| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* Levenshtein distance engine and candidate pre-filtering for the |
| 9 |
* spell-checking subsystem. |
| 10 |
* |
| 11 |
* Extracted from SpellCheckerTrait_LevenshteinEngine as a standalone class |
| 12 |
* with explicit dependency injection. |
| 13 |
*/ |
| 14 |
class ABJ_404_Solution_SpellLevenshteinEngine { |
| 15 |
|
| 16 |
const MAX_DIST = 2083; |
| 17 |
|
| 18 |
const MAX_LIKELY_DISTANCE = 300; |
| 19 |
|
| 20 |
const NGRAM_PREFILTER_THRESHOLD = 0.3; |
| 21 |
|
| 22 |
const NGRAM_PREFILTER_MAX_CANDIDATES = 500; |
| 23 |
|
| 24 |
const NGRAM_MIN_CACHE_ENTRIES = 50; |
| 25 |
|
| 26 |
const NGRAM_SECONDARY_THRESHOLD = 0.4; |
| 27 |
|
| 28 |
const NGRAM_SECONDARY_MAX_CANDIDATES = 100; |
| 29 |
|
| 30 |
const NGRAM_MIN_COVERAGE_RATIO = 0.8; |
| 31 |
|
| 32 |
const NGRAM_SECONDARY_MIN_CANDIDATES = 50; |
| 33 |
|
| 34 |
/** @var ABJ_404_Solution_Functions */ |
| 35 |
private $f; |
| 36 |
|
| 37 |
/** @var ABJ_404_Solution_PluginLogic */ |
| 38 |
private $logic; |
| 39 |
|
| 40 |
/** @var ABJ_404_Solution_Logging */ |
| 41 |
private $logger; |
| 42 |
|
| 43 |
/** @var ABJ_404_Solution_ContentRepository */ |
| 44 |
private $contentRepository; |
| 45 |
|
| 46 |
/** @var ABJ_404_Solution_NGramFilter */ |
| 47 |
private $ngramFilter; |
| 48 |
|
| 49 |
/** @var ABJ_404_Solution_SpellURLMatcher */ |
| 50 |
private $urlMatcher; |
| 51 |
|
| 52 |
/** @var array<int, string> */ |
| 53 |
private array $separatingCharacters; |
| 54 |
|
| 55 |
private bool $enablePerformanceCounters = false; |
| 56 |
|
| 57 |
private bool $skipNgramGate4 = false; |
| 58 |
|
| 59 |
private int $levenshteinCallCount = 0; |
| 60 |
|
| 61 |
private int $totalPagesConsidered = 0; |
| 62 |
|
| 63 |
/** @var ABJ_404_Solution_PublishedPostsProvider|null */ |
| 64 |
private ?ABJ_404_Solution_PublishedPostsProvider $publishedPostsProvider = null; |
| 65 |
|
| 66 |
/** |
| 67 |
* @param ABJ_404_Solution_Functions $functions |
| 68 |
* @param ABJ_404_Solution_PluginLogic $logic |
| 69 |
* @param ABJ_404_Solution_Logging $logger |
| 70 |
* @param ABJ_404_Solution_ContentRepository $contentRepository |
| 71 |
* @param ABJ_404_Solution_NGramFilter $ngramFilter |
| 72 |
* @param ABJ_404_Solution_SpellURLMatcher $urlMatcher |
| 73 |
* @param array<int, string> $separatingCharacters |
| 74 |
*/ |
| 75 |
public function __construct($functions, $logic, $logger, $contentRepository, $ngramFilter, $urlMatcher, array $separatingCharacters) { |
| 76 |
$this->f = $functions; |
| 77 |
$this->logic = $logic; |
| 78 |
$this->logger = $logger; |
| 79 |
$this->contentRepository = $contentRepository; |
| 80 |
$this->ngramFilter = $ngramFilter; |
| 81 |
$this->urlMatcher = $urlMatcher; |
| 82 |
$this->separatingCharacters = $separatingCharacters; |
