| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* Candidate filtering, scoring, and matching on posts/tags/categories |
| 9 |
* for the spell-checking subsystem. |
| 10 |
* |
| 11 |
* Extracted from SpellCheckerTrait_CandidateFiltering as a standalone class |
| 12 |
* with explicit dependency injection. |
| 13 |
*/ |
| 14 |
class ABJ_404_Solution_SpellCandidateFilter { |
| 15 |
|
| 16 |
/** @var ABJ_404_Solution_Functions */ |
| 17 |
private $f; |
| 18 |
|
| 19 |
/** @var ABJ_404_Solution_PluginLogic */ |
| 20 |
private $logic; |
| 21 |
|
| 22 |
/** @var ABJ_404_Solution_Logging */ |
| 23 |
private $logger; |
| 24 |
|
| 25 |
/** @var ABJ_404_Solution_ContentRepository */ |
| 26 |
private $contentRepository; |
| 27 |
|
| 28 |
/** @var ABJ_404_Solution_SpellURLMatcher */ |
| 29 |
private $urlMatcher; |
| 30 |
|
| 31 |
/** @var ABJ_404_Solution_SpellLevenshteinEngine */ |
| 32 |
private $levenshteinEngine; |
| 33 |
|
| 34 |
/** @var ABJ_404_Solution_SpellPostListeners */ |
| 35 |
private $postListeners; |
| 36 |
|
| 37 |
/** @var string|int|null */ |
| 38 |
private $custom404PageID; |
| 39 |
|
| 40 |
/** @var array<int, string> */ |
| 41 |
private array $separatingCharacters; |
| 42 |
|
| 43 |
/** @var array<int, string> */ |
| 44 |
private array $separatingCharactersForImages; |
| 45 |
|
| 46 |
/** |
| 47 |
* @param ABJ_404_Solution_Functions $functions |
| 48 |
* @param ABJ_404_Solution_PluginLogic $logic |
| 49 |
* @param ABJ_404_Solution_Logging $logger |
| 50 |
* @param ABJ_404_Solution_ContentRepository $contentRepository |
| 51 |
* @param ABJ_404_Solution_SpellURLMatcher $urlMatcher |
| 52 |
* @param ABJ_404_Solution_SpellLevenshteinEngine $levenshteinEngine |
| 53 |
* @param ABJ_404_Solution_SpellPostListeners $postListeners |
| 54 |
* @param string|int|null $custom404PageID |
| 55 |
* @param array<int, string> $separatingCharacters |
| 56 |
* @param array<int, string> $separatingCharactersForImages |
| 57 |
*/ |
| 58 |
public function __construct( |
| 59 |
$functions, $logic, $logger, $contentRepository, |
| 60 |
$urlMatcher, $levenshteinEngine, $postListeners, |
| 61 |
$custom404PageID, array $separatingCharacters, array $separatingCharactersForImages |
| 62 |
) { |
| 63 |
$this->f = $functions; |
| 64 |
$this->logic = $logic; |
| 65 |
$this->logger = $logger; |
| 66 |
$this->contentRepository = $contentRepository; |
| 67 |
$this->urlMatcher = $urlMatcher; |
| 68 |
$this->levenshteinEngine = $levenshteinEngine; |
| 69 |
$this->postListeners = $postListeners; |
| 70 |
$this->custom404PageID = $custom404PageID; |
| 71 |
$this->separatingCharacters = $separatingCharacters; |
| 72 |
$this->separatingCharactersForImages = $separatingCharactersForImages; |
| 73 |
} |
| 74 |
|
| 75 |
/** |
| 76 |
* @param string $requestedURLRaw |
| 77 |
* @param string $includeCats |
| 78 |
* @param string $includeTags |
| 79 |
* @return array<int, mixed> |
| 80 |
*/ |
| 81 |
function findMatchingPosts(string $requestedURLRaw, string $includeCats = '1', string $includeTags = '1') { |
| 82 |
|
| 83 |
