virtualx-engine/thirdparty/rvo2/rvo2_2d/Vector2.h

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

347 lines
11 KiB
C++
Raw Normal View History

2024-02-18 21:31:05 +01:00
/*
* Vector2.h
* RVO2 Library
*
* Copyright 2008 University of North Carolina at Chapel Hill
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*
* Please send all bug reports to <geom@cs.unc.edu>.
*
* The authors may be contacted via:
*
* Jur van den Berg, Stephen J. Guy, Jamie Snape, Ming C. Lin, Dinesh Manocha
* Dept. of Computer Science
* 201 S. Columbia St.
* Frederick P. Brooks, Jr. Computer Science Bldg.
* Chapel Hill, N.C. 27599-3175
* United States of America
*
* <http://gamma.cs.unc.edu/RVO2/>
*/
#ifndef RVO_VECTOR2_H_
#define RVO_VECTOR2_H_
/**
* \file Vector2.h
* \brief Contains the Vector2 class.
*/
#include <cmath>
#include <ostream>
namespace RVO2D {
/**
* \brief Defines a two-dimensional vector.
*/
class Vector2 {
public:
/**
* \brief Constructs and initializes a two-dimensional vector instance
* to (0.0, 0.0).
*/
inline Vector2() : x_(0.0f), y_(0.0f) { }
/**
* \brief Constructs and initializes a two-dimensional vector from
* the specified xy-coordinates.
* \param x The x-coordinate of the two-dimensional
* vector.
* \param y The y-coordinate of the two-dimensional
* vector.
*/
inline Vector2(float x, float y) : x_(x), y_(y) { }
inline Vector2(const Vector2 &vector)
{
x_ = vector.x();
y_ = vector.y();
}
/**
* \brief Returns the x-coordinate of this two-dimensional vector.
* \return The x-coordinate of the two-dimensional vector.
*/
inline float x() const { return x_; }
/**
* \brief Returns the y-coordinate of this two-dimensional vector.
* \return The y-coordinate of the two-dimensional vector.
*/
inline float y() const { return y_; }
/**
* \brief Computes the negation of this two-dimensional vector.
* \return The negation of this two-dimensional vector.
*/
inline Vector2 operator-() const
{
return Vector2(-x_, -y_);
}
/**
* \brief Computes the dot product of this two-dimensional vector with
* the specified two-dimensional vector.
* \param vector The two-dimensional vector with which the
* dot product should be computed.
* \return The dot product of this two-dimensional vector with a
* specified two-dimensional vector.
*/
inline float operator*(const Vector2 &vector) const
{
return x_ * vector.x() + y_ * vector.y();
}
/**
* \brief Computes the scalar multiplication of this
* two-dimensional vector with the specified scalar value.
* \param s The scalar value with which the scalar
* multiplication should be computed.
* \return The scalar multiplication of this two-dimensional vector
* with a specified scalar value.
*/
inline Vector2 operator*(float s) const
{
return Vector2(x_ * s, y_ * s);
}
/**
* \brief Computes the scalar division of this two-dimensional vector
* with the specified scalar value.
* \param s The scalar value with which the scalar
* division should be computed.
* \return The scalar division of this two-dimensional vector with a
* specified scalar value.
*/
inline Vector2 operator/(float s) const
{
const float invS = 1.0f / s;
return Vector2(x_ * invS, y_ * invS);
}
/**
* \brief Computes the vector sum of this two-dimensional vector with
* the specified two-dimensional vector.
* \param vector The two-dimensional vector with which the
* vector sum should be computed.
* \return The vector sum of this two-dimensional vector with a
* specified two-dimensional vector.
*/
inline Vector2 operator+(const Vector2 &vector) const
{
return Vector2(x_ + vector.x(), y_ + vector.y());
}
/**
* \brief Computes the vector difference of this two-dimensional
* vector with the specified two-dimensional vector.
* \param vector The two-dimensional vector with which the
* vector difference should be computed.
* \return The vector difference of this two-dimensional vector with a
* specified two-dimensional vector.
*/
inline Vector2 operator-(const Vector2 &vector) const
{
return Vector2(x_ - vector.x(), y_ - vector.y());
}
/**
* \brief Tests this two-dimensional vector for equality with the
* specified two-dimensional vector.
* \param vector The two-dimensional vector with which to
* test for equality.
* \return True if the two-dimensional vectors are equal.
*/
inline bool operator==(const Vector2 &vector) const
{
return x_ == vector.x() && y_ == vector.y();
}
/**
* \brief Tests this two-dimensional vector for inequality with the
