Revision control

Copy as Markdown

Other Tools

// Copyright 2013 The Servo Project Developers. See the COPYRIGHT
// file at the top-level directory of this distribution.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.
use super::UnknownUnit;
use crate::approxeq::ApproxEq;
use crate::approxord::{max, min};
use crate::length::Length;
use crate::num::*;
use crate::scale::Scale;
use crate::size::{Size2D, Size3D};
use crate::vector::{vec2, vec3, Vector2D, Vector3D};
use core::cmp::{Eq, PartialEq};
use core::fmt;
use core::hash::Hash;
use core::marker::PhantomData;
use core::ops::{Add, AddAssign, Div, DivAssign, Mul, MulAssign, Neg, Sub, SubAssign};
#[cfg(feature = "mint")]
use mint;
use num_traits::real::Real;
use num_traits::{Euclid, Float, NumCast};
#[cfg(feature = "serde")]
use serde;
#[cfg(feature = "bytemuck")]
use bytemuck::{Pod, Zeroable};
/// A 2d Point tagged with a unit.
#[repr(C)]
pub struct Point2D<T, U> {
pub x: T,
pub y: T,
#[doc(hidden)]
pub _unit: PhantomData<U>,
}
impl<T: Copy, U> Copy for Point2D<T, U> {}
impl<T: Clone, U> Clone for Point2D<T, U> {
fn clone(&self) -> Self {
Point2D {
x: self.x.clone(),
y: self.y.clone(),
_unit: PhantomData,
}
}
}
#[cfg(feature = "serde")]
impl<'de, T, U> serde::Deserialize<'de> for Point2D<T, U>
where
T: serde::Deserialize<'de>,
{
fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
where
D: serde::Deserializer<'de>,
{
let (x, y) = serde::Deserialize::deserialize(deserializer)?;
Ok(Point2D {
x,
y,
_unit: PhantomData,
})
}
}
#[cfg(feature = "serde")]
impl<T, U> serde::Serialize for Point2D<T, U>
where
T: serde::Serialize,
{
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: serde::Serializer,
{
(&self.x, &self.y).serialize(serializer)
}
}
#[cfg(feature = "arbitrary")]
impl<'a, T, U> arbitrary::Arbitrary<'a> for Point2D<T, U>
where
T: arbitrary::Arbitrary<'a>,
{
fn arbitrary(u: &mut arbitrary::Unstructured<'a>) -> arbitrary::Result<Self> {
let (x, y) = arbitrary::Arbitrary::arbitrary(u)?;
Ok(Point2D {
x,
y,
_unit: PhantomData,
})
}
}
#[cfg(feature = "bytemuck")]
unsafe impl<T: Zeroable, U> Zeroable for Point2D<T, U> {}
#[cfg(feature = "bytemuck")]
unsafe impl<T: Pod, U: 'static> Pod for Point2D<T, U> {}
impl<T, U> Eq for Point2D<T, U> where T: Eq {}
impl<T, U> PartialEq for Point2D<T, U>
where
T: PartialEq,
{
fn eq(&self, other: &Self) -> bool {
self.x == other.x && self.y == other.y
}
}
impl<T, U> Hash for Point2D<T, U>
where
T: Hash,
{
fn hash<H: core::hash::Hasher>(&self, h: &mut H) {
self.x.hash(h);
self.y.hash(h);
}
}
mint_vec!(Point2D[x, y] = Point2);
impl<T: fmt::Debug, U> fmt::Debug for Point2D<T, U> {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
f.debug_tuple("").field(&self.x).field(&self.y).finish()
}
}
impl<T: Default, U> Default for Point2D<T, U> {
fn default() -> Self {
Point2D::new(Default::default(), Default::default())
}
}
impl<T, U> Point2D<T, U> {
/// Constructor, setting all components to zero.
#[inline]
pub fn origin() -> Self
where
T: Zero,
{
point2(Zero::zero(), Zero::zero())
}
/// The same as [`origin()`](#method.origin).
#[inline]
pub fn zero() -> Self
where
T: Zero,
{
Self::origin()
}
/// Constructor taking scalar values directly.
#[inline]
pub const fn new(x: T, y: T) -> Self {
Point2D {
x,
y,
_unit: PhantomData,
}
}
/// Constructor taking properly Lengths instead of scalar values.
#[inline]
pub fn from_lengths(x: Length<T, U>, y: Length<T, U>) -> Self {
point2(x.0, y.0)
}
/// Constructor setting all components to the same value.
#[inline]
pub fn splat(v: T) -> Self
where
T: Clone,
{
Point2D {
x: v.clone(),
y: v,
_unit: PhantomData,
}
}
/// Tag a unitless value with units.
#[inline]
pub fn from_untyped(p: Point2D<T, UnknownUnit>) -> Self {
point2(p.x, p.y)
}
/// Apply the function `f` to each component of this point.
///
/// # Example
///
/// This may be used to perform unusual arithmetic which is not already offered as methods.
///
/// ```
/// use euclid::default::Point2D;
///
/// let p = Point2D::<u32>::new(5, 15);
/// assert_eq!(p.map(|coord| coord.saturating_sub(10)), Point2D::new(0, 5));
/// ```
#[inline]
pub fn map<V, F: FnMut(T) -> V>(self, mut f: F) -> Point2D<V, U> {
point2(f(self.x), f(self.y))
}
/// Apply the function `f` to each pair of components of this point and `rhs`.
///
/// # Example
///
/// This may be used to perform unusual arithmetic which is not already offered as methods.
///
/// ```
/// use euclid::{default::{Point2D, Vector2D}, point2};
///
/// let a: Point2D<u32> = point2(50, 200);
/// let b: Point2D<u32> = point2(100, 100);
/// assert_eq!(a.zip(b, u32::saturating_sub), Vector2D::new(0, 100));
/// ```
#[inline]
pub fn zip<V, F: FnMut(T, T) -> V>(self, rhs: Self, mut f: F) -> Vector2D<V, U> {
vec2(f(self.x, rhs.x), f(self.y, rhs.y))
}
}
impl<T: Copy, U> Point2D<T, U> {
/// Create a 3d point from this one, using the specified z value.
#[inline]
pub fn extend(self, z: T) -> Point3D<T, U> {
point3(self.x, self.y, z)
}
/// Cast this point into a vector.
///
/// Equivalent to subtracting the origin from this point.
#[inline]
pub fn to_vector(self) -> Vector2D<T, U> {
Vector2D {
x: self.x,
y: self.y,
_unit: PhantomData,
}
}
/// Swap x and y.
///
/// # Example
///
/// ```rust
/// # use euclid::{Point2D, point2};
/// enum Mm {}
///
/// let point: Point2D<_, Mm> = point2(1, -8);
///
/// assert_eq!(point.yx(), point2(-8, 1));
/// ```
#[inline]
pub fn yx(self) -> Self {
point2(self.y, self.x)
}
/// Drop the units, preserving only the numeric value.
///
/// # Example
///
/// ```rust
/// # use euclid::{Point2D, point2};
/// enum Mm {}
///
/// let point: Point2D<_, Mm> = point2(1, -8);
///
/// assert_eq!(point.x, point.to_untyped().x);
/// assert_eq!(point.y, point.to_untyped().y);
/// ```
#[inline]
pub fn to_untyped(self) -> Point2D<T, UnknownUnit> {
point2(self.x, self.y)
}
/// Cast the unit, preserving the numeric value.
///
/// # Example
///
/// ```rust
/// # use euclid::{Point2D, point2};
/// enum Mm {}
/// enum Cm {}
///
/// let point: Point2D<_, Mm> = point2(1, -8);
///
/// assert_eq!(point.x, point.cast_unit::<Cm>().x);
/// assert_eq!(point.y, point.cast_unit::<Cm>().y);
/// ```
#[inline]
pub fn cast_unit<V>(self) -> Point2D<T, V> {
point2(self.x, self.y)
}
/// Cast into an array with x and y.
