Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
51.09% |
47 / 92 |
|
41.67% |
5 / 12 |
CRAP | |
0.00% |
0 / 1 |
| StringMatcher | |
51.09% |
47 / 92 |
|
41.67% |
5 / 12 |
119.75 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| init | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
4 | |||
| getValidKeyChars | |
27.27% |
3 / 11 |
|
0.00% |
0 / 1 |
3.54 | |||
| setConf | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| matches | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
6 | |||
| mangle | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
3 | |||
| mangleList | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| mangleArray | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| unmangle | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
3 | |||
| unmangleList | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| unmangleArray | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| getExtraSchema | |
0.00% |
0 / 23 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace MediaWiki\Extension\Translate\MessageProcessing; |
| 5 | |
| 6 | use MediaWiki\Extension\Translate\MessageGroupConfiguration\MetaYamlSchemaExtender; |
| 7 | use MediaWiki\Title\Title; |
| 8 | |
| 9 | /** |
| 10 | * The versatile default implementation of StringMangler interface. |
| 11 | * It supports exact matches and patterns with any-wildcard (*). |
| 12 | * All matching strings are prefixed with the same prefix. |
| 13 | * |
| 14 | * @author Niklas Laxström |
| 15 | * @license GPL-2.0-or-later |
| 16 | */ |
| 17 | class StringMatcher implements StringMangler, MetaYamlSchemaExtender { |
| 18 | /** @var string Prefix for mangled message keys */ |
| 19 | protected string $sPrefix = ''; |
| 20 | /** @var string[] Exact message keys */ |
| 21 | protected array $aExact = []; |
| 22 | /** @var int[] Patterns of type foo* */ |
| 23 | protected array $aPrefix = []; |
| 24 | /** @var string[] Patterns that contain wildcard anywhere else than in the end */ |
| 25 | protected array $aRegex = []; |
| 26 | |
| 27 | public function __construct( string $prefix = '', array $patterns = [] ) { |
| 28 | $this->sPrefix = $prefix; |
| 29 | $this->init( $patterns ); |
| 30 | } |
| 31 | |
| 32 | /** |
| 33 | * Preprocesses the patterns. |
| 34 | * |
| 35 | * They are split into exact keys, prefix matches and pattern matches to |
| 36 | * speed up matching process. |
| 37 | * |
| 38 | * @param string[] $strings Key patterns. |
| 39 | */ |
| 40 | protected function init( array $strings ): void { |
| 41 | foreach ( $strings as $string ) { |
| 42 | $pos = strpos( $string, '*' ); |
| 43 | if ( $pos === false ) { |
| 44 | $this->aExact[] = $string; |
| 45 | } elseif ( $pos + 1 === strlen( $string ) ) { |
| 46 | $prefix = substr( $string, 0, -1 ); |
| 47 | $this->aPrefix[$prefix] = strlen( $prefix ); |
| 48 | } else { |
| 49 | $string = str_replace( '\\*', '.+', preg_quote( $string, '/' ) ); |
| 50 | $this->aRegex[] = "/^$string$/"; |
| 51 | } |
| 52 | } |
| 53 | } |
| 54 | |
| 55 | protected static function getValidKeyChars(): string { |
| 56 | static $valid = null; |
| 57 | if ( $valid === null ) { |
| 58 | $valid = strtr( |
| 59 | Title::legalChars(), |
| 60 | [ |
| 61 | '=' => '', // equals sign, which is itself usef for escaping |
| 62 | '&' => '', // ampersand, for entities |
| 63 | '%' => '', // percent sign, which is used in URL encoding |
| 64 | ] |
| 65 | ); |
| 66 | } |
| 67 | |
| 68 | return $valid; |
| 69 | } |
| 70 | |
| 71 | /** @inheritDoc */ |
| 72 | public function setConf( array $conf ): void { |
| 73 | $this->sPrefix = $conf['prefix']; |
| 74 | $this->init( $conf['patterns'] ); |
| 75 | } |
| 76 | |
| 77 | /** @inheritDoc */ |
| 78 | public function matches( string $key ): bool { |
| 79 | if ( in_array( $key, $this->aExact ) ) { |