* specified two-dimensional vector.
* \param vector The two-dimensional vector with which to
* test for inequality.
* \return True if the two-dimensional vectors are not equal.
*/
inline bool operator!=(const Vector2 &vector) const
{
return x_ != vector.x() || y_ != vector.y();
}
/**
* \brief Sets the value of this two-dimensional vector to the scalar
* multiplication of itself with the specified scalar value.
* \param s The scalar value with which the scalar
* multiplication should be computed.
* \return A reference to this two-dimensional vector.
*/
inline Vector2 &operator*=(float s)
{
x_ *= s;
y_ *= s;
return *this;
}
/**
* \brief Sets the value of this two-dimensional vector to the scalar
* division of itself with the specified scalar value.
* \param s The scalar value with which the scalar
* division should be computed.
* \return A reference to this two-dimensional vector.
*/
inline Vector2 &operator/=(float s)
{
const float invS = 1.0f / s;
x_ *= invS;
y_ *= invS;
return *this;
}
/**
* \brief Sets the value of this two-dimensional vector to the vector
* sum of itself with the specified two-dimensional vector.
* \param vector The two-dimensional vector with which the
* vector sum should be computed.
* \return A reference to this two-dimensional vector.
*/
inline Vector2 &operator+=(const Vector2 &vector)
{
x_ += vector.x();
y_ += vector.y();
return *this;
}
/**
* \brief Sets the value of this two-dimensional vector to the vector
* difference of itself with the specified two-dimensional
* vector.
* \param vector The two-dimensional vector with which the
* vector difference should be computed.
* \return A reference to this two-dimensional vector.
*/
inline Vector2 &operator-=(const Vector2 &vector)
{
x_ -= vector.x();
y_ -= vector.y();
return *this;
}
inline Vector2 &operator=(const Vector2 &vector)
{
x_ = vector.x();
y_ = vector.y();
return *this;
}
private:
float x_;
float y_;
};
/**
* \relates Vector2
* \brief Computes the scalar multiplication of the specified
* two-dimensional vector with the specified scalar value.
* \param s The scalar value with which the scalar
* multiplication should be computed.
* \param vector The two-dimensional vector with which the scalar
* multiplication should be computed.
* \return The scalar multiplication of the two-dimensional vector with the
* scalar value.
*/
inline Vector2 operator*(float s, const Vector2 &vector)
{
return Vector2(s * vector.x(), s * vector.y());
}
/**
* \relates Vector2
* \brief Inserts the specified two-dimensional vector into the specified
* output stream.
* \param os The output stream into which the two-dimensional
* vector should be inserted.
* \param vector The two-dimensional vector which to insert into
* the output stream.
* \return A reference to the output stream.
*/
inline std::ostream &operator<<(std::ostream &os, const Vector2 &vector)
{
os << "(" << vector.x() << "," << vector.y() << ")";
return os;
}
/**
* \relates Vector2
* \brief Computes the length of a specified two-dimensional vector.
* \param vector The two-dimensional vector whose length is to be
* computed.
* \return The length of the two-dimensional vector.
*/
inline float abs(const Vector2 &vector)
{
return std::sqrt(vector * vector);
}
/**
* \relates Vector2
* \brief Computes the squared length of a specified two-dimensional
* vector.
* \param vector The two-dimensional vector whose squared length
* is to be computed.
* \return The squared length of the two-dimensional vector.
*/
inline float absSq(const Vector2 &vector)
{
return vector * vector;
}
/**
* \relates Vector2
* \brief Computes the determinant of a two-dimensional square matrix with
* rows consisting of the specified two-dimensional vectors.
* \param vector1 The top row of the two-dimensional square
* matrix.
* \param vector2 The bottom row of the two-dimensional square
* matrix.
* \return The determinant of the two-dimensional square matrix.
*/
inline float det(const Vector2 &vector1, const Vector2 &vector2)
{
return vector1.x() * vector2.y() - vector1.y() * vector2.x();
}
/**
* \relates Vector2
* \brief Computes the normalization of the specified two-dimensional
* vector.
* \param vector The two-dimensional vector whose normalization
* is to be computed.
* \return The normalization of the two-dimensional vector.
*/
inline Vector2 normalize(const Vector2 &vector)
{
return vector / abs(vector);
}
}
#endif /* RVO_VECTOR2_H_ */