| 83 |
} |
| 84 |
|
| 85 |
/** @param ABJ_404_Solution_PublishedPostsProvider|null $provider */ |
| 86 |
public function setPublishedPostsProvider(?ABJ_404_Solution_PublishedPostsProvider $provider): void { |
| 87 |
$this->publishedPostsProvider = $provider; |
| 88 |
} |
| 89 |
|
| 90 |
public function enablePerformanceCounters(bool $enable = true): void { |
| 91 |
$this->enablePerformanceCounters = $enable; |
| 92 |
if ($enable) { |
| 93 |
$this->resetPerformanceCounters(); |
| 94 |
} |
| 95 |
} |
| 96 |
|
| 97 |
public function setSkipNgramGate4(bool $skip = true): void { |
| 98 |
$this->skipNgramGate4 = $skip; |
| 99 |
} |
| 100 |
|
| 101 |
public function resetPerformanceCounters(): void { |
| 102 |
$this->levenshteinCallCount = 0; |
| 103 |
$this->totalPagesConsidered = 0; |
| 104 |
} |
| 105 |
|
| 106 |
/** |
| 107 |
* @return array{levenshtein_calls: int, pages_considered: int, efficiency_percent: float} |
| 108 |
*/ |
| 109 |
public function getPerformanceCounters(): array { |
| 110 |
$efficiency = 0; |
| 111 |
if ($this->totalPagesConsidered > 0) { |
| 112 |
$efficiency = ($this->levenshteinCallCount / $this->totalPagesConsidered) * 100; |
| 113 |
} |
| 114 |
|
| 115 |
return [ |
| 116 |
'levenshtein_calls' => $this->levenshteinCallCount, |
| 117 |
'pages_considered' => $this->totalPagesConsidered, |
| 118 |
'efficiency_percent' => round($efficiency, 2) |
| 119 |
]; |
| 120 |
} |
| 121 |
|
| 122 |
/** |
| 123 |
* @param string $requestedURLCleaned |
| 124 |
* @param string $fullURLspaces |
| 125 |
* @param string $rowType |
| 126 |
* @param array<int, array<string, mixed>>|null $rows |
| 127 |
* @return array<int|string, mixed> |
| 128 |
*/ |
| 129 |
function getLikelyMatchIDs(string $requestedURLCleaned, string $fullURLspaces, string $rowType, ?array $rows = null) { |
| 130 |
|
| 131 |
$options = $this->logic->getOptions(); |
| 132 |
$suggestMaxLikely = isset($options['suggest_max']) && is_scalar($options['suggest_max']) ? $options['suggest_max'] : 5; |
| 133 |
$onlyNeedThisManyPages = min(5 * absint($suggestMaxLikely), 100); |
| 134 |
|
| 135 |
$ngramPrefilterResult = $this->tryApplyNgramPrefilter($rowType, $rows, $requestedURLCleaned); |
| 136 |
if ($ngramPrefilterResult === 'early_return') { |
| 137 |
return array(); |
| 138 |
} |
| 139 |
$ngramPrefilterApplied = ($ngramPrefilterResult === 'applied'); |
| 140 |
|
| 141 |
$minDistances = array(); |
| 142 |
$maxDistances = array(); |
| 143 |
for ($currentDistanceIndex = 0; $currentDistanceIndex <= self::MAX_DIST; $currentDistanceIndex++) { |
| 144 |
$maxDistances[$currentDistanceIndex] = array(); |
| 145 |
$minDistances[$currentDistanceIndex] = array(); |
| 146 |
} |
| 147 |
|
| 148 |
$requestedURLCleanedLength = $this->f->strlen($requestedURLCleaned); |
