diff options
| author | Oli Scherer <github35764891676564198441@oli-obk.de> | 2025-08-30 07:07:41 +0000 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-08-30 07:07:41 +0000 |
| commit | c8d20ceb1be87bc35035328aa1925cb3d95ad10b (patch) | |
| tree | f3ccfe3fdc0af4ed3cda4e0352ba5743c1c10536 /library/core/src | |
| parent | b00c449ef7330353c85e4b1bb4cd30cf5a10f919 (diff) | |
| parent | d269d234e033e2b52722ea78767fe10898184993 (diff) | |
| download | rust-c8d20ceb1be87bc35035328aa1925cb3d95ad10b.tar.gz rust-c8d20ceb1be87bc35035328aa1925cb3d95ad10b.zip | |
Merge pull request #4548 from rust-lang/rustup-2025-08-30
Automatic Rustup
Diffstat (limited to 'library/core/src')
| -rw-r--r-- | library/core/src/hint.rs | 56 | ||||
| -rw-r--r-- | library/core/src/iter/traits/accum.rs | 58 | ||||
| -rw-r--r-- | library/core/src/lib.rs | 2 | ||||
| -rw-r--r-- | library/core/src/marker.rs | 2 | ||||
| -rw-r--r-- | library/core/src/num/f128.rs | 23 | ||||
| -rw-r--r-- | library/core/src/num/f16.rs | 20 | ||||
| -rw-r--r-- | library/core/src/num/f32.rs | 20 | ||||
| -rw-r--r-- | library/core/src/num/f64.rs | 26 | ||||
| -rw-r--r-- | library/core/src/num/niche_types.rs | 15 | ||||
| -rw-r--r-- | library/core/src/option.rs | 36 | ||||
| -rw-r--r-- | library/core/src/panic/location.rs | 4 | ||||
| -rw-r--r-- | library/core/src/ptr/metadata.rs | 3 | ||||
| -rw-r--r-- | library/core/src/str/mod.rs | 10 | ||||
| -rw-r--r-- | library/core/src/sync/atomic.rs | 21 | ||||
| -rw-r--r-- | library/core/src/wtf8.rs | 597 |
15 files changed, 827 insertions, 66 deletions
diff --git a/library/core/src/hint.rs b/library/core/src/hint.rs index c72eeb9a9c9..535c5a8e77b 100644 --- a/library/core/src/hint.rs +++ b/library/core/src/hint.rs @@ -267,39 +267,29 @@ pub const unsafe fn assert_unchecked(cond: bool) { #[inline(always)] #[stable(feature = "renamed_spin_loop", since = "1.49.0")] pub fn spin_loop() { - #[cfg(target_arch = "x86")] - { - // SAFETY: the `cfg` attr ensures that we only execute this on x86 targets. - unsafe { crate::arch::x86::_mm_pause() }; - } - - #[cfg(target_arch = "x86_64")] - { - // SAFETY: the `cfg` attr ensures that we only execute this on x86_64 targets. - unsafe { crate::arch::x86_64::_mm_pause() }; - } - - #[cfg(target_arch = "riscv32")] - { - crate::arch::riscv32::pause(); - } - - #[cfg(target_arch = "riscv64")] - { - crate::arch::riscv64::pause(); - } - - #[cfg(any(target_arch = "aarch64", target_arch = "arm64ec"))] - { - // SAFETY: the `cfg` attr ensures that we only execute this on aarch64 targets. - unsafe { crate::arch::aarch64::__isb(crate::arch::aarch64::SY) }; - } - - #[cfg(all(target_arch = "arm", target_feature = "v6"))] - { - // SAFETY: the `cfg` attr ensures that we only execute this on arm targets - // with support for the v6 feature. - unsafe { crate::arch::arm::__yield() }; + crate::cfg_select! { + target_arch = "x86" => { + // SAFETY: the `cfg` attr ensures that we only execute this on x86 targets. + unsafe { crate::arch::x86::_mm_pause() } + } + target_arch = "x86_64" => { + // SAFETY: the `cfg` attr ensures that we only execute this on x86_64 targets. + unsafe { crate::arch::x86_64::_mm_pause() } + } + target_arch = "riscv32" => crate::arch::riscv32::pause(), + target_arch = "riscv64" => crate::arch::riscv64::pause(), + any(target_arch = "aarch64", target_arch = "arm64ec") => { + // SAFETY: the `cfg` attr ensures that we only execute this on aarch64 targets. + unsafe { crate::arch::aarch64::__isb(crate::arch::aarch64::SY) } + } + all(target_arch = "arm", target_feature = "v6") => { + // SAFETY: the `cfg` attr ensures that we only execute this on arm targets + // with support for the v6 feature. + unsafe { crate::arch::arm::__yield() } + } + target_arch = "loongarch32" => crate::arch::loongarch32::ibar::<0>(), + target_arch = "loongarch64" => crate::arch::loongarch64::ibar::<0>(), + _ => { /* do nothing */ } } } diff --git a/library/core/src/iter/traits/accum.rs b/library/core/src/iter/traits/accum.rs index 12e2b8b393a..73122369b41 100644 --- a/library/core/src/iter/traits/accum.rs +++ b/library/core/src/iter/traits/accum.rs @@ -1,5 +1,5 @@ use crate::iter; -use crate::num::Wrapping; +use crate::num::{Saturating, Wrapping}; /// Trait to represent types that can be created by summing up an iterator. /// @@ -98,6 +98,61 @@ macro_rules! integer_sum_product { ); } +macro_rules! saturating_integer_sum_product { + (@impls $zero:expr, $one:expr, $doc:expr, #[$attr:meta], $($a:ty)*) => ($( + #[$attr] + #[doc = $doc] + impl Sum for $a { + fn sum<I: Iterator<Item=Self>>(iter: I) -> Self { + iter.fold( + $zero, + |a, b| a + b, + ) + } + } + + #[$attr] + #[doc = $doc] + impl Product for $a { + fn product<I: Iterator<Item=Self>>(iter: I) -> Self { + iter.fold( + $one, + |a, b| a * b, + ) + } + } + + #[$attr] + #[doc = $doc] + impl<'a> Sum<&'a $a> for $a { + fn sum<I: Iterator<Item=&'a Self>>(iter: I) -> Self { + iter.fold( + $zero, + |a, b| a + b, + ) + } + } + + #[$attr] + #[doc = $doc] + impl<'a> Product<&'a $a> for $a { + fn product<I: Iterator<Item=&'a Self>>(iter: I) -> Self { + iter.fold( + $one, + |a, b| a * b, + ) + } + } + )*); + ($($a:ty)*) => ( + saturating_integer_sum_product!(@impls Saturating(0), Saturating(1), + "The short-circuiting behavior of this implementation is unspecified. If you care about \ + short-circuiting, use [`Iterator::fold`] directly.", + #[stable(feature = "saturating_iter_arith", since = "CURRENT_RUSTC_VERSION")], + $(Saturating<$a>)*); + ); +} + macro_rules! float_sum_product { ($($a:ident)*) => ($( #[stable(feature = "iter_arith_traits", since = "1.12.0")] @@ -147,6 +202,7 @@ macro_rules! float_sum_product { } integer_sum_product! { i8 i16 i32 i64 i128 isize u8 u16 u32 u64 u128 usize } +saturating_integer_sum_product! { u8 u16 u32 u64 u128 usize } float_sum_product! { f32 f64 } #[stable(feature = "iter_arith_traits_result", since = "1.16.0")] diff --git a/library/core/src/lib.rs b/library/core/src/lib.rs index f69d6a98592..82506379cac 100644 --- a/library/core/src/lib.rs +++ b/library/core/src/lib.rs @@ -363,6 +363,8 @@ pub mod slice; pub mod str; pub mod time; +pub mod wtf8; + pub mod unicode; /* Async */ diff --git a/library/core/src/marker.rs b/library/core/src/marker.rs index 8ad58599c68..73aad27afef 100644 --- a/library/core/src/marker.rs +++ b/library/core/src/marker.rs @@ -1049,7 +1049,7 @@ marker_impls! { /// A marker for types that can be dropped. /// -/// This should be used for `~const` bounds, +/// This should be used for `[const]` bounds, /// as non-const bounds will always hold for every type. #[unstable(feature = "const_destruct", issue = "133214")] #[rustc_const_unstable(feature = "const_destruct", issue = "133214")] diff --git a/library/core/src/num/f128.rs b/library/core/src/num/f128.rs index 69e6c100e76..4282b1af9f2 100644 --- a/library/core/src/num/f128.rs +++ b/library/core/src/num/f128.rs @@ -630,6 +630,13 @@ impl f128 { /// Converts radians to degrees. /// + /// # Unspecified precision + /// + /// The precision of this function is non-deterministic. This means it varies by platform, + /// Rust version, and can even differ within the same execution from one invocation to the next. + /// + /// # Examples + /// /// ``` /// #![feature(f128)] /// # // FIXME(f16_f128): remove when `eqtf2` is available @@ -645,13 +652,22 @@ impl f128 { #[unstable(feature = "f128", issue = "116909")] #[must_use = "this returns the result of the operation, without modifying the original"] pub const fn to_degrees(self) -> Self { - // Use a literal for better precision. - const PIS_IN_180: f128 = 57.2957795130823208767981548141051703324054724665643215491602_f128; + // The division here is correctly rounded with respect to the true value of 180/π. + // Although π is irrational and already rounded, the double rounding happens + // to produce correct result for f128. + const PIS_IN_180: f128 = 180.0 / consts::PI; self * PIS_IN_180 } /// Converts degrees to radians. /// + /// # Unspecified precision + /// + /// The precision of this function is non-deterministic. This means it varies by platform, + /// Rust version, and can even differ within the same execution from one invocation to the next. + /// + /// # Examples + /// /// ``` /// #![feature(f128)] /// # // FIXME(f16_f128): remove when `eqtf2` is available @@ -668,7 +684,8 @@ impl f128 { #[unstable(feature = "f128", issue = "116909")] #[must_use = "this returns the result of the operation, without modifying the original"] pub const fn to_radians(self) -> f128 { - // Use a literal for better precision. + // Use a literal to avoid double rounding, consts::PI is already rounded, + // and dividing would round again. const RADS_PER_DEG: f128 = 0.0174532925199432957692369076848861271344287188854172545609719_f128; self * RADS_PER_DEG diff --git a/library/core/src/num/f16.rs b/library/core/src/num/f16.rs index b66cef03d20..23a8661c14b 100644 --- a/library/core/src/num/f16.rs +++ b/library/core/src/num/f16.rs @@ -625,6 +625,13 @@ impl f16 { /// Converts radians to degrees. /// + /// # Unspecified precision + /// + /// The precision of this function is non-deterministic. This means it varies by platform, + /// Rust version, and can even differ within the same execution from one invocation to the next. + /// + /// # Examples + /// /// ``` /// #![feature(f16)] /// # // FIXME(f16_f128): extendhfsf2, truncsfhf2, __gnu_h2f_ieee, __gnu_f2h_ieee missing for many platforms @@ -640,13 +647,21 @@ impl f16 { #[unstable(feature = "f16", issue = "116909")] #[must_use = "this returns the result of the operation, without modifying the original"] pub const fn to_degrees(self) -> Self { - // Use a literal for better precision. + // Use a literal to avoid double rounding, consts::PI is already rounded, + // and dividing would round again. const PIS_IN_180: f16 = 57.2957795130823208767981548141051703_f16; self * PIS_IN_180 } /// Converts degrees to radians. /// + /// # Unspecified precision + /// + /// The precision of this function is non-deterministic. This means it varies by platform, + /// Rust version, and can even differ within the same execution from one invocation to the next. + /// + /// # Examples + /// /// ``` /// #![feature(f16)] /// # // FIXME(f16_f128): extendhfsf2, truncsfhf2, __gnu_h2f_ieee, __gnu_f2h_ieee missing for many platforms @@ -663,7 +678,8 @@ impl f16 { #[unstable(feature = "f16", issue = "116909")] #[must_use = "this returns the result of the operation, without modifying the original"] pub const fn to_radians(self) -> f16 { - // Use a literal for better precision. + // Use a literal to avoid double rounding, consts::PI is already rounded, + // and dividing would round again. const RADS_PER_DEG: f16 = 0.017453292519943295769236907684886_f16; self * RADS_PER_DEG } diff --git a/library/core/src/num/f32.rs b/library/core/src/num/f32.rs index f8344da79ad..da08bfc5950 100644 --- a/library/core/src/num/f32.rs +++ b/library/core/src/num/f32.rs @@ -839,6 +839,13 @@ impl f32 { /// Converts radians to degrees. /// + /// # Unspecified precision + /// + /// The precision of this function is non-deterministic. This means it varies by platform, + /// Rust version, and can even differ within the same execution from one invocation to the next. + /// + /// # Examples + /// /// ``` /// let angle = std::f32::consts::PI; /// @@ -852,13 +859,21 @@ impl f32 { #[rustc_const_stable(feature = "const_float_methods", since = "1.85.0")] #[inline] pub const fn to_degrees(self) -> f32 { - // Use a constant for better precision. + // Use a literal to avoid double rounding, consts::PI is already rounded, + // and dividing would round again. const PIS_IN_180: f32 = 57.2957795130823208767981548141051703_f32; self * PIS_IN_180 } /// Converts degrees to radians. /// + /// # Unspecified precision + /// + /// The precision of this function is non-deterministic. This means it varies by platform, + /// Rust version, and can even differ within the same execution from one invocation to the next. + /// + /// # Examples + /// /// ``` /// let angle = 180.0f32; /// @@ -872,6 +887,9 @@ impl f32 { #[rustc_const_stable(feature = "const_float_methods", since = "1.85.0")] #[inline] pub const fn to_radians(self) -> f32 { + // The division here is correctly rounded with respect to the true value of π/180. + // Although π is irrational and already rounded, the double rounding happens + // to produce correct result for f32. const RADS_PER_DEG: f32 = consts::PI / 180.0; self * RADS_PER_DEG } diff --git a/library/core/src/num/f64.rs b/library/core/src/num/f64.rs index 93da63c896e..dc0977cb147 100644 --- a/library/core/src/num/f64.rs +++ b/library/core/src/num/f64.rs @@ -856,6 +856,13 @@ impl f64 { /// Converts radians to degrees. /// + /// # Unspecified precision + /// + /// The precision of this function is non-deterministic. This means it varies by platform, + /// Rust version, and can even differ within the same execution from one invocation to the next. + /// + /// # Examples + /// /// ``` /// let angle = std::f64::consts::PI; /// @@ -869,14 +876,22 @@ impl f64 { #[rustc_const_stable(feature = "const_float_methods", since = "1.85.0")] #[inline] pub const fn to_degrees(self) -> f64 { - // The division here is correctly rounded with respect to the true - // value of 180/π. (This differs from f32, where a constant must be - // used to ensure a correctly rounded result.) - self * (180.0f64 / consts::PI) + // The division