sync-basalam / includes / Services / Products / DescriptionErrorSanitizer.php

DescriptionErrorSanitizer.php in ووسلام – همگام سازی ووکامرس و باسلام 1.10.19, at includes/Services/Products/DescriptionErrorSanitizer.php

146 lines 4.4 KB

Raw Download Zip

1	<?php
2
3	namespace SyncBasalam\Services\Products;
4
5	defined('ABSPATH') \|\| exit;
6
7	class DescriptionErrorSanitizer
8	{
9	private const SEPARATOR = '[\s\x{200c}\x{00a0}\-\_\.\–\—]*';
10
11	public static function extractDescriptionValues(?array $responseData): array
12	{
13	if (!is_array($responseData)) return [];
14
15	$messages = $responseData['messages'] ?? null;
16	if (!is_array($messages)) return [];
17
18	$values = [];
19
20	foreach ($messages as $message) {
21	if (!is_array($message)) continue;
22
23	$fields = $message['fields'] ?? [];
24	if (!is_array($fields) \|\| !in_array('description', $fields, true)) continue;
25
26	$detected = $message['detected'] ?? [];
27	if (!is_array($detected)) continue;
28
29	foreach ($detected as $item) {
30	foreach (self::valuesFromDetectedItem($item) as $value) {
31	$values[$value] = $value;
32	}
33	}
34	}
35
36	$values = array_values($values);
37	usort($values, static function ($a, $b) {
38	return mb_strlen($b) <=> mb_strlen($a);
39	});
40
41	return $values;
42	}
43
44	private static function valuesFromDetectedItem($item): array
45	{
46	if (is_string($item)) {
47	$item = trim($item);
48	return $item !== '' ? [$item] : [];
49	}
50
51	if (!is_array($item)) return [];
52
53	$found = [];
54
55	$snippet = $item['snippet'] ?? '';
56	if (is_string($snippet) && $snippet !== '') {
57	if (preg_match_all('/<em>(.*?)<\/em>/isu', $snippet, $matches)) {
58	foreach ($matches[1] as $match) {
59	$value = trim(wp_strip_all_tags(html_entity_decode($match, ENT_QUOTES \| ENT_HTML5, 'UTF-8')));
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	}
71	}
72	}
73	}
74
75	$value = $item['value'] ?? '';
76	if (is_string($value)) {
77	$value = trim($value);
78	if ($value !== '') $found[] = $value;
79	}
80
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	}));
97	}
98
99	public static function sanitize(string $description, array $values): string
100	{
101	$result = $description;
102
103	foreach ($values as $value) {
104	if (!is_string($value)) continue;
105	$result = self::removeValue($result, $value);
106	}
107
108	$result = preg_replace('/[ \t]{2,}/u', ' ', $result);
109	$result = preg_replace('/\n{3,}/u', "\n\n", $result);
110
111	return trim($result);
112	}
113
114	private static function removeValue(string $text, string $value): string
115	{
116	$value = trim($value);
117	if ($value === '') return $text;
118
119	$pattern = self::buildPattern($value);
120	if ($pattern !== null) {
121	$replaced = preg_replace($pattern, ' ', $text);
122	if (is_string($replaced) && $replaced !== $text) return $replaced;
123	}
124
125	$replaced = str_ireplace($value, ' ', $text);
126
127	return is_string($replaced) ? $replaced : $text;
128	}
129
130	private static function buildPattern(string $value): ?string
131	{
132	$segments = preg_split('/[\s\x{200c}\x{00a0}\-\_\.\–\—]+/u', $value, -1, PREG_SPLIT_NO_EMPTY);
133	if (empty($segments)) return null;
134
135	$escaped = array_map(static function ($segment) {
136	return preg_quote($segment, '/');
137	}, $segments);
138
139	$pattern = '/' . implode(self::SEPARATOR, $escaped) . '/iu';
140
141	if (@preg_match($pattern, '') === false) return null;
142
143	return $pattern;
144	}
145	}
146