$options = $this->logic->getOptions(); |
| 84 |
$excludePagesCount = 0; |
| 85 |
$excludePagesRaw = isset($options['excludePages[]']) && is_string($options['excludePages[]']) ? $options['excludePages[]'] : ''; |
| 86 |
if (trim($excludePagesRaw) !== '') { |
| 87 |
$jsonResult = json_decode($excludePagesRaw); |
| 88 |
if (!is_array($jsonResult)) { |
| 89 |
$jsonResult = array($jsonResult); |
| 90 |
} |
| 91 |
$excludePagesCount = count($jsonResult); |
| 92 |
} |
| 93 |
$suggestMaxRaw = isset($options['suggest_max']) && is_scalar($options['suggest_max']) ? $options['suggest_max'] : 5; |
| 94 |
$maxCacheCount = absint($suggestMaxRaw) + $excludePagesCount; |
| 95 |
|
| 96 |
$requestedURLSpaces = $this->f->str_replace($this->separatingCharacters, " ", $requestedURLRaw); |
| 97 |
$requestedURLCleaned = $this->urlMatcher->getLastURLPart($requestedURLSpaces); |
| 98 |
$fullURLspacesCleaned = $this->f->str_replace('/', " ", $requestedURLSpaces); |
| 99 |
if ($fullURLspacesCleaned == $requestedURLCleaned) { |
| 100 |
$fullURLspacesCleaned = ''; |
| 101 |
} |
| 102 |
|
| 103 |
$this->postListeners->initializePublishedPostsProvider(); |
| 104 |
$this->levenshteinEngine->setPublishedPostsProvider( |
| 105 |
$this->postListeners->getPublishedPostsProvider() |
| 106 |
); |
| 107 |
|
| 108 |
$rowType = 'pages'; |
| 109 |
$permalinks = array(); |
| 110 |
$permalinks = $this->matchOnPosts($permalinks, $requestedURLRaw, $requestedURLCleaned, |
| 111 |
$fullURLspacesCleaned, $rowType); |
| 112 |
|
| 113 |
if ($includeTags == "1") { |
| 114 |
$permalinks = $this->matchOnTags($permalinks, $requestedURLCleaned, $fullURLspacesCleaned, 'tags'); |
| 115 |
} |
| 116 |
|
| 117 |
if ($includeCats == "1") { |
| 118 |
$permalinks = $this->matchOnCats($permalinks, $requestedURLCleaned, $fullURLspacesCleaned, 'categories'); |
| 119 |
} |
| 120 |
|
| 121 |
$permalinks = $this->removeExcludedPages($options, $permalinks); |
| 122 |
|
| 123 |
arsort($permalinks); |
| 124 |
|
| 125 |
$permalinks = $this->removeExcludedPagesWithRegex($options, $permalinks, $maxCacheCount); |
| 126 |
|
| 127 |
$permalinks = array_splice($permalinks, 0, $maxCacheCount); |
| 128 |
|
| 129 |
$returnValue = array($permalinks,$rowType); |
| 130 |
$this->contentRepository->storeSpellingPermalinksToCache($requestedURLRaw, $returnValue); |
| 131 |
$ctx = abj_service('request_context'); |
| 132 |
$ctx->permalinks_found = (string)json_encode($returnValue); |
| 133 |
$ctx->permalinks_kept = (string)json_encode($permalinks); |
| 134 |
|
| 135 |
return $returnValue; |
| 136 |
} |
| 137 |
|
| 138 |
/** |
| 139 |
* @param array<string, mixed> $options |
| 140 |
* @param array<string, string> $permalinks |
| 141 |
* @return array<string, string> |
| 142 |
*/ |
| 143 |
function removeExcludedPages(array $options, array $permalinks): array { |
| 144 |
$excludePagesJsonRaw = isset($options['excludePages[]']) ? $options['excludePages[]'] : ''; |
| 145 |
$excludePagesJson = is_string($excludePagesJsonRaw) ? $excludePagesJsonRaw : ''; |
| 146 |
if (trim($excludePagesJson) == '' && $this->custom404PageID == null) { |
