Line data    Source code

       1             : // Copyright 2011 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_CHAR_PREDICATES_H_
       6             : #define V8_CHAR_PREDICATES_H_
       7             : 
       8             : #include "src/globals.h"
       9             : #include "src/unicode.h"
      10             : 
      11             : namespace v8 {
      12             : namespace internal {
      13             : 
      14             : // Unicode character predicates as defined by ECMA-262, 3rd,
      15             : // used for lexical analysis.
      16             : 
      17             : inline int AsciiAlphaToLower(uc32 c);
      18             : inline bool IsCarriageReturn(uc32 c);
      19             : inline bool IsLineFeed(uc32 c);
      20             : inline bool IsAsciiIdentifier(uc32 c);
      21             : inline bool IsAlphaNumeric(uc32 c);
      22             : inline bool IsDecimalDigit(uc32 c);
      23             : inline bool IsHexDigit(uc32 c);
      24             : inline bool IsOctalDigit(uc32 c);
      25             : inline bool IsBinaryDigit(uc32 c);
      26             : inline bool IsRegExpWord(uc32 c);
      27             : inline bool IsRegExpNewline(uc32 c);
      28             : 
      29             : struct V8_EXPORT_PRIVATE SupplementaryPlanes {
      30             :   static bool IsIDStart(uc32 c);
      31             :   static bool IsIDPart(uc32 c);
      32             : };
      33             : 
      34             : 
      35             : // ES6 draft section 11.6
      36             : // This includes '_', '$' and '\', and ID_Start according to
      37             : // http://www.unicode.org/reports/tr31/, which consists of categories
      38             : // 'Lu', 'Ll', 'Lt', 'Lm', 'Lo', 'Nl', but excluding properties
      39             : // 'Pattern_Syntax' or 'Pattern_White_Space'.
      40             : // For code points in the SMPs, we can resort to ICU (if available).
      41             : struct IdentifierStart {
      42     1636657 :   static inline bool Is(uc32 c) {
      43     1636657 :     if (c > 0xFFFF) return SupplementaryPlanes::IsIDStart(c);
      44     1636553 :     return unibrow::ID_Start::Is(c);
      45             :   }
      46             : };
      47             : 
      48             : 
      49             : // ES6 draft section 11.6
      50             : // This includes \u200c and \u200d, and ID_Continue according to
      51             : // http://www.unicode.org/reports/tr31/, which consists of ID_Start,
      52             : // the categories 'Mn', 'Mc', 'Nd', 'Pc', but excluding properties
      53             : // 'Pattern_Syntax' or 'Pattern_White_Space'.
      54             : // For code points in the SMPs, we can resort to ICU (if available).
      55             : struct IdentifierPart {
      56       28419 :   static inline bool Is(uc32 c) {
      57       28419 :     if (c > 0xFFFF) return SupplementaryPlanes::IsIDPart(c);
      58       28228 :     return unibrow::ID_Start::Is(c) || unibrow::ID_Continue::Is(c);
      59             :   }
      60             : };
      61             : 
      62             : 
      63             : // ES6 draft section 11.2
      64             : // This includes all code points of Unicode category 'Zs'.
      65             : // \u180e stops being one as of Unicode 6.3.0, but ES6 adheres to Unicode 5.1,
      66             : // so it is also included.
      67             : // Further included are \u0009, \u000b, \u0020, \u00a0, \u000c, and \ufeff.
      68             : // There are no category 'Zs' code points in the SMPs.
      69             : struct WhiteSpace {
      70     1158628 :   static inline bool Is(uc32 c) { return unibrow::WhiteSpace::Is(c); }
      71             : };
      72             : 
      73             : 
      74             : // WhiteSpace and LineTerminator according to ES6 draft section 11.2 and 11.3
      75             : // This consists of \000a, \000d, \u2028, and \u2029.
      76             : struct WhiteSpaceOrLineTerminator {
      77     1031204 :   static inline bool Is(uc32 c) {
      78     1031204 :     return WhiteSpace::Is(c) || unibrow::LineTerminator::Is(c);
      79             :   }
      80             : };
      81             : 
      82             : }  // namespace internal
      83             : }  // namespace v8
      84             : 
      85             : #endif  // V8_CHAR_PREDICATES_H_