blob: 0472da89007eaebc0db108fcbc591dc55ee6bfd0 [file] [log] [blame]
 // Copyright 2011 the V8 project authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. #ifndef V8_CONVERSIONS_H_ #define V8_CONVERSIONS_H_ #include #include "src/base/logging.h" #include "src/handles.h" #include "src/utils.h" namespace v8 { namespace internal { class UnicodeCache; // Maximum number of significant digits in decimal representation. // The longest possible double in decimal representation is // (2^53 - 1) * 2 ^ -1074 that is (2 ^ 53 - 1) * 5 ^ 1074 / 10 ^ 1074 // (768 digits). If we parse a number whose first digits are equal to a // mean of 2 adjacent doubles (that could have up to 769 digits) the result // must be rounded to the bigger one unless the tail consists of zeros, so // we don't need to preserve all the digits. const int kMaxSignificantDigits = 772; inline bool isDigit(int x, int radix) { return (x >= '0' && x <= '9' && x < '0' + radix) || (radix > 10 && x >= 'a' && x < 'a' + radix - 10) || (radix > 10 && x >= 'A' && x < 'A' + radix - 10); } inline bool isBinaryDigit(int x) { return x == '0' || x == '1'; } // The fast double-to-(unsigned-)int conversion routine does not guarantee // rounding towards zero. // If x is NaN, the result is INT_MIN. Otherwise the result is the argument x, // clamped to [INT_MIN, INT_MAX] and then rounded to an integer. inline int FastD2IChecked(double x) { if (!(x >= INT_MIN)) return INT_MIN; // Negation to catch NaNs. if (x > INT_MAX) return INT_MAX; return static_cast(x); } // The fast double-to-(unsigned-)int conversion routine does not guarantee // rounding towards zero. // The result is unspecified if x is infinite or NaN, or if the rounded // integer value is outside the range of type int. inline int FastD2I(double x) { return static_cast(x); } inline unsigned int FastD2UI(double x); inline double FastI2D(int x) { // There is no rounding involved in converting an integer to a // double, so this code should compile to a few instructions without // any FPU pipeline stalls. return static_cast(x); } inline double FastUI2D(unsigned x) { // There is no rounding involved in converting an unsigned integer to a // double, so this code should compile to a few instructions without // any FPU pipeline stalls. return static_cast(x); } // This function should match the exact semantics of ECMA-262 20.2.2.17. inline float DoubleToFloat32(double x); // This function should match the exact semantics of ECMA-262 9.4. inline double DoubleToInteger(double x); // This function should match the exact semantics of ECMA-262 9.5. inline int32_t DoubleToInt32(double x); // This function should match the exact semantics of ECMA-262 9.6. inline uint32_t DoubleToUint32(double x); // Enumeration for allowing octals and ignoring junk when converting // strings to numbers. enum ConversionFlags { NO_FLAGS = 0, ALLOW_HEX = 1, ALLOW_OCTAL = 2, ALLOW_IMPLICIT_OCTAL = 4, ALLOW_BINARY = 8, ALLOW_TRAILING_JUNK = 16 }; // Converts a string into a double value according to ECMA-262 9.3.1 double StringToDouble(UnicodeCache* unicode_cache, Vector str, int flags, double empty_string_val = 0); double StringToDouble(UnicodeCache* unicode_cache, Vector str, int flags, double empty_string_val = 0); // This version expects a zero-terminated character array. double StringToDouble(UnicodeCache* unicode_cache, const char* str, int flags, double empty_string_val = 0); // Converts a string into an integer. double StringToInt(UnicodeCache* unicode_cache, Vector vector, int radix); double StringToInt(UnicodeCache* unicode_cache, Vector vector, int radix); const int kDoubleToCStringMinBufferSize = 100; // Converts a double to a string value according to ECMA-262 9.8.1. // The buffer should be large enough for any floating point number. // 100 characters is enough. const char* DoubleToCString(double value, Vector buffer); // Convert an int to a null-terminated string. The returned string is // located inside the buffer, but not necessarily at the start. const char* IntToCString(int n, Vector buffer); // Additional number to string conversions for the number type. // The caller is responsible for calling free on the returned pointer. char* DoubleToFixedCString(double value, int f); char* DoubleToExponentialCString(double value, int f); char* DoubleToPrecisionCString(double value, int f); char* DoubleToRadixCString(double value, int radix); static inline bool IsMinusZero(double value) { return bit_cast(value) == bit_cast(-0.0); } // Returns true if value can be converted to a SMI, and returns the resulting // integer value of the SMI in |smi_int_value|. inline bool DoubleToSmiInteger(double value, int* smi_int_value); inline bool IsSmiDouble(double value); // Integer32 is an integer that can be represented as a signed 32-bit // integer. It has to be in the range [-2^31, 2^31 - 1]. // We also have to check for negative 0 as it is not an Integer32. inline bool IsInt32Double(double value); // UInteger32 is an integer that can be represented as an unsigned 32-bit // integer. It has to be in the range [0, 2^32 - 1]. // We also have to check for negative 0 as it is not a UInteger32. inline bool IsUint32Double(double value); // Convert from Number object to C integer. inline uint32_t PositiveNumberToUint32(Object* number); inline int32_t NumberToInt32(Object* number); inline uint32_t NumberToUint32(Object* number); inline int64_t NumberToInt64(Object* number); double StringToDouble(UnicodeCache* unicode_cache, Handle string, int flags, double empty_string_val = 0.0); inline bool TryNumberToSize(Object* number, size_t* result); // Converts a number into size_t. inline size_t NumberToSize(Object* number); // returns DoubleToString(StringToDouble(string)) == string bool IsSpecialIndex(UnicodeCache* unicode_cache, String* string); } // namespace internal } // namespace v8 #endif // V8_CONVERSIONS_H_