Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
77.73% |
171 / 220 |
|
14.29% |
1 / 7 |
CRAP | |
0.00% |
0 / 1 |
| JavaScriptMinifier | |
77.73% |
171 / 220 |
|
14.29% |
1 / 7 |
282.77 | |
0.00% |
0 / 1 |
| ensureExpandedStates | |
7.69% |
2 / 26 |
|
0.00% |
0 / 1 |
244.31 | |||
| minify | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| createMinifier | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| createSourceMapState | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| createIdentityMinifier | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| minifyInternal | |
97.11% |
168 / 173 |
|
0.00% |
0 / 1 |
94 | |||
| debug | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
42 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Copyright 2011 Paul Copperman <paul.copperman@gmail.com> |
| 4 | * Copyright 2018 Timo Tijhof |
| 5 | * Copyright 2021 Roan Kattouw <roan.kattouw@gmail.com> |
| 6 | * |
| 7 | * Licensed under the Apache License, Version 2.0 (the "License"); |
| 8 | * you may not use this file except in compliance with the License. |
| 9 | * You may obtain a copy of the License at |
| 10 | * |
| 11 | * http://www.apache.org/licenses/LICENSE-2.0 |
| 12 | * |
| 13 | * Unless required by applicable law or agreed to in writing, software |
| 14 | * distributed under the License is distributed on an "AS IS" BASIS, |
| 15 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 16 | * See the License for the specific language governing permissions and |
| 17 | * limitations under the License. |
| 18 | * |
| 19 | * @file |
| 20 | * @license Apache-2.0 |
| 21 | * @license MIT |
| 22 | * @license GPL-2.0-or-later |
| 23 | * @license LGPL-2.1-or-later |
| 24 | */ |
| 25 | |
| 26 | namespace Wikimedia\Minify; |
| 27 | |
| 28 | use ReflectionClass; |
| 29 | |
| 30 | /** |
| 31 | * JavaScript Minifier |
| 32 | * |
| 33 | * This class is meant to safely minify JavaScript code, while leaving syntactically correct |
| 34 | * programs intact. Other libraries, such as JSMin require a certain coding style to work |
| 35 | * correctly. OTOH, libraries like jsminplus, that do parse the code correctly are rather |
| 36 | * slow, because they construct a complete parse tree before outputting the code minified. |
| 37 | * So this class is meant to allow arbitrary (but syntactically correct) input, while being |
| 38 | * fast enough to be used for on-the-fly minifying. |
| 39 | * |
| 40 | * This class was written with ECMA-262 7th Edition in mind ("ECMAScript 2016"). Parsing features |
| 41 | * new to later editions of ECMAScript might not be supported. It's assumed that the input is |
| 42 | * syntactically correct; if it's not, this class may not detect that, and may produce incorrect |
| 43 | * output. |
| 44 | * |
| 45 | * This class has limited support for 8.0 spec ("ECMAScript 2017"), specifically, the await |
| 46 | * keyword, and most kinds of async functions are implemented. Other new parsing features of ES2017 |
| 47 | * are not yet supported. |
| 48 | * |
| 49 | * See also: |
| 50 | * - <https://262.ecma-international.org/7.0/> |
| 51 | * - <https://262.ecma-international.org/8.0/> |
| 52 | * - <https://262.ecma-international.org/11.0/> |
| 53 | */ |
| 54 | class JavaScriptMinifier { |
| 55 | |
| 56 | /* Parsing states. |
| 57 | * The state machine is necessary to decide whether to parse a slash as division |
| 58 | * operator or as regexp literal, and to know where semicolon insertion is possible. |
| 59 | * States are generally named after the next expected item. We only distinguish states when the |
| 60 | * distinction is relevant for our purpose. The meaning of these states is documented |
| 61 | * in $model below. |
| 62 | * |
| 63 | * Negative numbers are used to indicate that the state is inside a generator function, |
| 64 | * which changes the behavior of 'yield' |
| 65 | */ |
| 66 | private const STATEMENT = 1; |
| 67 | private const CONDITION = 2; |
| 68 | private const FUNC = 3; |
| 69 | private const GENFUNC = 4; |
| 70 | private const PROPERTY_ASSIGNMENT = 5; |
| 71 | private const EXPRESSION = 6; |
| 72 | private const EXPRESSION_NO_NL = 7; |
| 73 | private const EXPRESSION_OP = 8; |
| 74 | private const EXPRESSION_DOT = 9; |
| 75 | private const EXPRESSION_END = 10; |
| 76 | private const EXPRESSION_ARROWFUNC = 11; |
| 77 | private const EXPRESSION_TERNARY = 12; |
| 78 | private const EXPRESSION_TERNARY_OP = 13; |
| 79 | private const EXPRESSION_TERNARY_DOT = 14; |
| 80 | private const EXPRESSION_TERNARY_ARROWFUNC = 15; |
| 81 | private const PAREN_EXPRESSION = 16; |
| 82 | private const PAREN_EXPRESSION_OP = 17; |
| 83 | private const PAREN_EXPRESSION_DOT = 18; |
| 84 | private const PAREN_EXPRESSION_ARROWFUNC = 19; |
| 85 | private const PROPERTY_EXPRESSION = 20; |
| 86 | private const PROPERTY_EXPRESSION_OP = 21; |
| 87 | private const PROPERTY_EXPRESSION_DOT = 22; |
| 88 | private const PROPERTY_EXPRESSION_ARROWFUNC = 23; |
| 89 | private const CLASS_DEF = 24; |
| 90 | private const IMPORT_EXPORT = 25; |
| 91 | private const TEMPLATE_STRING_HEAD = 26; |
| 92 | private const TEMPLATE_STRING_TAIL = 27; |
| 93 | private const PAREN_EXPRESSION_OP_NO_NL = 28; |
| 94 | private const EXPRESSION_TERNARY_NO_NL = 29; |
| 95 | private const PAREN_EXPRESSION_NO_NL = 30; |
| 96 | private const PROPERTY_EXPRESSION_NO_NL = 31; |
| 97 | |
| 98 | /* Token types */ |
| 99 | |
| 100 | /** @var int unary operators */ |
| 101 | private const TYPE_UN_OP = 101; |
| 102 | |
| 103 | /** @var int ++ and -- */ |
| 104 | private const TYPE_INCR_OP = 102; |
| 105 | |
| 106 | /** @var int binary operators (except .) */ |
| 107 | private const TYPE_BIN_OP = 103; |
| 108 | |
| 109 | /** @var int + and - which can be either unary or binary ops */ |
| 110 | private const TYPE_ADD_OP = 104; |
| 111 | |
| 112 | /** @var int . */ |
| 113 | private const TYPE_DOT = 105; |
| 114 | |
| 115 | /** @var int ? */ |
| 116 | private const TYPE_HOOK = 106; |
| 117 | |
| 118 | /** @var int : */ |
| 119 | private const TYPE_COLON = 107; |
| 120 | |
| 121 | /** @var int , */ |
| 122 | private const TYPE_COMMA = 108; |
| 123 | |
| 124 | /** @var int ; */ |
| 125 | private const TYPE_SEMICOLON = 109; |
| 126 | |
| 127 | /** @var int { */ |
| 128 | private const TYPE_BRACE_OPEN = 110; |
| 129 | |
| 130 | /** @var int } */ |
| 131 | private const TYPE_BRACE_CLOSE = 111; |
| 132 | |
| 133 | /** @var int ( and [ */ |
| 134 | private const TYPE_PAREN_OPEN = 112; |
| 135 | |
| 136 | /** @var int ) and ] */ |
| 137 | private const TYPE_PAREN_CLOSE = 113; |
| 138 | |
| 139 | /** @var int => */ |
| 140 | private const TYPE_ARROW = 114; |
| 141 | |
| 142 | /** @var int keywords: break, continue, return, throw (and yield, if we're in a generator) */ |
| 143 | private const TYPE_RETURN = 115; |
| 144 | |
| 145 | /** @var int keywords: catch, for, with, switch, while, if */ |
| 146 | private const TYPE_IF = 116; |
| 147 | |
| 148 | /** @var int keywords: case, finally, else, do, try */ |
| 149 | private const TYPE_DO = 117; |
| 150 | |
| 151 | /** @var int keywords: var, let, const */ |
| 152 | private const TYPE_VAR = 118; |
| 153 | |
| 154 | /** @var int keywords: yield */ |
| 155 | private const TYPE_YIELD = 119; |
| 156 | |
| 157 | /** @var int keywords: function */ |
| 158 | private const TYPE_FUNC = 120; |
| 159 | |
| 160 | /** @var int keywords: class */ |
| 161 | private const TYPE_CLASS = 121; |
| 162 | |
| 163 | /** @var int all literals, identifiers, unrecognised tokens, and other keywords */ |
| 164 | private const TYPE_LITERAL = 122; |
| 165 | |
| 166 | /** @var int For special treatment of tokens that usually mean something else */ |
| 167 | private const TYPE_SPECIAL = 123; |
| 168 | |
| 169 | /** @var int keywords: async */ |
| 170 | private const TYPE_ASYNC = 124; |
| 171 | |
| 172 | /** @var int keywords: await */ |
| 173 | private const TYPE_AWAIT = 125; |
| 174 | |
| 175 | /** @var int Go to another state */ |
| 176 | private const ACTION_GOTO = 201; |
| 177 | |
| 178 | /** @var int Push a state to the stack */ |
| 179 | private const ACTION_PUSH = 202; |
| 180 | |
| 181 | /** @var int Pop the state from the top of the stack, and go to that state */ |
| 182 | private const ACTION_POP = 203; |
| 183 | |
| 184 | /** @var int Limit to avoid excessive memory usage */ |
| 185 | private const STACK_LIMIT = 1000; |
| 186 | |
| 187 | /** Length of the longest token in $tokenTypes made of punctuation characters, |
| 188 | * as defined in $opChars. Update this if you add longer tokens to $tokenTypes. |
| 189 | * |
| 190 | * Currently, the longest punctuation token is `>>>=`, which is 4 characters. |
| 191 | */ |
| 192 | private const LONGEST_PUNCTUATION_TOKEN = 4; |
| 193 | |
| 194 | /** |
| 195 | * @var int $maxLineLength |
| 196 | * |
| 197 | * Maximum line length |
| 198 | * |
| 199 | * This is not a strict maximum, but a guideline. Longer lines will be |
| 200 | * produced when literals (e.g. quoted strings) longer than this are |
| 201 | * encountered, or when required to guard against semicolon insertion. |
| 202 | * |
| 203 | * This is a private member (instead of constant) to allow tests to |
| 204 | * set it to 1, to verify ASI and line-breaking behaviour. |
| 205 | */ |
| 206 | private static $maxLineLength = 1000; |
| 207 | |
| 208 | private static bool $expandedStates = false; |
| 209 | |
| 210 | /** |
| 211 | * @var array $opChars |
| 212 | * |
| 213 | * Characters which can be combined without whitespace between them. |
| 214 | * Unlike the ECMAScript spec, we define these as individual symbols, not sequences. |
| 215 | */ |
| 216 | private static $opChars = [ |
| 217 | // ECMAScript 7.0 § 11.7 Punctuators |
| 218 | // |
| 219 | // Punctuator |
| 220 | // DivPunctuator |
| 221 | // RightBracePunctuator |
| 222 | // |
| 223 | '{' => true, |
| 224 | '(' => true, |
| 225 | ')' => true, |
| 226 | '[' => true, |
| 227 | ']' => true, |
| 228 | // Dots have a special case after $dotlessNum which require whitespace |
| 229 | '.' => true, |
| 230 | ';' => true, |
| 231 | ',' => true, |
| 232 | '<' => true, |
| 233 | '>' => true, |
| 234 | '=' => true, |
| 235 | '!' => true, |
| 236 | '+' => true, |
| 237 | '-' => true, |
| 238 | '*' => true, |
| 239 | '%' => true, |
| 240 | '&' => true, |
| 241 | '|' => true, |
| 242 | '^' => true, |
| 243 | '~' => true, |
| 244 | '?' => true, |
| 245 | ':' => true, |
| 246 | '/' => true, |
| 247 | '}' => true, |
| 248 | |
| 249 | // ECMAScript 7.0 § 11.8.4 String Literals |
| 250 | '"' => true, |
| 251 | "'" => true, |
| 252 | |
| 253 | // ECMAScript 7.0 § 11.8.6 Template Literal Lexical Components |
| 254 | '`' => true, |
| 255 | ]; |
| 256 | |
| 257 | /** |
| 258 | * @var array $tokenTypes |
| 259 | * |
| 260 | * Tokens and their types. |
| 261 | */ |
| 262 | private static $tokenTypes = [ |
| 263 | // ECMAScript 7.0 § 12.2 Primary Expression |