///
/// # Example
///
/// ```rust
/// # use euclid::{Point2D, point2};
/// enum Mm {}
///
/// let point: Point2D<_, Mm> = point2(1, -8);
///
/// assert_eq!(point.to_array(), [1, -8]);
/// ```
#[inline]
pub fn to_array(self) -> [T; 2] {
[self.x, self.y]
}
/// Cast into a tuple with x and y.
///
/// # Example
///
/// ```rust
/// # use euclid::{Point2D, point2};
/// enum Mm {}
///
/// let point: Point2D<_, Mm> = point2(1, -8);
///
/// assert_eq!(point.to_tuple(), (1, -8));
/// ```
#[inline]
pub fn to_tuple(self) -> (T, T) {
(self.x, self.y)
}
/// Convert into a 3d point with z-coordinate equals to zero.
#[inline]
pub fn to_3d(self) -> Point3D<T, U>
where
T: Zero,
{
point3(self.x, self.y, Zero::zero())
}
/// Rounds each component to the nearest integer value.
///
/// This behavior is preserved for negative values (unlike the basic cast).
///
/// ```rust
/// # use euclid::point2;
/// enum Mm {}
///
/// assert_eq!(point2::<_, Mm>(-0.1, -0.8).round(), point2::<_, Mm>(0.0, -1.0))
/// ```
#[inline]
#[must_use]
pub fn round(self) -> Self
where
T: Round,
{
point2(self.x.round(), self.y.round())
}
/// Rounds each component to the smallest integer equal or greater than the original value.
///
/// This behavior is preserved for negative values (unlike the basic cast).
///
/// ```rust
/// # use euclid::point2;
/// enum Mm {}
///
/// assert_eq!(point2::<_, Mm>(-0.1, -0.8).ceil(), point2::<_, Mm>(0.0, 0.0))
/// ```
#[inline]
#[must_use]
pub fn ceil(self) -> Self
where
T: Ceil,
{
point2(self.x.ceil(), self.y.ceil())
}
/// Rounds each component to the biggest integer equal or lower than the original value.
///
/// This behavior is preserved for negative values (unlike the basic cast).
///
/// ```rust
/// # use euclid::point2;
/// enum Mm {}
///
/// assert_eq!(point2::<_, Mm>(-0.1, -0.8).floor(), point2::<_, Mm>(-1.0, -1.0))
/// ```
#[inline]
#[must_use]
pub fn floor(self) -> Self
where
T: Floor,
{
point2(self.x.floor(), self.y.floor())
}
/// Linearly interpolate between this point and another point.
///
/// # Example
///
/// ```rust
/// use euclid::point2;
/// use euclid::default::Point2D;
///
/// let from: Point2D<_> = point2(0.0, 10.0);
/// let to: Point2D<_> = point2(8.0, -4.0);
///
/// assert_eq!(from.lerp(to, -1.0), point2(-8.0, 24.0));
/// assert_eq!(from.lerp(to, 0.0), point2( 0.0, 10.0));
/// assert_eq!(from.lerp(to, 0.5), point2( 4.0, 3.0));
/// assert_eq!(from.lerp(to, 1.0), point2( 8.0, -4.0));
/// assert_eq!(from.lerp(to, 2.0), point2(16.0, -18.0));
/// ```
#[inline]
pub fn lerp(self, other: Self, t: T) -> Self
where
T: One + Sub<Output = T> + Mul<Output = T> + Add<Output = T>,
{
let one_t = T::one() - t;
point2(one_t * self.x + t * other.x, one_t * self.y + t * other.y)
}
}
impl<T: PartialOrd, U> Point2D<T, U> {
#[inline]
pub fn min(self, other: Self) -> Self {
point2(min(self.x, other.x), min(self.y, other.y))
}
#[inline]
pub fn max(self, other: Self) -> Self {
point2(max(self.x, other.x), max(self.y, other.y))
}
/// Returns the point each component of which clamped by corresponding
/// components of `start` and `end`.
///
/// Shortcut for `self.max(start).min(end)`.
#[inline]
pub fn clamp(self, start: Self, end: Self) -> Self
where
T: Copy,
{
self.max(start).min(end)
}
}
impl<T: NumCast + Copy, U> Point2D<T, U> {
/// Cast from one numeric representation to another, preserving the units.
///
/// When casting from floating point to integer coordinates, the decimals are truncated
/// as one would expect from a simple cast, but this behavior does not always make sense
/// geometrically. Consider using `round()`, `ceil()` or `floor()` before casting.
#[inline]
pub fn cast<NewT: NumCast>(self) -> Point2D<NewT, U> {
self.try_cast().unwrap()
}
/// Fallible cast from one numeric representation to another, preserving the units.
///
/// When casting from floating point to integer coordinates, the decimals are truncated
/// as one would expect from a simple cast, but this behavior does not always make sense
/// geometrically. Consider using `round()`, `ceil()` or `floor()` before casting.
pub fn try_cast<NewT: NumCast>(self) -> Option<Point2D<NewT, U>> {
match (NumCast::from(self.x), NumCast::from(self.y)) {
(Some(x), Some(y)) => Some(point2(x, y)),
_ => None,
}
}
// Convenience functions for common casts
/// Cast into an `f32` point.
#[inline]
pub fn to_f32(self) -> Point2D<f32, U> {
self.cast()
}
/// Cast into an `f64` point.
#[inline]
pub fn to_f64(self) -> Point2D<f64, U> {
self.cast()
}
/// Cast into an `usize` point, truncating decimals if any.
///
/// When casting from floating point points, it is worth considering whether
/// to `round()`, `ceil()` or `floor()` before the cast in order to obtain
/// the desired conversion behavior.
#[inline]
pub fn to_usize(self) -> Point2D<usize, U> {
self.cast()
}
/// Cast into an `u32` point, truncating decimals if any.
///
/// When casting from floating point points, it is worth considering whether
/// to `round()`, `ceil()` or `floor()` before the cast in order to obtain
/// the desired conversion behavior.
#[inline]
pub fn to_u32(self) -> Point2D<u32, U> {
self.cast()
}
/// Cast into an i32 point, truncating decimals if any.
///
/// When casting from floating point points, it is worth considering whether
/// to `round()`, `ceil()` or `floor()` before the cast in order to obtain
/// the desired conversion behavior.
#[inline]
pub fn to_i32(self) -> Point2D<i32, U> {
self.cast()
}
/// Cast into an i64 point, truncating decimals if any.
///
/// When casting from floating point points, it is worth considering whether
/// to `round()`, `ceil()` or `floor()` before the cast in order to obtain
/// the desired conversion behavior.
#[inline]
pub fn to_i64(self) -> Point2D<i64, U> {
self.cast()
}
}
impl<T: Float, U> Point2D<T, U> {
/// Returns true if all members are finite.
