summaryrefslogtreecommitdiffstats
path: root/vendor/ordered-float/src
diff options
context:
space:
mode:
Diffstat (limited to '')
-rw-r--r--vendor/ordered-float/src/lib.rs2055
1 files changed, 2055 insertions, 0 deletions
diff --git a/vendor/ordered-float/src/lib.rs b/vendor/ordered-float/src/lib.rs
new file mode 100644
index 0000000..7532ce2
--- /dev/null
+++ b/vendor/ordered-float/src/lib.rs
@@ -0,0 +1,2055 @@
+#![no_std]
+#![cfg_attr(test, deny(warnings))]
+#![deny(missing_docs)]
+
+//! Wrappers for total order on Floats. See the [`OrderedFloat`] and [`NotNan`] docs for details.
+
+#[cfg(feature = "std")]
+extern crate std;
+#[cfg(feature = "std")]
+use std::error::Error;
+
+use core::borrow::Borrow;
+use core::cmp::Ordering;
+use core::convert::TryFrom;
+use core::fmt;
+use core::hash::{Hash, Hasher};
+use core::hint::unreachable_unchecked;
+use core::iter::{Product, Sum};
+use core::num::FpCategory;
+use core::ops::{
+ Add, AddAssign, Deref, DerefMut, Div, DivAssign, Mul, MulAssign, Neg, Rem, RemAssign, Sub,
+ SubAssign,
+};
+use core::str::FromStr;
+
+#[cfg(not(feature = "std"))]
+use num_traits::float::FloatCore as Float;
+#[cfg(feature = "std")]
+pub use num_traits::Float;
+use num_traits::{Bounded, FromPrimitive, Num, NumCast, One, Signed, ToPrimitive, Zero};
+
+// masks for the parts of the IEEE 754 float
+const SIGN_MASK: u64 = 0x8000000000000000u64;
+const EXP_MASK: u64 = 0x7ff0000000000000u64;
+const MAN_MASK: u64 = 0x000fffffffffffffu64;
+
+// canonical raw bit patterns (for hashing)
+const CANONICAL_NAN_BITS: u64 = 0x7ff8000000000000u64;
+
+#[inline(always)]
+fn canonicalize_signed_zero<T: Float>(x: T) -> T {
+ // -0.0 + 0.0 == +0.0 under IEEE754 roundTiesToEven rounding mode,
+ // which Rust guarantees. Thus by adding a positive zero we
+ // canonicalize signed zero without any branches in one instruction.
+ x + T::zero()
+}
+
+/// A wrapper around floats providing implementations of `Eq`, `Ord`, and `Hash`.
+///
+/// NaN is sorted as *greater* than all other values and *equal*
+/// to itself, in contradiction with the IEEE standard.
+///
+/// ```
+/// use ordered_float::OrderedFloat;
+/// use std::f32::NAN;
+///
+/// let mut v = [OrderedFloat(NAN), OrderedFloat(2.0), OrderedFloat(1.0)];
+/// v.sort();
+/// assert_eq!(v, [OrderedFloat(1.0), OrderedFloat(2.0), OrderedFloat(NAN)]);
+/// ```
+///
+/// Because `OrderedFloat` implements `Ord` and `Eq`, it can be used as a key in a `HashSet`,
+/// `HashMap`, `BTreeMap`, or `BTreeSet` (unlike the primitive `f32` or `f64` types):
+///
+/// ```
+/// # use ordered_float::OrderedFloat;
+/// # use std::collections::HashSet;
+/// # use std::f32::NAN;
+///
+/// let mut s: HashSet<OrderedFloat<f32>> = HashSet::new();
+/// s.insert(OrderedFloat(NAN));
+/// assert!(s.contains(&OrderedFloat(NAN)));
+/// ```
+#[derive(Debug, Default, Clone, Copy)]
+#[repr(transparent)]
+pub struct OrderedFloat<T>(pub T);
+
+impl<T: Float> OrderedFloat<T> {
+ /// Get the value out.
+ #[inline]
+ pub fn into_inner(self) -> T {
+ self.0
+ }
+}
+
+impl<T: Float> AsRef<T> for OrderedFloat<T> {
+ #[inline]
+ fn as_ref(&self) -> &T {
+ &self.0
+ }
+}
+
+impl<T: Float> AsMut<T> for OrderedFloat<T> {
+ #[inline]
+ fn as_mut(&mut self) -> &mut T {
+ &mut self.0
+ }
+}
+
+impl<'a, T: Float> From<&'a T> for &'a OrderedFloat<T> {
+ #[inline]
+ fn from(t: &'a T) -> &'a OrderedFloat<T> {
+ // Safety: OrderedFloat is #[repr(transparent)] and has no invalid values.
+ unsafe { &*(t as *const T as *const OrderedFloat<T>) }
+ }
+}
+
+impl<'a, T: Float> From<&'a mut T> for &'a mut OrderedFloat<T> {
+ #[inline]
+ fn from(t: &'a mut T) -> &'a mut OrderedFloat<T> {
+ // Safety: OrderedFloat is #[repr(transparent)] and has no invalid values.
+ unsafe { &mut *(t as *mut T as *mut OrderedFloat<T>) }
+ }
+}
+
+impl<T: Float> PartialOrd for OrderedFloat<T> {
+ #[inline]
+ fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+ Some(self.cmp(other))
+ }
+
+ #[inline]
+ fn lt(&self, other: &Self) -> bool {
+ !self.ge(other)
+ }
+
+ #[inline]
+ fn le(&self, other: &Self) -> bool {
+ other.ge(self)
+ }
+
+ #[inline]
+ fn gt(&self, other: &Self) -> bool {
+ !other.ge(self)
+ }
+
+ #[inline]
+ fn ge(&self, other: &Self) -> bool {
+ // We consider all NaNs equal, and NaN is the largest possible
+ // value. Thus if self is NaN we always return true. Otherwise
+ // self >= other is correct. If other is also not NaN it is trivially
+ // correct, and if it is we note that nothing can be greater or
+ // equal to NaN except NaN itself, which we already handled earlier.
+ self.0.is_nan() | (self.0 >= other.0)
+ }
+}
+
+impl<T: Float> Ord for OrderedFloat<T> {
+ #[inline]
+ #[allow(clippy::comparison_chain)]
+ fn cmp(&self, other: &Self) -> Ordering {
+ if self < other {
+ Ordering::Less
+ } else if self > other {
+ Ordering::Greater
+ } else {
+ Ordering::Equal
+ }
+ }
+}
+
+impl<T: Float> PartialEq for OrderedFloat<T> {
+ #[inline]
+ fn eq(&self, other: &OrderedFloat<T>) -> bool {
+ if self.0.is_nan() {
+ other.0.is_nan()
+ } else {
+ self.0 == other.0
+ }
+ }
+}
+
+impl<T: Float> PartialEq<T> for OrderedFloat<T> {
+ #[inline]
+ fn eq(&self, other: &T) -> bool {
+ self.0 == *other
+ }
+}
+
+impl<T: Float> Hash for OrderedFloat<T> {
+ fn hash<H: Hasher>(&self, state: &mut H) {
+ let bits = if self.is_nan() {
+ CANONICAL_NAN_BITS
+ } else {
+ raw_double_bits(&canonicalize_signed_zero(self.0))
+ };
+
+ bits.hash(state)
+ }
+}
+
+impl<T: Float + fmt::Display> fmt::Display for OrderedFloat<T> {
+ #[inline]
+ fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+ self.0.fmt(f)
+ }
+}
+