| 149 |
$fullURLspacesLength = $this->f->strlen($fullURLspaces); |
| 150 |
|
| 151 |
$userRequestedURLWords = explode(" ", (empty($fullURLspaces) ? $requestedURLCleaned : $fullURLspaces)); |
| 152 |
$idsWithWordsInCommon = array(); |
| 153 |
$observedPermalinksById = array(); |
| 154 |
$wasntReadyCount = 0; |
| 155 |
|
| 156 |
if ($this->publishedPostsProvider === null) { |
| 157 |
return array(); |
| 158 |
} |
| 159 |
if (!$ngramPrefilterApplied) { |
| 160 |
$this->publishedPostsProvider->resetBatch(); |
| 161 |
} |
| 162 |
if ($rows != null) { |
| 163 |
$this->publishedPostsProvider->useThisData($rows); |
| 164 |
} |
| 165 |
$currentBatch = $this->publishedPostsProvider->getNextBatch($requestedURLCleanedLength); |
| 166 |
|
| 167 |
$row = array_pop($currentBatch); |
| 168 |
while ($row != null) { |
| 169 |
$row = (array)$row; |
| 170 |
|
| 171 |
if ($this->enablePerformanceCounters) { |
| 172 |
$this->totalPagesConsidered++; |
| 173 |
} |
| 174 |
|
| 175 |
$id = null; |
| 176 |
$the_permalink = null; |
| 177 |
$urlParts = null; |
| 178 |
if ($rowType == 'pages') { |
| 179 |
$id = $row['id']; |
| 180 |
|
| 181 |
} else if ($rowType == 'tags') { |
| 182 |
$id = array_key_exists('term_id', $row) ? $row['term_id'] : null; |
| 183 |
|
| 184 |
} else if ($rowType == 'categories') { |
| 185 |
$id = array_key_exists('term_id', $row) ? $row['term_id'] : null; |
| 186 |
|
| 187 |
} else if ($rowType == 'image') { |
| 188 |
$id = $row['id']; |
| 189 |
|
| 190 |
} else { |
| 191 |
throw new \Exception("Unknown row type ... " . esc_html($rowType)); // allow-raw-error: assertion, should never reach user |
| 192 |
} |
| 193 |
|
| 194 |
if ($id === null) { |
| 195 |
$row = array_pop($currentBatch); |
| 196 |
continue; |
| 197 |
} |
| 198 |
$idInt = is_scalar($id) ? (int)$id : 0; |
| 199 |
|
| 200 |
if (array_key_exists('url', $row)) { |
| 201 |
$the_permalink = isset($row['url']) && is_string($row['url']) ? $row['url'] : ''; |
| 202 |
$the_permalink = $this->f->normalizeUrlString($the_permalink); |
| 203 |
$urlParts = parse_url($the_permalink); |
| 204 |
|
| 205 |
if (is_bool($urlParts)) { |
| 206 |
$this->contentRepository->removeFromPermalinkCache($idInt); |
| 207 |
} |
| 208 |
} |
| 209 |
if (!array_key_exists('url', $row) || (isset($urlParts) && is_bool($urlParts))) { |
| 210 |
$wasntReadyCount++; |
| 211 |
$the_permalink = $this->urlMatcher->getPermalink($idInt, $rowType); |
| 212 |
$the_permalink = $this->f->normalizeUrlString($the_permalink); |
| 213 |
$urlParts = parse_url($the_permalink); |
| 214 |
} |
| 215 |
|
| 216 |
abj_service('request_context')->debug_info = 'Likely match IDs processing permalink: ' . |
| 217 |
$the_permalink . ', $wasntReadyCount: ' . $wasntReadyCount; |
| 218 |
|
| 219 |
