435 lines
12 KiB
C++
435 lines
12 KiB
C++
// Copyright (c) 2013, NetEase Inc. All rights reserved.
|
|
//
|
|
// wrt(guangguang)
|
|
// 2013/8/28
|
|
//
|
|
// String number conversion
|
|
|
|
#include "base/util/string_number_conversions.h"
|
|
|
|
#include <assert.h>
|
|
#include <ctype.h>
|
|
#include <errno.h>
|
|
#include <stdlib.h>
|
|
#include <wctype.h>
|
|
#include <limits>
|
|
|
|
namespace nbase {
|
|
|
|
namespace {
|
|
|
|
template <typename STR, typename INT, typename UINT, bool NEG>
|
|
struct IntToStringT {
|
|
// This is to avoid a compiler warning about unary minus on unsigned type.
|
|
// For example, say you had the following code:
|
|
// template <typename INT>
|
|
// INT abs(INT value) { return value < 0 ? -value : value; }
|
|
// Even though if INT is unsigned, it's impossible for value < 0, so the
|
|
// unary minus will never be taken, the compiler will still generate a
|
|
// warning. We do a little specialization dance...
|
|
template <typename INT2, typename UINT2, bool NEG2>
|
|
struct ToUnsignedT {};
|
|
|
|
template <typename INT2, typename UINT2>
|
|
struct ToUnsignedT<INT2, UINT2, false> {
|
|
static UINT2 ToUnsigned(INT2 value) {
|
|
return static_cast<UINT2>(value);
|
|
}
|
|
};
|
|
|
|
template <typename INT2, typename UINT2>
|
|
struct ToUnsignedT<INT2, UINT2, true> {
|
|
static UINT2 ToUnsigned(INT2 value) {
|
|
return static_cast<UINT2>(value < 0 ? -value : value);
|
|
}
|
|
};
|
|
|
|
// This set of templates is very similar to the above templates, but
|
|
// for testing whether an integer is negative.
|
|
template <typename INT2, bool NEG2>
|
|
struct TestNegT {};
|
|
template <typename INT2>
|
|
struct TestNegT<INT2, false> {
|
|
static bool TestNeg(INT2 value) {
|
|
// value is unsigned, and can never be negative.
|
|
return false;
|
|
}
|
|
};
|
|
template <typename INT2>
|
|
struct TestNegT<INT2, true> {
|
|
static bool TestNeg(INT2 value) {
|
|
return value < 0;
|
|
}
|
|
};
|
|
|
|
static STR IntToString(INT value) {
|
|
// log10(2) ~= 0.3 bytes needed per bit or per byte log10(2**8) ~= 2.4.
|
|
// So round up to allocate 3 output characters per byte, plus 1 for '-'.
|
|
const int kOutputBufSize = 3 * sizeof(INT) + 1;
|
|
|
|
// Allocate the whole string right away, we will right back to front, and
|
|
// then return the substr of what we ended up using.
|
|
STR outbuf(kOutputBufSize, 0);
|
|
|
|
bool is_neg = TestNegT<INT, NEG>::TestNeg(value);
|
|
// Even though is_neg will never be true when INT is parameterized as
|
|
// unsigned, even the presence of the unary operation causes a warning.
|
|
UINT res = ToUnsignedT<INT, UINT, NEG>::ToUnsigned(value);
|
|
|
|
for (typename STR::iterator it = outbuf.end();;) {
|
|
--it;
|
|
assert(it != outbuf.begin());
|
|
*it = static_cast<typename STR::value_type>((res % 10) + '0');
|
|
res /= 10;
|
|
|
|
// We're done..