+impl From<OrderedFloat<f32>> for f32 {
+ #[inline]
+ fn from(f: OrderedFloat<f32>) -> f32 {
+ f.0
+ }
+}
+
+impl From<OrderedFloat<f64>> for f64 {
+ #[inline]
+ fn from(f: OrderedFloat<f64>) -> f64 {
+ f.0
+ }
+}
+
+impl<T: Float> From<T> for OrderedFloat<T> {
+ #[inline]
+ fn from(val: T) -> Self {
+ OrderedFloat(val)
+ }
+}
+
+impl<T: Float> Deref for OrderedFloat<T> {
+ type Target = T;
+
+ #[inline]
+ fn deref(&self) -> &Self::Target {
+ &self.0
+ }
+}
+
+impl<T: Float> DerefMut for OrderedFloat<T> {
+ #[inline]
+ fn deref_mut(&mut self) -> &mut Self::Target {
+ &mut self.0
+ }
+}
+
+impl<T: Float> Eq for OrderedFloat<T> {}
+
+macro_rules! impl_ordered_float_binop {
+ ($imp:ident, $method:ident, $assign_imp:ident, $assign_method:ident) => {
+ impl<T: $imp> $imp for OrderedFloat<T> {
+ type Output = OrderedFloat<T::Output>;
+
+ #[inline]
+ fn $method(self, other: Self) -> Self::Output {
+ OrderedFloat((self.0).$method(other.0))
+ }
+ }
+
+ impl<T: $imp> $imp<T> for OrderedFloat<T> {
+ type Output = OrderedFloat<T::Output>;
+
+ #[inline]
+ fn $method(self, other: T) -> Self::Output {
+ OrderedFloat((self.0).$method(other))
+ }
+ }
+
+ impl<'a, T> $imp<&'a T> for OrderedFloat<T>
+ where
+ T: $imp<&'a T>,
+ {
+ type Output = OrderedFloat<<T as $imp<&'a T>>::Output>;
+
+ #[inline]
+ fn $method(self, other: &'a T) -> Self::Output {
+ OrderedFloat((self.0).$method(other))
+ }
+ }
+
+ impl<'a, T> $imp<&'a Self> for OrderedFloat<T>
+ where
+ T: $imp<&'a T>,
+ {
+ type Output = OrderedFloat<<T as $imp<&'a T>>::Output>;
+
+ #[inline]
+ fn $method(self, other: &'a Self) -> Self::Output {
+ OrderedFloat((self.0).$method(&other.0))
+ }
+ }
+
+ impl<'a, T> $imp for &'a OrderedFloat<T>
+ where
+ &'a T: $imp,
+ {
+ type Output = OrderedFloat<<&'a T as $imp>::Output>;
+
+ #[inline]
+ fn $method(self, other: Self) -> Self::Output {
+ OrderedFloat((self.0).$method(&other.0))
+ }
+ }
+
+ impl<'a, T> $imp<OrderedFloat<T>> for &'a OrderedFloat<T>
+ where
+ &'a T: $imp<T>,
+ {
+ type Output = OrderedFloat<<&'a T as $imp<T>>::Output>;
+
+ #[inline]
+ fn $method(self, other: OrderedFloat<T>) -> Self::Output {
+ OrderedFloat((self.0).$method(other.0))
+ }
+ }
+
+ impl<'a, T> $imp<T> for &'a OrderedFloat<T>
+ where
+ &'a T: $imp<T>,
+ {
+ type Output = OrderedFloat<<&'a T as $imp<T>>::Output>;
+
+ #[inline]
+ fn $method(self, other: T) -> Self::Output {
+ OrderedFloat((self.0).$method(other))
+ }
+ }
+
+ impl<'a, T> $imp<&'a T> for &'a OrderedFloat<T>
+ where
+ &'a T: $imp,
+ {
+ type Output = OrderedFloat<<&'a T as $imp>::Output>;
+
+ #[inline]
+ fn $method(self, other: &'a T) -> Self::Output {
+ OrderedFloat((self.0).$method(other))
+ }
+ }
+
+ #[doc(hidden)] // Added accidentally; remove in next major version
+ impl<'a, T> $imp<&'a Self> for &'a OrderedFloat<T>
+ where
+ &'a T: $imp,
+ {
+ type Output = OrderedFloat<<&'a T as $imp>::Output>;
+
+ #[inline]
+ fn $method(self, other: &'a Self) -> Self::Output {
+ OrderedFloat((self.0).$method(&other.0))
+ }
+ }
+
+ impl<T: $assign_imp> $assign_imp<T> for OrderedFloat<T> {
+ #[inline]
+ fn $assign_method(&mut self, other: T) {
+ (self.0).$assign_method(other);
+ }
+ }
+
+ impl<'a, T: $assign_imp<&'a T>> $assign_imp<&'a T> for OrderedFloat<T> {
+ #[inline]
+ fn $assign_method(&mut self, other: &'a T) {
+ (self.0).$assign_method(other);
+ }
+ }
+
+ impl<T: $assign_imp> $assign_imp for OrderedFloat<T> {
+ #[inline]
+ fn $assign_method(&mut self, other: Self) {
+ (self.0).$assign_method(other.0);
+ }
+ }
+
+ impl<'a, T: $assign_imp<&'a T>> $assign_imp<&'a Self> for OrderedFloat<T> {
+ #[inline]
+ fn $assign_method(&mut self, other: &'a Self) {
+ (self.0).$assign_method(&other.0);
+ }
+ }
+ };
+}
+
+impl_ordered_float_binop! {Add, add, AddAssign, add_assign}
+impl_ordered_float_binop! {Sub, sub, SubAssign, sub_assign}
+impl_ordered_float_binop! {Mul, mul, MulAssign, mul_assign}
+impl_ordered_float_binop! {Div, div, DivAssign, div_assign}
+impl_ordered_float_binop! {Rem, rem, RemAssign, rem_assign}
+
+/// Adds a float directly.
+impl<T: Float + Sum> Sum for OrderedFloat<T> {
+ fn sum<I: Iterator<Item = OrderedFloat<T>>>(iter: I) -> Self {
+ OrderedFloat(iter.map(|v| v.0).sum())
+ }
+}
+
+impl<'a, T: Float + Sum + 'a> Sum<&'a OrderedFloat<T>> for OrderedFloat<T> {
+ #[inline]
+ fn sum<I: Iterator<Item = &'a OrderedFloat<T>>>(iter: I) -> Self {
+ iter.cloned().sum()
+ }
+}
+
+impl<T: Float + Product> Product for OrderedFloat<T> {
+ fn product<I: Iterator<Item = OrderedFloat<T>>>(iter: I) -> Self {
+ OrderedFloat(iter.map(|v| v.0).product())
+ }
+}
+
+impl<'a, T: Float + Product + 'a> Product<&'a OrderedFloat<T>> for OrderedFloat<T> {
+ #[inline]
+ fn product<I: Iterator<Item = &'a OrderedFloat<T>>>(iter: I) -> Self {
+ iter.cloned().product()
+ }
+}
+
+impl<T: Float + Signed> Signed for OrderedFloat<T> {
+ #[inline]
+ fn abs(&self) -> Self {
+ OrderedFloat(self.0.abs())
+ }
+
+ fn abs_sub(&self, other: &Self) -> Self {
+ OrderedFloat(Signed::abs_sub(&self.0, &other.0))
+ }
+
+ #[inline]
+ fn signum(&self) -> Self {
+ OrderedFloat(self.0.signum())
+ }
+ #[inline]
+ fn is_positive(&self) -> bool {
+ self.0.is_positive()
+ }
+ #[inline]
+ fn is_negative(&self) -> bool {
+ self.0.is_negative()
+ }
+}
+
+impl<T: Bounded> Bounded for OrderedFloat<T> {
+ #[inline]
+ fn min_value() -> Self {
+ OrderedFloat(T::min_value())
+ }
+
+ #[inline]
+ fn max_value() -> Self {
+ OrderedFloat(T::max_value())
+ }
+}
+
+impl<T: FromStr> FromStr for OrderedFloat<T> {
+ type Err = T::Err;
+
+ /// Convert a &str to `OrderedFloat`. Returns an error if the string fails to parse.
