src/Charsets.php

   1 <?php
   2 /**
   3  * MySQL charset metadata and manipulations
   4  */
   5
   6 declare(strict_types=1);
   7
   8 namespace PhpMyAdmin;
   9
  10 use PhpMyAdmin\Charsets\Charset;
  11 use PhpMyAdmin\Charsets\Collation;
  12 use PhpMyAdmin\Dbal\DatabaseInterface;
  13
  14 use function __;
  15 use function array_keys;
  16 use function explode;
  17 use function is_string;
  18 use function ksort;
  19
  20 use const SORT_STRING;
  21
  22 /**
  23  * Class used to manage MySQL charsets
  24  */
  25 class Charsets
  26 {
  27     /**
  28      * MySQL charsets map
  29      *
  30      * @var array<string, string>
  31      */
  32     public static array $mysqlCharsetMap = [
  33         'big5' => 'big5',
  34         'cp-866' => 'cp866',
  35         'euc-jp' => 'ujis',
  36         'euc-kr' => 'euckr',
  37         'gb2312' => 'gb2312',
  38         'gbk' => 'gbk',
  39         'iso-8859-1' => 'latin1',
  40         'iso-8859-2' => 'latin2',
  41         'iso-8859-7' => 'greek',
  42         'iso-8859-8' => 'hebrew',
  43         'iso-8859-8-i' => 'hebrew',
  44         'iso-8859-9' => 'latin5',
  45         'iso-8859-13' => 'latin7',
  46         'iso-8859-15' => 'latin1',
  47         'koi8-r' => 'koi8r',
  48         'shift_jis' => 'sjis',
  49         'tis-620' => 'tis620',
  50         'utf-8' => 'utf8',
  51         'windows-1250' => 'cp1250',
  52         'windows-1251' => 'cp1251',
  53         'windows-1252' => 'latin1',
  54         'windows-1256' => 'cp1256',
  55         'windows-1257' => 'cp1257',
  56     ];
  57
  58     /**
  59      * The charset for the server
  60      */
  61     private static Charset|null $serverCharset = null;
  62
  63     /** @var array<string, Charset> */
  64     private static array $charsets = [];
  65
  66     /** @var array<string, array<string, Collation>> */
  67     private static array $collations = [];
  68
  69     /**
  70      * Loads charset data from the server
  71      *
  72      * @param bool $disableIs Disable use of INFORMATION_SCHEMA
  73      */
  74     private static function loadCharsets(DatabaseInterface $dbi, bool $disableIs): void
  75     {
  76         /* Data already loaded */
  77         if (self::$charsets !== []) {
  78             return;
  79         }
  80
  81         $sql = 'SELECT `CHARACTER_SET_NAME` AS `Charset`,'
  82             . ' `DEFAULT_COLLATE_NAME` AS `Default collation`,'
  83             . ' `DESCRIPTION` AS `Description`,'
  84             . ' `MAXLEN` AS `Maxlen`'
  85             . ' FROM `information_schema`.`CHARACTER_SETS`';
  86
  87         if ($disableIs) {
  88             $sql = 'SHOW CHARACTER SET';
  89         }
  90
  91         $res = $dbi->query($sql);
  92
  93         self::$charsets = [];
  94         foreach ($res as $row) {
  95             self::$charsets[$row['Charset']] = Charset::fromServer($row);
  96         }
  97
  98         ksort(self::$charsets, SORT_STRING);
  99     }
 100
 101     /**
 102      * Loads collation data from the server
 103      *
 104      * @param bool $disableIs Disable use of INFORMATION_SCHEMA
 105      */
 106     private static function loadCollations(DatabaseInterface $dbi, bool $disableIs): void
 107     {
 108         /* Data already loaded */
 109         if (self::$collations !== []) {
 110             return;
 111         }
 112
 113         if ($dbi->isMariaDB() && $dbi->getVersion() >= 101000) {
 114             /* Use query to accommodate new structure of MariaDB collations.
