Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
100.00% |
152 / 152 |
|
100.00% |
4 / 4 |
CRAP | |
100.00% |
1 / 1 |
| Lexer | |
100.00% |
152 / 152 |
|
100.00% |
4 / 4 |
59 | |
100.00% |
1 / 1 |
| getTokenSyntax | |
100.00% |
50 / 50 |
|
100.00% |
1 / 1 |
25 | |||
| getCatchablePatterns | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
1 | |||
| getNonCatchablePatterns | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| getType | |
100.00% |
89 / 89 |
|
100.00% |
1 / 1 |
32 | |||
| 1 | <?php |
| 2 | |
| 3 | declare(strict_types=1); |
| 4 | |
| 5 | namespace Symftony\Xpression; |
| 6 | |
| 7 | use Doctrine\Common\Lexer\AbstractLexer; |
| 8 | use Symftony\Xpression\Exception\Lexer\UnknownTokenTypeException; |
| 9 | |
| 10 | class Lexer extends AbstractLexer |
| 11 | { |
| 12 | public const T_NONE = 0; |
| 13 | public const T_ALL = 2 ** 24 - 1; |
| 14 | |
| 15 | // Punctuation |
| 16 | public const T_COMMA = 2 ** 0; |
| 17 | |
| 18 | // Operand |
| 19 | public const T_OPERAND = 2 ** 1 + 2 ** 2 + 2 ** 3 + 2 ** 4; |
| 20 | public const T_INTEGER = 2 ** 1; |
| 21 | public const T_STRING = 2 ** 2; |
| 22 | public const T_INPUT_PARAMETER = 2 ** 3; |
| 23 | public const T_FLOAT = 2 ** 4; |
| 24 | |
| 25 | // Comparison operator |
| 26 | public const T_COMPARISON = 2 ** 5 + 2 ** 6 + 2 ** 7 + 2 ** 8 + 2 ** 9 + 2 ** 10 + 2 ** 18 + 2 ** 19 + 2 ** 21 + 2 ** 22; |
| 27 | public const T_EQUALS = 2 ** 5; |
| 28 | public const T_NOT_EQUALS = 2 ** 6; |
| 29 | public const T_GREATER_THAN = 2 ** 7; |
| 30 | public const T_GREATER_THAN_EQUALS = 2 ** 8; |
| 31 | public const T_LOWER_THAN = 2 ** 9; |
| 32 | public const T_LOWER_THAN_EQUALS = 2 ** 10; |
| 33 | |
| 34 | // Composite operator |
| 35 | public const T_COMPOSITE = 2 ** 11 + 2 ** 12 + 2 ** 13 + 2 ** 14 + 2 ** 15; |
| 36 | public const T_AND = 2 ** 11; |
| 37 | public const T_NOT_AND = 2 ** 12; |
| 38 | public const T_OR = 2 ** 13; |
| 39 | public const T_NOT_OR = 2 ** 14; |
| 40 | public const T_XOR = 2 ** 15; |
| 41 | |
| 42 | // Brace |
| 43 | public const T_OPEN_PARENTHESIS = 2 ** 16; |
| 44 | public const T_CLOSE_PARENTHESIS = 2 ** 17; |
| 45 | public const T_OPEN_SQUARE_BRACKET = 2 ** 18; |
| 46 | public const T_NOT_OPEN_SQUARE_BRACKET = 2 ** 19; |
| 47 | public const T_CLOSE_SQUARE_BRACKET = 2 ** 20; |
| 48 | public const T_DOUBLE_OPEN_CURLY_BRACKET = 2 ** 21; |
| 49 | public const T_NOT_DOUBLE_OPEN_CURLY_BRACKET = 2 ** 22; |
| 50 | public const T_DOUBLE_CLOSE_CURLY_BRACKET = 2 ** 23; |
| 51 | |
| 52 | /** |
| 53 | * @return string[] |
| 54 | */ |
| 55 | public static function getTokenSyntax(mixed $tokenType): array |
| 56 | { |
| 57 | $tokenSyntax = []; |
| 58 | // Punctuation |
| 59 | if ($tokenType & self::T_COMMA) { |
| 60 | $tokenSyntax[] = ','; |
| 61 | } |
| 62 | |
| 63 | // Recognize numeric values |
| 64 | if ($tokenType & self::T_FLOAT) { |
| 65 | $tokenSyntax[] = 'simple float'; |
| 66 | } |
| 67 | if ($tokenType & self::T_INTEGER) { |
| 68 | $tokenSyntax[] = 'simple integer'; |
| 69 | } |
| 70 | if ($tokenType & self::T_INPUT_PARAMETER) { |
| 71 | $tokenSyntax[] = '/[a-z_][a-z0-9_]*/'; |
| 72 | } |
| 73 | |
| 74 | // Recognize quoted strings |
| 75 | if ($tokenType & self::T_STRING) { |
| 76 | $tokenSyntax[] = '"value" or \'value\''; |
| 77 | } |
| 78 | |
| 79 | // Comparison operator |
| 80 | if ($tokenType & self::T_EQUALS) { |