+ ///
+ /// ```
+ /// use ordered_float::OrderedFloat;
+ ///
+ /// assert!("-10".parse::<OrderedFloat<f32>>().is_ok());
+ /// assert!("abc".parse::<OrderedFloat<f32>>().is_err());
+ /// assert!("NaN".parse::<OrderedFloat<f32>>().is_ok());
+ /// ```
+ fn from_str(s: &str) -> Result<Self, Self::Err> {
+ T::from_str(s).map(OrderedFloat)
+ }
+}
+
+impl<T: Neg> Neg for OrderedFloat<T> {
+ type Output = OrderedFloat<T::Output>;
+
+ #[inline]
+ fn neg(self) -> Self::Output {
+ OrderedFloat(-self.0)
+ }
+}
+
+impl<'a, T> Neg for &'a OrderedFloat<T>
+where
+ &'a T: Neg,
+{
+ type Output = OrderedFloat<<&'a T as Neg>::Output>;
+
+ #[inline]
+ fn neg(self) -> Self::Output {
+ OrderedFloat(-(&self.0))
+ }
+}
+
+impl<T: Zero> Zero for OrderedFloat<T> {
+ #[inline]
+ fn zero() -> Self {
+ OrderedFloat(T::zero())
+ }
+
+ #[inline]
+ fn is_zero(&self) -> bool {
+ self.0.is_zero()
+ }
+}
+
+impl<T: One> One for OrderedFloat<T> {
+ #[inline]
+ fn one() -> Self {
+ OrderedFloat(T::one())
+ }
+}
+
+impl<T: NumCast> NumCast for OrderedFloat<T> {
+ #[inline]
+ fn from<F: ToPrimitive>(n: F) -> Option<Self> {
+ T::from(n).map(OrderedFloat)
+ }
+}
+
+impl<T: FromPrimitive> FromPrimitive for OrderedFloat<T> {
+ fn from_i64(n: i64) -> Option<Self> {
+ T::from_i64(n).map(OrderedFloat)
+ }
+ fn from_u64(n: u64) -> Option<Self> {
+ T::from_u64(n).map(OrderedFloat)
+ }
+ fn from_isize(n: isize) -> Option<Self> {
+ T::from_isize(n).map(OrderedFloat)
+ }
+ fn from_i8(n: i8) -> Option<Self> {
+ T::from_i8(n).map(OrderedFloat)
+ }
+ fn from_i16(n: i16) -> Option<Self> {
+ T::from_i16(n).map(OrderedFloat)
+ }
+ fn from_i32(n: i32) -> Option<Self> {
+ T::from_i32(n).map(OrderedFloat)
+ }
+ fn from_usize(n: usize) -> Option<Self> {
+ T::from_usize(n).map(OrderedFloat)
+ }
+ fn from_u8(n: u8) -> Option<Self> {
+ T::from_u8(n).map(OrderedFloat)
+ }
+ fn from_u16(n: u16) -> Option<Self> {
+ T::from_u16(n).map(OrderedFloat)
+ }
+ fn from_u32(n: u32) -> Option<Self> {
+ T::from_u32(n).map(OrderedFloat)
+ }
+ fn from_f32(n: f32) -> Option<Self> {
+ T::from_f32(n).map(OrderedFloat)
+ }
+ fn from_f64(n: f64) -> Option<Self> {
+ T::from_f64(n).map(OrderedFloat)
+ }
+}
+
+impl<T: ToPrimitive> ToPrimitive for OrderedFloat<T> {
+ fn to_i64(&self) -> Option<i64> {
+ self.0.to_i64()
+ }
+ fn to_u64(&self) -> Option<u64> {
+ self.0.to_u64()
+ }
+ fn to_isize(&self) -> Option<isize> {
+ self.0.to_isize()
+ }
+ fn to_i8(&self) -> Option<i8> {
+ self.0.to_i8()
+ }
+ fn to_i16(&self) -> Option<i16> {
+ self.0.to_i16()
+ }
+ fn to_i32(&self) -> Option<i32> {
+ self.0.to_i32()
+ }
+ fn to_usize(&self) -> Option<usize> {
+ self.0.to_usize()
+ }
+ fn to_u8(&self) -> Option<u8> {
+ self.0.to_u8()
+ }
+ fn to_u16(&self) -> Option<u16> {
+ self.0.to_u16()
+ }
+ fn to_u32(&self) -> Option<u32> {
+ self.0.to_u32()
+ }
+ fn to_f32(&self) -> Option<f32> {
+ self.0.to_f32()
+ }
+ fn to_f64(&self) -> Option<f64> {
+ self.0.to_f64()
+ }
+}
+
+impl<T: Float> num_traits::float::FloatCore for OrderedFloat<T> {
+ fn nan() -> Self {
+ OrderedFloat(T::nan())
+ }
+ fn infinity() -> Self {
+ OrderedFloat(T::infinity())
+ }
+ fn neg_infinity() -> Self {
+ OrderedFloat(T::neg_infinity())
+ }
+ fn neg_zero() -> Self {
+ OrderedFloat(T::neg_zero())
+ }
+ fn min_value() -> Self {
+ OrderedFloat(T::min_value())
+ }
+ fn min_positive_value() -> Self {
+ OrderedFloat(T::min_positive_value())
+ }
+ fn max_value() -> Self {
+ OrderedFloat(T::max_value())
+ }
+ fn is_nan(self) -> bool {
+ self.0.is_nan()
+ }
+ fn is_infinite(self) -> bool {
+ self.0.is_infinite()
+ }
+ fn is_finite(self) -> bool {
+ self.0.is_finite()
+ }
+ fn is_normal(self) -> bool {
+ self.0.is_normal()
+ }
+ fn classify(self) -> FpCategory {
+ self.0.classify()
+ }
+ fn floor(self) -> Self {
+ OrderedFloat(self.0.floor())
+ }
+ fn ceil(self) -> Self {
+ OrderedFloat(self.0.ceil())
+ }
+ fn round(self) -> Self {
+ OrderedFloat(self.0.round())
+ }
+ fn trunc(self) -> Self {
+ OrderedFloat(self.0.trunc())
+ }
+ fn fract(self) -> Self {
+ OrderedFloat(self.0.fract())
+ }
+ fn abs(self) -> Self {
+ OrderedFloat(self.0.abs())
+ }
+ fn signum(self) -> Self {
+ OrderedFloat(self.0.signum())
+ }
+ fn is_sign_positive(self) -> bool {
+ self.0.is_sign_positive()
+ }
+ fn is_sign_negative(self) -> bool {
+ self.0.is_sign_negative()
+ }
+ fn recip(self) -> Self {
+ OrderedFloat(self.0.recip())
+ }
+ fn powi(self, n: i32) -> Self {
+ OrderedFloat(self.0.powi(n))
+ }
+ fn integer_decode(self) -> (u64, i16, i8) {
+ self.0.integer_decode()
+ }
+ fn epsilon() -> Self {
+ OrderedFloat(T::epsilon())
+ }
+ fn to_degrees(self) -> Self {
+ OrderedFloat(self.0.to_degrees())
+ }
+ fn to_radians(self) -> Self {
+ OrderedFloat(self.0.to_radians())
+ }
+}
+
+#[cfg(feature = "std")]
+impl<T: Float> Float for OrderedFloat<T> {
+ fn nan() -> Self {
+ OrderedFloat(T::nan())
+ }
+ fn infinity() -> Self {
+ OrderedFloat(T::infinity())
+ }
+ fn neg_infinity() -> Self {
+ OrderedFloat(T::neg_infinity())
+ }
+ fn neg_zero() -> Self {
+ OrderedFloat(T::neg_zero())
+ }
+ fn min_value() -> Self {
+ OrderedFloat(T::min_value())
+ }
+ fn min_positive_value() -> Self {
+ OrderedFloat(T::min_positive_value())
+ }
+ fn max_value() -> Self {
+ OrderedFloat(T::max_value())
+ }
+ fn is_nan(self) -> bool {
+ self.0.is_nan()
+ }
+ fn is_infinite(self) -> bool {
+ self.0.is_infinite()
+ }
+ fn is_finite(self) -> bool {
+ self.0.is_finite()
+ }
+ fn is_normal(self) -> bool {
+ self.0.is_normal()
+ }
+ fn classify(self) -> FpCategory {
+ self.0.classify()
+ }
+ fn floor(self) -> Self {
+ OrderedFloat(self.0.floor())
+ }
+ fn ceil(self) -> Self {
+ OrderedFloat(self.0.ceil())
+ }
+ fn round(self) -> Self {
+ OrderedFloat(self.0.round())
+ }
+ fn trunc(self) -> Self {
+ OrderedFloat(self.0.trunc())
+ }
+ fn fract(self) -> Self {
+ OrderedFloat(self.0.fract())
+ }
+ fn abs(self) -> Self {
+ OrderedFloat(self.0.abs())
+ }
+ fn signum(self) -> Self {
+ OrderedFloat(self.0.signum())
+ }
+ fn is_sign_positive(self) -> bool {
+ self.0.is_sign_positive()
+ }
+ fn is_sign_negative(self) -> bool {
+ self.0.is_sign_negative()
+ }
+ fn mul_add(self, a: Self, b: Self) -> Self {
+ OrderedFloat(self.0.mul_add(a.0, b.0))
+ }
+ fn recip(self) -> Self {
+ OrderedFloat(self.0.recip())
+ }
+ fn powi(self, n: i32) -> Self {
+ OrderedFloat(self.0.powi(n))
+ }
+ fn powf(self, n: Self) -> Self {
+ OrderedFloat(self.0.powf(n.0))
+ }
+ fn sqrt(self) -> Self {
+ OrderedFloat(self.0.sqrt())
+ }
+ fn exp(self) -> Self {
+ OrderedFloat(self.0.exp())
+ }
+ fn exp2(self) -> Self {
+ OrderedFloat(self.0.exp2())
+ }
+ fn ln(self) -> Self {
+ OrderedFloat(self.0.ln())
+ }
+ fn log(self, base: Self) -> Self {
+ OrderedFloat(self.0.log(base.0))
+ }
+ fn log2(self) -> Self {
+ OrderedFloat(self.0.log2())
+ }
+ fn log10(self) -> Self {
+ OrderedFloat(self.0.log10())
+ }
+ fn max(self, other: Self) -> Self {
+ OrderedFloat(self.0.max(other.0))
+ }
+ fn min(self, other: Self) -> Self {
+ OrderedFloat(self.0.min(other.0))
+ }
+ fn abs_sub(self, other: Self) -> Self {
+ OrderedFloat(self.0.abs_sub(other.0))
+ }
+ fn cbrt(self) -> Self {
+ OrderedFloat(self.0.cbrt())
+ }
+ fn hypot(self, other: Self) -> Self {
+ OrderedFloat(self.0.hypot(other.0))
+ }
+ fn sin(self) -> Self {
+ OrderedFloat(self.0.sin())
+ }
+ fn cos(self) -> Self {
+ OrderedFloat(self.0.cos())
+ }
+ fn tan(self) -> Self {
+ OrderedFloat(self.0.tan())
+ }
+ fn asin(self) -> Self {
+ OrderedFloat(self.0.asin())
+ }
+ fn acos(self) -> Self {
+ OrderedFloat(self.0.acos())
+ }
+ fn atan(self) -> Self {
+ OrderedFloat(self.0.atan())
+ }
+ fn atan2(self, other: Self) -> Self {
+ OrderedFloat(self.0.atan2(other.0))
+ }
+ fn sin_cos(self) -> (Self, Self) {
+ let (a, b) = self.0.sin_cos();
+ (OrderedFloat(a), OrderedFloat(b))
+ }
+ fn exp_m1(self) -> Self {
+ OrderedFloat(self.0.exp_m1())
+ }
+ fn ln_1p(self) -> Self {
+ OrderedFloat(self.0.ln_1p())
+ }
+ fn sinh(self) -> Self {
+ OrderedFloat(self.0.sinh())
+ }
+ fn cosh(self) -> Self {
+ OrderedFloat(self.0.cosh())
+ }
+ fn tanh(self) -> Self {
+ OrderedFloat(self.0.tanh())
+ }
+ fn asinh(self) -> Self {
+ OrderedFloat(self.0.asinh())
+ }
+ fn acosh(self) -> Self {
+ OrderedFloat(self.0.acosh())
+ }
+ fn atanh(self) -> Self {
+ OrderedFloat(self.0.atanh())
+ }
+ fn integer_decode(self) -> (u64, i16, i8) {
+ self.0.integer_decode()
+ }
+ fn epsilon() -> Self {
+ OrderedFloat(T::epsilon())
+ }
+ fn to_degrees(self) -> Self {
+ OrderedFloat(self.0.to_degrees())
+ }
+ fn to_radians(self) -> Self {
+ OrderedFloat(self.0.to_radians())
+ }
+}
+
+impl<T: Float + Num> Num for OrderedFloat<T> {
+ type FromStrRadixErr = T::FromStrRadixErr;
+ fn from_str_radix(str: &str, radix: u32) -> Result<Self, Self::FromStrRadixErr> {
+ T::from_str_radix(str, radix).map(OrderedFloat)
+ }
+}
+
+/// A wrapper around floats providing an implementation of `Eq`, `Ord` and `Hash`.