| 147 |
return $permalinks; |
| 148 |
} |
| 149 |
|
| 150 |
$excludePages = json_decode($excludePagesJson); |
| 151 |
if (!is_array($excludePages)) { |
| 152 |
$excludePages = array($excludePages); |
| 153 |
} |
| 154 |
|
| 155 |
if ($this->custom404PageID != null) { |
| 156 |
array_push($excludePages, $this->custom404PageID); |
| 157 |
} |
| 158 |
|
| 159 |
for ($i = 0; $i < count($excludePages); $i++) { |
| 160 |
$excludePage = $excludePages[$i]; |
| 161 |
if ($excludePage == null || trim($excludePage) == '') { |
| 162 |
continue; |
| 163 |
} |
| 164 |
unset($permalinks[(string)$excludePage]); |
| 165 |
} |
| 166 |
|
| 167 |
return $permalinks; |
| 168 |
} |
| 169 |
|
| 170 |
/** |
| 171 |
* @param array<string, mixed> $options |
| 172 |
* @param array<string, string> $permalinks |
| 173 |
* @param int $maxCacheCount |
| 174 |
* @return array<string, string> |
| 175 |
*/ |
| 176 |
function removeExcludedPagesWithRegex(array $options, array $permalinks, int $maxCacheCount): array { |
| 177 |
if (!isset($options['suggest_regex_exclusions_usable']) || |
| 178 |
!is_array($options['suggest_regex_exclusions_usable']) || |
| 179 |
empty($options['suggest_regex_exclusions_usable'])) { |
| 180 |
return $permalinks; |
| 181 |
} |
| 182 |
|
| 183 |
$suggestionsKeptSoFar = 0; |
| 184 |
$regexExclusions = $options['suggest_regex_exclusions_usable']; |
| 185 |
|
| 186 |
$keys_to_check = array_keys($permalinks); |
| 187 |
|
| 188 |
foreach ($keys_to_check as $key) { |
| 189 |
if (!array_key_exists($key, $permalinks)) { |
| 190 |
continue; |
| 191 |
} |
| 192 |
|
| 193 |
$keyParts = explode('|', $key); |
| 194 |
if (count($keyParts) !== 2 || !is_numeric($keyParts[0])) { |
| 195 |
$this->logger->debugMessage("Skipping invalid key format in removeExcludedPagesWithRegex: " . $key); |
| 196 |
continue; |
| 197 |
} |
| 198 |
|
| 199 |
$id = (int)$keyParts[0]; |
| 200 |
$typeConstant = $keyParts[1]; |
| 201 |
|
| 202 |
$rowTypeString = $this->mapTypeConstantToString($typeConstant); |
| 203 |
if ($rowTypeString === null) { |
| 204 |
$this->logger->debugMessage("Skipping unknown type constant in removeExcludedPagesWithRegex: " . $typeConstant . " for key: " . $key); |
| 205 |
continue; |
| 206 |
} |
| 207 |
|
| 208 |
$urlOfPage = $this->urlMatcher->getPermalink($id, $rowTypeString); |
| 209 |
if ($urlOfPage === null || trim($urlOfPage) === '') { |
| 210 |
$this->logger->debugMessage("Skipping null/empty URL for key in removeExcludedPagesWithRegex: " . $key); |
| 211 |
continue; |
| 212 |
} |
| 213 |
|
| 214 |
$urlParts = parse_url($urlOfPage); |
| 215 |
if (!is_array($urlParts) || !isset($urlParts['path'])) { |
| 216 |
$this->logger->debugMessage("Skipping URL that failed parse_url for key in removeExcludedPagesWithRegex: " . $key . ", URL: " . esc_url($urlOfPage)); |
| 217 |
continue; |
| 218 |
} |
| 219 |
$pathOnly = $this->logic->removeHomeDirectory($urlParts['path']); |
| 220 |
if ( $pathOnly !== '' && substr($pathOnly, 0, 1) !== '/' ) { |
| 221 |
$pathOnly = '/' . $pathOnly; |
| 222 |
} |
