OpenTTD/src/core/math_func.hpp

307 lines
8.6 KiB
C++

/*
* This file is part of OpenTTD.
* OpenTTD is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, version 2.
* OpenTTD is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
* See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with OpenTTD. If not, see <http://www.gnu.org/licenses/>.
*/
/** @file math_func.hpp Integer math functions */
#ifndef MATH_FUNC_HPP
#define MATH_FUNC_HPP
/**
* Returns the absolute value of (scalar) variable.
*
* @note assumes variable to be signed
* @param a The value we want to unsign
* @return The unsigned value
*/
template <typename T>
static inline T abs(const T a)
{
return (a < (T)0) ? -a : a;
}
/**
* Return the smallest multiple of n equal or greater than x
*
* @note n must be a power of 2
* @param x The min value
* @param n The base of the number we are searching
* @return The smallest multiple of n equal or greater than x
*/
template <typename T>
static inline T Align(const T x, uint n)
{
assert((n & (n - 1)) == 0 && n != 0);
n--;
return (T)((x + n) & ~((T)n));
}
/**
* Return the smallest multiple of n equal or greater than x
* Applies to pointers only
*
* @note n must be a power of 2
* @param x The min value
* @param n The base of the number we are searching
* @return The smallest multiple of n equal or greater than x
* @see Align()
*/
template <typename T>
static inline T *AlignPtr(T *x, uint n)
{
static_assert(sizeof(size_t) == sizeof(void *));
return reinterpret_cast<T *>(Align((size_t)x, n));
}
/**
* Clamp a value between an interval.
*
* This function returns a value which is between the given interval of
* min and max. If the given value is in this interval the value itself
* is returned otherwise the border of the interval is returned, according
* which side of the interval was 'left'.
*
* @note The min value must be less or equal of max or you get some
* unexpected results.
* @param a The value to clamp/truncate.
* @param min The minimum of the interval.
* @param max the maximum of the interval.
* @returns A value between min and max which is closest to a.
* @see ClampU(uint, uint, uint)
* @see Clamp(int, int, int)
*/
template <typename T>
static inline T Clamp(const T a, const T min, const T max)
{
assert(min <= max);
if (a <= min) return min;
if (a >= max) return max;
return a;
}
/**
* Clamp an integer between an interval.
*
* This function returns a value which is between the given interval of
* min and max. If the given value is in this interval the value itself
* is returned otherwise the border of the interval is returned, according
* which side of the interval was 'left'.
*
* @note The min value must be less or equal of max or you get some
* unexpected results.
* @param a The value to clamp/truncate.
* @param min The minimum of the interval.
* @param max the maximum of the interval.
* @returns A value between min and max which is closest to a.
* @see ClampU(uint, uint, uint)
*/
static inline int Clamp(const int a, const int min, const int max)
{
return Clamp<int>(a, min, max);
}
/**
* Clamp an unsigned integer between an interval.
*
* This function returns a value which is between the given interval of
* min and max. If the given value is in this interval the value itself
* is returned otherwise the border of the interval is returned, according
* which side of the interval was 'left'.
*
* @note The min value must be less or equal of max or you get some
* unexpected results.
* @param a The value to clamp/truncate.
* @param min The minimum of the interval.
* @param max the maximum of the interval.
* @returns A value between min and max which is closest to a.
* @see Clamp(int, int, int)
*/
static inline uint ClampU(const uint a, const uint min, const uint max)
{
return Clamp<uint>(a, min, max);
}
/**
* Reduce a signed 64-bit int to a signed 32-bit one
*
* This function clamps a 64-bit integer to a 32-bit integer.
* If the 64-bit value is smaller than the smallest 32-bit integer
* value 0x80000000 this value is returned (the left one bit is the sign bit).
* If the 64-bit value is greater than the greatest 32-bit integer value 0x7FFFFFFF
* this value is returned. In all other cases the 64-bit value 'fits' in a
* 32-bits integer field and so the value is casted to int32 and returned.