+///
+/// A NaN value cannot be stored in this type.
+///
+/// ```
+/// use ordered_float::NotNan;
+///
+/// let mut v = [
+/// NotNan::new(2.0).unwrap(),
+/// NotNan::new(1.0).unwrap(),
+/// ];
+/// v.sort();
+/// assert_eq!(v, [1.0, 2.0]);
+/// ```
+///
+/// Because `NotNan` implements `Ord` and `Eq`, it can be used as a key in a `HashSet`,
+/// `HashMap`, `BTreeMap`, or `BTreeSet` (unlike the primitive `f32` or `f64` types):
+///
+/// ```
+/// # use ordered_float::NotNan;
+/// # use std::collections::HashSet;
+///
+/// let mut s: HashSet<NotNan<f32>> = HashSet::new();
+/// let key = NotNan::new(1.0).unwrap();
+/// s.insert(key);
+/// assert!(s.contains(&key));
+/// ```
+///
+/// Arithmetic on NotNan values will panic if it produces a NaN value:
+///
+/// ```should_panic
+/// # use ordered_float::NotNan;
+/// let a = NotNan::new(std::f32::INFINITY).unwrap();
+/// let b = NotNan::new(std::f32::NEG_INFINITY).unwrap();
+///
+/// // This will panic:
+/// let c = a + b;
+/// ```
+#[derive(PartialOrd, PartialEq, Debug, Default, Clone, Copy)]
+#[repr(transparent)]
+pub struct NotNan<T>(T);
+
+impl<T: Float> NotNan<T> {
+ /// Create a `NotNan` value.
+ ///
+ /// Returns `Err` if `val` is NaN
+ pub fn new(val: T) -> Result<Self, FloatIsNan> {
+ match val {
+ ref val if val.is_nan() => Err(FloatIsNan),
+ val => Ok(NotNan(val)),
+ }
+ }
+}
+
+impl<T> NotNan<T> {
+ /// Get the value out.
+ #[inline]
+ pub fn into_inner(self) -> T {
+ self.0
+ }
+
+ /// Create a `NotNan` value from a value that is guaranteed to not be NaN
+ ///
+ /// # Safety
+ ///
+ /// Behaviour is undefined if `val` is NaN
+ #[inline]
+ pub const unsafe fn new_unchecked(val: T) -> Self {
+ NotNan(val)
+ }
+
+ /// Create a `NotNan` value from a value that is guaranteed to not be NaN
+ ///
+ /// # Safety
+ ///
+ /// Behaviour is undefined if `val` is NaN
+ #[deprecated(
+ since = "2.5.0",
+ note = "Please use the new_unchecked function instead."
+ )]
+ #[inline]
+ pub const unsafe fn unchecked_new(val: T) -> Self {
+ Self::new_unchecked(val)
+ }
+}
+
+impl<T: Float> AsRef<T> for NotNan<T> {
+ #[inline]
+ fn as_ref(&self) -> &T {
+ &self.0
+ }
+}
+
+impl Borrow<f32> for NotNan<f32> {
+ #[inline]
+ fn borrow(&self) -> &f32 {
+ &self.0
+ }
+}
+
+impl Borrow<f64> for NotNan<f64> {
+ #[inline]
+ fn borrow(&self) -> &f64 {
+ &self.0
+ }
+}
+
+#[allow(clippy::derive_ord_xor_partial_ord)]
+impl<T: Float> Ord for NotNan<T> {
+ fn cmp(&self, other: &NotNan<T>) -> Ordering {
+ match self.partial_cmp(&other) {
+ Some(ord) => ord,
+ None => unsafe { unreachable_unchecked() },
+ }
+ }
+}
+
+#[allow(clippy::derive_hash_xor_eq)]
+impl<T: Float> Hash for NotNan<T> {
+ #[inline]
+ fn hash<H: Hasher>(&self, state: &mut H) {
+ let bits = raw_double_bits(&canonicalize_signed_zero(self.0));
+ bits.hash(state)
+ }
+}
+
+impl<T: Float + fmt::Display> fmt::Display for NotNan<T> {
+ #[inline]
+ fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+ self.0.fmt(f)
+ }
+}
+
+impl From<NotNan<f32>> for f32 {
+ #[inline]
+ fn from(value: NotNan<f32>) -> Self {
+ value.0
+ }
+}
+
+impl From<NotNan<f64>> for f64 {
+ #[inline]
+ fn from(value: NotNan<f64>) -> Self {
+ value.0
+ }
+}
+
+impl TryFrom<f32> for NotNan<f32> {
+ type Error = FloatIsNan;
+ #[inline]
+ fn try_from(v: f32) -> Result<Self, Self::Error> {
+ NotNan::new(v)
+ }
+}
+
+impl TryFrom<f64> for NotNan<f64> {
+ type Error = FloatIsNan;
+ #[inline]
+ fn try_from(v: f64) -> Result<Self, Self::Error> {
+ NotNan::new(v)
+ }
+}
+
+macro_rules! impl_from_int_primitive {
+ ($primitive:ty, $inner:ty) => {
+ impl From<$primitive> for NotNan<$inner> {
+ fn from(source: $primitive) -> Self {
+ // the primitives with which this macro will be called cannot hold a value that
+ // f64::from would convert to NaN, so this does not hurt invariants
+ NotNan(<$inner as From<$primitive>>::from(source))
+ }
+ }
+ };
+}
+
+impl_from_int_primitive!(i8, f64);
+impl_from_int_primitive!(i16, f64);
+impl_from_int_primitive!(i32, f64);
+impl_from_int_primitive!(u8, f64);
+impl_from_int_primitive!(u16, f64);
+impl_from_int_primitive!(u32, f64);
+
+impl_from_int_primitive!(i8, f32);
+impl_from_int_primitive!(i16, f32);
+impl_from_int_primitive!(u8, f32);
+impl_from_int_primitive!(u16, f32);
+
+impl From<NotNan<f32>> for NotNan<f64> {
+ #[inline]
+ fn from(v: NotNan<f32>) -> NotNan<f64> {
+ unsafe { NotNan::new_unchecked(v.0 as f64) }
+ }
+}
+
+impl<T: Float> Deref for NotNan<T> {
+ type Target = T;
+
+ #[inline]
+ fn deref(&self) -> &Self::Target {
+ &self.0
+ }
+}
+
+impl<T: Float + PartialEq> Eq for NotNan<T> {}
+
+impl<T: Float> PartialEq<T> for NotNan<T> {
+ #[inline]
+ fn eq(&self, other: &T) -> bool {
+ self.0 == *other
+ }
+}
+
+/// Adds a float directly.
+///
+/// Panics if the provided value is NaN or the computation results in NaN
+impl<T: Float> Add<T> for NotNan<T> {
+ type Output = Self;
+
+ #[inline]
+ fn add(self, other: T) -> Self {
+ NotNan::new(self.0 + other).expect("Addition resulted in NaN")
+ }
+}
+
+/// Adds a float directly.