 115             Note, that SHOW COLLATION command is not applicable at the time of writing.
 116             Refer https://jira.mariadb.org/browse/MDEV-27009 */
 117             $sql = 'SELECT `collapp`.`FULL_COLLATION_NAME` AS `Collation`,'
 118                 . ' `collapp`.`CHARACTER_SET_NAME` AS `Charset`,'
 119                 . ' `collapp`.`ID` AS `Id`,'
 120                 . ' `collapp`.`IS_DEFAULT` AS `Default`,'
 121                 . ' `coll`.`IS_COMPILED` AS `Compiled`,'
 122                 . ' `coll`.`SORTLEN` AS `Sortlen`'
 123                 . ' FROM `information_schema`.`COLLATION_CHARACTER_SET_APPLICABILITY` `collapp`'
 124                 . ' LEFT JOIN `information_schema`.`COLLATIONS` `coll`'
 125                 . ' ON `collapp`.`COLLATION_NAME`=`coll`.`COLLATION_NAME`';
 126         } else {
 127             $sql = 'SELECT `COLLATION_NAME` AS `Collation`,'
 128                 . ' `CHARACTER_SET_NAME` AS `Charset`,'
 129                 . ' `ID` AS `Id`,'
 130                 . ' `IS_DEFAULT` AS `Default`,'
 131                 . ' `IS_COMPILED` AS `Compiled`,'
 132                 . ' `SORTLEN` AS `Sortlen`'
 133                 . ' FROM `information_schema`.`COLLATIONS`';
 134
 135             if ($disableIs) {
 136                 $sql = 'SHOW COLLATION';
 137             }
 138         }
 139
 140         $res = $dbi->query($sql);
 141
 142         self::$collations = [];
 143         foreach ($res as $row) {
 144             self::$collations[$row['Charset']][$row['Collation']] = Collation::fromServer($row);
 145         }
 146
 147         foreach (array_keys(self::$collations) as $charset) {
 148             ksort(self::$collations[$charset], SORT_STRING);
 149         }
 150     }
 151
 152     /**
 153      * Get current server charset
 154      *
 155      * @param bool $disableIs Disable use of INFORMATION_SCHEMA
 156      */
 157     public static function getServerCharset(DatabaseInterface $dbi, bool $disableIs): Charset
 158     {
 159         if (self::$serverCharset !== null) {
 160             return self::$serverCharset;
 161         }
 162
 163         self::loadCharsets($dbi, $disableIs);
 164         $serverCharset = $dbi->getVariable('character_set_server');
 165         if (! is_string($serverCharset)) {// MySQL 5.7.8 fallback, issue #15614
 166             $serverCharset = $dbi->fetchValue('SELECT @@character_set_server;');
 167         }
 168
 169         self::$serverCharset = self::$charsets[$serverCharset] ?? null;
 170
 171         // MySQL 8.0.11+ fallback, issue #16931
 172         if (self::$serverCharset === null && $serverCharset === 'utf8mb3') {
 173             // See: https://dev.mysql.com/doc/relnotes/mysql/8.0/en/news-8-0-11.html#mysqld-8-0-11-charset
 174             // The utf8mb3 character set will be replaced by utf8mb4 in a future MySQL version.
 175             // The utf8 character set is currently an alias for utf8mb3,
 176             // but will at that point become a reference to utf8mb4.
 177             // To avoid ambiguity about the meaning of utf8,
 178             // consider specifying utf8mb4 explicitly for character set references instead of utf8.
 179             // Warning: #3719 'utf8' is currently an alias for the character set UTF8MB3 [...]
 180             return self::$charsets['utf8'];
 181         }
 182
 183         return self::$serverCharset ?? Charset::fromServer(
 184             ['Charset' => __('Unknown'), 'Description' => __('Unknown')],
 185         );
 186     }
 187
 188     /**
 189      * Get all server charsets
 190      *
 191      * @param bool $disableIs Disable use of INFORMATION_SCHEMA
 192      *
 193      * @return array<string, Charset>
 194      */
 195     public static function getCharsets(DatabaseInterface $dbi, bool $disableIs): array
 196     {
 197         self::loadCharsets($dbi, $disableIs);
 198
 199         return self::$charsets;
 200     }
 201
 202     /**
 203      * Get all server collations
 204      *
 205      * @param bool $disableIs Disable use of INFORMATION_SCHEMA
 206      *
 207      * @return array<string, array<string, Collation>>
 208      */
 209     public static function getCollations(DatabaseInterface $dbi, bool $disableIs): array
 210     {
 211         self::loadCollations($dbi, $disableIs);
 212
 213         return self::$collations;
 214     }
 215
 216     /**
 217      * @param bool   $disableIs Disable use of INFORMATION_SCHEMA
 218      * @param string $name      Collation name
 219      */
 220     public static function findCollationByName(
 221         DatabaseInterface $dbi,
 222         bool $disableIs,
 223         string $name,
 224     ): Collation|null {
 225         $charset = explode('_', $name)[0];
 226         $collations = self::getCollations($dbi, $disableIs);
 227
 228         return $collations[$charset][$name] ?? null;
 229     }
 230 }