if (!is_array($urlParts) || !array_key_exists('path', $urlParts)) { |
| 220 |
continue; |
| 221 |
} |
| 222 |
if (is_string($the_permalink)) { |
| 223 |
$observedPermalinksById[$idInt] = $the_permalink; |
| 224 |
} |
| 225 |
$existingPageURL = $this->logic->removeHomeDirectory($urlParts['path']); |
| 226 |
$urlParts = null; |
| 227 |
|
| 228 |
$existingPageURLSpaces = $this->f->str_replace($this->separatingCharacters, " ", $existingPageURL); |
| 229 |
|
| 230 |
$existingPageURLCleaned = $this->urlMatcher->getLastURLPart($existingPageURLSpaces); |
| 231 |
$existingPageURLSpaces = null; |
| 232 |
|
| 233 |
$minDist = abs($this->f->strlen($existingPageURLCleaned) - $requestedURLCleanedLength); |
| 234 |
if ($fullURLspaces != '') { |
| 235 |
$minDist = min($minDist, abs($fullURLspacesLength - $requestedURLCleanedLength)); |
| 236 |
} |
| 237 |
$maxDist = $this->f->strlen($existingPageURLCleaned); |
| 238 |
if ($fullURLspaces != '') { |
| 239 |
$maxDist = min($maxDist, $fullURLspacesLength); |
| 240 |
} |
| 241 |
|
| 242 |
$existingPageURLCleanedWords = explode(" ", $existingPageURLCleaned); |
| 243 |
$wordsInCommon = array_intersect($userRequestedURLWords, $existingPageURLCleanedWords); |
| 244 |
$wordsInCommon = array_merge(array_unique($wordsInCommon, SORT_REGULAR), array()); |
| 245 |
if (count($wordsInCommon) > 0) { |
| 246 |
array_push($idsWithWordsInCommon, $id); |
| 247 |
$lengthOfTheLongestWordInCommon = max(array_map(array($this->f,'strlen'), $wordsInCommon)); |
| 248 |
$maxDist = $maxDist - $lengthOfTheLongestWordInCommon; |
| 249 |
} |
| 250 |
|
| 251 |
if (isset($minDistances[$minDist])) { |
| 252 |
array_push($minDistances[$minDist], $id); |
| 253 |
} else { |
| 254 |
$minDistances[$minDist] = [$id]; |
| 255 |
} |
| 256 |
|
| 257 |
if ($maxDist < 0) { |
| 258 |
$this->logger->errorMessage("maxDist is less than 0 (" . $maxDist . |
| 259 |
") for '" . $existingPageURLCleaned . "', wordsInCommon: " . |
| 260 |
json_encode($wordsInCommon) . ", "); |
| 261 |
$maxDist = 0; |
| 262 |
} else if ($maxDist > self::MAX_DIST) { |
| 263 |
$maxDist = self::MAX_DIST; |
| 264 |
} |
| 265 |
|
| 266 |
if (is_array($maxDistances[$maxDist])) { |
| 267 |
array_push($maxDistances[$maxDist], $id); |
| 268 |
} |
| 269 |
|
| 270 |
$row = array_pop($currentBatch); |
| 271 |
if ($row == null) { |
| 272 |
$maxAcceptableDistance = $this->getMaxAcceptableDistance($maxDistances, $onlyNeedThisManyPages); |
| 273 |
|
| 274 |
$currentBatch = $this->publishedPostsProvider->getNextBatch( |
| 275 |
$requestedURLCleanedLength, 1000, $maxAcceptableDistance); |
| 276 |
$row = array_pop($currentBatch); |
| 277 |
} |
| 278 |
} |
| 279 |
abj_service('request_context')->debug_info = ''; |
| 280 |
|
| 281 |
if ($wasntReadyCount > 0) { |
| 282 |