#[inline]
pub fn is_finite(self) -> bool {
self.x.is_finite() && self.y.is_finite()
}
}
impl<T: Copy + Add<T, Output = T>, U> Point2D<T, U> {
#[inline]
pub fn add_size(self, other: &Size2D<T, U>) -> Self {
point2(self.x + other.width, self.y + other.height)
}
}
impl<T: Real + Sub<T, Output = T>, U> Point2D<T, U> {
#[inline]
pub fn distance_to(self, other: Self) -> T {
(self - other).length()
}
}
impl<T: Neg, U> Neg for Point2D<T, U> {
type Output = Point2D<T::Output, U>;
#[inline]
fn neg(self) -> Self::Output {
point2(-self.x, -self.y)
}
}
impl<T: Add, U> Add<Size2D<T, U>> for Point2D<T, U> {
type Output = Point2D<T::Output, U>;
#[inline]
fn add(self, other: Size2D<T, U>) -> Self::Output {
point2(self.x + other.width, self.y + other.height)
}
}
impl<T: AddAssign, U> AddAssign<Size2D<T, U>> for Point2D<T, U> {
#[inline]
fn add_assign(&mut self, other: Size2D<T, U>) {
self.x += other.width;
self.y += other.height;
}
}
impl<T: Add, U> Add<Vector2D<T, U>> for Point2D<T, U> {
type Output = Point2D<T::Output, U>;
#[inline]
fn add(self, other: Vector2D<T, U>) -> Self::Output {
point2(self.x + other.x, self.y + other.y)
}
}
impl<T: Copy + Add<T, Output = T>, U> AddAssign<Vector2D<T, U>> for Point2D<T, U> {
#[inline]
fn add_assign(&mut self, other: Vector2D<T, U>) {
*self = *self + other
}
}
impl<T: Sub, U> Sub for Point2D<T, U> {
type Output = Vector2D<T::Output, U>;
#[inline]
fn sub(self, other: Self) -> Self::Output {
vec2(self.x - other.x, self.y - other.y)
}
}
impl<T: Sub, U> Sub<Size2D<T, U>> for Point2D<T, U> {
type Output = Point2D<T::Output, U>;
#[inline]
fn sub(self, other: Size2D<T, U>) -> Self::Output {
point2(self.x - other.width, self.y - other.height)
}
}
impl<T: SubAssign, U> SubAssign<Size2D<T, U>> for Point2D<T, U> {
#[inline]
fn sub_assign(&mut self, other: Size2D<T, U>) {
self.x -= other.width;
self.y -= other.height;
}
}
impl<T: Sub, U> Sub<Vector2D<T, U>> for Point2D<T, U> {
type Output = Point2D<T::Output, U>;
#[inline]
fn sub(self, other: Vector2D<T, U>) -> Self::Output {
point2(self.x - other.x, self.y - other.y)
}
}
impl<T: Copy + Sub<T, Output = T>, U> SubAssign<Vector2D<T, U>> for Point2D<T, U> {
#[inline]
fn sub_assign(&mut self, other: Vector2D<T, U>) {
*self = *self - other
}
}
impl<T: Copy + Mul, U> Mul<T> for Point2D<T, U> {
type Output = Point2D<T::Output, U>;
#[inline]
fn mul(self, scale: T) -> Self::Output {
point2(self.x * scale, self.y * scale)
}
}
impl<T: Copy + Mul<T, Output = T>, U> MulAssign<T> for Point2D<T, U> {
#[inline]
fn mul_assign(&mut self, scale: T) {
*self = *self * scale
}
}
impl<T: Copy + Mul, U1, U2> Mul<Scale<T, U1, U2>> for Point2D<T, U1> {
type Output = Point2D<T::Output, U2>;
#[inline]
fn mul(self, scale: Scale<T, U1, U2>) -> Self::Output {
point2(self.x * scale.0, self.y * scale.0)
}
}
impl<T: Copy + MulAssign, U> MulAssign<Scale<T, U, U>> for Point2D<T, U> {
#[inline]
fn mul_assign(&mut self, scale: Scale<T, U, U>) {
self.x *= scale.0;
self.y *= scale.0;
}
}
impl<T: Copy + Div, U> Div<T> for Point2D<T, U> {
type Output = Point2D<T::Output, U>;
#[inline]
fn div(self, scale: T) -> Self::Output {
point2(self.x / scale, self.y / scale)
}
}
impl<T: Copy + Div<T, Output = T>, U> DivAssign<T> for Point2D<T, U> {
#[inline]
fn div_assign(&mut self, scale: T) {
*self = *self / scale
}
}
impl<T: Copy + Div, U1, U2> Div<Scale<T, U1, U2>> for Point2D<T, U2> {
type Output = Point2D<T::Output, U1>;
#[inline]
fn div(self, scale: Scale<T, U1, U2>) -> Self::Output {
point2(self.x / scale.0, self.y / scale.0)
}
}
impl<T: Copy + DivAssign, U> DivAssign<Scale<T, U, U>> for Point2D<T, U> {
#[inline]
fn div_assign(&mut self, scale: Scale<T, U, U>) {
self.x /= scale.0;
self.y /= scale.0;
}
}
impl<T: Zero, U> Zero for Point2D<T, U> {
#[inline]
fn zero() -> Self {
Self::origin()
}
}
impl<T: Round, U> Round for Point2D<T, U> {
/// See [Point2D::round()](#method.round)
#[inline]
fn round(self) -> Self {
self.round()
}
}
impl<T: Ceil, U> Ceil for Point2D<T, U> {
/// See [Point2D::ceil()](#method.ceil)
#[inline]
fn ceil(self) -> Self {
self.ceil()
}
}
impl<T: Floor, U> Floor for Point2D<T, U> {
/// See [Point2D::floor()](#method.floor)
#[inline]
fn floor(self) -> Self {
self.floor()
}
}
impl<T: ApproxEq<T>, U> ApproxEq<Point2D<T, U>> for Point2D<T, U> {
#[inline]
fn approx_epsilon() -> Self {
point2(T::approx_epsilon(), T::approx_epsilon())
}
#[inline]
fn approx_eq_eps(&self, other: &Self, eps: &Self) -> bool {
self.x.approx_eq_eps(&other.x, &eps.x) && self.y.approx_eq_eps(&other.y, &eps.y)
}
}
impl<T: Euclid, U> Point2D<T, U> {
/// Calculates the least nonnegative remainder of `self (mod other)`.
///
/// # Example
///
/// ```rust
/// use euclid::point2;
/// use euclid::default::{Point2D, Size2D};
///
/// let p = Point2D::new(7.0, -7.0);
/// let s = Size2D::new(4.0, -4.0);
///
/// assert_eq!(p.rem_euclid(&s), point2(3.0, 1.0));
/// assert_eq!((-p).rem_euclid(&s), point2(1.0, 3.0));
/// assert_eq!(p.rem_euclid(&-s), point2(3.0, 1.0));
/// ```
#[inline]
pub fn rem_euclid(&self, other: &Size2D<T, U>) -> Self {
point2(
self.x.rem_euclid(&other.width),
self.y.rem_euclid(&other.height),
)
}
/// Calculates Euclidean division, the matching method for `rem_euclid`.
///
/// # Example
///
/// ```rust
/// use euclid::point2;
/// use euclid::default::{Point2D, Size2D};
///
/// let p = Point2D::new(7.0, -7.0);
/// let s = Size2D::new(4.0, -4.0);
///
/// assert_eq!(p.div_euclid(&s), point2(1.0, 2.0));
/// assert_eq!((-p).div_euclid(&s), point2(-2.0, -1.0));
/// assert_eq!(p.div_euclid(&-s), point2(-1.0, -2.0));
/// ```
#[inline]
pub fn div_euclid(&self, other: &Size2D<T, U>) -> Self {
point2(
self.x.div_euclid(&other.width),
self.y.div_euclid(&other.height),
)
}
}
impl<T, U> From<Point2D<T, U>> for [T; 2] {
fn from(p: Point2D<T, U>) -> Self {
[p.x, p.y]
}
}
impl<T, U> From<[T; 2]> for Point2D<T, U> {
fn from([x, y]: [T; 2]) -> Self {
point2(x, y)
}
}
impl<T, U> From<Point2D<T, U>> for (T, T) {
fn from(p: Point2D<T, U>) -> Self {
(p.x, p.y)
}
}
impl<T, U> From<(T, T)> for Point2D<T, U> {
fn from(tuple: (T, T)) -> Self {
point2(tuple.0, tuple.1)
}
}
/// A 3d Point tagged with a unit.