| 264 | // |
| 265 | // ...BindingIdentifier |
| 266 | // |
| 267 | '...' => self::TYPE_UN_OP, |
| 268 | |
| 269 | // ECMAScript 7.0 § 12.3 Left-Hand-Side Expressions |
| 270 | // |
| 271 | // MemberExpression |
| 272 | // |
| 273 | // A dot can also be part of a DecimalLiteral, but in that case we handle the entire |
| 274 | // DecimalLiteral as one token. A separate '.' token is always part of a MemberExpression. |
| 275 | '.' => self::TYPE_DOT, |
| 276 | |
| 277 | // ECMAScript 7.0 § 12.4 Update Expressions |
| 278 | // |
| 279 | // LeftHandSideExpression [no LineTerminator here] ++ |
| 280 | // LeftHandSideExpression [no LineTerminator here] -- |
| 281 | // ++ UnaryExpression |
| 282 | // -- UnaryExpression |
| 283 | // |
| 284 | // This is given a separate type from TYPE_UN_OP, |
| 285 | // because `++` and `--` require special handling |
| 286 | // around new lines and semicolon insertion. |
| 287 | // |
| 288 | '++' => self::TYPE_INCR_OP, |
| 289 | '--' => self::TYPE_INCR_OP, |
| 290 | |
| 291 | // ECMAScript 7.0 § 12.5 Unary Operators |
| 292 | // |
| 293 | // UnaryExpression |
| 294 | // includes UpdateExpression |
| 295 | // includes NewExpression, which defines 'new' |
| 296 | // |
| 297 | 'new' => self::TYPE_UN_OP, |
| 298 | 'delete' => self::TYPE_UN_OP, |
| 299 | 'void' => self::TYPE_UN_OP, |
| 300 | 'typeof' => self::TYPE_UN_OP, |
| 301 | '~' => self::TYPE_UN_OP, |
| 302 | '!' => self::TYPE_UN_OP, |
| 303 | |
| 304 | // These operators can be either binary or unary depending on context, |
| 305 | // and thus require separate type from TYPE_UN_OP and TYPE_BIN_OP. |
| 306 | // |
| 307 | // var z = +y; // unary (convert to number) |
| 308 | // var z = x + y; // binary (add operation) |
| 309 | // |
| 310 | // ECMAScript 7.0 § 12.5 Unary Operators |
| 311 | // |
| 312 | // + UnaryExpression |
| 313 | // - UnaryExpression |
| 314 | // |
| 315 | // ECMAScript 7.0 § 12.8 Additive Operators |
| 316 | // |
| 317 | // Expression + Expression |
| 318 | // Expression - Expression |
| 319 | // |
| 320 | '+' => self::TYPE_ADD_OP, |
| 321 | '-' => self::TYPE_ADD_OP, |
| 322 | |
| 323 | // These operators can be treated the same as binary operators. |
| 324 | // They are all defined in one of these two forms, and do |
| 325 | // not require special handling for preserving whitespace or |
| 326 | // line breaks. |
| 327 | // |
| 328 | // Expression operator Expression |
| 329 | // |
| 330 | // Defined in: |
| 331 | // - ECMAScript 7.0 § 12.6 Exponentiation Operator |
| 332 | // ExponentiationExpression |
| 333 | // - ECMAScript 7.0 § 12.7 Multiplicative Operators |
| 334 | // MultiplicativeOperator |
| 335 | // - ECMAScript 7.0 § 12.9 Bitwise Shift Operators |
| 336 | // ShiftExpression |
| 337 | // - ECMAScript 7.0 § 12.10 Relational Operators |
| 338 | // RelationalExpression |
| 339 | // - ECMAScript 7.0 § 12.11 Equality Operators |
| 340 | // EqualityExpression |
| 341 | '**' => self::TYPE_BIN_OP, |
| 342 | '*' => self::TYPE_BIN_OP, |
| 343 | '/' => self::TYPE_BIN_OP, |
| 344 | '%' => self::TYPE_BIN_OP, |
| 345 | '<<' => self::TYPE_BIN_OP, |
| 346 | '>>' => self::TYPE_BIN_OP, |
| 347 | '>>>' => self::TYPE_BIN_OP, |
| 348 | '<' => self::TYPE_BIN_OP, |
| 349 | '>' => self::TYPE_BIN_OP, |
| 350 | '<=' => self::TYPE_BIN_OP, |
| 351 | '>=' => self::TYPE_BIN_OP, |
| 352 | 'instanceof' => self::TYPE_BIN_OP, |
| 353 | 'in' => self::TYPE_BIN_OP, |
| 354 | '==' => self::TYPE_BIN_OP, |
| 355 | '!=' => self::TYPE_BIN_OP, |
| 356 | '===' => self::TYPE_BIN_OP, |
| 357 | '!==' => self::TYPE_BIN_OP, |
| 358 | |
| 359 | // ECMAScript 7.0 § 12.12 Binary Bitwise Operators |
| 360 | // |
| 361 | // BitwiseANDExpression |
| 362 | // BitwiseXORExpression |
| 363 | // BitwiseORExpression |
| 364 | // |
| 365 | '&' => self::TYPE_BIN_OP, |
| 366 | '^' => self::TYPE_BIN_OP, |
| 367 | '|' => self::TYPE_BIN_OP, |
| 368 | |
| 369 | // ECMAScript 7.0 § 12.13 Binary Logical Operators |
| 370 | // |
| 371 | // LogicalANDExpression |
| 372 | // LogicalORExpression |
| 373 | // |
| 374 | '&&' => self::TYPE_BIN_OP, |
| 375 | '||' => self::TYPE_BIN_OP, |
| 376 | |
| 377 | // ECMAScript 11.0 § 12.13 Binary Logical Operators |
| 378 | '??' => self::TYPE_BIN_OP, |
| 379 | |
| 380 | // ECMAScript 7.0 § 12.14 Conditional Operator |
| 381 | // |
| 382 | // ConditionalExpression: |
| 383 | // LogicalORExpression ? AssignmentExpression : AssignmentExpression |
| 384 | // |
| 385 | // Also known as ternary. |
| 386 | '?' => self::TYPE_HOOK, |
| 387 | ':' => self::TYPE_COLON, |
| 388 | |
| 389 | // ECMAScript 7.0 § 12.15 Assignment Operators |
| 390 | '=' => self::TYPE_BIN_OP, |
| 391 | '*=' => self::TYPE_BIN_OP, |
| 392 | '/=' => self::TYPE_BIN_OP, |
| 393 | '%=' => self::TYPE_BIN_OP, |
| 394 | '+=' => self::TYPE_BIN_OP, |
| 395 | '-=' => self::TYPE_BIN_OP, |
| 396 | '<<=' => self::TYPE_BIN_OP, |
| 397 | '>>=' => self::TYPE_BIN_OP, |
| 398 | '>>>=' => self::TYPE_BIN_OP, |
| 399 | '&=' => self::TYPE_BIN_OP, |
| 400 | '^=' => self::TYPE_BIN_OP, |
| 401 | '|=' => self::TYPE_BIN_OP, |
| 402 | '**=' => self::TYPE_BIN_OP, |
| 403 | |
| 404 | // ECMAScript 7.0 § 12.16 Comma Operator |
| 405 | ',' => self::TYPE_COMMA, |
| 406 | |
| 407 | // ECMAScript 7.0 § 11.9.1 Rules of Automatic Semicolon Insertion |
| 408 | // |
| 409 | // These keywords disallow LineTerminator before their (sometimes optional) |
| 410 | // Expression or Identifier. They are similar enough that we can treat |
| 411 | // them all the same way that we treat return, with regards to new line |
| 412 | // and semicolon insertion. |
| 413 | // |
| 414 | // keyword ; |
| 415 | // keyword [no LineTerminator here] Identifier ; |
| 416 | // keyword [no LineTerminator here] Expression ; |
| 417 | // |
| 418 | // See also ECMAScript 7.0: |
| 419 | // - § 13.8 The continue Statement |
| 420 | // - § 13.9 The break Statement |
| 421 | // - § 13.10 The return Statement |
| 422 | // - § 13.14 The throw Statement |
| 423 | // - § 14.4 Generator Function Definitions (yield) |
| 424 | 'continue' => self::TYPE_RETURN, |
| 425 | 'break' => self::TYPE_RETURN, |
| 426 | 'return' => self::TYPE_RETURN, |
| 427 | 'throw' => self::TYPE_RETURN, |
| 428 | // "yield" only counts as a keyword if when inside inside a generator functions, |
| 429 | // otherwise it is a regular identifier. |
| 430 | // This is handled with the negative states hack: if the state is negative, TYPE_YIELD |
| 431 | // is treated as TYPE_RETURN, if it's positive it's treated as TYPE_LITERAL |
| 432 | 'yield' => self::TYPE_YIELD, |
| 433 | |
| 434 | // These keywords require a parenthesised Expression or Identifier before the |
| 435 | // next Statement. They are similar enough to all treat like "if". |
| 436 | // |
| 437 | // keyword ( Expression ) Statement |
| 438 | // keyword ( Identifier ) Statement |
| 439 | // |
| 440 | // See also ECMAScript 7.0: |
| 441 | // - § 13.6 The if Statement |
| 442 | // - § 13.7 Iteration Statements (while, for) |
| 443 | // - § 13.11 The with Statement |
| 444 | // - § 13.12 The switch Statement |
| 445 | // - § 13.15 The try Statement (catch) |
| 446 | 'if' => self::TYPE_IF, |
| 447 | 'while' => self::TYPE_IF, |
| 448 | 'for' => self::TYPE_IF, |
| 449 | 'with' => self::TYPE_IF, |
| 450 | 'switch' => self::TYPE_IF, |
| 451 | 'catch' => self::TYPE_IF, |
| 452 | |
| 453 | // The keywords followed by a Statement, Expression, or Block. |
| 454 | // |
| 455 | // keyword Statement |
| 456 | // keyword Expression |
| 457 | // keyword Block |
| 458 | // |
| 459 | // See also ECMAScript 7.0: |
| 460 | // - § 13.6 The if Statement (else) |
| 461 | // - § 13.7 Iteration Statements (do) |
| 462 | // - § 13.12 The switch Statement (case) |
| 463 | // - § 13.15 The try Statement (try, finally) |
| 464 | 'else' => self::TYPE_DO, |
| 465 | 'do' => self::TYPE_DO, |
| 466 | 'case' => self::TYPE_DO, |
| 467 | 'try' => self::TYPE_DO, |
| 468 | 'finally' => self::TYPE_DO, |
| 469 | |
| 470 | // ECMAScript 7.0 § 13.3 Declarations and the Variable Statement |
| 471 | // |
| 472 | // LetOrConst |
| 473 | // VariableStatement |
| 474 | // |
| 475 | // These keywords are followed by a variable declaration statement. |
| 476 | // This needs to be treated differently from the TYPE_DO group, |
| 477 | // because for TYPE_VAR, when we see an "{" open curly brace, it |
| 478 | // begins object destructuring (ObjectBindingPattern), not a block. |
| 479 | 'var' => self::TYPE_VAR, |
| 480 | 'let' => self::TYPE_VAR, |
| 481 | 'const' => self::TYPE_VAR, |
| 482 | |
| 483 | // ECMAScript 7.0 § 14.1 Function Definitions |
| 484 | 'function' => self::TYPE_FUNC, |
| 485 | |
| 486 | // ECMAScript 7.0 § 14.2 Arrow Function Definitions |
| 487 | '=>' => self::TYPE_ARROW, |
| 488 | |
| 489 | // ECMAScript 7.0 § 14.5 Class Definitions |
| 490 | // |
| 491 | // class Identifier { ClassBody } |
| 492 | // class { ClassBody } |
| 493 | // class Identifier extends Expression { ClassBody } |
| 494 | // class extends Expression { ClassBody } |
| 495 | // |
| 496 | 'class' => self::TYPE_CLASS, |
| 497 | |
| 498 | // ECMAScript 8.0 § 14.6 AwaitExpression |
| 499 | // |
| 500 | // await UnaryExpression |
| 501 | // |
| 502 | 'await' => self::TYPE_AWAIT, |
| 503 | |
| 504 | // Can be one of: |
| 505 | // - Block (ECMAScript 7.0 § 13.2 Block) |
| 506 | // - ObjectLiteral (ECMAScript 7.0 § 12.2 Primary Expression) |
| 507 | '{' => self::TYPE_BRACE_OPEN, |
| 508 | '}' => self::TYPE_BRACE_CLOSE, |
| 509 | |
| 510 | // Can be one of: |
| 511 | // - Parenthesised Identifier or Expression after a |
| 512 | // TYPE_IF or TYPE_FUNC keyword. |
| 513 | // - PrimaryExpression (ECMAScript 7.0 § 12.2 Primary Expression) |
| 514 | // - CallExpression (ECMAScript 7.0 § 12.3 Left-Hand-Side Expressions) |
| 515 | // - Beginning of an ArrowFunction (ECMAScript 7.0 § 14.2 Arrow Function Definitions) |
| 516 | '(' => self::TYPE_PAREN_OPEN, |
| 517 | ')' => self::TYPE_PAREN_CLOSE, |
| 518 | |
| 519 | // Can be one of: |
| 520 | // - ArrayLiteral (ECMAScript 7.0 § 12.2 Primary Expressions) |
| 521 | // - ComputedPropertyName (ECMAScript 7.0 § 12.2.6 Object Initializer) |
| 522 | '[' => self::TYPE_PAREN_OPEN, |
| 523 | ']' => self::TYPE_PAREN_CLOSE, |
| 524 | |
| 525 | // Can be one of: |
| 526 | // - End of any statement |
| 527 | // - EmptyStatement (ECMAScript 7.0 § 13.4 Empty Statement) |
| 528 | ';' => self::TYPE_SEMICOLON, |
| 529 | |
| 530 | // ECMAScript 8.0 § 14.6 Async Function Definitions |
| 531 | // async [no LineTerminator here] function ... |
| 532 | // async [no LineTerminator here] propertyName() ... |
| 533 | 'async' => self::TYPE_ASYNC, |
| 534 | ]; |
| 535 | |
| 536 | /** |
| 537 | * @var array $model |
| 538 | * |
| 539 | * The main table for the state machine. Defines the desired action for every state/token pair. |
| 540 | * |
| 541 | * The state pushed onto the stack by ACTION_PUSH will be returned to by ACTION_POP. |