$this->logger->infoMessage("The permalink cache wasn't ready for " . $wasntReadyCount . " IDs."); |
| 283 |
} |
| 284 |
|
| 285 |
$candidateIds = $this->pruneAndPrioritizeCandidates( |
| 286 |
$maxDistances, $minDistances, $onlyNeedThisManyPages, |
| 287 |
$idsWithWordsInCommon, $ngramPrefilterApplied, $requestedURLCleaned |
| 288 |
); |
| 289 |
|
| 290 |
return $this->batchLookupPermalinks( |
| 291 |
array_values(array_unique($candidateIds)), $rowType, $observedPermalinksById |
| 292 |
); |
| 293 |
} |
| 294 |
|
| 295 |
/** |
| 296 |
* @param string $rowType |
| 297 |
* @param array<int, array<string, mixed>>|null $rows |
| 298 |
* @param string $requestedURLCleaned |
| 299 |
* @return string 'applied' if prefilter was used, 'early_return' if no matches exist, 'skipped' otherwise |
| 300 |
*/ |
| 301 |
private function tryApplyNgramPrefilter(string $rowType, ?array $rows, string $requestedURLCleaned): string { |
| 302 |
if ($rowType != 'pages' || $rows !== null) { |
| 303 |
return 'skipped'; |
| 304 |
} |
| 305 |
|
| 306 |
$cacheCount = $this->ngramFilter->getCacheCount(); |
| 307 |
|
| 308 |
if ($cacheCount < self::NGRAM_MIN_CACHE_ENTRIES) { |
| 309 |
$this->logger->debugMessage(sprintf( |
| 310 |
"N-gram prefilter skipped (gate 1: min entries): count=%d (need %d)", |
| 311 |
$cacheCount, self::NGRAM_MIN_CACHE_ENTRIES |
| 312 |
)); |
| 313 |
return 'skipped'; |
| 314 |
} |
| 315 |
if (!$this->ngramFilter->isCacheInitialized()) { |
| 316 |
$this->logger->debugMessage(sprintf( |
| 317 |
"N-gram prefilter skipped (gate 2: not initialized): count=%d", $cacheCount |
| 318 |
)); |
| 319 |
return 'skipped'; |
| 320 |
} |
| 321 |
$coverageRatio = $this->ngramFilter->getCacheCoverageRatio(); |
| 322 |
if ($coverageRatio < self::NGRAM_MIN_COVERAGE_RATIO) { |
| 323 |
$this->logger->debugMessage(sprintf( |
| 324 |
"N-gram prefilter skipped (gate 3: low coverage): ratio=%.2f (need %.2f)", |
| 325 |
$coverageRatio, self::NGRAM_MIN_COVERAGE_RATIO |
| 326 |
)); |
| 327 |
return 'skipped'; |
| 328 |
} |
| 329 |
|
| 330 |
$similarPages = $this->ngramFilter->findSimilarPages( |
| 331 |
$requestedURLCleaned, self::NGRAM_PREFILTER_THRESHOLD, self::NGRAM_PREFILTER_MAX_CANDIDATES |
| 332 |
); |
| 333 |
|
| 334 |
if (!empty($similarPages) && $this->publishedPostsProvider !== null) { |
| 335 |
$candidateIds = array_keys($similarPages); |
| 336 |
$this->publishedPostsProvider->resetBatch(); |
| 337 |
$this->publishedPostsProvider->restrictToIds($candidateIds); |
| 338 |
$this->logger->debugMessage(sprintf( |
| 339 |
"N-gram prefilter: Restricted to %d candidates (cache has %d entries, coverage=%.2f)", |
| 340 |
count($candidateIds), $cacheCount, $coverageRatio |
| 341 |
)); |
| 342 |
return 'applied'; |
| 343 |
} |
| 344 |
|
| 345 |
if ($this->skipNgramGate4) { |