here is correctly rounded with respect to the true value of 180/π. + // Although π is irrational and already rounded, the double rounding happens + // to produce correct result for f64. + const PIS_IN_180: f64 = 180.0 / consts::PI; + self * PIS_IN_180 } /// Converts degrees to radians. /// + /// # Unspecified precision + /// + /// The precision of this function is non-deterministic. This means it varies by platform, + /// Rust version, and can even differ within the same execution from one invocation to the next. + /// + /// # Examples + /// /// ``` /// let angle = 180.0_f64; /// @@ -890,6 +905,9 @@ impl f64 { #[rustc_const_stable(feature = "const_float_methods", since = "1.85.0")] #[inline] pub const fn to_radians(self) -> f64 { + // The division here is correctly rounded with respect to the true value of π/180. + // Although π is irrational and already rounded, the double rounding happens + // to produce correct result for f64. const RADS_PER_DEG: f64 = consts::PI / 180.0; self * RADS_PER_DEG } diff --git a/library/core/src/num/niche_types.rs b/library/core/src/num/niche_types.rs index d57b1d433e5..610d9d8cf92 100644 --- a/library/core/src/num/niche_types.rs +++ b/library/core/src/num/niche_types.rs @@ -178,3 +178,18 @@ impl NotAllOnesHelper for u64 { impl NotAllOnesHelper for i64 { type Type = I64NotAllOnes; } + +define_valid_range_type! { + pub struct CodePointInner(u32 as u32 in 0..=0x10ffff); +} + +impl CodePointInner { + pub const ZERO: Self = CodePointInner::new(0).unwrap(); +} + +impl Default for CodePointInner { + #[inline] + fn default() -> Self { + Self::ZERO + } +} diff --git a/library/core/src/option.rs b/library/core/src/option.rs index e83e77344cf..fa09409b6da 100644 --- a/library/core/src/option.rs +++ b/library/core/src/option.rs @@ -1961,6 +1961,42 @@ impl<T> Option<T> { _ => None, } } + + /// Reduces two options into one, using the provided function if both are `Some`. + /// + /// If `self` is `Some(s)` and `other` is `Some(o)`, this method returns `Some(f(s, o))`. + /// Otherwise, if only one of `self` and `other` is `Some`, that one is returned. + /// If both `self` and `other` are `None`, `None` is returned. + /// + /// # Examples + /// + /// ``` + /// #![feature(option_reduce)] + /// + /// let s12 = Some(12); + /// let s17 = Some(17); + /// let n = None; + /// let f = |a, b| a + b; + /// + /// assert_eq!(s12.reduce(s17, f), Some(29)); + /// assert_eq!(s12.reduce(n, f), Some(12)); + /// assert_eq!(n.reduce(s17, f), Some(17)); + /// assert_eq!(n.reduce(n, f), None); + /// ``` + #[unstable(feature = "option_reduce", issue = "144273")] + pub fn reduce<U, R, F>(self, other: Option<U>, f: F) -> Option<R> + where + T: Into<R>, + U: Into<R>, + F: FnOnce(T, U) -> R, + { + match (self, other) { + (Some(a), Some(b)) => Some(f(a, b)), + (Some(a), _) => Some(a.into()), + (_, Some(b)) => Some(b.into()), + _ => None, + } + } } impl<T, U> Option<(T, U)> { diff --git a/library/core/src/panic/location.rs b/library/core/src/panic/location.rs index cafdcfa2c2e..7a68d393906 100644 --- a/library/core/src/panic/location.rs +++ b/library/core/src/panic/location.rs @@ -39,7 +39,7 @@ use crate::ptr::NonNull; pub struct Location<'a> { // A raw pointer is used rather than a reference because the pointer is valid for one more byte // than the length stored in this pointer; the additional byte is the NUL-terminator used by - // `Location::file_with_nul`. + // `Location::file_as_c_str`. filename: NonNull<str>, line: u32, col: u32, @@ -195,7 +195,7 @@ impl<'a> Location<'a> { #[must_use] #[unstable(feature = "file_with_nul", issue = "141727")] #[inline] - pub const fn file_with_nul(&self) -> &'a CStr { + pub const fn file_as_c_str(&self) -> &'a CStr { let filename = self.filename.as_ptr(); // SAFETY: The filename is valid for `filename_len+1` bytes, so this addition can't diff --git a/library/core/src/ptr/metadata.rs b/library/core/src/ptr/metadata.rs index 0deac3621e8..dc3ec3fd199 100644 --- a/library/core/src/ptr/metadata.rs +++ b/library/core/src/ptr/metadata.rs @@ -106,6 +106,9 @@ pub const fn metadata<T: PointeeSized>(ptr: *const T) -> <T as Pointee>::Metadat /// For slices, see the documentation of [`slice::from_raw_parts`] for safety requirements. /// For trait objects, the metadata must come from a pointer to the same underlying erased type. /// +/// If you are attempting to deconstruct a DST in a generic context to be reconstructed later, +/// a thin pointer can always be obtained by casting `*const T` to `*const ()`. +/// /// [`slice::from_raw_parts`]: crate::slice::from_raw_parts #[unstable(feature = "ptr_metadata", issue = "81513")] #[inline] diff --git a/library/core/src/str/mod.rs b/library/core/src/str/mod.rs index 1b6e84175b9..f1db385e212 100644 --- a/library/core/src/str/mod.rs +++ b/library/core/src/str/mod.rs @@ -396,7 +396,6 @@ impl str { /// # Examples /// /// ``` - /// #![feature(round_char_boundary)] /// let s = "❤️🧡💛💚💙💜"; /// assert_eq!