|
|
if (res == 0) {
|
|
if (is_neg) {
|
|
--it;
|
|
assert(it != outbuf.begin());
|
|
*it = static_cast<typename STR::value_type>('-');
|
|
}
|
|
return STR(it, outbuf.end());
|
|
}
|
|
}
|
|
// Unreachable
|
|
}
|
|
};
|
|
|
|
// Utility to convert a character to a digit in a given base
|
|
template<typename CHAR, int BASE, bool BASE_LTE_10> class BaseCharToDigit {
|
|
};
|
|
|
|
// Faster specialization for bases <= 10
|
|
template<typename CHAR, int BASE> class BaseCharToDigit<CHAR, BASE, true> {
|
|
public:
|
|
static bool Convert(CHAR c, uint8_t* digit) {
|
|
if (c >= '0' && c < '0' + BASE) {
|
|
*digit = c - '0';
|
|
return true;
|
|
}
|
|
return false;
|
|
}
|
|
};
|
|
|
|
// Specialization for bases where 10 < base <= 36
|
|
template<typename CHAR, int BASE> class BaseCharToDigit<CHAR, BASE, false> {
|
|
public:
|
|
static bool Convert(CHAR c, uint8_t* digit) {
|
|
if (c >= '0' && c <= '9') {
|
|
*digit = c - '0';
|
|
} else if (c >= 'a' && c < 'a' + BASE - 10) {
|
|
*digit = c - 'a' + 10;
|
|
} else if (c >= 'A' && c < 'A' + BASE - 10) {
|
|
*digit = c - 'A' + 10;
|
|
} else {
|
|
return false;
|
|
}
|
|
return true;
|
|
}
|
|
};
|
|
|
|
template<int BASE, typename CHAR> bool CharToDigit(CHAR c, uint8_t* digit) {
|
|
return BaseCharToDigit<CHAR, BASE, BASE <= 10>::Convert(c, digit);
|
|
}
|
|
|
|
// There is an IsWhitespace for wchars defined in string_util.h, but it is
|
|
// locale independent, whereas the functions we are replacing were
|
|
// locale-dependent. TBD what is desired, but for the moment let's not introduce
|
|
// a change in behaviour.
|
|
template<typename CHAR> class WhitespaceHelper {
|
|
};
|
|
|
|
template<> class WhitespaceHelper<char> {
|
|
public:
|
|
static bool Invoke(char c) {
|
|
return 0 != isspace(static_cast<unsigned char>(c));
|
|
}
|
|
};
|
|
|
|
template<> class WhitespaceHelper<wchar_t> {
|
|
public:
|
|
static bool Invoke(wchar_t c) {
|
|
return 0 != iswspace(c);
|
|
}
|
|
};
|
|
|
|
template<typename CHAR> bool LocalIsWhitespace(CHAR c) {
|
|
return WhitespaceHelper<CHAR>::Invoke(c);
|
|
}
|
|
|
|
// IteratorRangeToNumberTraits should provide:
|
|
// - a typedef for iterator_type, the iterator type used as input.
|
|
// - a typedef for value_type, the target numeric type.
|
|
// - static functions min, max (returning the minimum and maximum permitted
|
|
// values)
|
|
// - constant kBase, the base in which to interpret the input
|
|
template<typename IteratorRangeToNumberTraits>
|
|
class IteratorRangeToNumber {
|
|
public:
|
|
typedef IteratorRangeToNumberTraits traits;
|
|
typedef typename traits::iterator_type const_iterator;
|
|
typedef typename traits::value_type value_type;
|
|
|
|
// Generalized iterator-range-to-number conversion.
|
|
//
|
|
static bool Invoke(const_iterator begin,
|
|
const_iterator end,
|
|
value_type* output) {
|
|
bool valid = true;
|
|
|
|
while (begin != end && LocalIsWhitespace(*begin)) {
|
|
valid = false;
|
|
++begin;
|
|
}
|
|
|
|
if (begin != end && *begin == '-') {
|
|
if (!Negative::Invoke(begin + 1, end, output)) {
|
|
valid = false;
|
|
}
|
|
} else {
|
|
if (begin != end && *begin == '+') {
|
|
++begin;
|
|
}
|
|
if (!Positive::Invoke(begin, end, output)) {
|
|
valid = false;
|
|
}
|
|
}
|
|
|
|
return valid;
|
|
}
|
|
|
|
private:
|
|
// Sign provides:
|
|
// - a static function, CheckBounds, that determines whether the next digit
|
|
// causes an overflow/underflow
|
|
// - a static function, Increment, that appends the next digit appropriately
|
|
// according to the sign of the number being parsed.