| 346 |
$this->logger->debugMessage( |
| 347 |
"N-gram prefilter: zero candidates at Dice >= 0.3 \xe2\x80\x94 skipNgramGate4 is set, falling through to full scan" |
| 348 |
); |
| 349 |
return 'skipped'; |
| 350 |
} |
| 351 |
|
| 352 |
$this->logger->debugMessage( |
| 353 |
"N-gram prefilter: zero candidates at Dice >= 0.3 \xe2\x80\x94 no similar pages exist, returning early" |
| 354 |
); |
| 355 |
return 'early_return'; |
| 356 |
} |
| 357 |
|
| 358 |
/** |
| 359 |
* @param array<int, array<int, mixed>> $maxDistances |
| 360 |
* @param array<int, array<int, mixed>> $minDistances |
| 361 |
* @param int $onlyNeedThisManyPages |
| 362 |
* @param array<int, mixed> $idsWithWordsInCommon |
| 363 |
* @param bool $ngramPrefilterApplied |
| 364 |
* @param string $requestedURLCleaned |
| 365 |
* @return array<int, mixed> |
| 366 |
*/ |
| 367 |
private function pruneAndPrioritizeCandidates( |
| 368 |
array $maxDistances, array $minDistances, int $onlyNeedThisManyPages, |
| 369 |
array $idsWithWordsInCommon, bool $ngramPrefilterApplied, string $requestedURLCleaned |
| 370 |
): array { |
| 371 |
$pagesSeenSoFar = 0; |
| 372 |
$maxDistFound = self::MAX_LIKELY_DISTANCE; |
| 373 |
for ($currentDistanceIndex = 0; $currentDistanceIndex <= self::MAX_LIKELY_DISTANCE; $currentDistanceIndex++) { |
| 374 |
$pagesSeenSoFar += sizeof($maxDistances[$currentDistanceIndex]); |
| 375 |
if ($pagesSeenSoFar >= $onlyNeedThisManyPages) { |
| 376 |
$maxDistFound = $currentDistanceIndex; |
| 377 |
break; |
| 378 |
} |
| 379 |
} |
| 380 |
|
| 381 |
$listOfIDsToReturn = array(); |
| 382 |
for ($currentDistanceIndex = 0; $currentDistanceIndex <= $maxDistFound; $currentDistanceIndex++) { |
| 383 |
$listOfMinDistanceIDs = $minDistances[$currentDistanceIndex]; |
| 384 |
$listOfIDsToReturn = array_merge($listOfIDsToReturn, $listOfMinDistanceIDs); |
| 385 |
} |
| 386 |
|
| 387 |
$idsWithWords = array_intersect($listOfIDsToReturn, $idsWithWordsInCommon); |
| 388 |
$idsWithoutWords = array_diff($listOfIDsToReturn, $idsWithWordsInCommon); |
| 389 |
$listOfIDsToReturn = array_merge($idsWithWords, $idsWithoutWords); |
| 390 |
|
| 391 |
$beforeNGramCount = count($listOfIDsToReturn); |
| 392 |
if (!$ngramPrefilterApplied |
| 393 |
&& $beforeNGramCount > self::NGRAM_SECONDARY_MIN_CANDIDATES |
| 394 |
&& $this->ngramFilter->getCacheCount() >= self::NGRAM_MIN_CACHE_ENTRIES |
| 395 |
&& $this->ngramFilter->isCacheInitialized() |
| 396 |
&& $this->ngramFilter->getCacheCoverageRatio() >= self::NGRAM_MIN_COVERAGE_RATIO) { |
| 397 |
$similarPages = $this->ngramFilter->findSimilarPages( |
| 398 |
$requestedURLCleaned, |
| 399 |
self::NGRAM_SECONDARY_THRESHOLD, |
| 400 |
min($beforeNGramCount, self::NGRAM_SECONDARY_MAX_CANDIDATES) |
| 401 |
); |
| 402 |
if (!empty($similarPages)) { |