| 542 | * A state/token pair may not specify both ACTION_POP and ACTION_GOTO. If that does happen, |
| 543 | * ACTION_POP takes precedence. |
| 544 | * |
| 545 | * This table is augmented by self::ensureExpandedStates(). |
| 546 | */ |
| 547 | private static $model = [ |
| 548 | // Statement - This is the initial state. |
| 549 | self::STATEMENT => [ |
| 550 | self::TYPE_UN_OP => [ |
| 551 | self::ACTION_GOTO => self::EXPRESSION, |
| 552 | ], |
| 553 | self::TYPE_INCR_OP => [ |
| 554 | self::ACTION_GOTO => self::EXPRESSION, |
| 555 | ], |
| 556 | self::TYPE_ADD_OP => [ |
| 557 | self::ACTION_GOTO => self::EXPRESSION, |
| 558 | ], |
| 559 | self::TYPE_BRACE_OPEN => [ |
| 560 | // Use of '{' in statement context, creates a Block. |
| 561 | self::ACTION_PUSH => self::STATEMENT, |
| 562 | ], |
| 563 | self::TYPE_BRACE_CLOSE => [ |
| 564 | // Ends a Block |
| 565 | self::ACTION_POP => true, |
| 566 | ], |
| 567 | self::TYPE_PAREN_OPEN => [ |
| 568 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 569 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 570 | ], |
| 571 | self::TYPE_RETURN => [ |
| 572 | self::ACTION_GOTO => self::EXPRESSION_NO_NL, |
| 573 | ], |
| 574 | self::TYPE_IF => [ |
| 575 | self::ACTION_GOTO => self::CONDITION, |
| 576 | ], |
| 577 | self::TYPE_VAR => [ |
| 578 | self::ACTION_GOTO => self::EXPRESSION, |
| 579 | ], |
| 580 | self::TYPE_FUNC => [ |
| 581 | self::ACTION_PUSH => self::STATEMENT, |
| 582 | self::ACTION_GOTO => self::FUNC, |
| 583 | ], |
| 584 | self::TYPE_CLASS => [ |
| 585 | self::ACTION_PUSH => self::STATEMENT, |
| 586 | self::ACTION_GOTO => self::CLASS_DEF, |
| 587 | ], |
| 588 | self::TYPE_SPECIAL => [ |
| 589 | 'import' => [ |
| 590 | self::ACTION_GOTO => self::IMPORT_EXPORT, |
| 591 | ], |
| 592 | 'export' => [ |
| 593 | self::ACTION_GOTO => self::IMPORT_EXPORT, |
| 594 | ], |
| 595 | ], |
| 596 | self::TYPE_LITERAL => [ |
| 597 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 598 | ], |
| 599 | self::TYPE_ASYNC => [ |
| 600 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 601 | ], |
| 602 | self::TYPE_AWAIT => [ |
| 603 | self::ACTION_GOTO => self::EXPRESSION, |
| 604 | ], |
| 605 | ], |
| 606 | // The state after if/catch/while/for/switch/with |
| 607 | // Waits for an expression in parentheses, then goes to STATEMENT |
| 608 | self::CONDITION => [ |
| 609 | self::TYPE_PAREN_OPEN => [ |
| 610 | self::ACTION_PUSH => self::STATEMENT, |
| 611 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 612 | ], |
| 613 | ], |
| 614 | // The state after the function keyword. Waits for {, then goes to STATEMENT. |
| 615 | // The function body's closing } will pop the stack, so the state to return to |
| 616 | // after the function should be pushed to the stack first |
| 617 | self::FUNC => [ |
| 618 | // Needed to prevent * in an expression in the argument list from improperly |
| 619 | // triggering GENFUNC |
| 620 | self::TYPE_PAREN_OPEN => [ |
| 621 | self::ACTION_PUSH => self::FUNC, |
| 622 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 623 | ], |
| 624 | self::TYPE_BRACE_OPEN => [ |
| 625 | self::ACTION_GOTO => self::STATEMENT, |
| 626 | ], |
| 627 | self::TYPE_SPECIAL => [ |
| 628 | '*' => [ |
| 629 | self::ACTION_GOTO => self::GENFUNC, |
| 630 | ], |
| 631 | ], |
| 632 | ], |
| 633 | // After function*. Waits for { , then goes to a generator function statement. |
| 634 | self::GENFUNC => [ |
| 635 | self::TYPE_BRACE_OPEN => [ |
| 636 | // Note negative value: generator function states are negative |
| 637 | self::ACTION_GOTO => -self::STATEMENT |
| 638 | ], |
| 639 | ], |
| 640 | // Property assignment - This is an object literal declaration. |
| 641 | // For example: `{ key: value, key2, [computedKey3]: value3, method4() { ... } }` |
| 642 | self::PROPERTY_ASSIGNMENT => [ |
| 643 | // Note that keywords like if, class, var, delete, instanceof etc. can be used as keys, |
| 644 | // and should be treated as literals here, as they are in EXPRESSION_DOT. In this state, |
| 645 | // that is implicitly true because TYPE_LITERAL has no action, so it stays in this state. |
| 646 | // If we later add a state transition for TYPE_LITERAL, that same transition should |
| 647 | // also be applied to TYPE_RETURN, TYPE_IF, TYPE_DO, TYPE_VAR, TYPE_FUNC and TYPE_CLASS. |
| 648 | self::TYPE_COLON => [ |
| 649 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION, |
| 650 | ], |
| 651 | // For {, which begins a method |
| 652 | self::TYPE_BRACE_OPEN => [ |
| 653 | self::ACTION_PUSH => self::PROPERTY_ASSIGNMENT, |
| 654 | // This is not flipped, see "Special cases" below |
| 655 | self::ACTION_GOTO => self::STATEMENT, |
| 656 | ], |
| 657 | self::TYPE_BRACE_CLOSE => [ |
| 658 | self::ACTION_POP => true, |
| 659 | ], |
| 660 | // For [, which begins a computed key |
| 661 | self::TYPE_PAREN_OPEN => [ |
| 662 | self::ACTION_PUSH => self::PROPERTY_ASSIGNMENT, |
| 663 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 664 | ], |
| 665 | self::TYPE_SPECIAL => [ |
| 666 | '*' => [ |
| 667 | self::ACTION_PUSH => self::PROPERTY_ASSIGNMENT, |
| 668 | self::ACTION_GOTO => self::GENFUNC, |
| 669 | ], |
| 670 | ], |
| 671 | ], |
| 672 | // Place in an expression where we expect an operand or a unary operator: the start |
| 673 | // of an expression or after an operator. Note that unary operators (including INCR_OP |
| 674 | // and ADD_OP) cause us to stay in this state, while operands take us to EXPRESSION_OP |
| 675 | self::EXPRESSION => [ |
| 676 | self::TYPE_SEMICOLON => [ |
| 677 | self::ACTION_GOTO => self::STATEMENT, |
| 678 | ], |
| 679 | self::TYPE_BRACE_OPEN => [ |
| 680 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 681 | self::ACTION_GOTO => self::PROPERTY_ASSIGNMENT, |
| 682 | ], |
| 683 | self::TYPE_BRACE_CLOSE => [ |
| 684 | self::ACTION_POP => true, |
| 685 | ], |
| 686 | self::TYPE_PAREN_OPEN => [ |
| 687 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 688 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 689 | ], |
| 690 | self::TYPE_FUNC => [ |
| 691 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 692 | self::ACTION_GOTO => self::FUNC, |
| 693 | ], |
| 694 | self::TYPE_CLASS => [ |
| 695 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 696 | self::ACTION_GOTO => self::CLASS_DEF, |
| 697 | ], |
| 698 | self::TYPE_LITERAL => [ |
| 699 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 700 | ], |
| 701 | self::TYPE_ASYNC => [ |
| 702 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 703 | ], |
| 704 | // 'return' can't appear here, but 'yield' can |
| 705 | self::TYPE_RETURN => [ |
| 706 | self::ACTION_GOTO => self::EXPRESSION_NO_NL, |
| 707 | ], |
| 708 | ], |
| 709 | // An expression immediately after return/throw/break/continue/yield, where a newline |
| 710 | // is not allowed. This state is identical to EXPRESSION, except that semicolon |
| 711 | // insertion can happen here, and we (almost) never stay here: in cases where EXPRESSION |
| 712 | // would do nothing, we go to EXPRESSION. We only stay here if there's a double yield, |
| 713 | // because 'yield yield foo' is a valid expression. |
| 714 | self::EXPRESSION_NO_NL => [ |
| 715 | self::TYPE_UN_OP => [ |
| 716 | self::ACTION_GOTO => self::EXPRESSION, |
| 717 | ], |
| 718 | self::TYPE_INCR_OP => [ |
| 719 | self::ACTION_GOTO => self::EXPRESSION, |
| 720 | ], |
| 721 | // BIN_OP seems impossible at the start of an expression, but it can happen in |
| 722 | // yield *foo |
| 723 | self::TYPE_BIN_OP => [ |
| 724 | self::ACTION_GOTO => self::EXPRESSION, |
| 725 | ], |
| 726 | self::TYPE_ADD_OP => [ |
| 727 | self::ACTION_GOTO => self::EXPRESSION, |
| 728 | ], |
| 729 | self::TYPE_SEMICOLON => [ |
| 730 | self::ACTION_GOTO => self::STATEMENT, |
| 731 | ], |
| 732 | self::TYPE_BRACE_OPEN => [ |
| 733 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 734 | self::ACTION_GOTO => self::PROPERTY_ASSIGNMENT, |
| 735 | ], |
| 736 | self::TYPE_BRACE_CLOSE => [ |
| 737 | self::ACTION_POP => true, |
| 738 | ], |
| 739 | self::TYPE_PAREN_OPEN => [ |
| 740 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 741 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 742 | ], |
| 743 | self::TYPE_FUNC => [ |
| 744 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 745 | self::ACTION_GOTO => self::FUNC, |
| 746 | ], |
| 747 | self::TYPE_CLASS => [ |
| 748 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 749 | self::ACTION_GOTO => self::CLASS_DEF, |
| 750 | ], |
| 751 | self::TYPE_LITERAL => [ |
| 752 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 753 | ], |
| 754 | self::TYPE_ASYNC => [ |
| 755 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 756 | ], |
| 757 | self::TYPE_AWAIT => [ |
| 758 | self::ACTION_GOTO => self::EXPRESSION, |
| 759 | ], |
| 760 | // 'return' can't appear here, because 'return return' isn't allowed |
| 761 | // But 'yield' can appear here, because 'yield yield' is allowed |
| 762 | self::TYPE_RETURN => [ |
| 763 | self::ACTION_GOTO => self::EXPRESSION_NO_NL, |
| 764 | ], |
| 765 | ], |
| 766 | // Place in an expression after an operand, where we expect an operator |
| 767 | self::EXPRESSION_OP => [ |
| 768 | self::TYPE_BIN_OP => [ |
| 769 | self::ACTION_GOTO => self::EXPRESSION, |
| 770 | ], |
| 771 | self::TYPE_ADD_OP => [ |
| 772 | self::ACTION_GOTO => self::EXPRESSION, |
| 773 | ], |
| 774 | self::TYPE_DOT => [ |
| 775 | self::ACTION_GOTO => self::EXPRESSION_DOT, |
| 776 | ], |
| 777 | self::TYPE_HOOK => [ |
| 778 | self::ACTION_PUSH => self::EXPRESSION, |
| 779 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 780 | ], |
| 781 | self::TYPE_COLON => [ |
| 782 | self::ACTION_GOTO => self::STATEMENT, |
| 783 | ], |
| 784 | self::TYPE_COMMA => [ |
| 785 | self::ACTION_GOTO => self::EXPRESSION, |
| 786 | ], |
| 787 | self::TYPE_SEMICOLON => [ |
| 788 | self::ACTION_GOTO => self::STATEMENT, |
| 789 | ], |
| 790 | self::TYPE_ARROW => [ |
| 791 | self::ACTION_GOTO => self::EXPRESSION_ARROWFUNC, |
| 792 | ], |
| 793 | self::TYPE_PAREN_OPEN => [ |
| 794 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 795 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 796 | ], |
| 797 | self::TYPE_BRACE_CLOSE => [ |
| 798 | self::ACTION_POP => true, |
| 799 | ], |
| 800 | self::TYPE_FUNC => [ |
| 801 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 802 | self::ACTION_GOTO => self::FUNC, |
| 803 | ], |
| 804 | ], |
| 805 | // State after a dot (.). Like EXPRESSION, except that many keywords behave like literals |
| 806 | // (e.g. class, if, else, var, function) because they're not valid as identifiers but are |
| 807 | // valid as property names. |
| 808 | self::EXPRESSION_DOT => [ |
| 809 | self::TYPE_LITERAL => [ |
| 810 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 811 | ], |
| 812 | // The following are keywords behaving as literals |
| 813 | self::TYPE_RETURN => [ |