*
* @param a The 64-bit value to clamps
* @return The 64-bit value reduced to a 32-bit value
* @see Clamp(int, int, int)
*/
static inline int32 ClampToI32(const int64 a)
{
return static_cast<int32>(Clamp<int64>(a, INT32_MIN, INT32_MAX));
}
/**
* Reduce an unsigned 64-bit int to an unsigned 16-bit one
*
* @param a The 64-bit value to clamp
* @return The 64-bit value reduced to a 16-bit value
* @see ClampU(uint, uint, uint)
*/
static inline uint16 ClampToU16(const uint64 a)
{
/* MSVC thinks, in its infinite wisdom, that int min(int, int) is a better
* match for min(uint64, uint) than uint64 min(uint64, uint64). As such we
* need to cast the UINT16_MAX to prevent MSVC from displaying its
* infinite loads of warnings. */
return static_cast<uint16>(std::min(a, static_cast<uint64>(UINT16_MAX)));
}
/**
* Returns the (absolute) difference between two (scalar) variables
*
* @param a The first scalar
* @param b The second scalar
* @return The absolute difference between the given scalars
*/
template <typename T>
static inline T Delta(const T a, const T b)
{
return (a < b) ? b - a : a - b;
}
/**
* Checks if a value is between a window started at some base point.
*
* This function checks if the value x is between the value of base
* and base+size. If x equals base this returns true. If x equals
* base+size this returns false.
*
* @param x The value to check
* @param base The base value of the interval
* @param size The size of the interval
* @return True if the value is in the interval, false else.
*/
template <typename T>
static inline bool IsInsideBS(const T x, const size_t base, const size_t size)
{
return (size_t)(x - base) < size;
}
/**
* Checks if a value is in an interval.
*
* Returns true if a value is in the interval of [min, max).
*
* @param x The value to check
* @param min The minimum of the interval
* @param max The maximum of the interval
* @see IsInsideBS()
*/
template <typename T>
static constexpr inline bool IsInsideMM(const T x, const size_t min, const size_t max) noexcept
{
return (size_t)(x - min) < (max - min);
}
/**
* Type safe swap operation
* @param a variable to swap with b
* @param b variable to swap with a
*/
template <typename T>
static inline void Swap(T &a, T &b)
{
T t = a;
a = b;
b = t;
}
/**
* Converts a "fract" value 0..255 to "percent" value 0..100
* @param i value to convert, range 0..255
* @return value in range 0..100
*/
static inline uint ToPercent8(uint i)
{
assert(i < 256);
return i * 101 >> 8;
}
/**
* Converts a "fract" value 0..65535 to "percent" value 0..100
* @param i value to convert, range 0..65535
* @return value in range 0..100
*/
static inline uint ToPercent16(uint i)
{
assert(i < 65536);
return i * 101 >> 16;
}
int LeastCommonMultiple(int a, int b);
int GreatestCommonDivisor(int a, int b);
int DivideApprox(int a, int b);
/**
* Computes ceil(a / b) for non-negative a and b.
* @param a Numerator
* @param b Denominator
* @return Quotient, rounded up
*/
static inline uint CeilDiv(uint a, uint b)
{
return (a + b - 1) / b;
}
/**
* Computes ceil(a / b) * b for non-negative a and b.
* @param a Numerator
* @param b Denominator
* @return a rounded up to the nearest multiple of b.
*/
static inline uint Ceil(uint a, uint b)
{
return CeilDiv(a, b) * b;
}
/**
* Computes round(a / b) for signed a and unsigned b.
* @param a Numerator
* @param b Denominator
* @return Quotient, rounded to nearest
*/
static inline int RoundDivSU(int a, uint b)
{
if (a > 0) {
/* 0.5 is rounded to 1 */
return (a + static_cast<int>(b) / 2) / static_cast<int>(b);
} else {
/* -0.5 is rounded to 0 */
return (a - (static_cast<int>(b) - 1) / 2) / static_cast<int>(b);
}
}
/**
* Computes (a / b) rounded away from zero.
* @param a Numerator
* @param b Denominator
* @return Quotient, rounded away from zero
*/
static inline int DivAwayFromZero(int a, uint b)
{
const int _b = static_cast<int>(b);
if (a > 0) {
return (a + _b - 1) / _b;
} else {
/* Note: Behaviour of negative numerator division is truncation toward zero. */
return (a - _b + 1) / _b;
}
}
uint32 IntSqrt(uint32 num);
#endif /* MATH_FUNC_HPP */