+///
+/// Panics if the provided value is NaN.
+impl<T: Float + Sum> Sum for NotNan<T> {
+ fn sum<I: Iterator<Item = NotNan<T>>>(iter: I) -> Self {
+ NotNan::new(iter.map(|v| v.0).sum()).expect("Sum resulted in NaN")
+ }
+}
+
+impl<'a, T: Float + Sum + 'a> Sum<&'a NotNan<T>> for NotNan<T> {
+ #[inline]
+ fn sum<I: Iterator<Item = &'a NotNan<T>>>(iter: I) -> Self {
+ iter.cloned().sum()
+ }
+}
+
+/// Subtracts a float directly.
+///
+/// Panics if the provided value is NaN or the computation results in NaN
+impl<T: Float> Sub<T> for NotNan<T> {
+ type Output = Self;
+
+ #[inline]
+ fn sub(self, other: T) -> Self {
+ NotNan::new(self.0 - other).expect("Subtraction resulted in NaN")
+ }
+}
+
+/// Multiplies a float directly.
+///
+/// Panics if the provided value is NaN or the computation results in NaN
+impl<T: Float> Mul<T> for NotNan<T> {
+ type Output = Self;
+
+ #[inline]
+ fn mul(self, other: T) -> Self {
+ NotNan::new(self.0 * other).expect("Multiplication resulted in NaN")
+ }
+}
+
+impl<T: Float + Product> Product for NotNan<T> {
+ fn product<I: Iterator<Item = NotNan<T>>>(iter: I) -> Self {
+ NotNan::new(iter.map(|v| v.0).product()).expect("Product resulted in NaN")
+ }
+}
+
+impl<'a, T: Float + Product + 'a> Product<&'a NotNan<T>> for NotNan<T> {
+ #[inline]
+ fn product<I: Iterator<Item = &'a NotNan<T>>>(iter: I) -> Self {
+ iter.cloned().product()
+ }
+}
+
+/// Divides a float directly.
+///
+/// Panics if the provided value is NaN or the computation results in NaN
+impl<T: Float> Div<T> for NotNan<T> {
+ type Output = Self;
+
+ #[inline]
+ fn div(self, other: T) -> Self {
+ NotNan::new(self.0 / other).expect("Division resulted in NaN")
+ }
+}
+
+/// Calculates `%` with a float directly.
+///
+/// Panics if the provided value is NaN or the computation results in NaN
+impl<T: Float> Rem<T> for NotNan<T> {
+ type Output = Self;
+
+ #[inline]
+ fn rem(self, other: T) -> Self {
+ NotNan::new(self.0 % other).expect("Rem resulted in NaN")
+ }
+}
+
+macro_rules! impl_not_nan_binop {
+ ($imp:ident, $method:ident, $assign_imp:ident, $assign_method:ident) => {
+ impl<T: Float> $imp for NotNan<T> {
+ type Output = Self;
+
+ #[inline]
+ fn $method(self, other: Self) -> Self {
+ self.$method(other.0)
+ }
+ }
+
+ impl<T: Float> $imp<&T> for NotNan<T> {
+ type Output = NotNan<T>;
+
+ #[inline]
+ fn $method(self, other: &T) -> Self::Output {
+ self.$method(*other)
+ }
+ }
+
+ impl<T: Float> $imp<&Self> for NotNan<T> {
+ type Output = NotNan<T>;
+
+ #[inline]
+ fn $method(self, other: &Self) -> Self::Output {
+ self.$method(other.0)
+ }
+ }
+
+ impl<T: Float> $imp for &NotNan<T> {
+ type Output = NotNan<T>;
+
+ #[inline]
+ fn $method(self, other: Self) -> Self::Output {
+ (*self).$method(other.0)
+ }
+ }
+
+ impl<T: Float> $imp<NotNan<T>> for &NotNan<T> {
+ type Output = NotNan<T>;
+
+ #[inline]
+ fn $method(self, other: NotNan<T>) -> Self::Output {
+ (*self).$method(other.0)
+ }
+ }
+
+ impl<T: Float> $imp<T> for &NotNan<T> {
+ type Output = NotNan<T>;
+
+ #[inline]
+ fn $method(self, other: T) -> Self::Output {
+ (*self).$method(other)
+ }
+ }
+
+ impl<T: Float> $imp<&T> for &NotNan<T> {
+ type Output = NotNan<T>;
+
+ #[inline]
+ fn $method(self, other: &T) -> Self::Output {
+ (*self).$method(*other)
+ }
+ }
+
+ impl<T: Float + $assign_imp> $assign_imp<T> for NotNan<T> {
+ #[inline]
+ fn $assign_method(&mut self, other: T) {
+ *self = (*self).$method(other);
+ }
+ }
+
+ impl<T: Float + $assign_imp> $assign_imp<&T> for NotNan<T> {
+ #[inline]
+ fn $assign_method(&mut self, other: &T) {
+ *self = (*self).$method(*other);
+ }
+ }
+
+ impl<T: Float + $assign_imp> $assign_imp for NotNan<T> {
+ #[inline]
+ fn $assign_method(&mut self, other: Self) {
+ (*self).$assign_method(other.0);
+ }
+ }
+
+ impl<T: Float + $assign_imp> $assign_imp<&Self> for NotNan<T> {
+ #[inline]
+ fn $assign_method(&mut self, other: &Self) {
+ (*self).$assign_method(other.0);
+ }
+ }
+ };
+}
+
+impl_not_nan_binop! {Add, add, AddAssign, add_assign}
+impl_not_nan_binop! {Sub, sub, SubAssign, sub_assign}
+impl_not_nan_binop! {Mul, mul, MulAssign, mul_assign}
+impl_not_nan_binop! {Div, div, DivAssign, div_assign}
+impl_not_nan_binop! {Rem, rem, RemAssign, rem_assign}
+
+impl<T: Float> Neg for NotNan<T> {
+ type Output = Self;
+
+ #[inline]
+ fn neg(self) -> Self {
+ NotNan(-self.0)
+ }
+}
+
+impl<T: Float> Neg for &NotNan<T> {
+ type Output = NotNan<T>;
+
+ #[inline]
+ fn neg(self) -> Self::Output {
+ NotNan(-self.0)
+ }
+}
+
+/// An error indicating an attempt to construct NotNan from a NaN
+#[derive(Copy, Clone, PartialEq, Eq, Debug)]
+pub struct FloatIsNan;
+
+#[cfg(feature = "std")]
+impl Error for FloatIsNan {
+ fn description(&self) -> &str {
+ "NotNan constructed with NaN"
+ }
+}
+
+impl fmt::Display for FloatIsNan {
+ fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+ write!(f, "NotNan constructed with NaN")
+ }
+}
+
+#[cfg(feature = "std")]
+impl From<FloatIsNan> for std::io::Error {
+ #[inline]
+ fn from(e: FloatIsNan) -> std::io::Error {
+ std::io::Error::new(std::io::ErrorKind::InvalidInput, e)
+ }
+}
+
+#[inline]
+fn raw_double_bits<F: Float>(f: &F) -> u64 {
+ let (man, exp, sign) = f.integer_decode();
+ let exp_u64 = exp as u16 as u64;
+ let sign_u64 = if sign > 0 { 1u64 } else { 0u64 };
+ (man & MAN_MASK) | ((exp_u64 << 52) & EXP_MASK) | ((sign_u64 << 63) & SIGN_MASK)
+}
+
+impl<T: Float> Zero for NotNan<T> {
+ #[inline]
+ fn zero() -> Self {
+ NotNan(T::zero())
+ }
+
+ #[inline]
+ fn is_zero(&self) -> bool {
+ self.0.is_zero()
+ }
+}
+
+impl<T: Float> One for NotNan<T> {
+ #[inline]
+ fn one() -> Self {
+ NotNan(T::one())
+ }
+}
+
+impl<T: Float> Bounded for NotNan<T> {
+ #[inline]
+ fn min_value() -> Self {
+ NotNan(T::min_value())
+ }
+
+ #[inline]
+ fn max_value() -> Self {
+ NotNan(T::max_value())
+ }
+}
+
+impl<T: Float + FromStr> FromStr for NotNan<T> {
+ type Err = ParseNotNanError<T::Err>;
+
+ /// Convert a &str to `NotNan`. Returns an error if the string fails to parse,
+ /// or if the resulting value is NaN
+ ///
+ /// ```
+ /// use ordered_float::NotNan;
+ ///
+ /// assert!("-10".parse::<NotNan<f32>>().is_ok());
+ /// assert!("abc".parse::<NotNan<f32>>().is_err());
+ /// assert!("NaN".parse::<NotNan<f32>>().is_err());
+ /// ```
+ fn from_str(src: &str) -> Result<Self, Self::Err> {
+ src.parse()
+ .map_err(ParseNotNanError::ParseFloatError)
+ .and_then(|f| NotNan::new(f).map_err(|_| ParseNotNanError::IsNaN))