| 814 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 815 | ], |
| 816 | self::TYPE_IF => [ |
| 817 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 818 | ], |
| 819 | self::TYPE_DO => [ |
| 820 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 821 | ], |
| 822 | self::TYPE_VAR => [ |
| 823 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 824 | ], |
| 825 | self::TYPE_FUNC => [ |
| 826 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 827 | ], |
| 828 | self::TYPE_CLASS => [ |
| 829 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 830 | ], |
| 831 | // We don't expect real unary/binary operators here, but some keywords |
| 832 | // (new, delete, void, typeof, instanceof, in) are classified as such, and they can be |
| 833 | // used as property names |
| 834 | self::TYPE_UN_OP => [ |
| 835 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 836 | ], |
| 837 | self::TYPE_BIN_OP => [ |
| 838 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 839 | ], |
| 840 | ], |
| 841 | // State after the } closing an arrow function body: like STATEMENT except |
| 842 | // that it has semicolon insertion, COMMA can continue the expression, and after |
| 843 | // a function we go to STATEMENT instead of EXPRESSION_OP |
| 844 | self::EXPRESSION_END => [ |
| 845 | self::TYPE_UN_OP => [ |
| 846 | self::ACTION_GOTO => self::EXPRESSION, |
| 847 | ], |
| 848 | self::TYPE_INCR_OP => [ |
| 849 | self::ACTION_GOTO => self::EXPRESSION, |
| 850 | ], |
| 851 | self::TYPE_ADD_OP => [ |
| 852 | self::ACTION_GOTO => self::EXPRESSION, |
| 853 | ], |
| 854 | self::TYPE_COMMA => [ |
| 855 | self::ACTION_GOTO => self::EXPRESSION, |
| 856 | ], |
| 857 | self::TYPE_SEMICOLON => [ |
| 858 | self::ACTION_GOTO => self::STATEMENT, |
| 859 | ], |
| 860 | self::TYPE_BRACE_OPEN => [ |
| 861 | self::ACTION_PUSH => self::STATEMENT, |
| 862 | self::ACTION_GOTO => self::STATEMENT, |
| 863 | ], |
| 864 | self::TYPE_BRACE_CLOSE => [ |
| 865 | self::ACTION_POP => true, |
| 866 | ], |
| 867 | self::TYPE_PAREN_OPEN => [ |
| 868 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 869 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 870 | ], |
| 871 | self::TYPE_RETURN => [ |
| 872 | self::ACTION_GOTO => self::EXPRESSION_NO_NL, |
| 873 | ], |
| 874 | self::TYPE_IF => [ |
| 875 | self::ACTION_GOTO => self::CONDITION, |
| 876 | ], |
| 877 | self::TYPE_VAR => [ |
| 878 | self::ACTION_GOTO => self::EXPRESSION, |
| 879 | ], |
| 880 | self::TYPE_FUNC => [ |
| 881 | self::ACTION_PUSH => self::STATEMENT, |
| 882 | self::ACTION_GOTO => self::FUNC, |
| 883 | ], |
| 884 | self::TYPE_CLASS => [ |
| 885 | self::ACTION_PUSH => self::STATEMENT, |
| 886 | self::ACTION_GOTO => self::CLASS_DEF, |
| 887 | ], |
| 888 | self::TYPE_LITERAL => [ |
| 889 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 890 | ], |
| 891 | self::TYPE_ASYNC => [ |
| 892 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 893 | ], |
| 894 | ], |
| 895 | // State after =>. Like EXPRESSION, except that { begins an arrow function body |
| 896 | // rather than an object literal. |
| 897 | self::EXPRESSION_ARROWFUNC => [ |
| 898 | self::TYPE_UN_OP => [ |
| 899 | self::ACTION_GOTO => self::EXPRESSION, |
| 900 | ], |
| 901 | self::TYPE_INCR_OP => [ |
| 902 | self::ACTION_GOTO => self::EXPRESSION, |
| 903 | ], |
| 904 | self::TYPE_ADD_OP => [ |
| 905 | self::ACTION_GOTO => self::EXPRESSION, |
| 906 | ], |
| 907 | self::TYPE_BRACE_OPEN => [ |
| 908 | self::ACTION_PUSH => self::EXPRESSION_END, |
| 909 | self::ACTION_GOTO => self::STATEMENT, |
| 910 | ], |
| 911 | self::TYPE_PAREN_OPEN => [ |
| 912 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 913 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 914 | ], |
| 915 | self::TYPE_FUNC => [ |
| 916 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 917 | self::ACTION_GOTO => self::FUNC, |
| 918 | ], |
| 919 | self::TYPE_CLASS => [ |
| 920 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 921 | self::ACTION_GOTO => self::CLASS_DEF, |
| 922 | ], |
| 923 | self::TYPE_LITERAL => [ |
| 924 | self::ACTION_GOTO => self::EXPRESSION_OP, |
| 925 | ], |
| 926 | ], |
| 927 | // Expression after a ? . This differs from EXPRESSION because a : ends the ternary |
| 928 | // rather than starting STATEMENT (outside a ternary, : comes after a goto label) |
| 929 | // The actual rule for : ending the ternary is in EXPRESSION_TERNARY_OP. |
| 930 | self::EXPRESSION_TERNARY => [ |
| 931 | self::TYPE_BRACE_OPEN => [ |
| 932 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 933 | self::ACTION_GOTO => self::PROPERTY_ASSIGNMENT, |
| 934 | ], |
| 935 | self::TYPE_PAREN_OPEN => [ |
| 936 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 937 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 938 | ], |
| 939 | self::TYPE_FUNC => [ |
| 940 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 941 | self::ACTION_GOTO => self::FUNC, |
| 942 | ], |
| 943 | self::TYPE_CLASS => [ |
| 944 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 945 | self::ACTION_GOTO => self::CLASS_DEF, |
| 946 | ], |
| 947 | self::TYPE_LITERAL => [ |
| 948 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_OP, |
| 949 | ], |
| 950 | // 'return' can't appear here, but 'yield' can |
| 951 | self::TYPE_RETURN => [ |
| 952 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_NO_NL, |
| 953 | ], |
| 954 | ], |
| 955 | // Like EXPRESSION_TERNARY, except that semicolon insertion can happen |
| 956 | // See also EXPRESSION_NO_NL |
| 957 | self::EXPRESSION_TERNARY_NO_NL => [ |
| 958 | self::TYPE_BRACE_OPEN => [ |
| 959 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 960 | self::ACTION_GOTO => self::PROPERTY_ASSIGNMENT, |
| 961 | ], |
| 962 | self::TYPE_PAREN_OPEN => [ |
| 963 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 964 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 965 | ], |
| 966 | self::TYPE_FUNC => [ |
| 967 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 968 | self::ACTION_GOTO => self::FUNC, |
| 969 | ], |
| 970 | self::TYPE_CLASS => [ |
| 971 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 972 | self::ACTION_GOTO => self::CLASS_DEF, |
| 973 | ], |
| 974 | self::TYPE_LITERAL => [ |
| 975 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_OP, |
| 976 | ], |
| 977 | // 'yield' can appear here, because 'yield yield' is allowed |
| 978 | self::TYPE_RETURN => [ |
| 979 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_NO_NL, |
| 980 | ], |
| 981 | self::TYPE_UN_OP => [ |
| 982 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 983 | ], |
| 984 | self::TYPE_INCR_OP => [ |
| 985 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 986 | ], |
| 987 | // BIN_OP seems impossible at the start of an expression, but it can happen in |
| 988 | // yield *foo |
| 989 | self::TYPE_BIN_OP => [ |
| 990 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 991 | ], |
| 992 | self::TYPE_ADD_OP => [ |
| 993 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 994 | ], |
| 995 | ], |
| 996 | // Like EXPRESSION_OP, but for ternaries, see EXPRESSION_TERNARY |
| 997 | self::EXPRESSION_TERNARY_OP => [ |
| 998 | self::TYPE_BIN_OP => [ |
| 999 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 1000 | ], |
| 1001 | self::TYPE_ADD_OP => [ |
| 1002 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 1003 | ], |
| 1004 | self::TYPE_DOT => [ |
| 1005 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_DOT, |
| 1006 | ], |
| 1007 | self::TYPE_HOOK => [ |
| 1008 | self::ACTION_PUSH => self::EXPRESSION_TERNARY, |
| 1009 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 1010 | ], |
| 1011 | self::TYPE_COMMA => [ |
| 1012 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 1013 | ], |
| 1014 | self::TYPE_ARROW => [ |
| 1015 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_ARROWFUNC, |
| 1016 | ], |
| 1017 | self::TYPE_PAREN_OPEN => [ |
| 1018 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 1019 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1020 | ], |
| 1021 | self::TYPE_COLON => [ |
| 1022 | self::ACTION_POP => true, |
| 1023 | ], |
| 1024 | ], |
| 1025 | // Like EXPRESSION_DOT, but for ternaries, see EXPRESSION_TERNARY |
| 1026 | self::EXPRESSION_TERNARY_DOT => [ |
| 1027 | self::TYPE_LITERAL => [ |
| 1028 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_OP, |
| 1029 | ], |
| 1030 | // The following are keywords behaving as literals |
| 1031 | self::TYPE_RETURN => [ |
| 1032 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_OP, |
| 1033 | ], |
| 1034 | self::TYPE_IF => [ |
| 1035 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_OP, |
| 1036 | ], |
| 1037 | self::TYPE_DO => [ |
| 1038 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_OP, |
| 1039 | ], |
| 1040 | self::TYPE_VAR => [ |
| 1041 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_OP, |
| 1042 | ], |
| 1043 | self::TYPE_FUNC => [ |
| 1044 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_OP, |
| 1045 | ], |
| 1046 | self::TYPE_CLASS => [ |
| 1047 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_OP, |
| 1048 | ], |
| 1049 | // We don't expect real unary/binary operators here, but some keywords |
| 1050 | // (new, delete, void, typeof, instanceof, in) are classified as such, and they can be |
| 1051 | // used as property names |
| 1052 | self::TYPE_UN_OP => [ |
| 1053 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_OP, |
| 1054 | ], |
| 1055 | self::TYPE_BIN_OP => [ |
| 1056 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_OP, |
| 1057 | ], |
| 1058 | ], |
| 1059 | // Like EXPRESSION_ARROWFUNC, but for ternaries, see EXPRESSION_TERNARY |
| 1060 | self::EXPRESSION_TERNARY_ARROWFUNC => [ |
| 1061 | self::TYPE_UN_OP => [ |
| 1062 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 1063 | ], |
| 1064 | self::TYPE_INCR_OP => [ |
| 1065 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 1066 | ], |
| 1067 | self::TYPE_ADD_OP => [ |
| 1068 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 1069 | ], |
| 1070 | self::TYPE_BRACE_OPEN => [ |
| 1071 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 1072 | self::ACTION_GOTO => self::STATEMENT, |
| 1073 | ], |
| 1074 | self::TYPE_PAREN_OPEN => [ |
| 1075 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 1076 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1077 | ], |
| 1078 | self::TYPE_FUNC => [ |
| 1079 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 1080 | self::ACTION_GOTO => self::FUNC, |
| 1081 | ], |
| 1082 | self::TYPE_CLASS => [ |
| 1083 | self::ACTION_PUSH => self::EXPRESSION_TERNARY_OP, |
| 1084 | self::ACTION_GOTO => self::CLASS_DEF, |
| 1085 | ], |
| 1086 | self::TYPE_LITERAL => [ |
| 1087 | self::ACTION_GOTO => self::EXPRESSION_TERNARY_OP, |
| 1088 | ], |
| 1089 | ], |
| 1090 | // Expression inside parentheses. Like EXPRESSION, except that ) ends this state |
