cookieadmin / includes / cookiecategorizer.php

cookiecategorizer.php in CookieAdmin – Cookie Consent Banner 1.0.2, at includes/cookiecategorizer.php

123 lines 4.0 KB

Raw Download Zip

1	<?php
2
3	namespace CookieAdmin;
4
5	if(!defined('COOKIEADMIN_VERSION') \|\| !defined('ABSPATH')){
6	die('Hacking Attempt');
7	}
8
9	class CookieCategorizer {
10
11	/**
12	* Takes the raw output from the scanner and enriches it with data from the knowledge base.
13	*
14	* @param array $found_cookies The array of cookies from the Scanner class.
15	* @return array The enriched list of cookies.
16	*/
17	static function categorize_cookies(array $found_cookies, array $categorized_cookies) {
18	if (empty($found_cookies)) {
19	return [];
20	}
21
22	$prefixes = self::generate_query_prefixes(array_keys($found_cookies));
23
24	if (empty($prefixes)) {
25	return $found_cookies;
26	}
27
28	$db_candidates = self::fetch_candidates_from_db($prefixes);
29
30	if (empty($db_candidates)) {
31	return $found_cookies;
32	}
33
34	return self::match_and_enrich_cookies($found_cookies, $db_candidates, $categorized_cookies);
35	}
36
37
38	static function generate_query_prefixes(array $cookie_names): array {
39	$prefixes = [];
40	foreach ($cookie_names as $name) {
41
42	$parts = explode('_', $name, 2);
43	$prefix = $parts[0];
44
45	// Add a simple rule to avoid tiny, useless prefixes like "a" or "b"
46	if (strlen($prefix) >= 3) {
47	$prefixes[] = $prefix;
48	} else {
49	// If prefix is too short, use the full name up to a limit
50	$prefixes[] = substr($name, 0, 3);
51	}
52	}
53	return array_unique($prefixes);
54	}
55
56
57	static function fetch_candidates_from_db(array $prefixes): array {
58
59	if (empty($prefixes)) {
60	return [];
61	}
62
63	$cookies_info = cookieadmin_load_cookies_csv($prefixes, 1);
64
65	if(empty($cookies_info) \|\| is_wp_error($cookies_info)){
66	return [];
67	}
68
69	return $cookies_info;
70	}
71
72
73	static function match_and_enrich_cookies(array $found_cookies, array $db_candidates, array $old_categorized_cookies): array {
74	$categorized_cookies = [];
75	$old_categorized_cookies = [];
76	$remove_cookies = [];
77
78	foreach ($found_cookies as $scanned_name => $cookie_data) {
79	$best_match = null;
80	$longest_match_length = 0;
81
82	foreach ($db_candidates as $candidate) {
83
84	$candidate_name = $candidate['cookie_name'];
85
86	// Rule 1: Wildcard Match
87	if ($candidate['wildcard'] == 1 && strpos($scanned_name, $candidate_name) === 0) {
88	// "Best" is defined as the longest matching prefix.
89	if (strlen($candidate_name) > $longest_match_length) {
90	$longest_match_length = strlen($candidate_name);
91	$best_match = $candidate;
92	}
93	}
94	// Rule 2: Exact Match
95	else if ($candidate['wildcard'] == 0 && $scanned_name === $candidate_name) {
96	if (strlen($candidate_name) > $longest_match_length) {
97	$longest_match_length = strlen($candidate_name);
98	$best_match = $candidate;
99	break;
100	}
101	}
102	}
103
104	if (!empty($best_match)) {
105
106	if(!empty($old_categorized_cookies) && array_key_exists($best_match['cookie_name'], $old_categorized_cookies)){
107	$remove_cookies[] = $scanned_name;
108	continue;
109	}
110
111	$cookie_name = $best_match['cookie_name'];
112	$categorized_cookies[$cookie_name]['cookie_name'] = $cookie_name;
113	$categorized_cookies[$cookie_name]['raw_name'] = $scanned_name;
114	$categorized_cookies[$cookie_name]['category'] = $best_match['category'];
115	$categorized_cookies[$cookie_name]['description'] = $best_match['description'];
116	$categorized_cookies[$cookie_name]['patterns'] = $best_match['patterns'];
117
118	}
119	}
120	$categorized_cookies['remove_cookies'] = $remove_cookies;
121	return $categorized_cookies;
122	}
123	}