404-solution / includes / logs / LogsRequestedUrlColumnMetadata.php

LogsRequestedUrlColumnMetadata.php in 404 Solution trunk, at includes/logs/LogsRequestedUrlColumnMetadata.php

133 lines 5.5 KB

Raw Download Zip

1	<?php
2
3	if (!defined('ABSPATH')) {
4	exit;
5	}
6
7	/**
8	* Resolves and caches metadata for logsv2.requested_url so log writes can
9	* adapt to legacy charsets and engine-specific collation reporting.
10	*/
11	class ABJ_404_Solution_LogsRequestedUrlColumnMetadata {
12
13	/** @var ABJ_404_Solution_Logging */
14	private $logger;
15
16	/** @param ABJ_404_Solution_Logging $logger */
17	public function __construct($logger) {
18	$this->logger = $logger;
19	}
20
21	/**
22	* Look up (and cache for a week) the character set and collation of the
23	* logsv2.requested_url column.
24	*
25	* @param string $logTableName
26	* @param mixed $wpdb
27	* @return array{charset_name: ?string, collation_name: ?string}\|null
28	*/
29	public function resolveRequestedUrlColumnMeta(string $logTableName, $wpdb): ?array {
30	static $cached = null;
31	if ($cached !== null) {
32	return $cached;
33	}
34
35	$cachedMeta = $this->readCachedColumnMeta();
36	if ($cachedMeta !== null) {
37	$cached = $cachedMeta;
38	return $cached;
39	}
40
41	$dbName = defined('DB_NAME') ? (string)DB_NAME : '';
42	if ($dbName === '' \|\| !is_object($wpdb)
43	\|\| !method_exists($wpdb, 'prepare') \|\| !method_exists($wpdb, 'get_results')) {
44	return null;
45	}
46
47	$meta = $this->queryRequestedUrlColumnMeta($logTableName, $dbName, $wpdb);
48	if ($meta === null) {
49	return null;
50	}
51	$this->cacheColumnMeta($meta);
52	$cached = $meta;
53	return $meta;
54	}
55
56	/**
57	* @return array{charset_name: ?string, collation_name: ?string}\|null
58	*/
59	private function readCachedColumnMeta(): ?array {
60	if (!function_exists('get_transient')) {
61	return null;
62	}
63	$existing = get_transient('abj404_logs_requested_url_column_meta');
64	if (is_array($existing)) {
65	return array(
66	'charset_name' => is_string($existing['charset_name'] ?? null) ? $existing['charset_name'] : null,
67	'collation_name' => is_string($existing['collation_name'] ?? null) ? $existing['collation_name'] : null,
68	);
69	}
70	$legacyCharset = get_transient('abj404_logs_requested_url_charset');
71	if (is_string($legacyCharset) && $legacyCharset !== '') {
72	return array('charset_name' => $legacyCharset, 'collation_name' => null);
73	}
74	return null;
75	}
76
77	/**
78	* @return array{charset_name: ?string, collation_name: ?string}\|null
79	*/
80	private function queryRequestedUrlColumnMeta(string $logTableName, string $dbName, object $wpdb): ?array {
81	if (!method_exists($wpdb, 'prepare') \|\| !method_exists($wpdb, 'get_results')) {
82	return null;
83	}
84	// DAO-bypass-approved: metadata probe needs wpdb prepare/get_results against information_schema and reads no plugin data.
85	$getCharsetQuery = $wpdb->prepare("SELECT character_set_name as charset_name, collation_name as collation_name \n FROM information_schema.columns \n WHERE lower(table_schema) = lower(%s) \n AND lower(table_name) = lower(%s) \n AND lower(column_name) = lower(%s) ", $dbName, $logTableName, 'requested_url');
86	// DAO-bypass-approved: paired information_schema metadata read for requested_url charset/collation probe.
87	$resultArray = $wpdb->get_results($getCharsetQuery, ARRAY_A);
88	if (empty($resultArray)) {
89	return null;
90	}
91	$firstRow = is_array($resultArray[0]) ? $resultArray[0] : array();
92	$row = array_change_key_case($firstRow, CASE_LOWER);
93	$meta = array(
94	'charset_name' => is_string($row['charset_name'] ?? null) ? $row['charset_name'] : null,
95	'collation_name' => is_string($row['collation_name'] ?? null) ? $row['collation_name'] : null,
96	);
97	return $meta;
98	}
99
100	/**
101	* @param array{charset_name: ?string, collation_name: ?string} $meta
102	*/
103	private function cacheColumnMeta(array $meta): void {
104	if (!function_exists('set_transient')) {
105	return;
106	}
107	// @cache-write-audit: opt-out - schema metadata probe cache, not user-query result data.
108	// allow-cache-empty: $meta is built from a non-empty $resultArray in queryRequestedUrlColumnMeta(), so charset_name / collation_name are populated DB metadata, not a failed-fetch sentinel.
109	$ttl = defined('WEEK_IN_SECONDS') ? WEEK_IN_SECONDS : 604800;
110	set_transient('abj404_logs_requested_url_column_meta', $meta, $ttl);
111	if (!empty($meta['charset_name'])) {
112	set_transient('abj404_logs_requested_url_charset', $meta['charset_name'], $ttl);
113	}
114	}
115
116	/** Emit a one-shot warning the first time a non-utf8 logs charset is observed. */
117	public function warnLogsCharsetMismatchOnce(string $logTableName, string $charset): void {
118	if (!function_exists('get_transient') \|\| !function_exists('set_transient')) {
119	return;
120	}
121	$warnKey = 'abj404_warned_logs_charset_mismatch';
122	$warnVal = $logTableName . '\|' . strtolower($charset);
123	$already = get_transient($warnKey);
124	if ($already === $warnVal) {
125	return;
126	}
127	$ttl = defined('WEEK_IN_SECONDS') ? WEEK_IN_SECONDS : 604800;
128	// @cache-write-audit: opt-out - charset warning dedup marker, not query result data.
129	set_transient($warnKey, $warnVal, $ttl);
130	$this->logger->warn("Logs table column charset is '{$charset}' for {$logTableName}. URL-encoding stored requested URLs to avoid charset issues.");
131	}
132	}
133