| 1091 | // This differs from EXPRESSION because semicolon insertion can't happen here |
| 1092 | self::PAREN_EXPRESSION => [ |
| 1093 | self::TYPE_BRACE_OPEN => [ |
| 1094 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1095 | self::ACTION_GOTO => self::PROPERTY_ASSIGNMENT, |
| 1096 | ], |
| 1097 | self::TYPE_PAREN_OPEN => [ |
| 1098 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1099 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1100 | ], |
| 1101 | self::TYPE_PAREN_CLOSE => [ |
| 1102 | self::ACTION_POP => true, |
| 1103 | ], |
| 1104 | self::TYPE_FUNC => [ |
| 1105 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1106 | self::ACTION_GOTO => self::FUNC, |
| 1107 | ], |
| 1108 | self::TYPE_CLASS => [ |
| 1109 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1110 | self::ACTION_GOTO => self::CLASS_DEF, |
| 1111 | ], |
| 1112 | self::TYPE_LITERAL => [ |
| 1113 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP, |
| 1114 | ], |
| 1115 | self::TYPE_ASYNC => [ |
| 1116 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP_NO_NL, |
| 1117 | ], |
| 1118 | // 'return' can't appear here, but 'yield' can |
| 1119 | self::TYPE_RETURN => [ |
| 1120 | self::ACTION_GOTO => self::PAREN_EXPRESSION_NO_NL, |
| 1121 | ], |
| 1122 | ], |
| 1123 | // Like PAREN_EXPRESSION, except that semicolon insertion can happen |
| 1124 | // See also EXPRESSION_NO_NL |
| 1125 | self::PAREN_EXPRESSION_NO_NL => [ |
| 1126 | self::TYPE_BRACE_OPEN => [ |
| 1127 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1128 | self::ACTION_GOTO => self::PROPERTY_ASSIGNMENT, |
| 1129 | ], |
| 1130 | self::TYPE_PAREN_OPEN => [ |
| 1131 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1132 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1133 | ], |
| 1134 | self::TYPE_PAREN_CLOSE => [ |
| 1135 | self::ACTION_POP => true, |
| 1136 | ], |
| 1137 | self::TYPE_FUNC => [ |
| 1138 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1139 | self::ACTION_GOTO => self::FUNC, |
| 1140 | ], |
| 1141 | self::TYPE_CLASS => [ |
| 1142 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1143 | self::ACTION_GOTO => self::CLASS_DEF, |
| 1144 | ], |
| 1145 | self::TYPE_LITERAL => [ |
| 1146 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP, |
| 1147 | ], |
| 1148 | self::TYPE_ASYNC => [ |
| 1149 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP_NO_NL, |
| 1150 | ], |
| 1151 | // 'yield' can appear here, because 'yield yield' is allowed |
| 1152 | self::TYPE_RETURN => [ |
| 1153 | self::ACTION_GOTO => self::PAREN_EXPRESSION_NO_NL, |
| 1154 | ], |
| 1155 | self::TYPE_UN_OP => [ |
| 1156 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1157 | ], |
| 1158 | self::TYPE_INCR_OP => [ |
| 1159 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1160 | ], |
| 1161 | // BIN_OP seems impossible at the start of an expression, but it can happen in |
| 1162 | // yield *foo |
| 1163 | self::TYPE_BIN_OP => [ |
| 1164 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1165 | ], |
| 1166 | self::TYPE_ADD_OP => [ |
| 1167 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1168 | ], |
| 1169 | self::TYPE_AWAIT => [ |
| 1170 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1171 | ], |
| 1172 | ], |
| 1173 | // Like EXPRESSION_OP, but in parentheses, see PAREN_EXPRESSION |
| 1174 | self::PAREN_EXPRESSION_OP => [ |
| 1175 | self::TYPE_BIN_OP => [ |
| 1176 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1177 | ], |
| 1178 | self::TYPE_ADD_OP => [ |
| 1179 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1180 | ], |
| 1181 | self::TYPE_DOT => [ |
| 1182 | self::ACTION_GOTO => self::PAREN_EXPRESSION_DOT, |
| 1183 | ], |
| 1184 | self::TYPE_HOOK => [ |
| 1185 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1186 | ], |
| 1187 | self::TYPE_COLON => [ |
| 1188 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1189 | ], |
| 1190 | self::TYPE_COMMA => [ |
| 1191 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1192 | ], |
| 1193 | self::TYPE_SEMICOLON => [ |
| 1194 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1195 | ], |
| 1196 | self::TYPE_ARROW => [ |
| 1197 | self::ACTION_GOTO => self::PAREN_EXPRESSION_ARROWFUNC, |
| 1198 | ], |
| 1199 | self::TYPE_PAREN_OPEN => [ |
| 1200 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1201 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1202 | ], |
| 1203 | self::TYPE_PAREN_CLOSE => [ |
| 1204 | self::ACTION_POP => true, |
| 1205 | ], |
| 1206 | ], |
| 1207 | // Like EXPRESSION_DOT, but in parentheses, see PAREN_EXPRESSION |
| 1208 | self::PAREN_EXPRESSION_DOT => [ |
| 1209 | self::TYPE_LITERAL => [ |
| 1210 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP, |
| 1211 | ], |
| 1212 | // The following are keywords behaving as literals |
| 1213 | self::TYPE_RETURN => [ |
| 1214 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP, |
| 1215 | ], |
| 1216 | self::TYPE_IF => [ |
| 1217 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP, |
| 1218 | ], |
| 1219 | self::TYPE_DO => [ |
| 1220 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP, |
| 1221 | ], |
| 1222 | self::TYPE_VAR => [ |
| 1223 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP, |
| 1224 | ], |
| 1225 | self::TYPE_FUNC => [ |
| 1226 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP, |
| 1227 | ], |
| 1228 | self::TYPE_CLASS => [ |
| 1229 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP, |
| 1230 | ], |
| 1231 | // We don't expect real unary/binary operators here, but some keywords |
| 1232 | // (new, delete, void, typeof, instanceof, in) are classified as such, and they can be |
| 1233 | // used as property names |
| 1234 | self::TYPE_UN_OP => [ |
| 1235 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP, |
| 1236 | ], |
| 1237 | self::TYPE_BIN_OP => [ |
| 1238 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP, |
| 1239 | ], |
| 1240 | ], |
| 1241 | // Like EXPRESSION_ARROWFUNC, but in parentheses, see PAREN_EXPRESSION |
| 1242 | self::PAREN_EXPRESSION_ARROWFUNC => [ |
| 1243 | self::TYPE_UN_OP => [ |
| 1244 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1245 | ], |
| 1246 | self::TYPE_INCR_OP => [ |
| 1247 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1248 | ], |
| 1249 | self::TYPE_ADD_OP => [ |
| 1250 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1251 | ], |
| 1252 | self::TYPE_BRACE_OPEN => [ |
| 1253 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1254 | self::ACTION_GOTO => self::STATEMENT, |
| 1255 | ], |
| 1256 | self::TYPE_PAREN_OPEN => [ |
| 1257 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1258 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1259 | ], |
| 1260 | self::TYPE_FUNC => [ |
| 1261 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1262 | self::ACTION_GOTO => self::FUNC, |
| 1263 | ], |
| 1264 | self::TYPE_CLASS => [ |
| 1265 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1266 | self::ACTION_GOTO => self::CLASS_DEF, |
| 1267 | ], |
| 1268 | self::TYPE_LITERAL => [ |
| 1269 | self::ACTION_GOTO => self::PAREN_EXPRESSION_OP, |
| 1270 | ], |
| 1271 | ], |
| 1272 | |
| 1273 | // Like PAREN_EXPRESSION_OP, for the state after "async" in a PAREN_EXPRESSION, |
| 1274 | // for use by the $semicolon model. |
| 1275 | self::PAREN_EXPRESSION_OP_NO_NL => [ |
| 1276 | self::TYPE_BIN_OP => [ |
| 1277 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1278 | ], |
| 1279 | self::TYPE_ADD_OP => [ |
| 1280 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1281 | ], |
| 1282 | self::TYPE_DOT => [ |
| 1283 | self::ACTION_GOTO => self::PAREN_EXPRESSION_DOT, |
| 1284 | ], |
| 1285 | self::TYPE_HOOK => [ |
| 1286 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1287 | ], |
| 1288 | self::TYPE_COLON => [ |
| 1289 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1290 | ], |
| 1291 | self::TYPE_COMMA => [ |
| 1292 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1293 | ], |
| 1294 | self::TYPE_SEMICOLON => [ |
| 1295 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1296 | ], |
| 1297 | self::TYPE_ARROW => [ |
| 1298 | self::ACTION_GOTO => self::PAREN_EXPRESSION_ARROWFUNC, |
| 1299 | ], |
| 1300 | self::TYPE_PAREN_OPEN => [ |
| 1301 | self::ACTION_PUSH => self::PAREN_EXPRESSION_OP, |
| 1302 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1303 | ], |
| 1304 | self::TYPE_PAREN_CLOSE => [ |
| 1305 | self::ACTION_POP => true, |
| 1306 | ], |
| 1307 | ], |
| 1308 | // Expression as the value of a key in an object literal. Like EXPRESSION, except that |
| 1309 | // a comma (in PROPERTY_EXPRESSION_OP) goes to PROPERTY_ASSIGNMENT instead |
| 1310 | self::PROPERTY_EXPRESSION => [ |
| 1311 | self::TYPE_BRACE_OPEN => [ |
| 1312 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1313 | self::ACTION_GOTO => self::PROPERTY_ASSIGNMENT, |
| 1314 | ], |
| 1315 | self::TYPE_BRACE_CLOSE => [ |
| 1316 | self::ACTION_POP => true, |
| 1317 | ], |
| 1318 | self::TYPE_PAREN_OPEN => [ |
| 1319 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1320 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1321 | ], |
| 1322 | self::TYPE_FUNC => [ |
| 1323 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1324 | self::ACTION_GOTO => self::FUNC, |
| 1325 | ], |
| 1326 | self::TYPE_CLASS => [ |
| 1327 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1328 | self::ACTION_GOTO => self::CLASS_DEF, |
| 1329 | ], |
| 1330 | self::TYPE_LITERAL => [ |
| 1331 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_OP, |
| 1332 | ], |
| 1333 | // 'return' can't appear here, but 'yield' can |
| 1334 | self::TYPE_RETURN => [ |
| 1335 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_NO_NL, |
| 1336 | ], |
| 1337 | ], |
| 1338 | // Like PROPERTY_EXPRESSION, except that semicolon insertion can happen |
| 1339 | // See also EXPRESSION_NO_NL |
| 1340 | self::PROPERTY_EXPRESSION_NO_NL => [ |
| 1341 | self::TYPE_BRACE_OPEN => [ |
| 1342 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1343 | self::ACTION_GOTO => self::PROPERTY_ASSIGNMENT, |
| 1344 | ], |
| 1345 | self::TYPE_BRACE_CLOSE => [ |
| 1346 | self::ACTION_POP => true, |
| 1347 | ], |
| 1348 | self::TYPE_PAREN_OPEN => [ |
| 1349 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1350 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1351 | ], |
| 1352 | self::TYPE_FUNC => [ |
| 1353 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1354 | self::ACTION_GOTO => self::FUNC, |
| 1355 | ], |
| 1356 | self::TYPE_CLASS => [ |
| 1357 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1358 | self::ACTION_GOTO => self::CLASS_DEF, |
| 1359 | ], |
| 1360 | self::TYPE_LITERAL => [ |
| 1361 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_OP, |
| 1362 | ], |
| 1363 | // 'yield' can appear here, because 'yield yield' is allowed |
| 1364 | self::TYPE_RETURN => [ |
| 1365 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_NO_NL, |