| 81 | $tokenSyntax[] = '='; |
| 82 | } |
| 83 | if ($tokenType & self::T_NOT_EQUALS) { |
| 84 | $tokenSyntax[] = '≠ or !='; |
| 85 | } |
| 86 | if ($tokenType & self::T_GREATER_THAN) { |
| 87 | $tokenSyntax[] = '>'; |
| 88 | } |
| 89 | if ($tokenType & self::T_GREATER_THAN_EQUALS) { |
| 90 | $tokenSyntax[] = '≥ or >='; |
| 91 | } |
| 92 | if ($tokenType & self::T_LOWER_THAN) { |
| 93 | $tokenSyntax[] = '<'; |
| 94 | } |
| 95 | if ($tokenType & self::T_LOWER_THAN_EQUALS) { |
| 96 | $tokenSyntax[] = '≤ or <='; |
| 97 | } |
| 98 | |
| 99 | // Composite operator |
| 100 | if ($tokenType & self::T_AND) { |
| 101 | $tokenSyntax[] = '&'; |
| 102 | } |
| 103 | if ($tokenType & self::T_NOT_AND) { |
| 104 | $tokenSyntax[] = '!&'; |
| 105 | } |
| 106 | if ($tokenType & self::T_OR) { |
| 107 | $tokenSyntax[] = '|'; |
| 108 | } |
| 109 | if ($tokenType & self::T_NOT_OR) { |
| 110 | $tokenSyntax[] = '!|'; |
| 111 | } |
| 112 | if ($tokenType & self::T_XOR) { |
| 113 | $tokenSyntax[] = '⊕ or ^|'; |
| 114 | } |
| 115 | |
| 116 | // Brace |
| 117 | if ($tokenType & self::T_OPEN_PARENTHESIS) { |
| 118 | $tokenSyntax[] = '('; |
| 119 | } |
| 120 | if ($tokenType & self::T_CLOSE_PARENTHESIS) { |
| 121 | $tokenSyntax[] = ')'; |
| 122 | } |
| 123 | if ($tokenType & self::T_OPEN_SQUARE_BRACKET) { |
| 124 | $tokenSyntax[] = '['; |
| 125 | } |
| 126 | if ($tokenType & self::T_NOT_OPEN_SQUARE_BRACKET) { |
| 127 | $tokenSyntax[] = '!['; |
| 128 | } |
| 129 | if ($tokenType & self::T_CLOSE_SQUARE_BRACKET) { |
| 130 | $tokenSyntax[] = ']'; |
| 131 | } |
| 132 | if ($tokenType & self::T_DOUBLE_OPEN_CURLY_BRACKET) { |
| 133 | $tokenSyntax[] = '{{'; |
| 134 | } |
| 135 | if ($tokenType & self::T_NOT_DOUBLE_OPEN_CURLY_BRACKET) { |
| 136 | $tokenSyntax[] = '!{{'; |
| 137 | } |
| 138 | if ($tokenType & self::T_DOUBLE_CLOSE_CURLY_BRACKET) { |
| 139 | $tokenSyntax[] = '}}'; |
| 140 | } |
| 141 | |
| 142 | return $tokenSyntax; |
| 143 | } |
| 144 | |
| 145 | /** |
| 146 | * @return array |
| 147 | */ |
| 148 | protected function getCatchablePatterns() |
| 149 | { |
| 150 | return [ |
| 151 | "'(?:[^']|'')*'", // quoted strings |
| 152 | '"(?:[^"]|"")*"', // quoted strings |
| 153 | '\^\||⊕|!&|&|!\||\|', // Composite operator |
| 154 | '≤|≥|≠|<=|>=|!=|<|>|=|\[|!\[|\]|!{{|{{|}}', // Comparison operator |
| 155 | '[a-z_][a-z0-9_\.\-]*', // identifier or qualified name |
| 156 | '(?:[+-]?[0-9]*(?:[\.][0-9]+)*)', // numbers |
| 157 | // '(?:[+-]?(?:(?:(?:[0-9]+|(?:[0-9]*[\.][0-9]+)|(?:[0-9]+[\.][0-9]*))[eE][+-]?[0-9]+)|(?:[0-9]*[\.][0-9]+)|(?:[0-9]+[\.][0-9]*)))', // number extended all float (.5 / 1.5 / -1.2e3) |
| 158 | ]; |
| 159 | } |
| 160 | |
| 161 | /** |
| 162 | * @return array |
| 163 | */ |
| 164 | protected function getNonCatchablePatterns() |
| 165 | { |
| 166 | return [ |
| 167 | '\s+', |
| 168 | '(.)', |
| 169 | ]; |
| 170 | } |
| 171 | |
| 172 | /** |
| 173 | * @param string $value |
| 174 | * |
| 175 | * @return int |
| 176 | * |
| 177 | * @throws UnknownTokenTypeException |
| 178 | */ |
| 179 | protected function getType(&$value) |
| 180 | { |
| 181 | switch (true) { |
| 182 | // Punctuation |
| 183 | case ',' === $value[0]: |
| 184 | $type = self::T_COMMA; |
| 185 | |
| 186 | break; |
| 187 | |
| 188 | // Recognize numeric values |
| 189 | case is_numeric($value): |
| 190 | if (str_contains($value, '.') || false !== stripos($value, 'e')) { |