|
|
template<typename Sign>
|
|
class Base {
|
|
public:
|
|
static bool Invoke(const_iterator begin, const_iterator end,
|
|
typename traits::value_type* output) {
|
|
*output = 0;
|
|
|
|
if (begin == end) {
|
|
return false;
|
|
}
|
|
|
|
// Note: no performance difference was found when using template
|
|
// specialization to remove this check in bases other than 16
|
|
if (traits::kBase == 16 && end - begin > 2 && *begin == '0' &&
|
|
(*(begin + 1) == 'x' || *(begin + 1) == 'X')) {
|
|
begin += 2;
|
|
}
|
|
|
|
for (const_iterator current = begin; current != end; ++current) {
|
|
uint8_t new_digit = 0;
|
|
|
|
if (!CharToDigit<traits::kBase>(*current, &new_digit)) {
|
|
return false;
|
|
}
|
|
|
|
if (current != begin) {
|
|
if (!Sign::CheckBounds(output, new_digit)) {
|
|
return false;
|
|
}
|
|
*output *= traits::kBase;
|
|
}
|
|
|
|
Sign::Increment(new_digit, output);
|
|
}
|
|
return true;
|
|
}
|
|
};
|
|
|
|
class Positive : public Base<Positive> {
|
|
public:
|
|
static bool CheckBounds(value_type* output, uint8_t new_digit) {
|
|
if (*output > static_cast<value_type>(traits::max() / traits::kBase) ||
|
|
(*output == static_cast<value_type>(traits::max() / traits::kBase) &&
|
|
new_digit > traits::max() % traits::kBase)) {
|
|
*output = traits::max();
|
|
return false;
|
|
}
|
|
return true;
|
|
}
|
|
static void Increment(uint8_t increment, value_type* output) {
|
|
*output += increment;
|
|
}
|
|
};
|
|
|
|
class Negative : public Base<Negative> {
|
|
public:
|
|
static bool CheckBounds(value_type* output, uint8_t new_digit) {
|
|
if (*output < traits::min() / traits::kBase ||
|
|
(*output == traits::min() / traits::kBase &&
|
|
new_digit > 0 - traits::min() % traits::kBase)) {
|
|
*output = traits::min();
|
|
return false;
|
|
}
|
|
return true;
|
|
}
|
|
static void Increment(uint8_t increment, value_type* output) {
|
|
*output -= increment;
|
|
}
|
|
};
|
|
};
|
|
|
|
template<typename ITERATOR, typename VALUE, int BASE>
|
|
class BaseIteratorRangeToNumberTraits {
|
|
public:
|
|
typedef ITERATOR iterator_type;
|
|
typedef VALUE value_type;
|
|
static value_type min() {
|
|
return std::numeric_limits<value_type>::min();
|
|
}
|
|
static value_type max() {
|
|
return std::numeric_limits<value_type>::max();
|
|
}
|
|
static const int kBase = BASE;
|
|
};
|
|
|
|
template<typename ITERATOR>
|
|
class BaseHexIteratorRangeToIntTraits
|
|
: public BaseIteratorRangeToNumberTraits<ITERATOR, int, 16> {
|
|
public:
|
|
// Allow parsing of 0xFFFFFFFF, which is technically an overflow
|
|
static unsigned int max() {
|
|
return std::numeric_limits<unsigned int>::max();
|
|
}
|
|
};
|
|
|
|
typedef BaseHexIteratorRangeToIntTraits<std::string::const_iterator>
|
|
HexIteratorRangeToIntTraits;
|
|
|
|
template <typename VALUE, int BASE>
|
|
class StringPieceToNumberTraits
|
|
: public BaseIteratorRangeToNumberTraits<std::string::const_iterator,
|
|
VALUE,
|
|
BASE> {};
|
|
|
|
template <typename VALUE>
|
|
bool StringToIntImpl(const std::string& input, VALUE* output) {
|
|
return IteratorRangeToNumber<StringPieceToNumberTraits<VALUE, 10> >::Invoke(
|
|
input.begin(), input.end(), output);
|
|
}
|
|
|
|
template <typename VALUE, int BASE>
|
|
class StringPiece16ToNumberTraits
|
|
: public BaseIteratorRangeToNumberTraits<std::wstring::const_iterator,
|
|
VALUE,
|
|
BASE> {};
|
|
|
|
