Home | History | Annotate | Download | only in parsing
      1 // Copyright 2012 the V8 project authors. All rights reserved.
      2 // Use of this source code is governed by a BSD-style license that can be
      3 // found in the LICENSE file.
      4 
      5 #ifndef V8_PARSING_TOKEN_H_
      6 #define V8_PARSING_TOKEN_H_
      7 
      8 #include "src/base/logging.h"
      9 #include "src/globals.h"
     10 
     11 namespace v8 {
     12 namespace internal {
     13 
     14 // TOKEN_LIST takes a list of 3 macros M, all of which satisfy the
     15 // same signature M(name, string, precedence), where name is the
     16 // symbolic token name, string is the corresponding syntactic symbol
     17 // (or NULL, for literals), and precedence is the precedence (or 0).
     18 // The parameters are invoked for token categories as follows:
     19 //
     20 //   T: Non-keyword tokens
     21 //   K: Keyword tokens
     22 
     23 // IGNORE_TOKEN is a convenience macro that can be supplied as
     24 // an argument (at any position) for a TOKEN_LIST call. It does
     25 // nothing with tokens belonging to the respective category.
     26 
     27 #define IGNORE_TOKEN(name, string, precedence)
     28 
     29 #define TOKEN_LIST(T, K)                                             \
     30   /* End of source indicator. */                                     \
     31   T(EOS, "EOS", 0)                                                   \
     32                                                                      \
     33   /* Punctuators (ECMA-262, section 7.7, page 15). */                \
     34   T(LPAREN, "(", 0)                                                  \
     35   T(RPAREN, ")", 0)                                                  \
     36   T(LBRACK, "[", 0)                                                  \
     37   T(RBRACK, "]", 0)                                                  \
     38   T(LBRACE, "{", 0)                                                  \
     39   T(RBRACE, "}", 0)                                                  \
     40   T(COLON, ":", 0)                                                   \
     41   T(SEMICOLON, ";", 0)                                               \
     42   T(PERIOD, ".", 0)                                                  \
     43   T(ELLIPSIS, "...", 0)                                              \
     44   T(CONDITIONAL, "?", 3)                                             \
     45   T(INC, "++", 0)                                                    \
     46   T(DEC, "--", 0)                                                    \
     47   T(ARROW, "=>", 0)                                                  \
     48                                                                      \
     49   /* Assignment operators. */                                        \
     50   /* IsAssignmentOp() and Assignment::is_compound() relies on */     \
     51   /* this block of enum values being contiguous and sorted in the */ \
     52   /* same order! */                                                  \
     53   T(INIT, "=init", 2) /* AST-use only. */                            \
     54   T(ASSIGN, "=", 2)                                                  \
     55   T(ASSIGN_BIT_OR, "|=", 2)                                          \
     56   T(ASSIGN_BIT_XOR, "^=", 2)                                         \
     57   T(ASSIGN_BIT_AND, "&=", 2)                                         \
     58   T(ASSIGN_SHL, "<<=", 2)                                            \
     59   T(ASSIGN_SAR, ">>=", 2)                                            \
     60   T(ASSIGN_SHR, ">>>=", 2)                                           \
     61   T(ASSIGN_ADD, "+=", 2)                                             \
     62   T(ASSIGN_SUB, "-=", 2)                                             \
     63   T(ASSIGN_MUL, "*=", 2)                                             \
     64   T(ASSIGN_DIV, "/=", 2)                                             \
     65   T(ASSIGN_MOD, "%=", 2)                                             \
     66   T(ASSIGN_EXP, "**=", 2)                                            \
     67                                                                      \
     68   /* Binary operators sorted by precedence. */                       \
     69   /* IsBinaryOp() relies on this block of enum values */             \
     70   /* being contiguous and sorted in the same order! */               \
     71   T(COMMA, ",", 1)                                                   \