#[repr(C)]
pub struct Point3D<T, U> {
pub x: T,
pub y: T,
pub z: T,
#[doc(hidden)]
pub _unit: PhantomData<U>,
}
mint_vec!(Point3D[x, y, z] = Point3);
impl<T: Copy, U> Copy for Point3D<T, U> {}
impl<T: Clone, U> Clone for Point3D<T, U> {
fn clone(&self) -> Self {
Point3D {
x: self.x.clone(),
y: self.y.clone(),
z: self.z.clone(),
_unit: PhantomData,
}
}
}
#[cfg(feature = "serde")]
impl<'de, T, U> serde::Deserialize<'de> for Point3D<T, U>
where
T: serde::Deserialize<'de>,
{
fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
where
D: serde::Deserializer<'de>,
{
let (x, y, z) = serde::Deserialize::deserialize(deserializer)?;
Ok(Point3D {
x,
y,
z,
_unit: PhantomData,
})
}
}
#[cfg(feature = "serde")]
impl<T, U> serde::Serialize for Point3D<T, U>
where
T: serde::Serialize,
{
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: serde::Serializer,
{
(&self.x, &self.y, &self.z).serialize(serializer)
}
}
#[cfg(feature = "bytemuck")]
unsafe impl<T: Zeroable, U> Zeroable for Point3D<T, U> {}
#[cfg(feature = "bytemuck")]
unsafe impl<T: Pod, U: 'static> Pod for Point3D<T, U> {}
impl<T, U> Eq for Point3D<T, U> where T: Eq {}
impl<T, U> PartialEq for Point3D<T, U>
where
T: PartialEq,
{
fn eq(&self, other: &Self) -> bool {
self.x == other.x && self.y == other.y && self.z == other.z
}
}
impl<T, U> Hash for Point3D<T, U>
where
T: Hash,
{
fn hash<H: core::hash::Hasher>(&self, h: &mut H) {
self.x.hash(h);
self.y.hash(h);
self.z.hash(h);
}
}
impl<T: fmt::Debug, U> fmt::Debug for Point3D<T, U> {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
f.debug_tuple("")
.field(&self.x)
.field(&self.y)
.field(&self.z)
.finish()
}
}
impl<T: Default, U> Default for Point3D<T, U> {
fn default() -> Self {
Point3D::new(Default::default(), Default::default(), Default::default())
}
}
impl<T, U> Point3D<T, U> {
/// Constructor, setting all components to zero.
#[inline]
pub fn origin() -> Self
where
T: Zero,
{
point3(Zero::zero(), Zero::zero(), Zero::zero())
}
/// The same as [`origin()`](#method.origin).
#[inline]
pub fn zero() -> Self
where
T: Zero,
{
Self::origin()
}
/// Constructor taking scalar values directly.
#[inline]
pub const fn new(x: T, y: T, z: T) -> Self {
Point3D {
x,
y,
z,
_unit: PhantomData,
}
}
/// Constructor taking properly Lengths instead of scalar values.
#[inline]
pub fn from_lengths(x: Length<T, U>, y: Length<T, U>, z: Length<T, U>) -> Self {
point3(x.0, y.0, z.0)
}
/// Constructor setting all components to the same value.
#[inline]
pub fn splat(v: T) -> Self
where
T: Clone,
{
Point3D {
x: v.clone(),
y: v.clone(),
z: v,
_unit: PhantomData,
}
}
/// Tag a unitless value with units.
#[inline]
pub fn from_untyped(p: Point3D<T, UnknownUnit>) -> Self {
point3(p.x, p.y, p.z)
}
/// Apply the function `f` to each component of this point.
///
/// # Example
///
/// This may be used to perform unusual arithmetic which is not already offered as methods.
///
/// ```
/// use euclid::default::Point3D;
///
/// let p = Point3D::<u32>::new(5, 11, 15);
/// assert_eq!(p.map(|coord| coord.saturating_sub(10)), Point3D::new(0, 1, 5));
/// ```
#[inline]
pub fn map<V, F: FnMut(T) -> V>(self, mut f: F) -> Point3D<V, U> {
point3(f(self.x), f(self.y), f(self.z))
}
/// Apply the function `f` to each pair of components of this point and `rhs`.
///
/// # Example
///
/// This may be used to perform unusual arithmetic which is not already offered as methods.
///
/// ```
/// use euclid::{default::{Point3D, Vector3D}, point2};
///
/// let a: Point3D<u32> = Point3D::new(50, 200, 400);
/// let b: Point3D<u32> = Point3D::new(100, 100, 150);
/// assert_eq!(a.zip(b, u32::saturating_sub), Vector3D::new(0, 100, 250));
/// ```
#[inline]
pub fn zip<V, F: FnMut(T, T) -> V>(self, rhs: Self, mut f: F) -> Vector3D<V, U> {
vec3(f(self.x, rhs.x), f(self.y, rhs.y), f(self.z, rhs.z))
}
}
impl<T: Copy, U> Point3D<T, U> {
/// Cast this point into a vector.
///
/// Equivalent to subtracting the origin to this point.
#[inline]
pub fn to_vector(self) -> Vector3D<T, U> {
Vector3D {
x: self.x,
y: self.y,
z: self.z,
_unit: PhantomData,
}
}
/// Returns a 2d point using this point's x and y coordinates
#[inline]
pub fn xy(self) -> Point2D<T, U> {
point2(self.x, self.y)
}
/// Returns a 2d point using this point's x and z coordinates
#[inline]
pub fn xz(self) -> Point2D<T, U> {
point2(self.x, self.z)
}
/// Returns a 2d point using this point's x and z coordinates
#[inline]
pub fn yz(self) -> Point2D<T, U> {
point2(self.y, self.z)
}
/// Cast into an array with x, y and z.
///
/// # Example
///
/// ```rust
/// # use euclid::{Point3D, point3};
/// enum Mm {}
///
/// let point: Point3D<_, Mm> = point3(1, -8, 0);
///
/// assert_eq!(point.to_array(), [1, -8, 0]);
/// ```
#[inline]
pub fn to_array(self) -> [T; 3] {
[self.x, self.y, self.z]
}
#[inline]
pub fn to_array_4d(self) -> [T; 4]
where
T: One,
{
[self.x, self.y, self.z, One::one()]
}
/// Cast into a tuple with x, y and z.
///
/// # Example
///
/// ```rust
/// # use euclid::{Point3D, point3};
/// enum Mm {}
///
/// let point: Point3D<_, Mm> = point3(1, -8, 0);
///
/// assert_eq!(point.to_tuple(), (1, -8, 0));
/// ```
#[inline]
pub fn to_tuple(self) -> (T, T, T) {
(self.x, self.y, self.z)
}
#[inline]
pub fn to_tuple_4d(self) -> (T, T, T, T)
where
T: One,
{
(self.x, self.y, self.z, One::one())
}
/// Drop the units, preserving only the numeric value.
///
/// # Example
///
/// ```rust
/// # use euclid::{Point3D, point3};
/// enum Mm {}
///
/// let point: Point3D<_, Mm> = point3(1, -8, 0);
///
/// assert_eq!(point.x, point.to_untyped().x);
/// assert_eq!(point.y, point.to_untyped().y);
/// assert_eq!(point.z, point.to_untyped().z);
/// ```
#[inline]
pub fn to_untyped(self) -> Point3D<T, UnknownUnit> {
point3(self.x, self.y, self.z)
}
/// Cast the unit, preserving the numeric value.
///
/// # Example
///
/// ```rust
/// # use euclid::{Point3D, point3};
/// enum Mm {}
/// enum Cm {}
///
/// let point: Point3D<_, Mm> = point3(1, -8, 0);
///
/// assert_eq!(point.x, point.cast_unit::<Cm>().x);
/// assert_eq!(point.y, point.cast_unit::<Cm>().y);
/// assert_eq!(point.z, point.cast_unit::<Cm>().z);
/// ```
#[inline]
pub fn cast_unit<V>(self) -> Point3D<T, V> {
point3(self.x, self.y, self.z)
}
/// Convert into a 2d point.
#[inline]
pub fn to_2d(self) -> Point2D<T, U> {
self.xy()
}
/// Rounds each component to the nearest integer value.
///
/// This behavior is preserved for negative values (unlike the basic cast).
///
/// ```rust
/// # use euclid::point3;
/// enum Mm {}
///
/// assert_eq!(point3::<_, Mm>(-0.1, -0.8, 0.4).round(), point3::<_, Mm>(0.0, -1.0, 0.0))
/// ```
#[inline]
#[must_use]
pub fn round(self) -> Self
where
T: Round,
{
point3(self.x.round(), self.y.round(), self.z.round())
}
/// Rounds each component to the smallest integer equal or greater than the original value.
///
/// This behavior is preserved for negative values (unlike the basic cast).
