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/unicode.h" 9 10 namespace v8 { 11 namespace internal { 12 13 // Unicode character predicates as defined by ECMA-262, 3rd, 14 // used for lexical analysis. 15 16 inline bool IsCarriageReturn(uc32 c); 17 inline bool IsLineFeed(uc32 c); 18 inline bool IsDecimalDigit(uc32 c); 19 inline bool IsHexDigit(uc32 c); 20 inline bool IsOctalDigit(uc32 c); 21 inline bool IsBinaryDigit(uc32 c); 22 inline bool IsRegExpWord(uc32 c); 23 inline bool IsRegExpNewline(uc32 c); 24 25 struct IdentifierStart { IsIdentifierStart26 static inline bool Is(uc32 c) { 27 switch (c) { 28 case '$': case '_': case '\\': return true; 29 default: return unibrow::Letter::Is(c); 30 } 31 } 32 }; 33 34 35 struct IdentifierPart { IsIdentifierPart36 static inline bool Is(uc32 c) { 37 return IdentifierStart::Is(c) 38 || unibrow::Number::Is(c) 39 || c == 0x200C // U+200C is Zero-Width Non-Joiner. 40 || c == 0x200D // U+200D is Zero-Width Joiner. 41 || unibrow::CombiningMark::Is(c) 42 || unibrow::ConnectorPunctuation::Is(c); 43 } 44 }; 45 46 47 // WhiteSpace according to ECMA-262 5.1, 7.2. 48 struct WhiteSpace { IsWhiteSpace49 static inline bool Is(uc32 c) { 50 return c == 0x0009 || // <TAB> 51 c == 0x000B || // <VT> 52 c == 0x000C || // <FF> 53 c == 0xFEFF || // <BOM> 54 // \u0020 and \u00A0 are included in unibrow::WhiteSpace. 55 unibrow::WhiteSpace::Is(c); 56 } 57 }; 58 59 60 // WhiteSpace and LineTerminator according to ECMA-262 5.1, 7.2 and 7.3. 61 struct WhiteSpaceOrLineTerminator { IsWhiteSpaceOrLineTerminator62 static inline bool Is(uc32 c) { 63 return WhiteSpace::Is(c) || unibrow::LineTerminator::Is(c); 64 } 65 }; 66 67 } } // namespace v8::internal 68 69 #endif // V8_CHAR_PREDICATES_H_ 70