| 1366 | ], |
| 1367 | self::TYPE_UN_OP => [ |
| 1368 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION, |
| 1369 | ], |
| 1370 | self::TYPE_INCR_OP => [ |
| 1371 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION, |
| 1372 | ], |
| 1373 | // BIN_OP seems impossible at the start of an expression, but it can happen in |
| 1374 | // yield *foo |
| 1375 | self::TYPE_BIN_OP => [ |
| 1376 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION, |
| 1377 | ], |
| 1378 | self::TYPE_ADD_OP => [ |
| 1379 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION, |
| 1380 | ], |
| 1381 | ], |
| 1382 | // Like EXPRESSION_OP, but in a property expression, see PROPERTY_EXPRESSION |
| 1383 | self::PROPERTY_EXPRESSION_OP => [ |
| 1384 | self::TYPE_BIN_OP => [ |
| 1385 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION, |
| 1386 | ], |
| 1387 | self::TYPE_ADD_OP => [ |
| 1388 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION, |
| 1389 | ], |
| 1390 | self::TYPE_DOT => [ |
| 1391 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_DOT, |
| 1392 | ], |
| 1393 | self::TYPE_HOOK => [ |
| 1394 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION, |
| 1395 | self::ACTION_GOTO => self::EXPRESSION_TERNARY, |
| 1396 | ], |
| 1397 | self::TYPE_COMMA => [ |
| 1398 | self::ACTION_GOTO => self::PROPERTY_ASSIGNMENT, |
| 1399 | ], |
| 1400 | self::TYPE_ARROW => [ |
| 1401 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_ARROWFUNC, |
| 1402 | ], |
| 1403 | self::TYPE_BRACE_OPEN => [ |
| 1404 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1405 | ], |
| 1406 | self::TYPE_BRACE_CLOSE => [ |
| 1407 | self::ACTION_POP => true, |
| 1408 | ], |
| 1409 | self::TYPE_PAREN_OPEN => [ |
| 1410 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1411 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1412 | ], |
| 1413 | ], |
| 1414 | // Like EXPRESSION_DOT, but in a property expression, see PROPERTY_EXPRESSION |
| 1415 | self::PROPERTY_EXPRESSION_DOT => [ |
| 1416 | self::TYPE_LITERAL => [ |
| 1417 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_OP, |
| 1418 | ], |
| 1419 | // The following are keywords behaving as literals |
| 1420 | self::TYPE_RETURN => [ |
| 1421 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_OP, |
| 1422 | ], |
| 1423 | self::TYPE_IF => [ |
| 1424 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_OP, |
| 1425 | ], |
| 1426 | self::TYPE_DO => [ |
| 1427 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_OP, |
| 1428 | ], |
| 1429 | self::TYPE_VAR => [ |
| 1430 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_OP, |
| 1431 | ], |
| 1432 | self::TYPE_FUNC => [ |
| 1433 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_OP, |
| 1434 | ], |
| 1435 | self::TYPE_CLASS => [ |
| 1436 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_OP, |
| 1437 | ], |
| 1438 | // We don't expect real unary/binary operators here, but some keywords |
| 1439 | // (new, delete, void, typeof, instanceof, in) are classified as such, and they can be |
| 1440 | // used as property names |
| 1441 | self::TYPE_UN_OP => [ |
| 1442 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_OP, |
| 1443 | ], |
| 1444 | self::TYPE_BIN_OP => [ |
| 1445 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_OP, |
| 1446 | ], |
| 1447 | ], |
| 1448 | // Like EXPRESSION_ARROWFUNC, but in a property expression, see PROPERTY_EXPRESSION |
| 1449 | self::PROPERTY_EXPRESSION_ARROWFUNC => [ |
| 1450 | self::TYPE_UN_OP => [ |
| 1451 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION, |
| 1452 | ], |
| 1453 | self::TYPE_INCR_OP => [ |
| 1454 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION, |
| 1455 | ], |
| 1456 | self::TYPE_ADD_OP => [ |
| 1457 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION, |
| 1458 | ], |
| 1459 | self::TYPE_BRACE_OPEN => [ |
| 1460 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1461 | self::ACTION_GOTO => self::STATEMENT, |
| 1462 | ], |
| 1463 | self::TYPE_PAREN_OPEN => [ |
| 1464 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1465 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1466 | ], |
| 1467 | self::TYPE_FUNC => [ |
| 1468 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1469 | self::ACTION_GOTO => self::FUNC, |
| 1470 | ], |
| 1471 | self::TYPE_CLASS => [ |
| 1472 | self::ACTION_PUSH => self::PROPERTY_EXPRESSION_OP, |
| 1473 | self::ACTION_GOTO => self::CLASS_DEF, |
| 1474 | ], |
| 1475 | self::TYPE_LITERAL => [ |
| 1476 | self::ACTION_GOTO => self::PROPERTY_EXPRESSION_OP, |
| 1477 | ], |
| 1478 | ], |
| 1479 | // Class definition (after the class keyword). Expects an identifier, or the extends |
| 1480 | // keyword followed by an expression (or both), followed by {, which starts an object |
| 1481 | // literal. The object literal's closing } will pop the stack, so the state to return |
| 1482 | // to after the class definition should be pushed to the stack first. |
| 1483 | self::CLASS_DEF => [ |
| 1484 | self::TYPE_BRACE_OPEN => [ |
| 1485 | self::ACTION_GOTO => self::PROPERTY_ASSIGNMENT, |
| 1486 | ], |
| 1487 | self::TYPE_PAREN_OPEN => [ |
| 1488 | self::ACTION_PUSH => self::CLASS_DEF, |
| 1489 | self::ACTION_GOTO => self::PAREN_EXPRESSION, |
| 1490 | ], |
| 1491 | self::TYPE_FUNC => [ |
| 1492 | self::ACTION_PUSH => self::CLASS_DEF, |
| 1493 | self::ACTION_GOTO => self::FUNC, |
| 1494 | ], |
| 1495 | ], |
| 1496 | // Import or export declaration |
| 1497 | self::IMPORT_EXPORT => [ |
| 1498 | self::TYPE_SEMICOLON => [ |
| 1499 | self::ACTION_GOTO => self::STATEMENT, |
| 1500 | ], |
| 1501 | self::TYPE_VAR => [ |
| 1502 | self::ACTION_GOTO => self::EXPRESSION, |
| 1503 | ], |
| 1504 | self::TYPE_FUNC => [ |
| 1505 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 1506 | self::ACTION_GOTO => self::FUNC, |
| 1507 | ], |
| 1508 | self::TYPE_CLASS => [ |
| 1509 | self::ACTION_PUSH => self::EXPRESSION_OP, |
| 1510 | self::ACTION_GOTO => self::CLASS_DEF, |
| 1511 | ], |
| 1512 | self::TYPE_SPECIAL => [ |
| 1513 | 'default' => [ |
| 1514 | self::ACTION_GOTO => self::EXPRESSION, |
| 1515 | ], |
| 1516 | // Stay in this state for *, as, from |
| 1517 | '*' => [], |
| 1518 | 'as' => [], |
| 1519 | 'from' => [], |
| 1520 | ], |
| 1521 | ], |
| 1522 | // Used in template string-specific code below |
| 1523 | self::TEMPLATE_STRING_HEAD => [ |
| 1524 | self::TYPE_LITERAL => [ |
| 1525 | self::ACTION_PUSH => self::TEMPLATE_STRING_TAIL, |
| 1526 | self::ACTION_GOTO => self::EXPRESSION, |
| 1527 | ], |
| 1528 | ], |
| 1529 | ]; |
| 1530 | |
| 1531 | /** |
| 1532 | * @var array $semicolon |
| 1533 | * |
| 1534 | * Rules for when semicolon insertion is appropriate. Semicolon insertion happens if we are |
| 1535 | * in one of these states, and encounter one of these tokens preceded by a newline. |
| 1536 | * |
| 1537 | * This array is augmented by ensureExpandedStates(). |
| 1538 | */ |
| 1539 | private static $semicolon = [ |
| 1540 | self::EXPRESSION_NO_NL => [ |
| 1541 | self::TYPE_UN_OP => true, |
| 1542 | // BIN_OP seems impossible at the start of an expression, but it can happen in |
| 1543 | // yield *foo |
| 1544 | self::TYPE_BIN_OP => true, |
| 1545 | self::TYPE_INCR_OP => true, |
| 1546 | self::TYPE_ADD_OP => true, |
| 1547 | self::TYPE_BRACE_OPEN => true, |
| 1548 | self::TYPE_PAREN_OPEN => true, |
| 1549 | self::TYPE_RETURN => true, |
| 1550 | self::TYPE_IF => true, |
| 1551 | self::TYPE_DO => true, |
| 1552 | self::TYPE_VAR => true, |
| 1553 | self::TYPE_FUNC => true, |
| 1554 | self::TYPE_CLASS => true, |
| 1555 | self::TYPE_LITERAL => true, |
| 1556 | self::TYPE_ASYNC => true, |
| 1557 | ], |
| 1558 | self::EXPRESSION_TERNARY_NO_NL => [ |
| 1559 | self::TYPE_UN_OP => true, |
| 1560 | // BIN_OP seems impossible at the start of an expression, but it can happen in |
| 1561 | // yield *foo |
| 1562 | self::TYPE_BIN_OP => true, |
| 1563 | self::TYPE_INCR_OP => true, |
| 1564 | self::TYPE_ADD_OP => true, |
| 1565 | self::TYPE_BRACE_OPEN => true, |
| 1566 | self::TYPE_PAREN_OPEN => true, |
| 1567 | self::TYPE_RETURN => true, |
| 1568 | self::TYPE_IF => true, |
| 1569 | self::TYPE_DO => true, |
| 1570 | self::TYPE_VAR => true, |
| 1571 | self::TYPE_FUNC => true, |
| 1572 | self::TYPE_CLASS => true, |
| 1573 | self::TYPE_LITERAL => true, |
| 1574 | self::TYPE_ASYNC => true, |
| 1575 | ], |
| 1576 | self::PAREN_EXPRESSION_NO_NL => [ |
| 1577 | self::TYPE_UN_OP => true, |
| 1578 | // BIN_OP seems impossible at the start of an expression, but it can happen in |
| 1579 | // yield *foo |
| 1580 | self::TYPE_BIN_OP => true, |
| 1581 | self::TYPE_INCR_OP => true, |
| 1582 | self::TYPE_ADD_OP => true, |
| 1583 | self::TYPE_BRACE_OPEN => true, |
| 1584 | self::TYPE_PAREN_OPEN => true, |
| 1585 | self::TYPE_RETURN => true, |
| 1586 | self::TYPE_IF => true, |
| 1587 | self::TYPE_DO => true, |
| 1588 | self::TYPE_VAR => true, |
| 1589 | self::TYPE_FUNC => true, |
| 1590 | self::TYPE_CLASS => true, |
| 1591 | self::TYPE_LITERAL => true, |
| 1592 | self::TYPE_ASYNC => true, |
| 1593 | ], |
| 1594 | self::PROPERTY_EXPRESSION_NO_NL => [ |
| 1595 | self::TYPE_UN_OP => true, |
| 1596 | // BIN_OP seems impossible at the start of an expression, but it can happen in |
| 1597 | // yield *foo |
| 1598 | self::TYPE_BIN_OP => true, |
| 1599 | self::TYPE_INCR_OP => true, |
| 1600 | self::TYPE_ADD_OP => true, |
| 1601 | self::TYPE_BRACE_OPEN => true, |
| 1602 | self::TYPE_PAREN_OPEN => true, |
| 1603 | self::TYPE_RETURN => true, |
| 1604 | self::TYPE_IF => true, |
| 1605 | self::TYPE_DO => true, |
| 1606 | self::TYPE_VAR => true, |
| 1607 | self::TYPE_FUNC => true, |
| 1608 | self::TYPE_CLASS => true, |
| 1609 | self::TYPE_LITERAL => true, |
| 1610 | self::TYPE_ASYNC => true, |
| 1611 | ], |
| 1612 | self::EXPRESSION_OP => [ |
| 1613 | self::TYPE_UN_OP => true, |
| 1614 | self::TYPE_INCR_OP => true, |
| 1615 | self::TYPE_BRACE_OPEN => true, |
| 1616 | self::TYPE_RETURN => true, |
| 1617 | self::TYPE_IF => true, |
| 1618 | self::TYPE_DO => true, |
| 1619 | self::TYPE_VAR => true, |
| 1620 | self::TYPE_FUNC => true, |
| 1621 | self::TYPE_CLASS => true, |
| 1622 | self::TYPE_LITERAL => true, |
| 1623 | self::TYPE_ASYNC => true, |
| 1624 | ], |
| 1625 | self::EXPRESSION_END => [ |
| 1626 | self::TYPE_UN_OP => true, |
| 1627 | self::TYPE_INCR_OP => true, |
| 1628 | self::TYPE_ADD_OP => true, |
| 1629 | self::TYPE_BRACE_OPEN => true, |
| 1630 | self::TYPE_PAREN_OPEN => true, |
| 1631 | self::TYPE_RETURN => true, |
| 1632 | self::TYPE_IF => true, |
| 1633 | self::TYPE_DO => true, |
| 1634 | self::TYPE_VAR => true, |
| 1635 | self::TYPE_FUNC => true, |
| 1636 | self::TYPE_CLASS => true, |
| 1637 | self::TYPE_LITERAL => true, |
| 1638 | self::TYPE_ASYNC => true, |
| 1639 | ], |
| 1640 | self::PAREN_EXPRESSION_OP_NO_NL => [ |
| 1641 | self::TYPE_FUNC => true, |
| 1642 | ] |
| 1643 | ]; |
| 1644 | |
| 1645 | /** |