///
/// ```rust
/// # use euclid::point3;
/// enum Mm {}
///
/// assert_eq!(point3::<_, Mm>(-0.1, -0.8, 0.4).ceil(), point3::<_, Mm>(0.0, 0.0, 1.0))
/// ```
#[inline]
#[must_use]
pub fn ceil(self) -> Self
where
T: Ceil,
{
point3(self.x.ceil(), self.y.ceil(), self.z.ceil())
}
/// Rounds each component to the biggest integer equal or lower than the original value.
///
/// This behavior is preserved for negative values (unlike the basic cast).
///
/// ```rust
/// # use euclid::point3;
/// enum Mm {}
///
/// assert_eq!(point3::<_, Mm>(-0.1, -0.8, 0.4).floor(), point3::<_, Mm>(-1.0, -1.0, 0.0))
/// ```
#[inline]
#[must_use]
pub fn floor(self) -> Self
where
T: Floor,
{
point3(self.x.floor(), self.y.floor(), self.z.floor())
}
/// Linearly interpolate between this point and another point.
///
/// # Example
///
/// ```rust
/// use euclid::point3;
/// use euclid::default::Point3D;
///
/// let from: Point3D<_> = point3(0.0, 10.0, -1.0);
/// let to: Point3D<_> = point3(8.0, -4.0, 0.0);
///
/// assert_eq!(from.lerp(to, -1.0), point3(-8.0, 24.0, -2.0));
/// assert_eq!(from.lerp(to, 0.0), point3( 0.0, 10.0, -1.0));
/// assert_eq!(from.lerp(to, 0.5), point3( 4.0, 3.0, -0.5));
/// assert_eq!(from.lerp(to, 1.0), point3( 8.0, -4.0, 0.0));
/// assert_eq!(from.lerp(to, 2.0), point3(16.0, -18.0, 1.0));
/// ```
#[inline]
pub fn lerp(self, other: Self, t: T) -> Self
where
T: One + Sub<Output = T> + Mul<Output = T> + Add<Output = T>,
{
let one_t = T::one() - t;
point3(
one_t * self.x + t * other.x,
one_t * self.y + t * other.y,
one_t * self.z + t * other.z,
)
}
}
impl<T: PartialOrd, U> Point3D<T, U> {
#[inline]
pub fn min(self, other: Self) -> Self {
point3(
min(self.x, other.x),
min(self.y, other.y),
min(self.z, other.z),
)
}
#[inline]
pub fn max(self, other: Self) -> Self {
point3(
max(self.x, other.x),
max(self.y, other.y),
max(self.z, other.z),
)
}
/// Returns the point each component of which clamped by corresponding
/// components of `start` and `end`.
///
/// Shortcut for `self.max(start).min(end)`.
#[inline]
pub fn clamp(self, start: Self, end: Self) -> Self
where
T: Copy,
{
self.max(start).min(end)
}
}
impl<T: NumCast + Copy, U> Point3D<T, U> {
/// Cast from one numeric representation to another, preserving the units.
///
/// When casting from floating point to integer coordinates, the decimals are truncated
/// as one would expect from a simple cast, but this behavior does not always make sense
/// geometrically. Consider using `round()`, `ceil()` or `floor()` before casting.
#[inline]
pub fn cast<NewT: NumCast>(self) -> Point3D<NewT, U> {
self.try_cast().unwrap()
}
/// Fallible cast from one numeric representation to another, preserving the units.
///
/// When casting from floating point to integer coordinates, the decimals are truncated
/// as one would expect from a simple cast, but this behavior does not always make sense
/// geometrically. Consider using `round()`, `ceil()` or `floor()` before casting.
pub fn try_cast<NewT: NumCast>(self) -> Option<Point3D<NewT, U>> {
match (
NumCast::from(self.x),
NumCast::from(self.y),
NumCast::from(self.z),
) {
(Some(x), Some(y), Some(z)) => Some(point3(x, y, z)),
_ => None,
}
}
// Convenience functions for common casts
/// Cast into an `f32` point.
#[inline]
pub fn to_f32(self) -> Point3D<f32, U> {
self.cast()
}
/// Cast into an `f64` point.
#[inline]
pub fn to_f64(self) -> Point3D<f64, U> {
self.cast()
}
/// Cast into an `usize` point, truncating decimals if any.
///
/// When casting from floating point points, it is worth considering whether
/// to `round()`, `ceil()` or `floor()` before the cast in order to obtain
/// the desired conversion behavior.
#[inline]
pub fn to_usize(self) -> Point3D<usize, U> {
self.cast()
}
/// Cast into an `u32` point, truncating decimals if any.
///
/// When casting from floating point points, it is worth considering whether
/// to `round()`, `ceil()` or `floor()` before the cast in order to obtain
/// the desired conversion behavior.
#[inline]
pub fn to_u32(self) -> Point3D<u32, U> {
self.cast()
}
/// Cast into an `i32` point, truncating decimals if any.
///
/// When casting from floating point points, it is worth considering whether
/// to `round()`, `ceil()` or `floor()` before the cast in order to obtain
/// the desired conversion behavior.
#[inline]
pub fn to_i32(self) -> Point3D<i32, U> {
self.cast()
}
/// Cast into an `i64` point, truncating decimals if any.
///
/// When casting from floating point points, it is worth considering whether
/// to `round()`, `ceil()` or `floor()` before the cast in order to obtain
/// the desired conversion behavior.
#[inline]
pub fn to_i64(self) -> Point3D<i64, U> {
self.cast()
}
}
impl<T: Float, U> Point3D<T, U> {
/// Returns true if all members are finite.
#[inline]
pub fn is_finite(self) -> bool {
self.x.is_finite() && self.y.is_finite() && self.z.is_finite()
}
}
impl<T: Copy + Add<T, Output = T>, U> Point3D<T, U> {
#[inline]
pub fn add_size(self, other: Size3D<T, U>) -> Self {
point3(
self.x + other.width,
self.y + other.height,
self.z + other.depth,
)
}
}
impl<T: Real + Sub<T, Output = T>, U> Point3D<T, U> {
#[inline]
pub fn distance_to(self, other: Self) -> T {
(self - other).length()
}
}
impl<T: Neg, U> Neg for Point3D<T, U> {
type Output = Point3D<T::Output, U>;
#[inline]
fn neg(self) -> Self::Output {
point3(-self.x, -self.y, -self.z)
}
}
impl<T: Add, U> Add<Size3D<T, U>> for Point3D<T, U> {
type Output = Point3D<T::Output, U>;
#[inline]
fn add(self, other: Size3D<T, U>) -> Self::Output {
point3(
self.x + other.width,
self.y + other.height,
self.z + other.depth,
)
}
}
impl<T: AddAssign, U> AddAssign<Size3D<T, U>> for Point3D<T, U> {
#[inline]
fn add_assign(&mut self, other: Size3D<T, U>) {
self.x += other.width;
self.y += other.height;
