| 1 |
<?php |
| 2 |
|
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* Matching engine that delegates to SpellChecker::getPermalinkUsingSpelling(). |
| 9 |
* |
| 10 |
* Uses Levenshtein distance / N-gram filtering to find the closest matching |
| 11 |
* published page. Expensive, so shouldRun() filters out URL shapes that are |
| 12 |
* unlikely to be useful typo corrections. |
| 13 |
*/ |
| 14 |
class ABJ_404_Solution_SpellingMatchingEngine implements ABJ_404_Solution_MatchingEngine { |
| 15 |
|
| 16 |
/** @var ABJ_404_Solution_SpellChecker */ |
| 17 |
private $spellChecker; |
| 18 |
|
| 19 |
/** |
| 20 |
* @param ABJ_404_Solution_SpellChecker $spellChecker |
| 21 |
*/ |
| 22 |
public function __construct(ABJ_404_Solution_SpellChecker $spellChecker) { |
| 23 |
$this->spellChecker = $spellChecker; |
| 24 |
} |
| 25 |
|
| 26 |
/** |
| 27 |
* The label stored in the redirect row's `engine` column for anything this |
| 28 |
* engine scored. Static because SpellChecker records a rejected match under |
| 29 |
* the same name (see ABJ_404_Solution_NearMissRecorder) before this engine |
| 30 |
* ever sees the result, and the two must not drift apart: an automatic |
| 31 |
* redirect and a captured near miss produced by the same code have to read |
| 32 |
* the same in the admin Engine column. |
| 33 |
* |
| 34 |
* @return string |
| 35 |
*/ |
| 36 |
public static function engineName(): string { |
| 37 |
return __('spell check', '404-solution'); |
| 38 |
} |
| 39 |
|
| 40 |
/** @return string */ |
| 41 |
public function getName(): string { |
| 42 |
return self::engineName(); |
| 43 |
} |
| 44 |
|
| 45 |
/** @param ABJ_404_Solution_MatchRequest $request */ |
| 46 |
public function shouldRun(ABJ_404_Solution_MatchRequest $request): bool { |
| 47 |
$urlSlugOnly = $request->getUrlSlugOnly(); |
| 48 |
|
| 49 |
if ($urlSlugOnly === '') { |
| 50 |
return false; |
| 51 |
} |
| 52 |
|
| 53 |
$segments = array_values(array_filter(explode('/', $urlSlugOnly))); |
| 54 |
if (count($segments) === 0) { |
| 55 |
return false; |
| 56 |
} |
| 57 |
|
| 58 |
$lastSegment = (string)end($segments); |
| 59 |
|
| 60 |
if (strlen($lastSegment) > 80) { |
| 61 |
return false; |
| 62 |
} |
| 63 |
if (substr_count($lastSegment, '-') >= 6) { |
| 64 |
return false; |
| 65 |
} |
| 66 |
if (preg_match('/\d{4,}/', $lastSegment)) { |
| 67 |
return false; |
| 68 |
} |
| 69 |
if (!preg_match('/[a-zA-Z]/', $lastSegment)) { |
| 70 |
return false; |
| 71 |
} |
| 72 |
|
| 73 |
return true; |
| 74 |
} |
| 75 |
|
| 76 |
/** @param ABJ_404_Solution_MatchRequest $request */ |
| 77 |
public function match(ABJ_404_Solution_MatchRequest $request): ?ABJ_404_Solution_MatchResult { |
| 78 |
$permalink = $this->spellChecker->getPermalinkUsingSpelling( |
| 79 |
$request->getUrlSlugOnly(), |
| 80 |
$request->getRequestedURL(), |
| 81 |
$request->getOptions() |
| 82 |
); |
| 83 |
|
| 84 |
if (empty($permalink)) { |
| 85 |
return null; |
| 86 |
} |
| 87 |
|
| 88 |
$id = isset($permalink['id']) && is_scalar($permalink['id']) ? (string)$permalink['id'] : ''; |
| 89 |
$type = isset($permalink['type']) && is_scalar($permalink['type']) ? (string)$permalink['type'] : ''; |
| 90 |
$link = isset($permalink['link']) && is_string($permalink['link']) ? $permalink['link'] : ''; |
| 91 |
$title = isset($permalink['title']) && is_string($permalink['title']) ? $permalink['title'] : ''; |
| 92 |
$score = isset($permalink['score']) && is_scalar($permalink['score']) ? (float)$permalink['score'] : 0.0; |
| 93 |
|
| 94 |
return new ABJ_404_Solution_MatchResult($id, $type, $link, $title, $score, $this->getName()); |
| 95 |
} |
| 96 |
} |
| 97 |
|