| 1646 | * @var array $divStates |
| 1647 | * |
| 1648 | * States in which a / is a division operator. In all other states, it's the start of a regex. |
| 1649 | * |
| 1650 | * This array is augmented by self::ensureExpandedStates(). |
| 1651 | */ |
| 1652 | private static $divStates = [ |
| 1653 | self::EXPRESSION_OP => true, |
| 1654 | self::EXPRESSION_TERNARY_OP => true, |
| 1655 | self::PAREN_EXPRESSION_OP => true, |
| 1656 | self::PROPERTY_EXPRESSION_OP => true |
| 1657 | ]; |
| 1658 | |
| 1659 | /** |
| 1660 | * Add copies of all states but with negative numbers to self::$model (if not already present), |
| 1661 | * to represent generator function states. |
| 1662 | */ |
| 1663 | private static function ensureExpandedStates() { |
| 1664 | // Already done? |
| 1665 | if ( self::$expandedStates ) { |
| 1666 | return; |
| 1667 | } |
| 1668 | self::$expandedStates = true; |
| 1669 | |
| 1670 | // Add copies of all states (except FUNC and GENFUNC) with negative numbers. |
| 1671 | // These negative states represent states inside generator functions. When in these states, |
| 1672 | // TYPE_YIELD is treated as TYPE_RETURN, otherwise as TYPE_LITERAL |
| 1673 | foreach ( self::$model as $state => $transitions ) { |
| 1674 | if ( $state === self::FUNC || $state === self::GENFUNC ) { |
| 1675 | continue; |
| 1676 | } |
| 1677 | foreach ( $transitions as $tokenType => $actions ) { |
| 1678 | foreach ( $actions as $action => $target ) { |
| 1679 | if ( !is_array( $target ) ) { |
| 1680 | self::$model[-$state][$tokenType][$action] = ( |
| 1681 | $target === self::FUNC || |
| 1682 | $target === true || |
| 1683 | $target === self::GENFUNC |
| 1684 | ) ? $target : -$target; |
| 1685 | continue; |
| 1686 | } |
| 1687 | |
| 1688 | foreach ( $target as $subaction => $subtarget ) { |
| 1689 | self::$model[-$state][$tokenType][$action][$subaction] = ( |
| 1690 | $subtarget === self::FUNC || |
| 1691 | $subtarget === true || |
| 1692 | $subtarget === self::GENFUNC |
| 1693 | ) ? $subtarget : -$subtarget; |
| 1694 | } |
| 1695 | } |
| 1696 | } |
| 1697 | } |
| 1698 | // Special cases: |
| 1699 | // '{' in a property assignment starts a method, so it shouldn't be flipped |
| 1700 | self::$model[-self::PROPERTY_ASSIGNMENT][self::TYPE_BRACE_OPEN][self::ACTION_GOTO] = self::STATEMENT; |
| 1701 | |
| 1702 | // Also add negative versions of states to the other arrays |
| 1703 | foreach ( self::$semicolon as $state => $value ) { |
| 1704 | self::$semicolon[-$state] = $value; |
| 1705 | } |
| 1706 | foreach ( self::$divStates as $state => $value ) { |
| 1707 | self::$divStates[-$state] = $value; |
| 1708 | } |
| 1709 | } |
| 1710 | |
| 1711 | /** |
| 1712 | * Returns minified JavaScript code. |
| 1713 | * |
| 1714 | * @see MinifierState::setErrorHandler |
| 1715 | * @param string $s JavaScript code to minify |
| 1716 | * @param callable|null $onError Called with a ParseError object |
| 1717 | * @return string Minified code |
| 1718 | */ |
| 1719 | public static function minify( $s, $onError = null ) { |
| 1720 | return self::minifyInternal( $s, null, $onError ); |
| 1721 | } |
| 1722 | |
| 1723 | /** |
| 1724 | * Create a minifier state object without source map capabilities |
| 1725 | * |
| 1726 | * Example: |
| 1727 | * |
| 1728 | * JavaScriptMinifier::createMinifier() |
| 1729 | * ->addSourceFile( 'file.js', $source ) |
| 1730 | * ->getMinifiedOutput(); |
| 1731 | * |
| 1732 | * @return JavaScriptMinifierState |
| 1733 | */ |
| 1734 | public static function createMinifier() { |
| 1735 | return new JavaScriptMinifierState; |
| 1736 | } |
| 1737 | |
| 1738 | /** |
| 1739 | * Create a minifier state object with source map capabilities |
| 1740 | * |
| 1741 | * Example: |
| 1742 | * |
| 1743 | * $mapper = JavaScriptMinifier::createSourceMapState() |
| 1744 | * ->addSourceFile( 'file1.js', $source1 ) |
| 1745 | * ->addOutput( "\n\n" ) |
| 1746 | * ->addSourceFile( 'file2.js', $source2 ); |
| 1747 | * $out = $mapper->getMinifiedOutput(); |
| 1748 | * $map = $mapper->getSourceMap() |
| 1749 | * |
| 1750 | * @return JavaScriptMapperState |
| 1751 | */ |
| 1752 | public static function createSourceMapState() { |
| 1753 | return new JavaScriptMapperState; |
| 1754 | } |
| 1755 | |
| 1756 | /** |
| 1757 | * Create a MinifierState that doesn't actually minify |
| 1758 | * |
| 1759 | * @return IdentityMinifierState |
| 1760 | */ |
| 1761 | public static function createIdentityMinifier() { |
| 1762 | return new IdentityMinifierState; |
| 1763 | } |
| 1764 | |
| 1765 | /** |
| 1766 | * Minify with optional source map. |
| 1767 | * |
| 1768 | * @internal |
| 1769 | * |
| 1770 | * @param string $s |
| 1771 | * @param MappingsGenerator|null $mapGenerator |
| 1772 | * @param callable|null $onError |
| 1773 | * @return string |
| 1774 | */ |
| 1775 | public static function minifyInternal( $s, $mapGenerator = null, $onError = null ) { |
| 1776 | self::ensureExpandedStates(); |
| 1777 | |
| 1778 | // Here's where the minifying takes place: Loop through the input, looking for tokens |
| 1779 | // and output them to $out, taking actions to the above defined rules when appropriate. |
| 1780 | $error = null; |
| 1781 | $out = ''; |
| 1782 | $pos = 0; |
| 1783 | $length = strlen( $s ); |
| 1784 | $lineLength = 0; |
| 1785 | $dotlessNum = false; |
| 1786 | $lastDotlessNum = false; |
| 1787 | $newlineFound = true; |
| 1788 | $state = self::STATEMENT; |
| 1789 | $stack = []; |
| 1790 | // Optimization: calling end( $stack ) repeatedly is expensive |
| 1791 | $topOfStack = null; |
| 1792 | // Pretend that we have seen a semicolon yet |
| 1793 | $last = ';'; |
| 1794 | while ( $pos < $length ) { |
| 1795 | // First, skip over any whitespace and multiline comments, recording whether we |
| 1796 | // found any newline character |
| 1797 | $skip = strspn( $s, " \t\n\r\xb\xc", $pos ); |
| 1798 | if ( !$skip ) { |
| 1799 | $ch = $s[$pos]; |
| 1800 | if ( $ch === '/' && substr( $s, $pos, 2 ) === '/*' ) { |
| 1801 | // Multiline comment. Search for the end token or EOT. |
| 1802 | $end = strpos( $s, '*/', $pos + 2 ); |
| 1803 | $skip = $end === false ? $length - $pos : $end - $pos + 2; |
| 1804 | } |
| 1805 | } |
| 1806 | if ( $skip ) { |
| 1807 | // The semicolon insertion mechanism needs to know whether there was a newline |
| 1808 | // between two tokens, so record it now. |
| 1809 | if ( !$newlineFound && strcspn( $s, "\r\n", $pos, $skip ) !== $skip ) { |
| 1810 | $newlineFound = true; |
| 1811 | } |
| 1812 | if ( $mapGenerator ) { |
| 1813 | $mapGenerator->consumeSource( $skip ); |
| 1814 | } |
| 1815 | $pos += $skip; |
| 1816 | continue; |
| 1817 | } |
| 1818 | // Handle C++-style comments and html comments, which are treated as single line |
| 1819 | // comments by the browser, regardless of whether the end tag is on the same line. |
| 1820 | // Handle --> the same way, but only if it's at the beginning of the line |
| 1821 | // @phan-suppress-next-line PhanPossiblyUndeclaredVariable |
| 1822 | if ( ( $ch === '/' && substr( $s, $pos, 2 ) === '//' ) |
| 1823 | || ( $ch === '<' && substr( $s, $pos, 4 ) === '<!--' ) |
| 1824 | || ( $ch === '-' && $newlineFound && substr( $s, $pos, 3 ) === '-->' ) |
| 1825 | ) { |
| 1826 | $skip = strcspn( $s, "\r\n", $pos ); |
| 1827 | if ( $mapGenerator ) { |
| 1828 | $mapGenerator->consumeSource( $skip ); |
| 1829 | } |
| 1830 | $pos += $skip; |
| 1831 | continue; |
| 1832 | } |
| 1833 | |
| 1834 | // Find out which kind of token we're handling. |
| 1835 | // Note: $end must point past the end of the current token |
| 1836 | // so that `substr($s, $pos, $end - $pos)` would be the entire token. |
| 1837 | // In order words, $end will be the offset of the last relevant character |
| 1838 | // in the stream + 1, or simply put: The offset of the first character |
| 1839 | // of any next token in the stream. |
| 1840 | $end = $pos + 1; |
| 1841 | // Handle string literals |
| 1842 | if ( $ch === "'" || $ch === '"' ) { |
| 1843 | // Search to the end of the string literal, skipping over backslash escapes |
| 1844 | $search = $ch . '\\'; |
| 1845 | do { |
| 1846 | // Speculatively add 2 to the end so that if we see a backslash, |
| 1847 | // the next iteration will start 2 characters further (one for the |
| 1848 | // backslash, one for the escaped character). |
| 1849 | // We'll correct this outside the loop. |
| 1850 | $end += strcspn( $s, $search, $end ) + 2; |
| 1851 | // If the last character in our search for a quote or a backlash |
| 1852 | // matched a backslash and we haven't reached the end, keep searching.. |
| 1853 | } while ( $end - 2 < $length && $s[$end - 2] === '\\' ); |
| 1854 | // Correction (1): Undo speculative add, keep only one (end of string literal) |
| 1855 | $end--; |
| 1856 | if ( $end > $length ) { |
| 1857 | // Correction (2): Loop wrongly assumed an end quote ended the search, |
| 1858 | // but search ended because we've reached the end. Correct $end. |
| 1859 | // TODO: This is invalid and should throw. |
| 1860 | $end--; |
| 1861 | } |
| 1862 | |
| 1863 | // Handle template strings, either from "`" to begin a new string, |
| 1864 | // or continuation after the "}" that ends a "${"-expression. |
| 1865 | } elseif ( $ch === '`' || ( $ch === '}' && $topOfStack === self::TEMPLATE_STRING_TAIL ) ) { |
| 1866 | if ( $ch === '}' ) { |
| 1867 | // Pop the TEMPLATE_STRING_TAIL state off the stack |
| 1868 | // We don't let it get popped off the stack the normal way, to avoid the newline |
| 1869 | // and comment stripping code above running on the continuation of the literal |
| 1870 | array_pop( $stack ); |
| 1871 | // Also pop the previous state off the stack |
| 1872 | $state = array_pop( $stack ); |
| 1873 | $topOfStack = end( $stack ); |
| 1874 | } |
| 1875 | // Search until we reach either a closing ` or a ${, skipping over backslash escapes |
| 1876 | // and $ characters followed by something other than { or ` |
| 1877 | do { |
| 1878 | $end += strcspn( $s, '`$\\', $end ) + 1; |
| 1879 | if ( $end - 1 < $length && $s[$end - 1] === '`' ) { |
| 1880 | // End of the string, stop |
| 1881 | // We don't do this in the while() condition because the $end++ in the |
| 1882 | // backslash escape branch makes it difficult to do so without incorrectly |
| 1883 | // considering an escaped backtick (\`) the end of the string |
| 1884 | break; |
| 1885 | } |
| 1886 | if ( $end - 1 < $length && $s[$end - 1] === '\\' ) { |
| 1887 | // Backslash escape. Skip the next character, and keep going |
| 1888 | $end++; |
| 1889 | continue; |
| 1890 | } |
| 1891 | if ( $end < $length && $s[$end - 1] === '$' && $s[$end] === '{' ) { |
| 1892 | // Beginning of an expression in ${ ... }. Skip the {, and stop |
| 1893 | $end++; |
| 1894 | // Push the current state to the stack. We'll pop this off later when hitting |
| 1895 | // the end of this template string |
| 1896 | $stack[] = $state; |