self.z += other.depth;
}
}
impl<T: Add, U> Add<Vector3D<T, U>> for Point3D<T, U> {
type Output = Point3D<T::Output, U>;
#[inline]
fn add(self, other: Vector3D<T, U>) -> Self::Output {
point3(self.x + other.x, self.y + other.y, self.z + other.z)
}
}
impl<T: Copy + Add<T, Output = T>, U> AddAssign<Vector3D<T, U>> for Point3D<T, U> {
#[inline]
fn add_assign(&mut self, other: Vector3D<T, U>) {
*self = *self + other
}
}
impl<T: Sub, U> Sub for Point3D<T, U> {
type Output = Vector3D<T::Output, U>;
#[inline]
fn sub(self, other: Self) -> Self::Output {
vec3(self.x - other.x, self.y - other.y, self.z - other.z)
}
}
impl<T: Sub, U> Sub<Size3D<T, U>> for Point3D<T, U> {
type Output = Point3D<T::Output, U>;
#[inline]
fn sub(self, other: Size3D<T, U>) -> Self::Output {
point3(
self.x - other.width,
self.y - other.height,
self.z - other.depth,
)
}
}
impl<T: SubAssign, U> SubAssign<Size3D<T, U>> for Point3D<T, U> {
#[inline]
fn sub_assign(&mut self, other: Size3D<T, U>) {
self.x -= other.width;
self.y -= other.height;
self.z -= other.depth;
}
}
impl<T: Sub, U> Sub<Vector3D<T, U>> for Point3D<T, U> {
type Output = Point3D<T::Output, U>;
#[inline]
fn sub(self, other: Vector3D<T, U>) -> Self::Output {
point3(self.x - other.x, self.y - other.y, self.z - other.z)
}
}
impl<T: Copy + Sub<T, Output = T>, U> SubAssign<Vector3D<T, U>> for Point3D<T, U> {
#[inline]
fn sub_assign(&mut self, other: Vector3D<T, U>) {
*self = *self - other
}
}
impl<T: Copy + Mul, U> Mul<T> for Point3D<T, U> {
type Output = Point3D<T::Output, U>;
#[inline]
fn mul(self, scale: T) -> Self::Output {
point3(self.x * scale, self.y * scale, self.z * scale)
}
}
impl<T: Copy + MulAssign, U> MulAssign<T> for Point3D<T, U> {
#[inline]
fn mul_assign(&mut self, scale: T) {
self.x *= scale;
self.y *= scale;
self.z *= scale;
}
}
impl<T: Copy + Mul, U1, U2> Mul<Scale<T, U1, U2>> for Point3D<T, U1> {
type Output = Point3D<T::Output, U2>;
#[inline]
fn mul(self, scale: Scale<T, U1, U2>) -> Self::Output {
point3(self.x * scale.0, self.y * scale.0, self.z * scale.0)
}
}
impl<T: Copy + MulAssign, U> MulAssign<Scale<T, U, U>> for Point3D<T, U> {
#[inline]
fn mul_assign(&mut self, scale: Scale<T, U, U>) {
*self *= scale.0;
}
}
impl<T: Copy + Div, U> Div<T> for Point3D<T, U> {
type Output = Point3D<T::Output, U>;
#[inline]
fn div(self, scale: T) -> Self::Output {
point3(self.x / scale, self.y / scale, self.z / scale)
}
}
impl<T: Copy + DivAssign, U> DivAssign<T> for Point3D<T, U> {
#[inline]
fn div_assign(&mut self, scale: T) {
self.x /= scale;
self.y /= scale;
self.z /= scale;
}
}
impl<T: Copy + Div, U1, U2> Div<Scale<T, U1, U2>> for Point3D<T, U2> {
type Output = Point3D<T::Output, U1>;
#[inline]
fn div(self, scale: Scale<T, U1, U2>) -> Self::Output {
point3(self.x / scale.0, self.y / scale.0, self.z / scale.0)
}
}
impl<T: Copy + DivAssign, U> DivAssign<Scale<T, U, U>> for Point3D<T, U> {
#[inline]
fn div_assign(&mut self, scale: Scale<T, U, U>) {
*self /= scale.0;
}
}
impl<T: Zero, U> Zero for Point3D<T, U> {
#[inline]
fn zero() -> Self {
Self::origin()
}
}
impl<T: Round, U> Round for Point3D<T, U> {
/// See [Point3D::round()](#method.round)
#[inline]
fn round(self) -> Self {
self.round()
}
}
impl<T: Ceil, U> Ceil for Point3D<T, U> {
/// See [Point3D::ceil()](#method.ceil)
#[inline]
fn ceil(self) -> Self {
self.ceil()
}
}
impl<T: Floor, U> Floor for Point3D<T, U> {
/// See [Point3D::floor()](#method.floor)
#[inline]
fn floor(self) -> Self {
self.floor()
}
}
impl<T: ApproxEq<T>, U> ApproxEq<Point3D<T, U>> for Point3D<T, U> {
#[inline]
fn approx_epsilon() -> Self {
point3(
T::approx_epsilon(),
T::approx_epsilon(),
T::approx_epsilon(),
)
}
#[inline]
fn approx_eq_eps(&self, other: &Self, eps: &Self) -> bool {
self.x.approx_eq_eps(&other.x, &eps.x)
&& self.y.approx_eq_eps(&other.y, &eps.y)
&& self.z.approx_eq_eps(&other.z, &eps.z)
}
}
impl<T: Euclid, U> Point3D<T, U> {
/// Calculates the least nonnegative remainder of `self (mod other)`.
///
/// # Example
///
/// ```rust
/// use euclid::point3;
/// use euclid::default::{Point3D, Size3D};
///
/// let p = Point3D::new(7.0, -7.0, 0.0);
/// let s = Size3D::new(4.0, -4.0, 12.0);
/// assert_eq!(p.rem_euclid(&s), point3(3.0, 1.0, 0.0));
/// assert_eq!((-p).rem_euclid(&s), point3(1.0, 3.0, 0.0));
/// assert_eq!(p.rem_euclid(&-s), point3(3.0, 1.0, 0.0));
/// ```
#[inline]
pub fn rem_euclid(&self, other: &Size3D<T, U>) -> Self {
point3(
self.x.rem_euclid(&other.width),
self.y.rem_euclid(&other.height),
self.z.rem_euclid(&other.depth),
)
}
/// Calculates Euclidean division, the matching method for `rem_euclid`.
///
/// # Example
///
/// ```rust
/// use euclid::point3;
/// use euclid::default::{Point3D, Size3D};
///
/// let p = Point3D::new(7.0, -7.0, 0.0);
/// let s = Size3D::new(4.0, -4.0, 12.0);
///
/// assert_eq!(p.div_euclid(&s), point3(1.0, 2.0, 0.0));
/// assert_eq!((-p).div_euclid(&s), point3(-2.0, -1.0, 0.0));
/// assert_eq!(p.div_euclid(&-s), point3(-1.0, -2.0, 0.0));
/// ```
#[inline]
pub fn div_euclid(&self, other: &Size3D<T, U>) -> Self {
point3(
self.x.div_euclid(&other.width),
self.y.div_euclid(&other.height),
self.z.div_euclid(&other.depth),
)
}
}
impl<T, U> From<Point3D<T, U>> for [T; 3] {
fn from(p: Point3D<T, U>) -> Self {
[p.x, p.y, p.z]
}
}
impl<T, U> From<[T; 3]> for Point3D<T, U> {
fn from([x, y, z]: [T; 3]) -> Self {
point3(x, y, z)
}
}
impl<T, U> From<Point3D<T, U>> for (T, T, T) {
fn from(p: Point3D<T, U>) -> Self {
(p.x, p.y, p.z)
}
}
impl<T, U> From<(T, T, T)> for Point3D<T, U> {
fn from(tuple: (T, T, T)) -> Self {
point3(tuple.0, tuple.1, tuple.2)
}
}
/// Shorthand for `Point2D::new(x, y)`.
#[inline]
pub const fn point2<T, U>(x: T, y: T) -> Point2D<T, U> {
Point2D {
x,
y,
_unit: PhantomData,
}
}
/// Shorthand for `Point3D::new(x, y)`.