+ }
+}
+
+impl<T: Float + FromPrimitive> FromPrimitive for NotNan<T> {
+ fn from_i64(n: i64) -> Option<Self> {
+ T::from_i64(n).and_then(|n| NotNan::new(n).ok())
+ }
+ fn from_u64(n: u64) -> Option<Self> {
+ T::from_u64(n).and_then(|n| NotNan::new(n).ok())
+ }
+
+ fn from_isize(n: isize) -> Option<Self> {
+ T::from_isize(n).and_then(|n| NotNan::new(n).ok())
+ }
+ fn from_i8(n: i8) -> Option<Self> {
+ T::from_i8(n).and_then(|n| NotNan::new(n).ok())
+ }
+ fn from_i16(n: i16) -> Option<Self> {
+ T::from_i16(n).and_then(|n| NotNan::new(n).ok())
+ }
+ fn from_i32(n: i32) -> Option<Self> {
+ T::from_i32(n).and_then(|n| NotNan::new(n).ok())
+ }
+ fn from_usize(n: usize) -> Option<Self> {
+ T::from_usize(n).and_then(|n| NotNan::new(n).ok())
+ }
+ fn from_u8(n: u8) -> Option<Self> {
+ T::from_u8(n).and_then(|n| NotNan::new(n).ok())
+ }
+ fn from_u16(n: u16) -> Option<Self> {
+ T::from_u16(n).and_then(|n| NotNan::new(n).ok())
+ }
+ fn from_u32(n: u32) -> Option<Self> {
+ T::from_u32(n).and_then(|n| NotNan::new(n).ok())
+ }
+ fn from_f32(n: f32) -> Option<Self> {
+ T::from_f32(n).and_then(|n| NotNan::new(n).ok())
+ }
+ fn from_f64(n: f64) -> Option<Self> {
+ T::from_f64(n).and_then(|n| NotNan::new(n).ok())
+ }
+}
+
+impl<T: Float> ToPrimitive for NotNan<T> {
+ fn to_i64(&self) -> Option<i64> {
+ self.0.to_i64()
+ }
+ fn to_u64(&self) -> Option<u64> {
+ self.0.to_u64()
+ }
+
+ fn to_isize(&self) -> Option<isize> {
+ self.0.to_isize()
+ }
+ fn to_i8(&self) -> Option<i8> {
+ self.0.to_i8()
+ }
+ fn to_i16(&self) -> Option<i16> {
+ self.0.to_i16()
+ }
+ fn to_i32(&self) -> Option<i32> {
+ self.0.to_i32()
+ }
+ fn to_usize(&self) -> Option<usize> {
+ self.0.to_usize()
+ }
+ fn to_u8(&self) -> Option<u8> {
+ self.0.to_u8()
+ }
+ fn to_u16(&self) -> Option<u16> {
+ self.0.to_u16()
+ }
+ fn to_u32(&self) -> Option<u32> {
+ self.0.to_u32()
+ }
+ fn to_f32(&self) -> Option<f32> {
+ self.0.to_f32()
+ }
+ fn to_f64(&self) -> Option<f64> {
+ self.0.to_f64()
+ }
+}
+
+/// An error indicating a parse error from a string for `NotNan`.
+#[derive(Copy, Clone, PartialEq, Eq, Debug)]
+pub enum ParseNotNanError<E> {
+ /// A plain parse error from the underlying float type.
+ ParseFloatError(E),
+ /// The parsed float value resulted in a NaN.
+ IsNaN,
+}
+
+#[cfg(feature = "std")]
+impl<E: fmt::Debug + Error + 'static> Error for ParseNotNanError<E> {
+ fn description(&self) -> &str {
+ "Error parsing a not-NaN floating point value"
+ }
+
+ fn source(&self) -> Option<&(dyn Error + 'static)> {
+ match self {
+ ParseNotNanError::ParseFloatError(e) => Some(e),
+ ParseNotNanError::IsNaN => None,
+ }
+ }
+}
+
+impl<E: fmt::Display> fmt::Display for ParseNotNanError<E> {
+ fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+ match self {
+ ParseNotNanError::ParseFloatError(e) => write!(f, "Parse error: {}", e),
+ ParseNotNanError::IsNaN => write!(f, "NotNan parser encounter a NaN"),
+ }
+ }
+}
+
+impl<T: Float> Num for NotNan<T> {
+ type FromStrRadixErr = ParseNotNanError<T::FromStrRadixErr>;
+
+ fn from_str_radix(src: &str, radix: u32) -> Result<Self, Self::FromStrRadixErr> {
+ T::from_str_radix(src, radix)
+ .map_err(ParseNotNanError::ParseFloatError)
+ .and_then(|n| NotNan::new(n).map_err(|_| ParseNotNanError::IsNaN))
+ }
+}
+
+impl<T: Float + Signed> Signed for NotNan<T> {
+ #[inline]
+ fn abs(&self) -> Self {
+ NotNan(self.0.abs())
+ }
+
+ fn abs_sub(&self, other: &Self) -> Self {
+ NotNan::new(Signed::abs_sub(&self.0, &other.0)).expect("Subtraction resulted in NaN")
+ }
+
+ #[inline]
+ fn signum(&self) -> Self {
+ NotNan(self.0.signum())
+ }
+ #[inline]
+ fn is_positive(&self) -> bool {
+ self.0.is_positive()
+ }
+ #[inline]
+ fn is_negative(&self) -> bool {
+ self.0.is_negative()
+ }
+}
+
+impl<T: Float> NumCast for NotNan<T> {
+ fn from<F: ToPrimitive>(n: F) -> Option<Self> {
+ T::from(n).and_then(|n| NotNan::new(n).ok())
+ }
+}
+
+#[cfg(feature = "serde")]
+mod impl_serde {
+ extern crate serde;
+ use self::serde::de::{Error, Unexpected};
+ use self::serde::{Deserialize, Deserializer, Serialize, Serializer};
+ use super::{NotNan, OrderedFloat};
+ use core::f64;
+ #[cfg(not(feature = "std"))]
+ use num_traits::float::FloatCore as Float;
+ #[cfg(feature = "std")]
+ use num_traits::Float;
+
+ #[cfg(test)]
+ extern crate serde_test;
+ #[cfg(test)]
+ use self::serde_test::{assert_de_tokens_error, assert_tokens, Token};
+
+ impl<T: Float + Serialize> Serialize for OrderedFloat<T> {
+ #[inline]
+ fn serialize<S: Serializer>(&self, s: S) -> Result<S::Ok, S::Error> {
+ self.0.serialize(s)
+ }
+ }
+
+ impl<'de, T: Float + Deserialize<'de>> Deserialize<'de> for OrderedFloat<T> {
+ #[inline]
+ fn deserialize<D: Deserializer<'de>>(d: D) -> Result<Self, D::Error> {
+ T::deserialize(d).map(OrderedFloat)
+ }
+ }
+
+ impl<T: Float + Serialize> Serialize for NotNan<T> {
+ #[inline]
+ fn serialize<S: Serializer>(&self, s: S) -> Result<S::Ok, S::Error> {
+ self.0.serialize(s)
+ }
+ }
+
+ impl<'de, T: Float + Deserialize<'de>> Deserialize<'de> for NotNan<T> {
+ fn deserialize<D: Deserializer<'de>>(d: D) -> Result<Self, D::Error> {
+ let float = T::deserialize(d)?;
+ NotNan::new(float).map_err(|_| {
+ Error::invalid_value(Unexpected::Float(f64::NAN), &"float (but not NaN)")
+ })
+ }
+ }
+
+ #[test]
+ fn test_ordered_float() {
+ let float = OrderedFloat(1.0f64);
+ assert_tokens(&float, &[Token::F64(1.0)]);
+ }
+
+ #[test]
+ fn test_not_nan() {
+ let float = NotNan(1.0f64);
+ assert_tokens(&float, &[Token::F64(1.0)]);
+ }
+
+ #[test]
+ fn test_fail_on_nan() {
+ assert_de_tokens_error::<NotNan<f64>>(
+ &[Token::F64(f64::NAN)],
+ "invalid value: floating point `NaN`, expected float (but not NaN)",
+ );
+ }
+}
+
+#[cfg(feature = "rkyv")]
+mod impl_rkyv {
+ use super::{NotNan, OrderedFloat};
+ #[cfg(not(feature = "std"))]
+ use num_traits::float::FloatCore as Float;
+ #[cfg(feature = "std")]
+ use num_traits::Float;
+ #[cfg(test)]
+ use rkyv::{archived_root, ser::Serializer};
+ use rkyv::{from_archived, Archive, Deserialize, Fallible, Serialize};
+
+ #[cfg(test)]
+ type DefaultSerializer = rkyv::ser::serializers::CoreSerializer<16, 16>;
+ #[cfg(test)]
+ type DefaultDeserializer = rkyv::Infallible;
+
+ impl<T: Float + Archive> Archive for OrderedFloat<T> {
+ type Archived = OrderedFloat<T>;
+
+ type Resolver = ();
+
+ unsafe fn resolve(&self, _: usize, _: Self::Resolver, out: *mut Self::Archived) {