     72   T(OR, "||", 4)                                                     \
     73   T(AND, "&&", 5)                                                    \
     74   T(BIT_OR, "|", 6)                                                  \
     75   T(BIT_XOR, "^", 7)                                                 \
     76   T(BIT_AND, "&", 8)                                                 \
     77   T(SHL, "<<", 11)                                                   \
     78   T(SAR, ">>", 11)                                                   \
     79   T(SHR, ">>>", 11)                                                  \
     80   T(ROR, "rotate right", 11) /* only used by Crankshaft */           \
     81   T(ADD, "+", 12)                                                    \
     82   T(SUB, "-", 12)                                                    \
     83   T(MUL, "*", 13)                                                    \
     84   T(DIV, "/", 13)                                                    \
     85   T(MOD, "%", 13)                                                    \
     86   T(EXP, "**", 14)                                                   \
     87                                                                      \
     88   /* Compare operators sorted by precedence. */                      \
     89   /* IsCompareOp() relies on this block of enum values */            \
     90   /* being contiguous and sorted in the same order! */               \
     91   T(EQ, "==", 9)                                                     \
     92   T(NE, "!=", 9)                                                     \
     93   T(EQ_STRICT, "===", 9)                                             \
     94   T(NE_STRICT, "!==", 9)                                             \
     95   T(LT, "<", 10)                                                     \
     96   T(GT, ">", 10)                                                     \
     97   T(LTE, "<=", 10)                                                   \
     98   T(GTE, ">=", 10)                                                   \
     99   K(INSTANCEOF, "instanceof", 10)                                    \
    100   K(IN, "in", 10)                                                    \
    101                                                                      \
    102   /* Unary operators. */                                             \
    103   /* IsUnaryOp() relies on this block of enum values */              \
    104   /* being contiguous and sorted in the same order! */               \
    105   T(NOT, "!", 0)                                                     \
    106   T(BIT_NOT, "~", 0)                                                 \
    107   K(DELETE, "delete", 0)                                             \
    108   K(TYPEOF, "typeof", 0)                                             \
    109   K(VOID, "void", 0)                                                 \
    110                                                                      \
    111   /* Keywords (ECMA-262, section 7.5.2, page 13). */                 \
    112   K(BREAK, "break", 0)                                               \
    113   K(CASE, "case", 0)                                                 \
    114   K(CATCH, "catch", 0)                                               \
    115   K(CONTINUE, "continue", 0)                                         \
    116   K(DEBUGGER, "debugger", 0)                                         \
    117   K(DEFAULT, "default", 0)                                           \
    118   /* DELETE */                                                       \
    119   K(DO, "do", 0)                                                     \
    120   K(ELSE, "else", 0)                                                 \
    121   K(FINALLY, "finally", 0)                                           \
    122   K(FOR, "for", 0)                                                   \
    123   K(FUNCTION, "function", 0)                                         \
    124   K(IF, "if", 0)                                                     \
    125   /* IN */                                                           \
    126   /* INSTANCEOF */                                                   \
    127   K(NEW, "new", 0)                                                   \
    128   K(RETURN, "return", 0)                                             \
    129   K(SWITCH, "switch", 0)                                             \
    130   K(THIS, "this", 0)                                                 \