#[inline]
pub const fn point3<T, U>(x: T, y: T, z: T) -> Point3D<T, U> {
Point3D {
x,
y,
z,
_unit: PhantomData,
}
}
#[cfg(test)]
mod point2d {
use crate::default::Point2D;
use crate::point2;
#[cfg(feature = "mint")]
use mint;
#[test]
pub fn test_min() {
let p1 = Point2D::new(1.0, 3.0);
let p2 = Point2D::new(2.0, 2.0);
let result = p1.min(p2);
assert_eq!(result, Point2D::new(1.0, 2.0));
}
#[test]
pub fn test_max() {
let p1 = Point2D::new(1.0, 3.0);
let p2 = Point2D::new(2.0, 2.0);
let result = p1.max(p2);
assert_eq!(result, Point2D::new(2.0, 3.0));
}
#[cfg(feature = "mint")]
#[test]
pub fn test_mint() {
let p1 = Point2D::new(1.0, 3.0);
let pm: mint::Point2<_> = p1.into();
let p2 = Point2D::from(pm);
assert_eq!(p1, p2);
}
#[test]
pub fn test_conv_vector() {
for i in 0..100 {
// We don't care about these values as long as they are not the same.
let x = i as f32 * 0.012345;
let y = i as f32 * 0.987654;
let p: Point2D<f32> = point2(x, y);
assert_eq!(p.to_vector().to_point(), p);
}
}
#[test]
pub fn test_swizzling() {
let p: Point2D<i32> = point2(1, 2);
assert_eq!(p.yx(), point2(2, 1));
}
#[test]
pub fn test_distance_to() {
let p1 = Point2D::new(1.0, 2.0);
let p2 = Point2D::new(2.0, 2.0);
assert_eq!(p1.distance_to(p2), 1.0);
let p1 = Point2D::new(1.0, 2.0);
let p2 = Point2D::new(1.0, 4.0);
assert_eq!(p1.distance_to(p2), 2.0);
}
mod ops {
use crate::default::Point2D;
use crate::scale::Scale;
use crate::{size2, vec2, Vector2D};
pub enum Mm {}
pub enum Cm {}
pub type Point2DMm<T> = crate::Point2D<T, Mm>;
pub type Point2DCm<T> = crate::Point2D<T, Cm>;
#[test]
pub fn test_neg() {
assert_eq!(-Point2D::new(1.0, 2.0), Point2D::new(-1.0, -2.0));
assert_eq!(-Point2D::new(0.0, 0.0), Point2D::new(-0.0, -0.0));
assert_eq!(-Point2D::new(-1.0, -2.0), Point2D::new(1.0, 2.0));
}
#[test]
pub fn test_add_size() {
let p1 = Point2DMm::new(1.0, 2.0);
let p2 = size2(3.0, 4.0);
let result = p1 + p2;
assert_eq!(result, Point2DMm::new(4.0, 6.0));
}
#[test]
pub fn test_add_assign_size() {
let mut p1 = Point2DMm::new(1.0, 2.0);
p1 += size2(3.0, 4.0);
assert_eq!(p1, Point2DMm::new(4.0, 6.0));
}
#[test]
pub fn test_add_vec() {
let p1 = Point2DMm::new(1.0, 2.0);
let p2 = vec2(3.0, 4.0);
let result = p1 + p2;
assert_eq!(result, Point2DMm::new(4.0, 6.0));
}
#[test]
pub fn test_add_assign_vec() {
let mut p1 = Point2DMm::new(1.0, 2.0);
p1 += vec2(3.0, 4.0);
assert_eq!(p1, Point2DMm::new(4.0, 6.0));
}
#[test]
pub fn test_sub() {
let p1 = Point2DMm::new(1.0, 2.0);
let p2 = Point2DMm::new(3.0, 4.0);
let result = p1 - p2;
assert_eq!(result, Vector2D::<_, Mm>::new(-2.0, -2.0));
}
#[test]
pub fn test_sub_size() {
let p1 = Point2DMm::new(1.0, 2.0);
let p2 = size2(3.0, 4.0);
let result = p1 - p2;
assert_eq!(result, Point2DMm::new(-2.0, -2.0));
}
#[test]
pub fn test_sub_assign_size() {
let mut p1 = Point2DMm::new(1.0, 2.0);
p1 -= size2(3.0, 4.0);
assert_eq!(p1, Point2DMm::new(-2.0, -2.0));
}
#[test]
pub fn test_sub_vec() {
let p1 = Point2DMm::new(1.0, 2.0);
let p2 = vec2(3.0, 4.0);
let result = p1 - p2;
assert_eq!(result, Point2DMm::new(-2.0, -2.0));
}
#[test]
pub fn test_sub_assign_vec() {
let mut p1 = Point2DMm::new(1.0, 2.0);
p1 -= vec2(3.0, 4.0);
assert_eq!(p1, Point2DMm::new(-2.0, -2.0));
}
#[test]
pub fn test_mul_scalar() {
let p1: Point2D<f32> = Point2D::new(3.0, 5.0);
let result = p1 * 5.0;
assert_eq!(result, Point2D::new(15.0, 25.0));
}
#[test]
pub fn test_mul_assign_scalar() {
let mut p1 = Point2D::new(3.0, 5.0);
p1 *= 5.0;
assert_eq!(p1, Point2D::new(15.0, 25.0));
}
#[test]
pub fn test_mul_scale() {
let p1 = Point2DMm::new(1.0, 2.0);
let cm_per_mm: Scale<f32, Mm, Cm> = Scale::new(0.1);
let result = p1 * cm_per_mm;
assert_eq!(result, Point2DCm::new(0.1, 0.2));
}
#[test]
pub fn test_mul_assign_scale() {
let mut p1 = Point2DMm::new(1.0, 2.0);
let scale: Scale<f32, Mm, Mm> = Scale::new(0.1);
p1 *= scale;
assert_eq!(p1, Point2DMm::new(0.1, 0.2));
}
#[test]
pub fn test_div_scalar() {
let p1: Point2D<f32> = Point2D::new(15.0, 25.0);
let result = p1 / 5.0;
assert_eq!(result, Point2D::new(3.0, 5.0));
}
#[test]
pub fn test_div_assign_scalar() {
let mut p1: Point2D<f32> = Point2D::new(15.0, 25.0);
p1 /= 5.0;
assert_eq!(p1, Point2D::new(3.0, 5.0));
}
#[test]
pub fn test_div_scale() {
let p1 = Point2DCm::new(0.1, 0.2);
let cm_per_mm: Scale<f32, Mm, Cm> = Scale::new(0.1);
let result = p1 / cm_per_mm;
assert_eq!(result, Point2DMm::new(1.0, 2.0));
}
#[test]
pub fn test_div_assign_scale() {
let mut p1 = Point2DMm::new(0.1, 0.2);
let scale: Scale<f32, Mm, Mm> = Scale::new(0.1);
p1 /= scale;
assert_eq!(p1, Point2DMm::new(1.0, 2.0));
}
#[test]
pub fn test_point_debug_formatting() {
let n = 1.23456789;
let p1 = Point2D::new(n, -n);
let should_be = format!("({:.4}, {:.4})", n, -n);
let got = format!("{:.4?}", p1);
assert_eq!(got, should_be);
}
}
mod euclid {
use crate::default::{Point2D, Size2D};
use crate::point2;
#[test]
pub fn test_rem_euclid() {
let p = Point2D::new(7.0, -7.0);
let s = Size2D::new(4.0, -4.0);
assert_eq!(p.rem_euclid(&s), point2(3.0, 1.0));
assert_eq!((-p).rem_euclid(&s), point2(1.0, 3.0));
assert_eq!(p.rem_euclid(&-s), point2(3.0, 1.0));
}
#[test]
pub fn test_div_euclid() {
let p = Point2D::new(7.0, -7.0);
let s = Size2D::new(4.0, -4.0);
assert_eq!(p.div_euclid(&s), point2(1.0, 2.0));
assert_eq!((-p).div_euclid(&s), point2(-2.0, -1.0));
assert_eq!(p.div_euclid(&-s), point2(-1.0, -2.0));
}
}
}
#[cfg(test)]
mod point3d {
use crate::default;
use crate::default::Point3D;
use crate::{point2, point3};
#[cfg(feature = "mint")]
use mint;
#[test]
pub fn test_min() {
let p1 = Point3D::new(1.0, 3.0, 5.0);
let p2 = Point3D::new(2.0, 2.0, -1.0);
let result = p1.min(p2);
assert_eq!(result, Point3D::new(1.0, 2.0, -1.0));
}
#[test]
pub fn test_max() {
let p1 = Point3D::new(1.0, 3.0, 5.0);
let p2 = Point3D::new(2.0, 2.0, -1.0);
let result = p1.max(p2);
assert_eq!(result, Point3D::new(2.0, 3.0, 5.0));
}
#[test]
pub fn test_conv_vector() {
use crate::point3;
for i in 0..100 {
// We don't care about these values as long as they are not the same.