template <typename VALUE>
|
|
bool String16ToIntImpl(const std::wstring& input, VALUE* output) {
|
|
return IteratorRangeToNumber<StringPiece16ToNumberTraits<VALUE, 10> >::Invoke(
|
|
input.begin(), input.end(), output);
|
|
}
|
|
|
|
} // namespace
|
|
|
|
std::string IntToString(int value) {
|
|
return IntToStringT<std::string, int, unsigned int, true>::
|
|
IntToString(value);
|
|
}
|
|
|
|
std::wstring IntToString16(int value) {
|
|
return IntToStringT<std::wstring, int, unsigned int, true>::
|
|
IntToString(value);
|
|
}
|
|
|
|
std::string UintToString(unsigned int value) {
|
|
return IntToStringT<std::string, unsigned int, unsigned int, false>::
|
|
IntToString(value);
|
|
}
|
|
|
|
std::wstring UintToString16(unsigned int value) {
|
|
return IntToStringT<std::wstring, unsigned int, unsigned int, false>::
|
|
IntToString(value);
|
|
}
|
|
|
|
std::string Int64ToString(int64_t value) {
|
|
return IntToStringT<std::string, int64_t, uint64_t, true>::
|
|
IntToString(value);
|
|
}
|
|
|
|
std::wstring Int64ToString16(int64_t value) {
|
|
return IntToStringT<std::wstring, int64_t, uint64_t, true>::IntToString(value);
|
|
}
|
|
|
|
std::string Uint64ToString(uint64_t value) {
|
|
return IntToStringT<std::string, uint64_t, uint64_t, false>::
|
|
IntToString(value);
|
|
}
|
|
|
|
std::wstring Uint64ToString16(uint64_t value) {
|
|
return IntToStringT<std::wstring, uint64_t, uint64_t, false>::
|
|
IntToString(value);
|
|
}
|
|
|
|
std::string DoubleToString(double value) {
|
|
char buffer[64];
|
|
#ifdef COMPILER_MSVC
|
|
_snprintf(buffer, sizeof(buffer), "%lf", value);
|
|
#else
|
|
snprintf(buffer, sizeof(buffer), "%lf", value);
|
|
#endif
|
|
return std::string(buffer);
|
|
}
|
|
|
|
bool StringToInt(const std::string& input, int* output) {
|
|
return StringToIntImpl(input, output);
|
|
}
|
|
|
|
bool StringToInt(const std::wstring& input, int* output) {
|
|
return String16ToIntImpl(input, output);
|
|
}
|
|
|
|
bool StringToUint(const std::string& input, unsigned* output) {
|
|
return StringToIntImpl(input, output);
|
|
}
|
|
|
|
bool StringToUint(const std::wstring& input, unsigned* output) {
|
|
return String16ToIntImpl(input, output);
|
|
}
|
|
|
|
bool StringToInt64(const std::string& input, int64_t* output) {
|
|
return StringToIntImpl(input, output);
|
|
}
|
|
|
|
bool StringToInt64(const std::wstring& input, int64_t* output) {
|
|
return String16ToIntImpl(input, output);
|
|
}
|
|
|
|
bool StringToUint64(const std::string& input, uint64_t* output) {
|
|
return StringToIntImpl(input, output);
|
|
}
|
|
|
|
bool StringToUint64(const std::wstring& input, uint64_t* output) {
|
|
return String16ToIntImpl(input, output);
|
|
}
|
|
|
|
bool StringToSizeT(const std::string& input, size_t* output) {
|
|
return StringToIntImpl(input, output);
|
|
}
|
|
|
|
bool StringToSizeT(const std::wstring& input, size_t* output) {
|
|
return String16ToIntImpl(input, output);
|
|
}
|
|
|
|
bool StringToDouble(const std::string& input, double* output) {
|
|
if (input.empty() || !output)
|
|
return false;
|
|
if (sscanf(input.c_str(), "%lf", output) == 1)
|
|
return true;
|
|
return false;
|
|
}
|
|
|
|
bool HexStringToInt(const std::string& input, int* output) {
|
|
return IteratorRangeToNumber<HexIteratorRangeToIntTraits>::Invoke(
|
|
input.begin(), input.end(), output);
|
|
}
|
|
|
|
} // namespace nbase
|