| 223 |
if ( $pathOnly === '' ) { |
| 224 |
$pathOnly = '/'; |
| 225 |
} |
| 226 |
|
| 227 |
$stringToMatch = $pathOnly; |
| 228 |
|
| 229 |
$kept = true; |
| 230 |
foreach ($regexExclusions as $pattern) { |
| 231 |
$patternToExcludeNoSlashes = stripslashes($pattern); |
| 232 |
$matches = array(); |
| 233 |
|
| 234 |
if ($this->f->regexMatch($patternToExcludeNoSlashes, $stringToMatch, $matches)) { |
| 235 |
unset($permalinks[$key]); |
| 236 |
$this->logger->debugMessage("Regex excluded suggestion. Key: " . $key . |
| 237 |
", Path: '" . esc_html($stringToMatch) . "', Pattern: '" . esc_html($patternToExcludeNoSlashes) . "'"); |
| 238 |
$kept = false; |
| 239 |
break; |
| 240 |
} |
| 241 |
} |
| 242 |
|
| 243 |
if ($kept) { |
| 244 |
$suggestionsKeptSoFar++; |
| 245 |
} |
| 246 |
if ($suggestionsKeptSoFar >= $maxCacheCount) { |
| 247 |
break; |
| 248 |
} |
| 249 |
} |
| 250 |
|
| 251 |
return $permalinks; |
| 252 |
} |
| 253 |
|
| 254 |
/** |
| 255 |
* @param mixed $typeConstant |
| 256 |
* @return string|null |
| 257 |
*/ |
| 258 |
private function mapTypeConstantToString($typeConstant) { |
| 259 |
if (!defined('ABJ404_TYPE_POST')) define('ABJ404_TYPE_POST', 1); |
| 260 |
if (!defined('ABJ404_TYPE_CAT')) define('ABJ404_TYPE_CAT', 2); |
| 261 |
if (!defined('ABJ404_TYPE_TAG')) define('ABJ404_TYPE_TAG', 3); |
| 262 |
|
| 263 |
$typeConstantStr = is_scalar($typeConstant) ? (string)$typeConstant : ''; |
| 264 |
switch ($typeConstantStr) { |
| 265 |
case ABJ404_TYPE_POST: |
| 266 |
return 'pages'; |
| 267 |
case ABJ404_TYPE_TAG: |
| 268 |
return 'tags'; |
| 269 |
case ABJ404_TYPE_CAT: |
| 270 |
return 'categories'; |
| 271 |
default: |
| 272 |
return null; |
| 273 |
} |
| 274 |
} |
| 275 |
|
| 276 |
/** |
| 277 |
* @param array<string, string> $permalinks |
| 278 |
* @param string $requestedURLCleaned |
| 279 |
* @param string $fullURLspacesCleaned |
| 280 |
* @param string $rowType |
| 281 |
* @return array<string, string> |
| 282 |
*/ |
| 283 |
function matchOnCats(array $permalinks, string $requestedURLCleaned, string $fullURLspacesCleaned, string $rowType): array { |
| 284 |
|
| 285 |
$rows = $this->contentRepository->getPublishedCategories(); |
| 286 |
$rows = $this->urlMatcher->getOnlyIDandTermID($rows); |
| 287 |
|
| 288 |
$likelyMatchIDsAndPermalinks = $this->levenshteinEngine->getLikelyMatchIDs($requestedURLCleaned, $fullURLspacesCleaned, 'categories', $rows); |
| 289 |
$likelyMatchIDs = array_keys($likelyMatchIDsAndPermalinks); |
| 290 |
|
| 291 |
$options = $this->logic->getOptions(); |
| 292 |
$suggestMaxRaw = isset($options['suggest_max']) && is_scalar($options['suggest_max']) ? $options['suggest_max'] : 5; |
| 293 |
$suggestMax = absint($suggestMaxRaw); |
| 294 |
$topKScores = new SplMinHeap(); |
| 295 |
$requestedURLCleanedLength = $this->f->strlen($requestedURLCleaned); |
| 296 |
|
| 297 |
foreach ($likelyMatchIDs as $id) { |
| 298 |
$the_permalink = $this->urlMatcher->getPermalink((int)$id, 'categories'); |
| 299 |
$urlParts = parse_url(is_string($the_permalink) ? $the_permalink : ''); |
| 300 |