+ out.write(*self);
+ }
+ }
+
+ impl<T: Float + Serialize<S>, S: Fallible + ?Sized> Serialize<S> for OrderedFloat<T> {
+ fn serialize(&self, _: &mut S) -> Result<Self::Resolver, S::Error> {
+ Ok(())
+ }
+ }
+
+ impl<T: Float + Deserialize<T, D>, D: Fallible + ?Sized> Deserialize<OrderedFloat<T>, D>
+ for OrderedFloat<T>
+ {
+ fn deserialize(&self, _: &mut D) -> Result<OrderedFloat<T>, D::Error> {
+ Ok(from_archived!(*self))
+ }
+ }
+
+ impl<T: Float + Archive> Archive for NotNan<T> {
+ type Archived = NotNan<T>;
+
+ type Resolver = ();
+
+ unsafe fn resolve(&self, _: usize, _: Self::Resolver, out: *mut Self::Archived) {
+ out.write(*self);
+ }
+ }
+
+ impl<T: Float + Serialize<S>, S: Fallible + ?Sized> Serialize<S> for NotNan<T> {
+ fn serialize(&self, _: &mut S) -> Result<Self::Resolver, S::Error> {
+ Ok(())
+ }
+ }
+
+ impl<T: Float + Deserialize<T, D>, D: Fallible + ?Sized> Deserialize<NotNan<T>, D> for NotNan<T> {
+ fn deserialize(&self, _: &mut D) -> Result<NotNan<T>, D::Error> {
+ Ok(from_archived!(*self))
+ }
+ }
+
+ #[test]
+ fn test_ordered_float() {
+ let float = OrderedFloat(1.0f64);
+ let mut serializer = DefaultSerializer::default();
+ serializer
+ .serialize_value(&float)
+ .expect("failed to archive value");
+ let len = serializer.pos();
+ let buffer = serializer.into_serializer().into_inner();
+
+ let archived_value = unsafe { archived_root::<OrderedFloat<f64>>(&buffer[0..len]) };
+ assert_eq!(archived_value, &float);
+ let mut deserializer = DefaultDeserializer::default();
+ let deser_float: OrderedFloat<f64> = archived_value.deserialize(&mut deserializer).unwrap();
+ assert_eq!(deser_float, float);
+ }
+
+ #[test]
+ fn test_not_nan() {
+ let float = NotNan(1.0f64);
+ let mut serializer = DefaultSerializer::default();
+ serializer
+ .serialize_value(&float)
+ .expect("failed to archive value");
+ let len = serializer.pos();
+ let buffer = serializer.into_serializer().into_inner();
+
+ let archived_value = unsafe { archived_root::<NotNan<f64>>(&buffer[0..len]) };
+ assert_eq!(archived_value, &float);
+ let mut deserializer = DefaultDeserializer::default();
+ let deser_float: NotNan<f64> = archived_value.deserialize(&mut deserializer).unwrap();
+ assert_eq!(deser_float, float);
+ }
+}
+
+#[cfg(all(feature = "std", feature = "schemars"))]
+mod impl_schemars {
+ extern crate schemars;
+ use self::schemars::gen::SchemaGenerator;
+ use self::schemars::schema::{InstanceType, Schema, SchemaObject};
+ use super::{NotNan, OrderedFloat};
+
+ macro_rules! primitive_float_impl {
+ ($type:ty, $schema_name:literal) => {
+ impl schemars::JsonSchema for $type {
+ fn is_referenceable() -> bool {
+ false
+ }
+
+ fn schema_name() -> std::string::String {
+ std::string::String::from($schema_name)
+ }
+
+ fn json_schema(_: &mut SchemaGenerator) -> Schema {
+ SchemaObject {
+ instance_type: Some(InstanceType::Number.into()),
+ format: Some(std::string::String::from($schema_name)),
+ ..Default::default()
+ }
+ .into()
+ }
+ }
+ };
+ }
+
+ primitive_float_impl!(OrderedFloat<f32>, "float");
+ primitive_float_impl!(OrderedFloat<f64>, "double");
+ primitive_float_impl!(NotNan<f32>, "float");
+ primitive_float_impl!(NotNan<f64>, "double");
+
+ #[test]
+ fn schema_generation_does_not_panic_for_common_floats() {
+ {
+ let schema = schemars::gen::SchemaGenerator::default()
+ .into_root_schema_for::<OrderedFloat<f32>>();
+ assert_eq!(
+ schema.schema.instance_type,
+ Some(schemars::schema::SingleOrVec::Single(std::boxed::Box::new(
+ schemars::schema::InstanceType::Number
+ )))
+ );
+ assert_eq!(
+ schema.schema.metadata.unwrap().title.unwrap(),
+ std::string::String::from("float")
+ );
+ }
+ {
+ let schema = schemars::gen::SchemaGenerator::default()
+ .into_root_schema_for::<OrderedFloat<f64>>();
+ assert_eq!(
+ schema.schema.instance_type,
+ Some(schemars::schema::SingleOrVec::Single(std::boxed::Box::new(
+ schemars::schema::InstanceType::Number
+ )))
+ );
+ assert_eq!(
+ schema.schema.metadata.unwrap().title.unwrap(),
+ std::string::String::from("double")
+ );
+ }
+ {
+ let schema =
+ schemars::gen::SchemaGenerator::default().into_root_schema_for::<NotNan<f32>>();
+ assert_eq!(
+ schema.schema.instance_type,
+ Some(schemars::schema::SingleOrVec::Single(std::boxed::Box::new(
+ schemars::schema::InstanceType::Number
+ )))
+ );
+ assert_eq!(
+ schema.schema.metadata.unwrap().title.unwrap(),
+ std::string::String::from("float")
+ );
+ }
+ {
+ let schema =
+ schemars::gen::SchemaGenerator::default().into_root_schema_for::<NotNan<f64>>();
+ assert_eq!(
+ schema.schema.instance_type,
+ Some(schemars::schema::SingleOrVec::Single(std::boxed::Box::new(
+ schemars::schema::InstanceType::Number
+ )))
+ );
+ assert_eq!(
+ schema.schema.metadata.unwrap().title.unwrap(),
+ std::string::String::from("double")
+ );
+ }
+ }
+ #[test]
+ fn ordered_float_schema_match_primitive_schema() {
+ {
+ let of_schema = schemars::gen::SchemaGenerator::default()
+ .into_root_schema_for::<OrderedFloat<f32>>();
+ let prim_schema =
+ schemars::gen::SchemaGenerator::default().into_root_schema_for::<f32>();
+ assert_eq!(of_schema, prim_schema);
+ }
+ {
+ let of_schema = schemars::gen::SchemaGenerator::default()
+ .into_root_schema_for::<OrderedFloat<f64>>();
+ let prim_schema =
+ schemars::gen::SchemaGenerator::default().into_root_schema_for::<f64>();
+ assert_eq!(of_schema, prim_schema);
+ }
+ {
+ let of_schema =
+ schemars::gen::SchemaGenerator::default().into_root_schema_for::<NotNan<f32>>();
+ let prim_schema =
+ schemars::gen::SchemaGenerator::default().into_root_schema_for::<f32>();
+ assert_eq!(of_schema, prim_schema);
+ }
+ {
+ let of_schema =
+ schemars::gen::SchemaGenerator::default().into_root_schema_for::<NotNan<f64>>();
+ let prim_schema =
+ schemars::gen::SchemaGenerator::default().into_root_schema_for::<f64>();
+ assert_eq!(of_schema, prim_schema);
+ }
+ }
+}
+
+#[cfg(feature = "rand")]
+mod impl_rand {
+ use super::{NotNan, OrderedFloat};
+ use rand::distributions::uniform::*;
+ use rand::distributions::{Distribution, Open01, OpenClosed01, Standard};
+ use rand::Rng;
+
+ macro_rules! impl_distribution {
+ ($dist:ident, $($f:ty),+) => {
+ $(
+ impl Distribution<NotNan<$f>> for $dist {
+ fn sample<R: Rng + ?Sized>(&self, rng: &mut R) -> NotNan<$f> {
+ // 'rand' never generates NaN values in the Standard, Open01, or
+ // OpenClosed01 distributions. Using 'new_unchecked' is therefore
+ // safe.