(s.len(), 26); /// assert!(!s.is_char_boundary(13)); @@ -405,7 +404,8 @@ impl str { /// assert_eq!(closest, 10); /// assert_eq!(&s[..closest], "❤️🧡"); /// ``` - #[unstable(feature = "round_char_boundary", issue = "93743")] + #[stable(feature = "round_char_boundary", since = "CURRENT_RUSTC_VERSION")] + #[rustc_const_stable(feature = "round_char_boundary", since = "CURRENT_RUSTC_VERSION")] #[inline] pub const fn floor_char_boundary(&self, index: usize) -> usize { if index >= self.len() { @@ -439,7 +439,6 @@ impl str { /// # Examples /// /// ``` - /// #![feature(round_char_boundary)] /// let s = "❤️🧡💛💚💙💜"; /// assert_eq!(s.len(), 26); /// assert!(!s.is_char_boundary(13)); @@ -448,7 +447,8 @@ impl str { /// assert_eq!(closest, 14); /// assert_eq!(&s[..closest], "❤️🧡💛"); /// ``` - #[unstable(feature = "round_char_boundary", issue = "93743")] + #[stable(feature = "round_char_boundary", since = "CURRENT_RUSTC_VERSION")] + #[rustc_const_stable(feature = "round_char_boundary", since = "CURRENT_RUSTC_VERSION")] #[inline] pub const fn ceil_char_boundary(&self, index: usize) -> usize { if index >= self.len() { @@ -3072,7 +3072,7 @@ impl str { /// for example references to `Box<str>` or `Arc<str>`. #[inline] #[unstable(feature = "str_as_str", issue = "130366")] - pub fn as_str(&self) -> &str { + pub const fn as_str(&self) -> &str { self } } diff --git a/library/core/src/sync/atomic.rs b/library/core/src/sync/atomic.rs index 44a6895f90a..7bd68bcd0bc 100644 --- a/library/core/src/sync/atomic.rs +++ b/library/core/src/sync/atomic.rs @@ -2199,7 +2199,6 @@ impl<T> AtomicPtr<T> { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let atom = AtomicPtr::<i64>::new(core::ptr::null_mut()); @@ -2209,7 +2208,7 @@ impl<T> AtomicPtr<T> { /// ``` #[inline] #[cfg(target_has_atomic = "ptr")] - #[unstable(feature = "strict_provenance_atomic_ptr", issue = "99108")] + #[stable(feature = "strict_provenance_atomic_ptr", since = "CURRENT_RUSTC_VERSION")] #[cfg_attr(miri, track_caller)] // even without panics, this helps for Miri backtraces pub fn fetch_ptr_add(&self, val: usize, order: Ordering) -> *mut T { self.fetch_byte_add(val.wrapping_mul(size_of::<T>()), order) @@ -2240,7 +2239,6 @@ impl<T> AtomicPtr<T> { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let array = [1i32, 2i32]; @@ -2254,7 +2252,7 @@ impl<T> AtomicPtr<T> { /// ``` #[inline] #[cfg(target_has_atomic = "ptr")] - #[unstable(feature = "strict_provenance_atomic_ptr", issue = "99108")] + #[stable(feature = "strict_provenance_atomic_ptr", since = "CURRENT_RUSTC_VERSION")] #[cfg_attr(miri, track_caller)] // even without panics, this helps for Miri backtraces pub fn fetch_ptr_sub(&self, val: usize, order: Ordering) -> *mut T { self.fetch_byte_sub(val.wrapping_mul(size_of::<T>()), order) @@ -2279,7 +2277,6 @@ impl<T> AtomicPtr<T> { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let atom = AtomicPtr::<i64>::new(core::ptr::null_mut()); @@ -2289,7 +2286,7 @@ impl<T> AtomicPtr<T> { /// ``` #[inline] #[cfg(target_has_atomic = "ptr")] - #[unstable(feature = "strict_provenance_atomic_ptr", issue = "99108")] + #[stable(feature = "strict_provenance_atomic_ptr", since = "CURRENT_RUSTC_VERSION")] #[cfg_attr(miri, track_caller)] // even without panics, this helps for Miri backtraces pub fn fetch_byte_add(&self, val: usize, order: Ordering) -> *mut T { // SAFETY: data races are prevented by atomic intrinsics. @@ -2315,7 +2312,6 @@ impl<T> AtomicPtr<T> { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let mut arr = [0i64, 1]; @@ -2325,7 +2321,7 @@ impl<T> AtomicPtr<T> { /// ``` #[inline] #[cfg(target_has_atomic = "ptr")] - #[unstable(feature = "strict_provenance_atomic_ptr", issue = "99108")] + #[stable(feature = "strict_provenance_atomic_ptr", since = "CURRENT_RUSTC_VERSION")] #[cfg_attr(miri, track_caller)] // even without panics, this helps for Miri backtraces pub fn fetch_byte_sub(&self, val: usize, order: Ordering) -> *mut T { // SAFETY: data races are prevented by atomic intrinsics. @@ -2361,7 +2357,6 @@ impl<T> AtomicPtr<T> { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let pointer = &mut 3i64 as *mut i64; @@ -2376,7 +2371,7 @@ impl<T> AtomicPtr<T> { /// ``` #[inline] #[cfg(target_has_atomic = "ptr")] - #[unstable(feature = "strict_provenance_atomic_ptr", issue = "99108")] + #[stable(feature = "strict_provenance_atomic_ptr", since = "CURRENT_RUSTC_VERSION")] #[cfg_attr(miri, track_caller)] // even without panics, this helps for Miri backtraces pub fn fetch_or(&self, val: usize, order: Ordering) -> *mut T { // SAFETY: data races are prevented by atomic intrinsics. @@ -2412,7 +2407,6 @@ impl<T> AtomicPtr<T> { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let pointer = &mut 3i64 as *mut i64; @@ -2426,7 +2420,7 @@ impl<T> AtomicPtr<T> { /// ``` #[inline] #[cfg(target_has_atomic = "ptr")] - #[unstable(feature = "strict_provenance_atomic_ptr", issue = "99108")] + #[stable(feature = "strict_provenance_atomic_ptr", since = "CURRENT_RUSTC_VERSION")] #[cfg_attr(miri, track_caller)] // even without panics, this helps for Miri backtraces pub fn fetch_and(&self, val: usize, order: Ordering) -> *mut T { // SAFETY: data races are prevented by atomic intrinsics. @@ -2462,7 +2456,6 @@ impl<T> AtomicPtr<T> { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let pointer = &mut 3i64 as *mut i64; @@ -2474,7 +2467,7 @@ impl<T> AtomicPtr<T> { /// ``` #[inline] #[cfg(target_has_atomic = "ptr")] - #[unstable(feature = "strict_provenance_atomic_ptr", issue = "99108")] + #[stable(feature = "strict_provenance_atomic_ptr", since = "CURRENT_RUSTC_VERSION")] #[cfg_attr(miri, track_caller)] // even without panics, this helps for Miri backtraces pub fn fetch_xor(&self, val: usize, order: Ordering) -> *mut T { // SAFETY: data races are prevented by atomic intrinsics. diff --git a/library/core/src/wtf8.rs b/library/core/src/wtf8.rs new file mode 100644 index 00000000000..de0dfa560a3 --- /dev/null +++ b/library/core/src/wtf8.rs @@ -0,0 +1,597 @@ +//! Implementation of [the WTF-8 encoding](https://simonsapin.github.io/wtf-8/). +//! +//! This library uses Rust’s type system to maintain +//! [well-formedness](https://simonsapin.github.io/wtf-8/#well-formed), +//! like the `String` and `&str` types do for UTF-8. +//! +//! Since [WTF-8 must not be used +//! for interchange](https://simonsapin.github.io/wtf-8/#intended-audience), +//! this library deliberately does not provide access to the underlying bytes +//! of WTF-8 strings, +//! nor can it decode WTF-8 from arbitrary bytes. +//! WTF-8 strings can be obtained from UTF-8, UTF-16, or code points. +#![unstable( + feature = "wtf8_internals", + issue = "none", + reason = "this is internal code for representing OsStr on some platforms and not a public API" +)] +// rustdoc bug: doc(hidden) on the module won't stop types in the module from showing up in trait +// implementations, so, we'll have to add more doc(hidden)s anyway +#![doc(hidden)] + +use crate::char::{MAX_LEN_UTF16, encode_utf16_raw}; +use crate::clone::CloneToUninit; +use crate::fmt::{self, Write}; +use crate::hash::{Hash, Hasher}; +use crate::iter::FusedIterator; +use crate::num::niche_types::CodePointInner; +use crate::str::next_code_point; +use crate::{ops, slice, str}; + +/// A Unicode code point: from U+0000 to U+10FFFF. +/// +/// Compares with the `char` type, +/// which represents a Unicode scalar value: +/// a code point that is not a surrogate (U+D800 to U+DFFF). +#[derive(Eq, PartialEq, Ord, PartialOrd, Clone, Copy)] +#[doc(hidden)] +pub struct CodePoint(CodePointInner); + +/// Format the code point as `U+` followed by four to six hexadecimal digits. +/// Example: `U+1F4A9` +impl fmt::Debug for CodePoint { + #[inline] + fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { + write!(formatter, "U+{:04X}", self.0.as_inner()) + } +} + +impl CodePoint { + /// Unsafely creates a new `CodePoint` without checking the value. + /// + /// Only use when `value` is known to be less than or equal to 0x10FFFF. + #[inline] + pub unsafe fn from_u32_unchecked(value: u32) -> CodePoint { + // SAFETY: Guaranteed by caller. + CodePoint(unsafe { CodePointInner::new_unchecked(value) }) + } + + /// Creates a new `CodePoint` if the value is a valid code point. + /// + /// Returns `None` if `value` is above 0x10FFFF. + #[inline] + pub fn from_u32(value: u32) -> Option<CodePoint> { + Some(CodePoint(CodePointInner::new(value)?)) + } + + /// Creates a new `CodePoint` from a `char`. + /// + /// Since all Unicode scalar values are code points, this always succeeds. + #[inline] + pub fn from_char(value: char) -> CodePoint { + // SAFETY: All char are valid for this type. + unsafe { CodePoint::from_u32_unchecked(value as u32) } + } + + /// Returns the numeric value of the code point. + #[inline] + pub fn to_u32(&self) -> u32 { + self.0.as_inner() + } + + /// Returns the numeric value of the code point if it is a leading surrogate. + #[inline] + pub fn to_lead_surrogate(&self) -> Option<u16> { + match self.to_u32() { + lead @ 0xD800..