| 403 |
$ngramFilteredIDs = array_keys($similarPages); |
| 404 |
$listOfIDsToReturn = array_intersect($listOfIDsToReturn, $ngramFilteredIDs); |
| 405 |
usort($listOfIDsToReturn, function($a, $b) use ($similarPages) { |
| 406 |
$simA = isset($similarPages[$a]) ? $similarPages[$a] : 0; |
| 407 |
$simB = isset($similarPages[$b]) ? $similarPages[$b] : 0; |
| 408 |
return $simB <=> $simA; |
| 409 |
}); |
| 410 |
$this->logger->debugMessage(sprintf( |
| 411 |
"N-gram filter (secondary): %d to %d candidates (%.1f%% reduction)", |
| 412 |
$beforeNGramCount, count($listOfIDsToReturn), |
| 413 |
100 * (1 - count($listOfIDsToReturn) / max(1, $beforeNGramCount)) |
| 414 |
)); |
| 415 |
} |
| 416 |
} |
| 417 |
|
| 418 |
if (count($listOfIDsToReturn) > 300 && count($idsWithWordsInCommon) >= $onlyNeedThisManyPages) { |
| 419 |
$maybeOKguesses = array_intersect($listOfIDsToReturn, $idsWithWordsInCommon); |
| 420 |
return (count($maybeOKguesses) >= $onlyNeedThisManyPages) |
| 421 |
? $maybeOKguesses : $idsWithWordsInCommon; |
| 422 |
} |
| 423 |
return $listOfIDsToReturn; |
| 424 |
} |
| 425 |
|
| 426 |
/** |
| 427 |
* @param array<int, mixed> $ids |
| 428 |
* @param string $rowType |
| 429 |
* @param array<int, string> $observedPermalinksById |
| 430 |
* @return array<int|string, string> |
| 431 |
*/ |
| 432 |
private function batchLookupPermalinks(array $ids, string $rowType, array $observedPermalinksById = array()): array { |
| 433 |
if (empty($ids)) { |
| 434 |
return []; |
| 435 |
} |
| 436 |
$result = []; |
| 437 |
if ($rowType === 'pages' || $rowType === 'image') { |
| 438 |
$intIds = array_map(function($v) { return is_scalar($v) ? (int)$v : 0; }, $ids); |
| 439 |
$rows = $this->contentRepository->getPermalinksByIds($intIds); |
| 440 |
foreach ($rows as $row) { |
| 441 |
$row = (array)$row; |
| 442 |
if (isset($row['id'], $row['url']) && is_string($row['url'])) { |
| 443 |
$result[(int)$row['id']] = $this->f->normalizeUrlString($row['url']); |
| 444 |
} |
| 445 |
} |
| 446 |
foreach ($intIds as $id) { |
| 447 |
if (!isset($result[$id]) && isset($observedPermalinksById[$id])) { |
| 448 |
$result[$id] = $this->f->normalizeUrlString($observedPermalinksById[$id]); |
| 449 |
} |
| 450 |
} |
| 451 |
} else { |
| 452 |
foreach ($ids as $id) { |
| 453 |
$idInt = is_scalar($id) ? (int)$id : 0; |
| 454 |
$permalink = $this->urlMatcher->getPermalink($idInt, $rowType); |
| 455 |
if (is_string($permalink) && $permalink !== '') { |
| 456 |
$result[$id] = $permalink; |
| 457 |
} |
| 458 |
} |
| 459 |
} |
| 460 |
return $result; |
| 461 |
} |
| 462 |
|
| 463 |
/** |
| 464 |
* @param array<int, array<int, mixed>> $maxDistances |
| 465 |
* @param int $onlyNeedThisManyPages |
| 466 |
* @return int |
| 467 |
*/ |
| 468 |
