← All changes | includes/Services/Products/DescriptionErrorSanitizer.php +26 -1 1.10.2 → 1.10.23 View file →

		@@ -56,9 +56,19 @@
56	56	if (is_string($snippet) && $snippet !== '') {
57	57	if (preg_match_all('/<em>(.*?)<\/em>/isu', $snippet, $matches)) {
58	58	foreach ($matches[1] as $match) {
59	59	$value = trim(wp_strip_all_tags(html_entity_decode($match, ENT_QUOTES | ENT_HTML5, 'UTF-8')));
60		- if ($value !== '') $found[] = $value;
	60	+ if ($value === '') continue;
	61	+
	62	+ $found[] = $value;
	63	+
	64	+ // Basalam highlights a whole window of text and returns it normalized
	65	+ // (e.g. "یک" comes back as "1"), so the full phrase often matches nothing
	66	+ // in the original description. The part that actually triggers the
	67	+ // "social id" signal is the handle-like token, so target it separately.
	68	+ foreach (self::handleLikeTokens($value) as $token) {
	69	+ $found[] = $token;
	70	+ }
61	71	}
62	72	}
63	73	}
64	74
		@@ -68,8 +78,23 @@
68	78	if ($value !== '') $found[] = $value;
69	79	}
70	80
71	81	return $found;
	82	+ }
	83	+
	84	+ /**
	85	+ * Pulls out handle/domain-like tokens ("s.exy", "س.کسی", "insta.gram") from a
	86	+ * highlighted snippet.
	87	+ *
	88	+ * @return string[]
	89	+ */
	90	+ private static function handleLikeTokens(string $text): array
	91	+ {
	92	+ if (!preg_match_all('/[\p{L}\p{N}_]+(?:\.[\p{L}\p{N}_]+)+/u', $text, $matches)) return [];
	93	+
	94	+ return array_values(array_filter($matches[0], static function ($token) {
	95	+ return mb_strlen($token) >= 3;
	96	+ }));
72	97	}
73	98
74	99	public static function sanitize(string $description, array $values): string
75	100	{