let x = i as f32 * 0.012345;
let y = i as f32 * 0.987654;
let z = x * y;
let p: Point3D<f32> = point3(x, y, z);
assert_eq!(p.to_vector().to_point(), p);
}
}
#[test]
pub fn test_swizzling() {
let p: default::Point3D<i32> = point3(1, 2, 3);
assert_eq!(p.xy(), point2(1, 2));
assert_eq!(p.xz(), point2(1, 3));
assert_eq!(p.yz(), point2(2, 3));
}
#[test]
pub fn test_distance_to() {
let p1 = Point3D::new(1.0, 2.0, 3.0);
let p2 = Point3D::new(2.0, 2.0, 3.0);
assert_eq!(p1.distance_to(p2), 1.0);
let p1 = Point3D::new(1.0, 2.0, 3.0);
let p2 = Point3D::new(1.0, 4.0, 3.0);
assert_eq!(p1.distance_to(p2), 2.0);
let p1 = Point3D::new(1.0, 2.0, 3.0);
let p2 = Point3D::new(1.0, 2.0, 6.0);
assert_eq!(p1.distance_to(p2), 3.0);
}
#[cfg(feature = "mint")]
#[test]
pub fn test_mint() {
let p1 = Point3D::new(1.0, 3.0, 5.0);
let pm: mint::Point3<_> = p1.into();
let p2 = Point3D::from(pm);
assert_eq!(p1, p2);
}
mod ops {
use crate::default::Point3D;
use crate::scale::Scale;
use crate::{size3, vec3, Vector3D};
pub enum Mm {}
pub enum Cm {}
pub type Point3DMm<T> = crate::Point3D<T, Mm>;
pub type Point3DCm<T> = crate::Point3D<T, Cm>;
#[test]
pub fn test_neg() {
assert_eq!(-Point3D::new(1.0, 2.0, 3.0), Point3D::new(-1.0, -2.0, -3.0));
assert_eq!(-Point3D::new(0.0, 0.0, 0.0), Point3D::new(-0.0, -0.0, -0.0));
assert_eq!(-Point3D::new(-1.0, -2.0, -3.0), Point3D::new(1.0, 2.0, 3.0));
}
#[test]
pub fn test_add_size() {
let p1 = Point3DMm::new(1.0, 2.0, 3.0);
let p2 = size3(4.0, 5.0, 6.0);
let result = p1 + p2;
assert_eq!(result, Point3DMm::new(5.0, 7.0, 9.0));
}
#[test]
pub fn test_add_assign_size() {
let mut p1 = Point3DMm::new(1.0, 2.0, 3.0);
p1 += size3(4.0, 5.0, 6.0);
assert_eq!(p1, Point3DMm::new(5.0, 7.0, 9.0));
}
#[test]
pub fn test_add_vec() {
let p1 = Point3DMm::new(1.0, 2.0, 3.0);
let p2 = vec3(4.0, 5.0, 6.0);
let result = p1 + p2;
assert_eq!(result, Point3DMm::new(5.0, 7.0, 9.0));
}
#[test]
pub fn test_add_assign_vec() {
let mut p1 = Point3DMm::new(1.0, 2.0, 3.0);
p1 += vec3(4.0, 5.0, 6.0);
assert_eq!(p1, Point3DMm::new(5.0, 7.0, 9.0));
}
#[test]
pub fn test_sub() {
let p1 = Point3DMm::new(1.0, 2.0, 3.0);
let p2 = Point3DMm::new(4.0, 5.0, 6.0);
let result = p1 - p2;
assert_eq!(result, Vector3D::<_, Mm>::new(-3.0, -3.0, -3.0));
}
#[test]
pub fn test_sub_size() {
let p1 = Point3DMm::new(1.0, 2.0, 3.0);
let p2 = size3(4.0, 5.0, 6.0);
let result = p1 - p2;
assert_eq!(result, Point3DMm::new(-3.0, -3.0, -3.0));
}
#[test]
pub fn test_sub_assign_size() {
let mut p1 = Point3DMm::new(1.0, 2.0, 3.0);
p1 -= size3(4.0, 5.0, 6.0);
assert_eq!(p1, Point3DMm::new(-3.0, -3.0, -3.0));
}
#[test]
pub fn test_sub_vec() {
let p1 = Point3DMm::new(1.0, 2.0, 3.0);
let p2 = vec3(4.0, 5.0, 6.0);
let result = p1 - p2;
assert_eq!(result, Point3DMm::new(-3.0, -3.0, -3.0));
}
#[test]
pub fn test_sub_assign_vec() {
let mut p1 = Point3DMm::new(1.0, 2.0, 3.0);
p1 -= vec3(4.0, 5.0, 6.0);
assert_eq!(p1, Point3DMm::new(-3.0, -3.0, -3.0));
}
#[test]
pub fn test_mul_scalar() {
let p1: Point3D<f32> = Point3D::new(3.0, 5.0, 7.0);
let result = p1 * 5.0;
assert_eq!(result, Point3D::new(15.0, 25.0, 35.0));
}
#[test]
pub fn test_mul_assign_scalar() {
let mut p1: Point3D<f32> = Point3D::new(3.0, 5.0, 7.0);
p1 *= 5.0;
assert_eq!(p1, Point3D::new(15.0, 25.0, 35.0));
}
#[test]
pub fn test_mul_scale() {
let p1 = Point3DMm::new(1.0, 2.0, 3.0);
let cm_per_mm: Scale<f32, Mm, Cm> = Scale::new(0.1);
let result = p1 * cm_per_mm;
assert_eq!(result, Point3DCm::new(0.1, 0.2, 0.3));
}
#[test]
pub fn test_mul_assign_scale() {
let mut p1 = Point3DMm::new(1.0, 2.0, 3.0);
let scale: Scale<f32, Mm, Mm> = Scale::new(0.1);
p1 *= scale;
assert_eq!(p1, Point3DMm::new(0.1, 0.2, 0.3));
}
#[test]
pub fn test_div_scalar() {
let p1: Point3D<f32> = Point3D::new(15.0, 25.0, 35.0);
let result = p1 / 5.0;
assert_eq!(result, Point3D::new(3.0, 5.0, 7.0));
}
#[test]
pub fn test_div_assign_scalar() {
let mut p1: Point3D<f32> = Point3D::new(15.0, 25.0, 35.0);
p1 /= 5.0;
assert_eq!(p1, Point3D::new(3.0, 5.0, 7.0));
}
#[test]
pub fn test_div_scale() {
let p1 = Point3DCm::new(0.1, 0.2, 0.3);
let cm_per_mm: Scale<f32, Mm, Cm> = Scale::new(0.1);
let result = p1 / cm_per_mm;
assert_eq!(result, Point3DMm::new(1.0, 2.0, 3.0));
}
#[test]
pub fn test_div_assign_scale() {
let mut p1 = Point3DMm::new(0.1, 0.2, 0.3);
let scale: Scale<f32, Mm, Mm> = Scale::new(0.1);
p1 /= scale;
assert_eq!(p1, Point3DMm::new(1.0, 2.0, 3.0));
}
}
mod euclid {
use crate::default::{Point3D, Size3D};
use crate::point3;
#[test]
pub fn test_rem_euclid() {
let p = Point3D::new(7.0, -7.0, 0.0);
let s = Size3D::new(4.0, -4.0, 12.0);
assert_eq!(p.rem_euclid(&s), point3(3.0, 1.0, 0.0));
assert_eq!((-p).rem_euclid(&s), point3(1.0, 3.0, 0.0));
assert_eq!(p.rem_euclid(&-s), point3(3.0, 1.0, 0.0));
}
#[test]
pub fn test_div_euclid() {
let p = Point3D::new(7.0, -7.0, 0.0);
let s = Size3D::new(4.0, -4.0, 12.0);
assert_eq!(p.div_euclid(&s), point3(1.0, 2.0, 0.0));
assert_eq!((-p).div_euclid(&s), point3(-2.0, -1.0, 0.0));
assert_eq!(p.div_euclid(&-s), point3(-1.0, -2.0, 0.0));
}
}
}