if (!is_array($urlParts) || !isset($urlParts['path'])) { |
| 301 |
continue; |
| 302 |
} |
| 303 |
$pathOnly = $this->logic->removeHomeDirectory($urlParts['path']); |
| 304 |
$scoreBasis = $this->f->strlen($pathOnly); |
| 305 |
if ($scoreBasis == 0) { |
| 306 |
continue; |
| 307 |
} |
| 308 |
|
| 309 |
if ($topKScores->count() >= $suggestMax) { |
| 310 |
$worstAcceptableScore = $topKScores->top(); |
| 311 |
|
| 312 |
$maxAllowedLevenshtein = ((100 - $worstAcceptableScore) * $scoreBasis) / 100; |
| 313 |
$pathOnlyLength = $this->f->strlen($pathOnly); |
| 314 |
$minPossibleDistance = abs($requestedURLCleanedLength - $pathOnlyLength); |
| 315 |
|
| 316 |
if ($minPossibleDistance > $maxAllowedLevenshtein) { |
| 317 |
continue; |
| 318 |
} |
| 319 |
} |
| 320 |
|
| 321 |
$levscore = $this->levenshteinEngine->customLevenshtein($requestedURLCleaned, $pathOnly); |
| 322 |
|
| 323 |
if ($fullURLspacesCleaned != '') { |
| 324 |
$tentativeScore = 100 - (($levscore / $scoreBasis) * 100); |
| 325 |
if ($tentativeScore < 95) { |
| 326 |
$pathOnlySpaces = $this->f->str_replace($this->separatingCharacters, " ", $pathOnly); |
| 327 |
$pathOnlySpaces = trim($this->f->str_replace('/', " ", $pathOnlySpaces)); |
| 328 |
$levscore = min($levscore, $this->levenshteinEngine->customLevenshtein($fullURLspacesCleaned, $pathOnlySpaces)); |
| 329 |
} |
| 330 |
} |
| 331 |
|
| 332 |
$onlyLastPart = $this->urlMatcher->getLastURLPart($pathOnly); |
| 333 |
if ($onlyLastPart != '' && $onlyLastPart != $pathOnly) { |
| 334 |
$levscore = min($levscore, $this->levenshteinEngine->customLevenshtein($requestedURLCleaned, $onlyLastPart)); |
| 335 |
} |
| 336 |
|
| 337 |
$score = 100 - (($levscore / $scoreBasis) * 100); |
| 338 |
$permalinks[$id . "|" . ABJ404_TYPE_CAT] = number_format($score, 4, '.', ''); |
| 339 |
|
| 340 |
$topKScores->insert($score); |
| 341 |
if ($topKScores->count() > $suggestMax) { |
| 342 |
$topKScores->extract(); |
| 343 |
} |
| 344 |
} |
| 345 |
|
| 346 |
return $permalinks; |
| 347 |
} |
| 348 |
|
| 349 |
/** |
| 350 |
* @param array<string, string> $permalinks |
| 351 |
* @param string $requestedURLCleaned |
| 352 |
* @param string $fullURLspacesCleaned |
| 353 |
* @param string $rowType |
| 354 |
* @return array<string, string> |
| 355 |
*/ |
| 356 |
function matchOnTags(array $permalinks, string $requestedURLCleaned, string $fullURLspacesCleaned, string $rowType): array { |
| 357 |
|
| 358 |
$rows = $this->contentRepository->getPublishedTags(); |
| 359 |
$rows = $this->urlMatcher->getOnlyIDandTermID($rows); |
| 360 |
|
| 361 |
$likelyMatchIDsAndPermalinks = $this->levenshteinEngine->getLikelyMatchIDs($requestedURLCleaned, $fullURLspacesCleaned, 'tags', $rows); |
| 362 |
$likelyMatchIDs = array_keys($likelyMatchIDsAndPermalinks); |
| 363 |
|
| 364 |
$options = $this->logic->getOptions(); |
| 365 |
$suggestMaxRawT = isset($options['suggest_max']) && is_scalar($options['suggest_max']) ? $options['suggest_max'] : 5; |
| 366 |
$suggestMax = absint($suggestMaxRawT); |
| 367 |
$topKScores = new SplMinHeap(); |
| 368 |