function getMaxAcceptableDistance(array $maxDistances, int $onlyNeedThisManyPages): int { |
| 469 |
$pagesSeenSoFar = 0; |
| 470 |
$currentDistanceIndex = 0; |
| 471 |
$maxDistFound = self::MAX_LIKELY_DISTANCE; |
| 472 |
for ($currentDistanceIndex = 0; $currentDistanceIndex <= self::MAX_LIKELY_DISTANCE; $currentDistanceIndex++) { |
| 473 |
$pagesSeenSoFar += sizeof($maxDistances[$currentDistanceIndex]); |
| 474 |
|
| 475 |
if ($pagesSeenSoFar >= $onlyNeedThisManyPages) { |
| 476 |
$maxDistFound = $currentDistanceIndex; |
| 477 |
break; |
| 478 |
} |
| 479 |
} |
| 480 |
|
| 481 |
$acceptableDistance = (int)($maxDistFound * 1.1); |
| 482 |
return $acceptableDistance; |
| 483 |
} |
| 484 |
|
| 485 |
/** |
| 486 |
* @param string $str1 |
| 487 |
* @param string $str2 |
| 488 |
* @return int |
| 489 |
* @throws Exception |
| 490 |
*/ |
| 491 |
function customLevenshtein($str1, $str2) { |
| 492 |
if ($this->enablePerformanceCounters) { |
| 493 |
$this->levenshteinCallCount++; |
| 494 |
} |
| 495 |
abj_service('request_context')->debug_info = 'customLevenshtein. str1: ' . esc_html($str1) . ', str2: ' . esc_html($str2); |
| 496 |
|
| 497 |
$RowLen = $this->f->strlen($str1); |
| 498 |
$ColLen = $this->f->strlen($str2); |
| 499 |
$cost = 0; |
| 500 |
|
| 501 |
if (max($RowLen, $ColLen) > ABJ404_MAX_URL_LENGTH) { |
| 502 |
throw new \Exception("Maximum string length in customLevenshtein is " . // allow-raw-error: assertion, should never reach user |
| 503 |
ABJ404_MAX_URL_LENGTH . ". Yours is " . max($RowLen, $ColLen) . "."); |
| 504 |
} |
| 505 |
|
| 506 |
if (strlen($str1) <= 255 && strlen($str2) <= 255) { |
| 507 |
return levenshtein($str1, $str2); |
| 508 |
} |
| 509 |
|
| 510 |
if ($RowLen == 0) { |
| 511 |
return $ColLen; |
| 512 |
} else if ($ColLen == 0) { |
| 513 |
return $RowLen; |
| 514 |
} |
| 515 |
|
| 516 |
$chars1 = mb_str_split($str1, 1, 'UTF-8'); |
| 517 |
$chars2 = mb_str_split($str2, 1, 'UTF-8'); |
| 518 |
|
| 519 |
$v0 = array_fill(0, $RowLen + 1, 0); |
| 520 |
$v1 = array_fill(0, $RowLen + 1, 0); |
| 521 |
|
| 522 |
for ($RowIdx = 1; $RowIdx <= $RowLen; $RowIdx++) { |
| 523 |
$v0[$RowIdx] = $RowIdx; |
| 524 |
} |
| 525 |
|
| 526 |
for ($ColIdx = 1; $ColIdx <= $ColLen; $ColIdx++) { |
| 527 |
$v1[0] = $ColIdx; |
| 528 |
|
| 529 |
for ($RowIdx = 1; $RowIdx <= $RowLen; $RowIdx++) { |
| 530 |
$cost = ($chars1[$RowIdx - 1] === $chars2[$ColIdx - 1]) ? 0 : 1; |
| 531 |
$v1[$RowIdx] = min($v0[$RowIdx] + 1, $v1[$RowIdx - 1] + 1, $v0[$RowIdx - 1] + $cost); |
| 532 |
} |
| 533 |
|
| 534 |
$vTmp = $v0; |
| 535 |
$v0 = $v1; |
| 536 |
$v1 = $vTmp; |
| 537 |
} |
| 538 |
|
| 539 |
abj_service('request_context')->debug_info = 'Cleared after customLevenshtein.'; |
| 540 |
return $v0[$RowLen]; |
| 541 |
} |
| 542 |
|
| 543 |
} |
| 544 |
|