=0xDBFF => Some(lead as u16), + _ => None, + } + } + + /// Returns the numeric value of the code point if it is a trailing surrogate. + #[inline] + pub fn to_trail_surrogate(&self) -> Option<u16> { + match self.to_u32() { + trail @ 0xDC00..=0xDFFF => Some(trail as u16), + _ => None, + } + } + + /// Optionally returns a Unicode scalar value for the code point. + /// + /// Returns `None` if the code point is a surrogate (from U+D800 to U+DFFF). + #[inline] + pub fn to_char(&self) -> Option<char> { + match self.to_u32() { + 0xD800..=0xDFFF => None, + // SAFETY: We explicitly check that the char is valid. + valid => Some(unsafe { char::from_u32_unchecked(valid) }), + } + } + + /// Returns a Unicode scalar value for the code point. + /// + /// Returns `'\u{FFFD}'` (the replacement character “�”) + /// if the code point is a surrogate (from U+D800 to U+DFFF). + #[inline] + pub fn to_char_lossy(&self) -> char { + self.to_char().unwrap_or(char::REPLACEMENT_CHARACTER) + } +} + +/// A borrowed slice of well-formed WTF-8 data. +/// +/// Similar to `&str`, but can additionally contain surrogate code points +/// if they’re not in a surrogate pair. +#[derive(Eq, Ord, PartialEq, PartialOrd)] +#[repr(transparent)] +#[rustc_has_incoherent_inherent_impls] +#[doc(hidden)] +pub struct Wtf8 { + bytes: [u8], +} + +impl AsRef<[u8]> for Wtf8 { + #[inline] + fn as_ref(&self) -> &[u8] { + &self.bytes + } +} + +/// Formats the string in double quotes, with characters escaped according to +/// [`char::escape_debug`] and unpaired surrogates represented as `\u{xxxx}`, +/// where each `x` is a hexadecimal digit. +impl fmt::Debug for Wtf8 { + fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { + fn write_str_escaped(f: &mut fmt::Formatter<'_>, s: &str) -> fmt::Result { + use crate::fmt::Write; + for c in s.chars().flat_map(|c| c.escape_debug()) { + f.write_char(c)? + } + Ok(()) + } + + formatter.write_str("\"")?; + let mut pos = 0; + while let Some((surrogate_pos, surrogate)) = self.next_surrogate(pos) { + // SAFETY: next_surrogate provides an index for a range of valid UTF-8 bytes. + write_str_escaped(formatter, unsafe { + str::from_utf8_unchecked(&self.bytes[pos..surrogate_pos]) + })?; + write!(formatter, "\\u{{{:x}}}", surrogate)?; + pos = surrogate_pos + 3; + } + + // SAFETY: after next_surrogate returns None, the remainder is valid UTF-8. + write_str_escaped(formatter, unsafe { str::from_utf8_unchecked(&self.bytes[pos..]) })?; + formatter.write_str("\"") + } +} + +/// Formats the string with unpaired surrogates substituted with the replacement +/// character, U+FFFD. +impl fmt::Display for Wtf8 { + fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { + let wtf8_bytes = &self.bytes; + let mut pos = 0; + loop { + match self.next_surrogate(pos) { + Some((surrogate_pos, _)) => { + // SAFETY: next_surrogate provides an index for a range of valid UTF-8 bytes. + formatter.write_str(unsafe { + str::from_utf8_unchecked(&wtf8_bytes[pos..surrogate_pos]) + })?; + formatter.write_char(char::REPLACEMENT_CHARACTER)?; + pos = surrogate_pos + 3; + } + None => { + // SAFETY: after next_surrogate returns None, the remainder is valid UTF-8. + let s = unsafe { str::from_utf8_unchecked(&wtf8_bytes[pos..]) }; + if pos == 0 { return s.fmt(formatter) } else { return formatter.write_str(s) } + } + } + } + } +} + +impl Wtf8 { + /// Creates a WTF-8 slice from a UTF-8 `&str` slice. + #[inline] + pub fn from_str(value: &str) -> &Wtf8 { + // SAFETY: Since WTF-8 is a superset of UTF-8, this always is valid. + unsafe { Wtf8::from_bytes_unchecked(value.as_bytes()) } + } + + /// Creates a WTF-8 slice from a WTF-8 byte slice. + /// + /// Since the byte slice is not checked for valid WTF-8, this functions is + /// marked unsafe. + #[inline] + pub unsafe fn from_bytes_unchecked(value: &[u8]) -> &Wtf8 { + // SAFETY: start with &[u8], end with fancy &[u8] + unsafe { &*(value as *const [u8] as *const