$requestedURLCleanedLength = $this->f->strlen($requestedURLCleaned); |
| 369 |
|
| 370 |
foreach ($likelyMatchIDs as $id) { |
| 371 |
$the_permalink = $this->urlMatcher->getPermalink((int)$id, 'tags'); |
| 372 |
$urlParts = parse_url(is_string($the_permalink) ? $the_permalink : ''); |
| 373 |
if (!is_array($urlParts) || !isset($urlParts['path'])) { |
| 374 |
continue; |
| 375 |
} |
| 376 |
$pathOnly = $this->logic->removeHomeDirectory($urlParts['path']); |
| 377 |
$scoreBasis = $this->f->strlen($pathOnly); |
| 378 |
if ($scoreBasis == 0) { |
| 379 |
continue; |
| 380 |
} |
| 381 |
|
| 382 |
if ($topKScores->count() >= $suggestMax) { |
| 383 |
$worstAcceptableScore = $topKScores->top(); |
| 384 |
|
| 385 |
$maxAllowedLevenshtein = ((100 - $worstAcceptableScore) * $scoreBasis) / 100; |
| 386 |
$pathOnlyLength = $this->f->strlen($pathOnly); |
| 387 |
$minPossibleDistance = abs($requestedURLCleanedLength - $pathOnlyLength); |
| 388 |
|
| 389 |
if ($minPossibleDistance > $maxAllowedLevenshtein) { |
| 390 |
continue; |
| 391 |
} |
| 392 |
} |
| 393 |
|
| 394 |
$levscore = $this->levenshteinEngine->customLevenshtein($requestedURLCleaned, $pathOnly); |
| 395 |
|
| 396 |
if ($fullURLspacesCleaned != '') { |
| 397 |
$tentativeScore = 100 - (($levscore / $scoreBasis) * 100); |
| 398 |
if ($tentativeScore < 95) { |
| 399 |
$pathOnlySpaces = $this->f->str_replace($this->separatingCharacters, " ", $pathOnly); |
| 400 |
$pathOnlySpaces = trim($this->f->str_replace('/', " ", $pathOnlySpaces)); |
| 401 |
$levscore = min($levscore, $this->levenshteinEngine->customLevenshtein($fullURLspacesCleaned, $pathOnlySpaces)); |
| 402 |
} |
| 403 |
} |
| 404 |
$score = 100 - (($levscore / $scoreBasis) * 100); |
| 405 |
$permalinks[$id . "|" . ABJ404_TYPE_TAG] = number_format($score, 4, '.', ''); |
| 406 |
|
| 407 |
$topKScores->insert($score); |
| 408 |
if ($topKScores->count() > $suggestMax) { |
| 409 |
$topKScores->extract(); |
| 410 |
} |
| 411 |
} |
| 412 |
|
| 413 |
return $permalinks; |
| 414 |
} |
| 415 |
|
| 416 |
/** |
| 417 |
* @param array<string, string> $permalinks |
| 418 |
* @param string $requestedURLRaw |
| 419 |
* @param string $requestedURLCleaned |
| 420 |
* @param string $fullURLspacesCleaned |
| 421 |
* @param string $rowType |
| 422 |
* @return array<string, string> |
| 423 |
*/ |
| 424 |
function matchOnPosts(array $permalinks, string $requestedURLRaw, string $requestedURLCleaned, string $fullURLspacesCleaned, string $rowType): array { |
| 425 |
|
| 426 |
$likelyMatchIDsAndPermalinks = $this->levenshteinEngine->getLikelyMatchIDs($requestedURLCleaned, $fullURLspacesCleaned, $rowType); |
| 427 |
$likelyMatchIDs = array_keys($likelyMatchIDsAndPermalinks); |
| 428 |
|
| 429 |
$this->logger->debugMessage("Found " . count($likelyMatchIDs) . " likely match IDs."); |
| 430 |
|
| 431 |
$options = $this->logic->getOptions(); |
| 432 |
$suggestMaxRawP = isset($options['suggest_max']) && is_scalar($options['suggest_max']) ? $options['suggest_max'] : 5; |
| 433 |
$suggestMax = absint($suggestMaxRawP); |
| 434 |