| 80 | return true; |
| 81 | } |
| 82 | |
| 83 | foreach ( $this->aPrefix as $prefix => $len ) { |
| 84 | if ( strncmp( $key, $prefix, $len ) === 0 ) { |
| 85 | return true; |
| 86 | } |
| 87 | } |
| 88 | |
| 89 | foreach ( $this->aRegex as $regex ) { |
| 90 | if ( preg_match( $regex, $key ) ) { |
| 91 | return true; |
| 92 | } |
| 93 | } |
| 94 | |
| 95 | return false; |
| 96 | } |
| 97 | |
| 98 | /** @inheritDoc */ |
| 99 | public function mangle( string $key ): string { |
| 100 | if ( $this->matches( $key ) ) { |
| 101 | $key = $this->sPrefix . $key; |
| 102 | } |
| 103 | |
| 104 | $escaper = static function ( $match ) { |
| 105 | $esc = ''; |
| 106 | foreach ( str_split( $match[0] ) as $c ) { |
| 107 | $esc .= '=' . sprintf( '%02X', ord( $c ) ); |
| 108 | } |
| 109 | return $esc; |
| 110 | }; |
| 111 | |
| 112 | // Apply a "quoted-printable"-like escaping |
| 113 | $valid = self::getValidKeyChars(); |
| 114 | $key = preg_replace_callback( "/[^$valid]/", $escaper, $key ); |
| 115 | // Additional limitations in MediaWiki, see MediaWikiTitleCodec::splitTitleString |
| 116 | $key = preg_replace_callback( '/(~~~|^[ _]|[ _]$|[ _]{2,}|^:)/', $escaper, $key ); |
| 117 | // TODO: length check + truncation |
| 118 | // TODO: forbid path travels |
| 119 | |
| 120 | return $key; |
| 121 | } |
| 122 | |
| 123 | /** @inheritDoc */ |
| 124 | public function mangleList( array $list ): array { |
| 125 | return array_map( [ $this, 'mangle' ], $list ); |
| 126 | } |
| 127 | |
| 128 | /** @inheritDoc */ |
| 129 | public function mangleArray( array $array ): array { |
| 130 | $out = []; |
| 131 | foreach ( $array as $key => $value ) { |
| 132 | $out[$this->mangle( (string)$key )] = $value; |
| 133 | } |
| 134 | |
| 135 | return $out; |
| 136 | } |
| 137 | |
| 138 | /** @inheritDoc */ |
| 139 | public function unmangle( string $key ): string { |
| 140 | // Unescape the "quoted-printable"-like escaping, |
| 141 | // which is applied in mangle |
| 142 | $unescapedString = preg_replace_callback( |
| 143 | '/=([A-F0-9]{2})/', |
| 144 | static function ( $match ) { |
| 145 | return chr( hexdec( $match[1] ) ); |
| 146 | }, |
| 147 | $key |
| 148 | ); |
| 149 | |
| 150 | if ( strncmp( $unescapedString, $this->sPrefix, strlen( $this->sPrefix ) ) === 0 ) { |
| 151 | $unmangled = substr( $unescapedString, strlen( $this->sPrefix ) ); |
| 152 | |
| 153 | // Check if this string should be mangled / un-mangled to begin with |
| 154 | if ( $this->matches( $unmangled ) ) { |
| 155 | return $unmangled; |
| 156 | } |
| 157 | } |
| 158 | return $unescapedString; |
| 159 | } |
| 160 | |
| 161 | /** @inheritDoc */ |
| 162 | public function unmangleList( array $list ): array { |
| 163 | foreach ( $list as $index => $key ) { |
| 164 | $list[$index] = $this->unmangle( $key ); |
| 165 | } |
| 166 | |
| 167 | return $list; |
| 168 | } |
| 169 | |
| 170 | /** @inheritDoc */ |
| 171 | public function unmangleArray( array $array ): array { |
| 172 | $out = []; |
| 173 | foreach ( $array as $key => $value ) { |
| 174 | $out[$this->unmangle( $key )] = $value; |
| 175 | } |
| 176 | |
| 177 | return $out; |
| 178 | } |
| 179 | |
| 180 | /** @inheritDoc */ |
| 181 | public static function getExtraSchema(): array { |
| 182 | $schema = [ |
| 183 | 'root' => [ |
| 184 | '_type' => 'array', |
| 185 | '_children' => [ |
| 186 | 'MANGLER' => [ |
| 187 | '_type' => 'array', |
| 188 | '_children' => [ |
| 189 | 'prefix' => [ |
| 190 | '_type' => 'text', |
| 191 | '_not_empty' => true, |
| 192 | ], |
| 193 | 'patterns' => [ |
| 194 | '_type' => 'array', |
| 195 | '_required' => true, |
| 196 | '_ignore_extra_keys' => true, |
| 197 | '_children' => [], |
| 198 | ], |
| 199 | ], |
| 200 | ], |
| 201 | ], |
| 202 | ], |
| 203 | ]; |
| 204 | |
| 205 | return $schema; |
| 206 | } |
| 207 | } |