| 1897 | $topOfStack = $state; |
| 1898 | // Change the state to TEMPLATE_STRING_HEAD. The token type will be detected |
| 1899 | // as TYPE_LITERAL, and this will cause the state machine to expect an |
| 1900 | // expression, then go to the TEMPLATE_STRING_TAIL state when it hits the } |
| 1901 | $state = self::TEMPLATE_STRING_HEAD; |
| 1902 | break; |
| 1903 | } |
| 1904 | } while ( $end - 1 < $length ); |
| 1905 | if ( $end > $length ) { |
| 1906 | // Loop wrongly assumed an end quote or ${ ended the search, |
| 1907 | // but search ended because we've reached the end. Correct $end. |
| 1908 | // TODO: This is invalid and should throw. |
| 1909 | $end--; |
| 1910 | } |
| 1911 | |
| 1912 | // We have to distinguish between regexp literals and division operators |
| 1913 | // A division operator is only possible in certain states |
| 1914 | } elseif ( $ch === '/' && !isset( self::$divStates[$state] ) ) { |
| 1915 | // Regexp literal |
| 1916 | for ( ; ; ) { |
| 1917 | // Search until we find "/" (end of regexp), "\" (backslash escapes), |
| 1918 | // or "[" (start of character classes). |
| 1919 | do { |
| 1920 | // Speculatively add 2 to ensure next iteration skips |
| 1921 | // over backslash and escaped character. |
| 1922 | // We'll correct this outside the loop. |
| 1923 | $end += strcspn( $s, '/[\\', $end ) + 2; |
| 1924 | // If backslash escape, keep searching... |
| 1925 | } while ( $end - 2 < $length && $s[$end - 2] === '\\' ); |
| 1926 | // Correction (1): Undo speculative add, keep only one (end of regexp) |
| 1927 | $end--; |
| 1928 | if ( $end > $length ) { |
| 1929 | // Correction (2): Loop wrongly assumed end slash was seen |
| 1930 | // String ended without end of regexp. Correct $end. |
| 1931 | // TODO: This is invalid and should throw. |
| 1932 | $end--; |
| 1933 | break; |
| 1934 | } |
| 1935 | if ( $s[$end - 1] === '/' ) { |
| 1936 | break; |
| 1937 | } |
| 1938 | // (Implicit else), we must've found the start of a char class, |
| 1939 | // skip until we find "]" (end of char class), or "\" (backslash escape) |
| 1940 | do { |
| 1941 | // Speculatively add 2 for backslash escape. |
| 1942 | // We'll substract one outside the loop. |
| 1943 | $end += strcspn( $s, ']\\', $end ) + 2; |
| 1944 | // If backslash escape, keep searching... |
| 1945 | } while ( $end - 2 < $length && $s[$end - 2] === '\\' ); |
| 1946 | // Correction (1): Undo speculative add, keep only one (end of regexp) |
| 1947 | $end--; |
| 1948 | if ( $end > $length ) { |
| 1949 | // Correction (2): Loop wrongly assumed "]" was seen |
| 1950 | // String ended without ending char class or regexp. Correct $end. |
| 1951 | // TODO: This is invalid and should throw. |
| 1952 | $end--; |
| 1953 | break; |
| 1954 | } |
| 1955 | } |
| 1956 | // Search past the regexp modifiers (gi) |
| 1957 | while ( $end < $length && ctype_alpha( $s[$end] ) ) { |
| 1958 | $end++; |
| 1959 | } |
| 1960 | } elseif ( |
| 1961 | $ch === '0' |
| 1962 | && ( $pos + 1 < $length ) && ( $s[$pos + 1] === 'x' || $s[$pos + 1] === 'X' ) |
| 1963 | ) { |
| 1964 | // Hex numeric literal |
| 1965 | // x or X |
| 1966 | $end++; |
| 1967 | $len = strspn( $s, '0123456789ABCDEFabcdef', $end ); |
| 1968 | if ( !$len && !$error ) { |
| 1969 | $error = new ParseError( |
| 1970 | 'Expected a hexadecimal number but found ' . substr( $s, $pos, 5 ), |
| 1971 | $pos, |
| 1972 | ); |
| 1973 | } |
| 1974 | $end += $len; |
| 1975 | } elseif ( |
| 1976 | // Optimisation: This check must accept only ASCII digits 0-9. |
| 1977 | // Avoid ctype_digit() because it is slower and also accepts locale-specific digits. |
| 1978 | // Using is_numeric() might seem wrong also as it accepts negative numbers, decimal |
| 1979 | // numbers, and exponents (e.g. strings like "+012.34e6"). But, it is fine here |
| 1980 | // because we know $ch is a single character, and we believe the only single |
| 1981 | // characters that is_numeric() accepts are ASCII digits 0-9. |
| 1982 | is_numeric( $ch ) |
| 1983 | || ( $ch === '.' && $pos + 1 < $length && is_numeric( $s[$pos + 1] ) ) |
| 1984 | ) { |
| 1985 | $end += strspn( $s, '0123456789', $end ); |
| 1986 | $decimal = strspn( $s, '.', $end ); |
| 1987 | if ( $decimal ) { |
| 1988 | if ( $decimal > 2 && !$error ) { |
| 1989 | $error = new ParseError( 'Too many decimal points', $end ); |
| 1990 | } |
| 1991 | $end += strspn( $s, '0123456789', $end + 1 ) + $decimal; |
| 1992 | } else { |
| 1993 | $dotlessNum = true; |
| 1994 | } |
| 1995 | $exponent = strspn( $s, 'eE', $end ); |
| 1996 | if ( $exponent ) { |
| 1997 | if ( $exponent > 1 && !$error ) { |
| 1998 | $error = new ParseError( 'Number with several E', $end ); |
| 1999 | } |
| 2000 | $end += $exponent; |
| 2001 | |
| 2002 | // + sign is optional; - sign is required. |
| 2003 | $end += strspn( $s, '-+', $end ); |
| 2004 | $len = strspn( $s, '0123456789', $end ); |
| 2005 | if ( !$len && !$error ) { |
| 2006 | $error = new ParseError( |
| 2007 | 'Missing decimal digits after exponent', |
| 2008 | $pos |
| 2009 | ); |
| 2010 | } |
| 2011 | $end += $len; |
| 2012 | } |
| 2013 | } elseif ( isset( self::$opChars[$ch] ) ) { |
| 2014 | // Punctuation character. Search for the longest matching operator. |
| 2015 | for ( $tokenLength = self::LONGEST_PUNCTUATION_TOKEN; $tokenLength > 1; $tokenLength-- ) { |
| 2016 | if ( |
| 2017 | $pos + $tokenLength <= $length && |
| 2018 | isset( self::$tokenTypes[ substr( $s, $pos, $tokenLength ) ] ) |
| 2019 | ) { |
| 2020 | $end = $pos + $tokenLength; |
| 2021 | break; |
| 2022 | } |
| 2023 | } |
| 2024 | } else { |
| 2025 | // Identifier or reserved word. Search for the end by excluding whitespace and |
| 2026 | // punctuation. |
| 2027 | $end += strcspn( $s, " \t\n.;,=<>+-{}()[]?:*/%'\"`!&|^~\xb\xc\r", $end ); |
| 2028 | } |
| 2029 | |
| 2030 | // Now get the token type from our type array |
| 2031 | // so $end - $pos == strlen( $token ) |
| 2032 | $token = substr( $s, $pos, $end - $pos ); |
| 2033 | $type = isset( self::$model[$state][self::TYPE_SPECIAL][$token] ) |
| 2034 | ? self::TYPE_SPECIAL |
| 2035 | : self::$tokenTypes[$token] ?? self::TYPE_LITERAL; |
| 2036 | if ( $type === self::TYPE_YIELD ) { |
| 2037 | // yield is treated as TYPE_RETURN inside a generator function (negative state) |
| 2038 | // but as TYPE_LITERAL when not in a generator function (positive state) |
| 2039 | $type = $state < 0 ? self::TYPE_RETURN : self::TYPE_LITERAL; |
| 2040 | } |
| 2041 | |
| 2042 | $pad = ''; |
| 2043 | |
| 2044 | if ( $newlineFound && isset( self::$semicolon[$state][$type] ) ) { |
| 2045 | // This token triggers the semicolon insertion mechanism of javascript. While we |
| 2046 | // could add the ; token here ourselves, keeping the newline has a few advantages. |
| 2047 | $pad = "\n"; |
| 2048 | $state = $state < 0 ? -self::STATEMENT : self::STATEMENT; |
| 2049 | $lineLength = 0; |
| 2050 | // This check adds a new line if we have exceeded the max length and only does this if |
| 2051 | // a newline was found in this this position, if it wasn't, it uses the next available |
| 2052 | // line break |
| 2053 | } elseif ( $newlineFound && |
| 2054 | $lineLength + $end - $pos > self::$maxLineLength && |
| 2055 | !isset( self::$semicolon[$state][$type] ) && |
| 2056 | $type !== self::TYPE_INCR_OP && |
| 2057 | $type !== self::TYPE_ARROW |
| 2058 | ) { |
| 2059 | $pad = "\n"; |
| 2060 | $lineLength = 0; |
| 2061 | // Check, whether we have to separate the token from the last one with whitespace |
| 2062 | } elseif ( !isset( self::$opChars[$last] ) && !isset( self::$opChars[$ch] ) ) { |
| 2063 | $pad = ' '; |
| 2064 | $lineLength++; |
| 2065 | // Don't accidentally create ++, -- or // tokens |
| 2066 | } elseif ( $last === $ch && ( $ch === '+' || $ch === '-' || $ch === '/' ) ) { |
| 2067 | $pad = ' '; |
| 2068 | $lineLength++; |
| 2069 | // Don't create invalid dot notation after number literal (T303827). |
| 2070 | // Keep whitespace in "42. foo". |
| 2071 | // But keep minifying "foo.bar", "42..foo", and "42.0.foo" per $opChars. |
| 2072 | } elseif ( $lastDotlessNum && $type === self::TYPE_DOT ) { |
| 2073 | $pad = ' '; |
| 2074 | $lineLength++; |
| 2075 | } |
| 2076 | |
| 2077 | // self::debug( $topOfStack, $last, $state, $ch, $token, $type ); |
| 2078 | |
| 2079 | if ( $mapGenerator ) { |
| 2080 | $mapGenerator->outputSpace( $pad ); |
| 2081 | $mapGenerator->outputToken( $token ); |
| 2082 | $mapGenerator->consumeSource( $end - $pos ); |
| 2083 | } |
| 2084 | $out .= $pad; |
| 2085 | $out .= $token; |
| 2086 | $lineLength += $end - $pos; |
| 2087 | $last = $s[$end - 1]; |
| 2088 | $pos = $end; |
| 2089 | $newlineFound = false; |
| 2090 | $lastDotlessNum = $dotlessNum; |
| 2091 | $dotlessNum = false; |
| 2092 | |
| 2093 | // Now that we have output our token, transition into the new state. |
| 2094 | $actions = $type === self::TYPE_SPECIAL ? |
| 2095 | self::$model[$state][$type][$token] : |
| 2096 | self::$model[$state][$type] ?? []; |
| 2097 | if ( isset( $actions[self::ACTION_PUSH] ) && |
| 2098 | count( $stack ) < self::STACK_LIMIT |
| 2099 | ) { |
| 2100 | $topOfStack = $actions[self::ACTION_PUSH]; |
| 2101 | $stack[] = $topOfStack; |
| 2102 | } |
| 2103 | if ( $stack && isset( $actions[self::ACTION_POP] ) ) { |
| 2104 | $state = array_pop( $stack ); |
| 2105 | $topOfStack = end( $stack ); |
| 2106 | } elseif ( isset( $actions[self::ACTION_GOTO] ) ) { |
| 2107 | $state = $actions[self::ACTION_GOTO]; |
| 2108 | } |
| 2109 | } |
| 2110 | if ( $onError && $error ) { |
| 2111 | $onError( $error ); |
| 2112 | } |
| 2113 | return $out; |
| 2114 | } |
| 2115 | |
| 2116 | /** |
| 2117 | * @param null|false|int $top |
| 2118 | * @param string $last |
| 2119 | * @param int $state |
| 2120 | * @param string $ch |
| 2121 | * @param string $token |
| 2122 | * @param int $type |
| 2123 | */ |
| 2124 | private static function debug( |
| 2125 | $top, string $last, |
| 2126 | int $state, string $ch, string $token, int $type |
| 2127 | ) { |
| 2128 | static $first = true; |
| 2129 | $self = new ReflectionClass( self::class ); |
| 2130 | |
| 2131 | foreach ( $self->getConstants() as $name => $value ) { |
| 2132 | if ( $value === $top ) { |
| 2133 | $top = $name; |
| 2134 | } |
| 2135 | if ( $value === $state ) { |
| 2136 | $state = $name; |
| 2137 | } |
| 2138 | if ( $value === $type ) { |
| 2139 | $type = $name; |
| 2140 | } |
| 2141 | } |
| 2142 | |
| 2143 | if ( $first ) { |
| 2144 | print sprintf( "| %-29s | %-4s | %-29s | %-2s | %-10s | %-29s\n", |
| 2145 | 'topOfStack', 'last', 'state', 'ch', 'token', 'type' ); |
| 2146 | print sprintf( "| %'-29s | %'-4s | %'-29s | %'-2s | %'-10s | %'-29s\n", |
| 2147 | '', '', '', '', '', '' ); |
| 2148 | $first = false; |
| 2149 | } |
| 2150 | print sprintf( "| %-29s | %-4s | %-29s | %-2s | %-10s | %-29s\n", |
| 2151 | (string)$top, $last, $state, $ch, $token, $type ); |
| 2152 | } |
| 2153 | } |