$topKScores = new SplMinHeap(); |
| 435 |
$requestedURLCleanedLength = $this->f->strlen($requestedURLCleaned); |
| 436 |
|
| 437 |
while (count($likelyMatchIDs) > 0) { |
| 438 |
$id = array_shift($likelyMatchIDs); |
| 439 |
|
| 440 |
$the_permalink = $likelyMatchIDsAndPermalinks[$id]; |
| 441 |
$thePermalinkStr = is_string($the_permalink) ? $the_permalink : ''; |
| 442 |
$urlParts = parse_url($thePermalinkStr); |
| 443 |
if (!is_array($urlParts) || !isset($urlParts['path'])) { |
| 444 |
continue; |
| 445 |
} |
| 446 |
$existingPageURL = $this->logic->removeHomeDirectory($urlParts['path']); |
| 447 |
$existingPageURLSpaces = $this->f->str_replace($this->separatingCharacters, " ", $existingPageURL); |
| 448 |
|
| 449 |
$existingPageURLCleaned = $this->urlMatcher->getLastURLPart($existingPageURLSpaces); |
| 450 |
$scoreBasis = $this->f->strlen($existingPageURLCleaned) * 3; |
| 451 |
if ($scoreBasis == 0) { |
| 452 |
continue; |
| 453 |
} |
| 454 |
|
| 455 |
if ($topKScores->count() >= $suggestMax) { |
| 456 |
$worstAcceptableScore = $topKScores->top(); |
| 457 |
|
| 458 |
$maxAllowedLevenshtein = ((100 - $worstAcceptableScore) * $scoreBasis) / 100; |
| 459 |
|
| 460 |
$existingURLCleanedLength = $this->f->strlen($existingPageURLCleaned); |
| 461 |
$minPossibleDistance = abs($requestedURLCleanedLength - $existingURLCleanedLength); |
| 462 |
|
| 463 |
if ($minPossibleDistance > $maxAllowedLevenshtein) { |
| 464 |
continue; |
| 465 |
} |
| 466 |
} |
| 467 |
|
| 468 |
$levscore = $this->levenshteinEngine->customLevenshtein($requestedURLCleaned, $existingPageURLCleaned); |
| 469 |
|
| 470 |
if ($fullURLspacesCleaned != '') { |
| 471 |
$tentativeScore = 100 - (($levscore / $scoreBasis) * 100); |
| 472 |
if ($tentativeScore < 95) { |
| 473 |
$levscore = min($levscore, $this->levenshteinEngine->customLevenshtein($fullURLspacesCleaned, $existingPageURLCleaned)); |
| 474 |
} |
| 475 |
} |
| 476 |
|
| 477 |
if ($rowType == 'image') { |
| 478 |
$strippedImageName = $this->f->regexReplace('(.+)([-]\d{1,5}[x]\d{1,5})([.].+)', |
| 479 |
'\\1\\3', $requestedURLRaw); |
| 480 |
|
| 481 |
if (($strippedImageName != null) && ($strippedImageName != $requestedURLRaw)) { |
| 482 |
$strippedImageName = $this->f->str_replace($this->separatingCharactersForImages, " ", $strippedImageName); |
| 483 |
$levscore = min($levscore, $this->levenshteinEngine->customLevenshtein($strippedImageName, $existingPageURL)); |
| 484 |
|
| 485 |
$strippedImageName = $this->urlMatcher->getLastURLPart($strippedImageName); |
| 486 |
$levscore = min($levscore, $this->levenshteinEngine->customLevenshtein($strippedImageName, $existingPageURLCleaned)); |
| 487 |
} |
| 488 |
} |
| 489 |
$score = 100 - (($levscore / $scoreBasis) * 100); |
| 490 |
$permalinks[$id . "|" . ABJ404_TYPE_POST] = number_format($score, 4, '.', ''); |
| 491 |
|
| 492 |
$topKScores->insert($score); |
| 493 |
if ($topKScores->count() > $suggestMax) { |
| 494 |
$topKScores->extract(); |
| 495 |
} |
| 496 |
} |
| 497 |
|
| 498 |
return $permalinks; |
| 499 |
} |
| 500 |
|
| 501 |
} |
| 502 |
|