    131   K(THROW, "throw", 0)                                               \
    132   K(TRY, "try", 0)                                                   \
    133   /* TYPEOF */                                                       \
    134   K(VAR, "var", 0)                                                   \
    135   /* VOID */                                                         \
    136   K(WHILE, "while", 0)                                               \
    137   K(WITH, "with", 0)                                                 \
    138                                                                      \
    139   /* Literals (ECMA-262, section 7.8, page 16). */                   \
    140   K(NULL_LITERAL, "null", 0)                                         \
    141   K(TRUE_LITERAL, "true", 0)                                         \
    142   K(FALSE_LITERAL, "false", 0)                                       \
    143   T(NUMBER, NULL, 0)                                                 \
    144   T(SMI, NULL, 0)                                                    \
    145   T(STRING, NULL, 0)                                                 \
    146                                                                      \
    147   /* Identifiers (not keywords or future reserved words). */         \
    148   T(IDENTIFIER, NULL, 0)                                             \
    149                                                                      \
    150   /* Future reserved words (ECMA-262, section 7.6.1.2). */           \
    151   T(FUTURE_STRICT_RESERVED_WORD, NULL, 0)                            \
    152   K(ASYNC, "async", 0)                                               \
    153   /* `await` is a reserved word in module code only */               \
    154   K(AWAIT, "await", 0)                                               \
    155   K(CLASS, "class", 0)                                               \
    156   K(CONST, "const", 0)                                               \
    157   K(ENUM, "enum", 0)                                                 \
    158   K(EXPORT, "export", 0)                                             \
    159   K(EXTENDS, "extends", 0)                                           \
    160   K(IMPORT, "import", 0)                                             \
    161   K(LET, "let", 0)                                                   \
    162   K(STATIC, "static", 0)                                             \
    163   K(YIELD, "yield", 0)                                               \
    164   K(SUPER, "super", 0)                                               \
    165                                                                      \
    166   /* Illegal token - not able to scan. */                            \
    167   T(ILLEGAL, "ILLEGAL", 0)                                           \
    168   T(ESCAPED_KEYWORD, NULL, 0)                                        \
    169   T(ESCAPED_STRICT_RESERVED_WORD, NULL, 0)                           \
    170                                                                      \
    171   /* Scanner-internal use only. */                                   \
    172   T(WHITESPACE, NULL, 0)                                             \
    173   T(UNINITIALIZED, NULL, 0)                                          \
    174                                                                      \
    175   /* ES6 Template Literals */                                        \
    176   T(TEMPLATE_SPAN, NULL, 0)                                          \
    177   T(TEMPLATE_TAIL, NULL, 0)
    178 
    179 class Token {
    180  public:
    181   // All token values.
    182 #define T(name, string, precedence) name,
    183   enum Value {
    184     TOKEN_LIST(T, T)
    185     NUM_TOKENS
    186   };
    187 #undef T
    188 
    189   // Returns a string corresponding to the C++ token name
    190   // (e.g. "LT" for the token LT).
    191   static const char* Name(Value tok) {
    192     DCHECK(tok < NUM_TOKENS);  // tok is unsigned
    193     return name_[tok];
    194   }
    195 
    196   // Predicates
    197   static bool IsKeyword(Value tok) {
    198     return token_type[tok] == 'K';
    199   }
    200 
    201   static bool IsIdentifier(Value tok, LanguageMode language_mode,
    202                            bool is_generator, bool is_module) {
    203     switch (tok) {
    204       case IDENTIFIER:
    205       case ASYNC:
    206         return true;
    207       case ESCAPED_STRICT_RESERVED_WORD:
    208       case FUTURE_STRICT_RESERVED_WORD:
    209       case LET:
    210       case STATIC:
    211         return is_sloppy(language_mode);
    212       case YIELD:
    213         return !is_generator && is_sloppy(language_mode);
    214       case AWAIT:
    215         return !is_module;
    216       default:
    217         return false;
    218     }
    219     UNREACHABLE();
    220     return false;
    221   }
    222 
    223   static bool IsAssignmentOp(Value tok) {
    224     return INIT <= tok && tok <= ASSIGN_EXP;
    225   }
    226 
    227   static bool IsBinaryOp(Value op) { return COMMA <= op && op <= EXP; }
    228 
    229   static bool IsTruncatingBinaryOp(Value op) {
    230     return BIT_OR <= op && op <= ROR;
    231   }
    232 
    233   static bool IsCompareOp(Value op) {
    234     return EQ <= op && op <= IN;
    235   }
    236 
    237   static bool IsOrderedRelationalCompareOp(Value op) {
    238     return op == LT || op == LTE || op == GT || op == GTE;
    239   }
    240 
    241   static bool IsEqualityOp(Value op) {
    242     return op == EQ || op == EQ_STRICT;
    243   }
    244 
    245   static bool IsInequalityOp(Value op) {
    246     return op == NE || op == NE_STRICT;
    247   }
    248 
    249   static bool IsArithmeticCompareOp(Value op) {
    250     return IsOrderedRelationalCompareOp(op) ||
    251         IsEqualityOp(op) || IsInequalityOp(op);
    252   }
    253 
    254   static Value NegateCompareOp(Value op) {
    255     DCHECK(IsArithmeticCompareOp(op));
    256     switch (op) {
    257       case EQ: return NE;
    258       case NE: return EQ;
    259       case EQ_STRICT: return NE_STRICT;
    260       case NE_STRICT: return EQ_STRICT;
    261       case LT: return GTE;
    262       case GT: return LTE;
    263       case LTE: return GT;
    264       case GTE: return LT;
    265       default:
    266         UNREACHABLE();
    267         return op;
    268     }
    269   }
    270 
    271   static Value ReverseCompareOp(Value op) {
    272     DCHECK(IsArithmeticCompareOp(op));
    273     switch (op) {
    274       case EQ: return EQ;
    275       case NE: return NE;
    276       case EQ_STRICT: return EQ_STRICT;
    277       case NE_STRICT: return NE_STRICT;
    278       case LT: return GT;
    279       case GT: return LT;
    280       case LTE: return GTE;
    281       case GTE: return LTE;
    282       default:
    283         UNREACHABLE();
    284         return op;
    285     }
    286   }
    287 
    288   static bool EvalComparison(Value op, double op1, double op2) {
    289     DCHECK(IsArithmeticCompareOp(op));
    290     switch (op) {
    291       case Token::EQ:
    292       case Token::EQ_STRICT: return (op1 == op2);
    293       case Token::NE: return (op1 != op2);
    294       case Token::LT: return (op1 < op2);
    295       case Token::GT: return (op1 > op2);
    296       case Token::LTE: return (op1 <= op2);
    297       case Token::GTE: return (op1 >= op2);
    298       default:
    299         UNREACHABLE();
    300         return false;
    301     }
    302   }
    303 
    304   static bool IsBitOp(Value op) {
    305     return (BIT_OR <= op && op <= SHR) || op == BIT_NOT;
    306   }
    307 
    308   static bool IsUnaryOp(Value op) {
    309     return (NOT <= op && op <= VOID) || op == ADD || op == SUB;
    310   }
    311 
    312   static bool IsCountOp(Value op) {
    313     return op == INC || op == DEC;
    314   }
    315 
    316   static bool IsShiftOp(Value op) {
    317     return (SHL <= op) && (op <= SHR);
    318   }
    319 
    320   // Returns a string corresponding to the JS token string
    321   // (.e., "<" for the token LT) or NULL if the token doesn't
    322   // have a (unique) string (e.g. an IDENTIFIER).
    323   static const char* String(Value tok) {
    324     DCHECK(tok < NUM_TOKENS);  // tok is unsigned.
    325     return string_[tok];
    326   }
    327 
    328   // Returns the precedence > 0 for binary and compare
    329   // operators; returns 0 otherwise.
    330   static int Precedence(Value tok) {
    331     DCHECK(tok < NUM_TOKENS);  // tok is unsigned.
    332     return precedence_[tok];
    333   }
    334 
    335  private:
    336   static const char* const name_[NUM_TOKENS];
    337   static const char* const string_[NUM_TOKENS];
    338   static const int8_t precedence_[NUM_TOKENS];
    339   static const char token_type[NUM_TOKENS];
    340 };
    341 
    342 }  // namespace internal
    343 }  // namespace v8
    344 
    345 #endif  // V8_PARSING_TOKEN_H_
    346