+ unsafe { NotNan::new_unchecked(self.sample(rng)) }
+ }
+ }
+
+ impl Distribution<OrderedFloat<$f>> for $dist {
+ fn sample<R: Rng + ?Sized>(&self, rng: &mut R) -> OrderedFloat<$f> {
+ OrderedFloat(self.sample(rng))
+ }
+ }
+ )*
+ }
+ }
+
+ impl_distribution! { Standard, f32, f64 }
+ impl_distribution! { Open01, f32, f64 }
+ impl_distribution! { OpenClosed01, f32, f64 }
+
+ pub struct UniformNotNan<T>(UniformFloat<T>);
+ impl SampleUniform for NotNan<f32> {
+ type Sampler = UniformNotNan<f32>;
+ }
+ impl SampleUniform for NotNan<f64> {
+ type Sampler = UniformNotNan<f64>;
+ }
+
+ pub struct UniformOrdered<T>(UniformFloat<T>);
+ impl SampleUniform for OrderedFloat<f32> {
+ type Sampler = UniformOrdered<f32>;
+ }
+ impl SampleUniform for OrderedFloat<f64> {
+ type Sampler = UniformOrdered<f64>;
+ }
+
+ macro_rules! impl_uniform_sampler {
+ ($f:ty) => {
+ impl UniformSampler for UniformNotNan<$f> {
+ type X = NotNan<$f>;
+ fn new<B1, B2>(low: B1, high: B2) -> Self
+ where
+ B1: SampleBorrow<Self::X> + Sized,
+ B2: SampleBorrow<Self::X> + Sized,
+ {
+ UniformNotNan(UniformFloat::<$f>::new(low.borrow().0, high.borrow().0))
+ }
+ fn new_inclusive<B1, B2>(low: B1, high: B2) -> Self
+ where
+ B1: SampleBorrow<Self::X> + Sized,
+ B2: SampleBorrow<Self::X> + Sized,
+ {
+ UniformSampler::new(low, high)
+ }
+ fn sample<R: Rng + ?Sized>(&self, rng: &mut R) -> Self::X {
+ // UniformFloat.sample() will never return NaN.
+ unsafe { NotNan::new_unchecked(self.0.sample(rng)) }
+ }
+ }
+
+ impl UniformSampler for UniformOrdered<$f> {
+ type X = OrderedFloat<$f>;
+ fn new<B1, B2>(low: B1, high: B2) -> Self
+ where
+ B1: SampleBorrow<Self::X> + Sized,
+ B2: SampleBorrow<Self::X> + Sized,
+ {
+ UniformOrdered(UniformFloat::<$f>::new(low.borrow().0, high.borrow().0))
+ }
+ fn new_inclusive<B1, B2>(low: B1, high: B2) -> Self
+ where
+ B1: SampleBorrow<Self::X> + Sized,
+ B2: SampleBorrow<Self::X> + Sized,
+ {
+ UniformSampler::new(low, high)
+ }
+ fn sample<R: Rng + ?Sized>(&self, rng: &mut R) -> Self::X {
+ OrderedFloat(self.0.sample(rng))
+ }
+ }
+ };
+ }
+
+ impl_uniform_sampler! { f32 }
+ impl_uniform_sampler! { f64 }
+
+ #[cfg(all(test, feature = "randtest"))]
+ mod tests {
+ use super::*;
+
+ fn sample_fuzz<T>()
+ where
+ Standard: Distribution<NotNan<T>>,
+ Open01: Distribution<NotNan<T>>,
+ OpenClosed01: Distribution<NotNan<T>>,
+ Standard: Distribution<OrderedFloat<T>>,
+ Open01: Distribution<OrderedFloat<T>>,
+ OpenClosed01: Distribution<OrderedFloat<T>>,
+ T: crate::Float,
+ {
+ let mut rng = rand::thread_rng();
+ let f1: NotNan<T> = rng.sample(Standard);
+ let f2: NotNan<T> = rng.sample(Open01);
+ let f3: NotNan<T> = rng.sample(OpenClosed01);
+ let _: OrderedFloat<T> = rng.sample(Standard);
+ let _: OrderedFloat<T> = rng.sample(Open01);
+ let _: OrderedFloat<T> = rng.sample(OpenClosed01);
+ assert!(!f1.into_inner().is_nan());
+ assert!(!f2.into_inner().is_nan());
+ assert!(!f3.into_inner().is_nan());
+ }
+
+ #[test]
+ fn sampling_f32_does_not_panic() {
+ sample_fuzz::<f32>();
+ }
+
+ #[test]
+ fn sampling_f64_does_not_panic() {
+ sample_fuzz::<f64>();
+ }
+
+ #[test]
+ #[should_panic]
+ fn uniform_sampling_panic_on_infinity_notnan() {
+ let (low, high) = (
+ NotNan::new(0f64).unwrap(),
+ NotNan::new(core::f64::INFINITY).unwrap(),
+ );
+ let uniform = Uniform::new(low, high);
+ let _ = uniform.sample(&mut rand::thread_rng());
+ }
+
+ #[test]
+ #[should_panic]
+ fn uniform_sampling_panic_on_infinity_ordered() {
+ let (low, high) = (OrderedFloat(0f64), OrderedFloat(core::f64::INFINITY));
+ let uniform = Uniform::new(low, high);
+ let _ = uniform.sample(&mut rand::thread_rng());
+ }
+
+ #[test]
+ #[should_panic]
+ fn uniform_sampling_panic_on_nan_ordered() {
+ let (low, high) = (OrderedFloat(0f64), OrderedFloat(core::f64::NAN));
+ let uniform = Uniform::new(low, high);
+ let _ = uniform.sample(&mut rand::thread_rng());
+ }
+ }
+}
+
+#[cfg(feature = "proptest")]
+mod impl_proptest {
+ use super::{NotNan, OrderedFloat};
+ use proptest::arbitrary::{Arbitrary, StrategyFor};
+ use proptest::num::{f32, f64};
+ use proptest::strategy::{FilterMap, Map, Strategy};
+ use std::convert::TryFrom;
+
+ macro_rules! impl_arbitrary {
+ ($($f:ident),+) => {
+ $(
+ impl Arbitrary for NotNan<$f> {
+ type Strategy = FilterMap<StrategyFor<$f>, fn(_: $f) -> Option<NotNan<$f>>>;
+ type Parameters = <$f as Arbitrary>::Parameters;
+ fn arbitrary_with(params: Self::Parameters) -> Self::Strategy {
+ <$f>::arbitrary_with(params)
+ .prop_filter_map("filter nan values", |f| NotNan::try_from(f).ok())
+ }
+ }
+
+ impl Arbitrary for OrderedFloat<$f> {
+ type Strategy = Map<StrategyFor<$f>, fn(_: $f) -> OrderedFloat<$f>>;
+ type Parameters = <$f as Arbitrary>::Parameters;
+ fn arbitrary_with(params: Self::Parameters) -> Self::Strategy {
+ <$f>::arbitrary_with(params).prop_map(|f| OrderedFloat::from(f))
+ }
+ }
+ )*
+ }
+ }
+ impl_arbitrary! { f32, f64 }
+}
+
+#[cfg(feature = "arbitrary")]
+mod impl_arbitrary {
+ use super::{FloatIsNan, NotNan, OrderedFloat};
+ use arbitrary::{Arbitrary, Unstructured};
+ use num_traits::FromPrimitive;
+
+ macro_rules! impl_arbitrary {
+ ($($f:ident),+) => {
+ $(
+ impl<'a> Arbitrary<'a> for NotNan<$f> {
+ fn arbitrary(u: &mut Unstructured<'a>) -> arbitrary::Result<Self> {
+ let float: $f = u.arbitrary()?;
+ match NotNan::new(float) {
+ Ok(notnan_value) => Ok(notnan_value),
+ Err(FloatIsNan) => {
+ // If our arbitrary float input was a NaN (encoded by exponent = max
+ // value), then replace it with a finite float, reusing the mantissa
+ // bits.
+ //
+ // This means the output is not uniformly distributed among all
+ // possible float values, but Arbitrary makes no promise that that
+ // is true.
+ //
+ // An alternative implementation would be to return an
+ // `arbitrary::Error`, but that is not as useful since it forces the
+ // caller to retry with new random/fuzzed data; and the precendent of
+ // `arbitrary`'s built-in implementations is to prefer the approach of
+ // mangling the input bits to fit.
+
+ let (mantissa, _exponent, sign) =
+ num_traits::Float::integer_decode(float);
+ let revised_float = <$f>::from_i64(
+ i64::from(sign) * mantissa as i64
+ ).unwrap();
+
+ // If this unwrap() fails, then there is a bug in the above code.
+ Ok(NotNan::new(revised_float).unwrap())
+ }
+ }
+ }
+
+ fn size_hint(depth: usize) -> (usize, Option<usize>) {
+ <$f as Arbitrary>::size_hint(depth)
+ }
+ }
+
+ impl<'a> Arbitrary<'a> for OrderedFloat<$f> {
+ fn arbitrary(u: &mut Unstructured<'a>) -> arbitrary::Result<Self> {
+ let float: $f = u.arbitrary()?;
+ Ok(OrderedFloat::from(float))
+ }
+
+ fn size_hint(depth: usize) -> (usize, Option<usize>) {
+ <$f as Arbitrary>::size_hint(depth)
+ }
+ }
+ )*
+ }
+ }
+ impl_arbitrary! { f32, f64 }
+}