacket the getFromPermalinkCache() return value |
| 317 |
* @return bool |
| 318 |
*/ |
| 319 |
private function isCachedNoMatchResult($cachedPacket): bool { |
| 320 |
return is_array($cachedPacket) && count($cachedPacket) === 2 |
| 321 |
&& array_key_exists(0, $cachedPacket) && is_array($cachedPacket[0]) |
| 322 |
&& empty($cachedPacket[0]); |
| 323 |
} |
| 324 |
|
| 325 |
/** |
| 326 |
* @return array<string, mixed>|null |
| 327 |
*/ |
| 328 |
function getPermalinkUsingSpelling(string $requestedURL, ?string $fullRequestedURL = null, $optionsOverride = null) { |
| 329 |
$abj404spellChecker = abj_service('spell_checker'); |
| 330 |
|
| 331 |
$options = is_array($optionsOverride) ? $optionsOverride : abj_service('options_repository')->getOptions(); |
| 332 |
|
| 333 |
if (@$options['auto_redirects'] == '1') { |
| 334 |
$autoCats = isset($options['auto_cats']) && is_string($options['auto_cats']) ? $options['auto_cats'] : '1'; |
| 335 |
$autoTags = isset($options['auto_tags']) && is_string($options['auto_tags']) ? $options['auto_tags'] : '1'; |
| 336 |
|
| 337 |
// Negative-result memoization. A repeated 404 to a URL with no |
| 338 |
// spelling match would otherwise re-run the full Levenshtein scan |
| 339 |
// on every hit: captured rows are excluded from the pre-match |
| 340 |
// redirect lookup (status/type filter in getPermalinkFromURL.sql), |
| 341 |
// so an already-seen no-match URL never short-circuits before this |
| 342 |
// point. We short-circuit ONLY on a cached no-match (an empty |
| 343 |
// permalink list). The outcome is identical to recomputing a |
| 344 |
// no-match (return null), so a stale entry can at worst miss a |
| 345 |
// redirect (graceful 404), never cause a wrong one. The spelling |
| 346 |
// cache is invalidated on any content change (SpellPostListeners: |
| 347 |
// save/delete post, created/edited/deleted category or tag), so a |
| 348 |
// URL that becomes matchable is recomputed on its next hit. A |
| 349 |
// cached POSITIVE result is intentionally ignored and recomputed |
| 350 |
// fresh so a deleted or unpublished destination can never be |
| 351 |
// served from cache. Positive matches do not recur here in any |
| 352 |
// case: a successful match is promoted to a stored AUTO redirect, |
| 353 |
// so the next hit short-circuits at getActiveRedirectForURL. |
| 354 |
$cachedPacket = $abj404spellChecker->getFromPermalinkCache($requestedURL); |
| 355 |
if ($this->isCachedNoMatchResult($cachedPacket)) { |
| 356 |
return null; |
| 357 |
} |
| 358 |
|
| 359 |
$permalinksPacket = $abj404spellChecker->findMatchingPosts($requestedURL, |
| 360 |
$autoCats, $autoTags); |
| 361 |
|
| 362 |
$permalinks = $permalinksPacket[0]; |
| 363 |
$rowType = $permalinksPacket[1]; |
| 364 |
|
| 365 |
$minScore = $options['auto_score']; |
| 366 |
|
| 367 |
if (!is_array($permalinks) || empty($permalinks)) { |
| 368 |
return null; |
| 369 |
} |
| 370 |
$linkScore = reset($permalinks); |
| 371 |
$idAndType = key($permalinks); |
| 372 |
$idAndTypeStr = is_string($idAndType) ? $idAndType : (string)$idAndType; |
| 373 |
$linkScoreInt = is_scalar($linkScore) ? (int)$linkScore : 0; |
| 374 |
$permalink = ABJ_404_Solution_PermalinkResolver::permalinkInfoToArray($idAndTypeStr, $linkScoreInt, |
| 375 |
is_string($rowType) ? $rowType : null, $options); |
| 376 |
|
| 377 |
if ($permalink['score'] >= $minScore) { |
| 378 |
$redirectType = $permalink['type']; |
| 379 |
if (('' . $redirectType != ABJ404_TYPE_404_DISPLAYED) && ('' . $redirectType != ABJ404_TYPE_HOME)) { |
| 380 |
return $permalink; |
| 381 |
|
| 382 |
} else { |
| 383 |
$permalinkJson = json_encode($permalink); |
| 384 |
$this->logger->errorMessage("Unhandled permalink type: " . |
| 385 |
wp_kses_post(is_string($permalinkJson) ? $permalinkJson : '{}')); |
| 386 |
return null; |
| 387 |
} |
| 388 |
} |
| 389 |
|
| 390 |
// A match was found and rejected only for scoring under the |
| 391 |
// admin's threshold. That score is the one number that explains why |
| 392 |
// this URL is about to be captured instead of redirected, so hand |
| 393 |
// it to the near-miss recorder before the losing branch discards |
| 394 |
// the packet; the captured-404 insert reads it back and stores it |
| 395 |
// on the row (NotFoundResponseService::sendTo404Page). |
| 396 |
// |
| 397 |
// Keyed by the full requested URL because that is the string the |
| 398 |
// pipeline carries into sendTo404Page(). Callers that pass no full |
| 399 |
// URL (nothing in the frontend pipeline does) fall back to the slug, |
| 400 |
// which simply will not match at read time: no score, never a wrong |
| 401 |
// one. |
| 402 |
abj_service('near_miss_recorder')->record(array( |
| 403 |
'requestedURL' => $fullRequestedURL !== null ? $fullRequestedURL : $requestedURL, |
| 404 |
'score' => is_numeric($permalink['score']) ? (float)$permalink['score'] : 0.0, |
| 405 |
'engineName' => ABJ_404_Solution_SpellingMatchingEngine::engineName(), |
| 406 |
)); |
| 407 |
|
| 408 |
if ($fullRequestedURL !== null) { |
| 409 |
$this->suggestionPublisher->cacheComputedSuggestionsForShortcode( |
| 410 |
$fullRequestedURL, $permalinksPacket); |
| 411 |
} |
| 412 |
} |
| 413 |
|
| 414 |
return null; |
| 415 |
} |
| 416 |
|
| 417 |
/** |
| 418 |
* Ask for background suggestions for this URL, rolling back the pending |
| 419 |
* marker if the dispatch fails. Delegates to the suggestion publisher, which |
| 420 |
* owns the transient and the loopback request. |
| 421 |
* |
| 422 |
* @param string $requestedURL |
| 423 |
* @return bool True when a background computation was dispatched. |
| 424 |
*/ |
| 425 |
public function triggerAndCleanupOnFailure(string $requestedURL): bool { |
| 426 |
return $this->suggestionPublisher->triggerAsyncSuggestions($requestedURL); |
| 427 |
} |
| 428 |
|
| 429 |
public function does404PageHaveSuggestionsShortcode() { |
| 430 |
return $this->shortcodeDetector->does404PageHaveSuggestionsShortcode(); |
| 431 |
} |
| 432 |
|
| 433 |
} |
| 434 |
|