| 191 | $value = (float)$value; |
| 192 | $type = self::T_FLOAT; |
| 193 | |
| 194 | break; |
| 195 | } |
| 196 | |
| 197 | $value = (int)$value; |
| 198 | $type = self::T_INTEGER; |
| 199 | |
| 200 | break; |
| 201 | |
| 202 | // Recognize quoted strings |
| 203 | case '"' === $value[0]: |
| 204 | $value = str_replace('""', '"', substr($value, 1, \strlen($value) - 2)); |
| 205 | |
| 206 | $type = self::T_STRING; |
| 207 | |
| 208 | break; |
| 209 | |
| 210 | case "'" === $value[0]: |
| 211 | $value = str_replace("''", "'", substr($value, 1, \strlen($value) - 2)); |
| 212 | |
| 213 | $type = self::T_STRING; |
| 214 | |
| 215 | break; |
| 216 | |
| 217 | case preg_match('/[a-z_][a-z0-9_]*/i', $value): |
| 218 | $type = self::T_INPUT_PARAMETER; |
| 219 | |
| 220 | break; |
| 221 | |
| 222 | // Comparison operator |
| 223 | case '=' === $value: |
| 224 | $type = self::T_EQUALS; |
| 225 | |
| 226 | break; |
| 227 | |
| 228 | case '≠' === $value: |
| 229 | case '!=' === $value: |
| 230 | $value = '≠'; |
| 231 | $type = self::T_NOT_EQUALS; |
| 232 | |
| 233 | break; |
| 234 | |
| 235 | case '>' === $value: |
| 236 | $type = self::T_GREATER_THAN; |
| 237 | |
| 238 | break; |
| 239 | |
| 240 | case '>=' === $value: |
| 241 | case '≥' === $value: |
| 242 | $value = '≥'; |
| 243 | $type = self::T_GREATER_THAN_EQUALS; |
| 244 | |
| 245 | break; |
| 246 | |
| 247 | case '<' === $value: |
| 248 | $type = self::T_LOWER_THAN; |
| 249 | |
| 250 | break; |
| 251 | |
| 252 | case '<=' === $value: |
| 253 | case '≤' === $value: |
| 254 | $value = '≤'; |
| 255 | $type = self::T_LOWER_THAN_EQUALS; |
| 256 | |
| 257 | break; |
| 258 | |
| 259 | // Composite operator |
| 260 | case '&' === $value: |
| 261 | $type = self::T_AND; |
| 262 | |
| 263 | break; |
| 264 | |
| 265 | case '!&' === $value: |
| 266 | $type = self::T_NOT_AND; |
| 267 | |
| 268 | break; |
| 269 | |
| 270 | case '|' === $value: |
| 271 | $type = self::T_OR; |
| 272 | |
| 273 | break; |
| 274 | |
| 275 | case '!|' === $value: |
| 276 | $type = self::T_NOT_OR; |
| 277 | |
| 278 | break; |
| 279 | |
| 280 | case '^|' === $value: |
| 281 | case '⊕' === $value: |
| 282 | $value = '⊕'; |
| 283 | $type = self::T_XOR; |
| 284 | |
| 285 | break; |
| 286 | |
| 287 | // Brace |
| 288 | case '(' === $value: |
| 289 | $type = self::T_OPEN_PARENTHESIS; |
| 290 | |
| 291 | break; |
| 292 | |
| 293 | case ')' === $value: |
| 294 | $type = self::T_CLOSE_PARENTHESIS; |
| 295 | |
| 296 | break; |
| 297 | |
| 298 | case '[' === $value: |
| 299 | $type = self::T_OPEN_SQUARE_BRACKET; |
| 300 | |
| 301 | break; |
| 302 | |
| 303 | case '![' === $value: |
| 304 | $type = self::T_NOT_OPEN_SQUARE_BRACKET; |
| 305 | |
| 306 | break; |
| 307 | |
| 308 | case ']' === $value: |
| 309 | $type = self::T_CLOSE_SQUARE_BRACKET; |
| 310 | |
| 311 | break; |
| 312 | |
| 313 | case '{{' === $value: |
| 314 | $type = self::T_DOUBLE_OPEN_CURLY_BRACKET; |
| 315 | |
| 316 | break; |
| 317 | |
| 318 | case '!{{' === $value: |
| 319 | $type = self::T_NOT_DOUBLE_OPEN_CURLY_BRACKET; |
| 320 | |
| 321 | break; |
| 322 | |
| 323 | case '}}' === $value: |
| 324 | $type = self::T_DOUBLE_CLOSE_CURLY_BRACKET; |
| 325 | |
| 326 | break; |
| 327 | |
| 328 | // Default |
| 329 | default: |
| 330 | throw new UnknownTokenTypeException($value); |
| 331 | } |
| 332 | |
| 333 | return $type; |
| 334 | } |
| 335 | } |