Wtf8) } + } + + /// Creates a mutable WTF-8 slice from a mutable WTF-8 byte slice. + /// + /// Since the byte slice is not checked for valid WTF-8, this functions is + /// marked unsafe. + #[inline] + pub unsafe fn from_mut_bytes_unchecked(value: &mut [u8]) -> &mut Wtf8 { + // SAFETY: start with &mut [u8], end with fancy &mut [u8] + unsafe { &mut *(value as *mut [u8] as *mut Wtf8) } + } + + /// Returns the length, in WTF-8 bytes. + #[inline] + pub fn len(&self) -> usize { + self.bytes.len() + } + + #[inline] + pub fn is_empty(&self) -> bool { + self.bytes.is_empty() + } + + /// Returns the code point at `position` if it is in the ASCII range, + /// or `b'\xFF'` otherwise. + /// + /// # Panics + /// + /// Panics if `position` is beyond the end of the string. + #[inline] + pub fn ascii_byte_at(&self, position: usize) -> u8 { + match self.bytes[position] { + ascii_byte @ 0x00..=0x7F => ascii_byte, + _ => 0xFF, + } + } + + /// Returns an iterator for the string’s code points. + #[inline] + pub fn code_points(&self) -> Wtf8CodePoints<'_> { + Wtf8CodePoints { bytes: self.bytes.iter() } + } + + /// Access raw bytes of WTF-8 data + #[inline] + pub fn as_bytes(&self) -> &[u8] { + &self.bytes + } + + /// Tries to convert the string to UTF-8 and return a `&str` slice. + /// + /// Returns `None` if the string contains surrogates. + /// + /// This does not copy the data. + #[inline] + pub fn as_str(&self) -> Result<&str, str::Utf8Error> { + str::from_utf8(&self.bytes) + } + + /// Converts the WTF-8 string to potentially ill-formed UTF-16 + /// and return an iterator of 16-bit code units. + /// + /// This is lossless: + /// calling `Wtf8Buf::from_ill_formed_utf16` on the resulting code units + /// would always return the original WTF-8 string. + #[inline] + pub fn encode_wide(&self) -> EncodeWide<'_> { + EncodeWide { code_points: self.code_points(), extra: 0 } + } + + #[inline] + pub fn next_surrogate(&self, mut pos: usize) -> Option<(usize, u16)> { + let mut iter = self.bytes[pos..].iter(); + loop { + let b = *iter.next()?; + if b < 0x80 { + pos += 1; + } else if b < 0xE0 { + iter.next(); + pos += 2; + } else if b == 0xED { + match (iter.next(), iter.next()) { + (Some(&b2), Some(&b3)) if b2 >= 0xA0 => { + return Some((pos, decode_surrogate(b2, b3))); + } + _ => pos += 3, + } + } else if b < 0xF0 { + iter.next(); + iter.next(); + pos += 3; + } else { + iter.next(); + iter.next(); + iter.next(); + pos += 4; + } + } + } + + #[inline] + pub fn final_lead_surrogate(&self) -> Option<u16> { + match self.bytes { + [.., 0xED, b2 @ 0xA0..=0xAF, b3] => Some(decode_surrogate(b2, b3)), + _ => None, + } + } + + #[inline] + pub fn initial_trail_surrogate(&self) -> Option<u16> { + match self.bytes { + [0xED, b2 @ 0xB0..=0xBF, b3, ..] => Some(decode_surrogate(b2, b3)), + _ => None, + } + } + + #[inline] + pub fn make_ascii_lowercase(&mut self) { + self.bytes.make_ascii_lowercase() + } + + #[inline] + pub fn make_ascii_uppercase(&mut self) { + self.bytes.make_ascii_uppercase() + } + + #[inline] + pub fn is_ascii(&self) -> bool { + self.bytes.is_ascii() + } + + #[inline] + pub fn eq_ignore_ascii_case(&self, other: &Self) -> bool { + self.bytes.eq_ignore_ascii_case(&other.bytes) + } +} + +/// Returns a slice of the given string for the byte range \[`begin`..`end`). +/// +/// # Panics +/// +/// Panics when `begin` and `end` do not point to code point boundaries, +/// or point beyond the end of the string. +impl ops::Index<ops::Range<usize>> for Wtf8 { + type Output = Wtf8; + + #[inline] + fn index(&self, range: ops::Range<usize>) -> &Wtf8 { + if range.start <= range.end + && self.is_code_point_boundary(range.start) + && self.is_code_point_boundary(range.end) + { + // SAFETY: is_code_point_boundary checks that the index is valid + unsafe { slice_unchecked(self, range.start, range.end) } + } else { + slice_error_fail(self, range.start, range.end) + } + } +} + +/// Returns a slice of the given string from byte `begin` to its end. +/// +/// # Panics +/// +/// Panics when `begin` is not at a code point boundary, +/// or is beyond the end of the string. +impl ops::Index<ops::RangeFrom<usize>> for Wtf8 { + type Output = Wtf8; + + #[inline] + fn index(&self, range: ops::RangeFrom<usize>) -> &Wtf8 { + if self.is_code_point_boundary(range.start) { + // SAFETY: is_code_point_boundary checks that the index is valid + unsafe { slice_unchecked(self, range.start, self.len()) } + } else { + slice_error_fail(self, range.start, self.len()) + } + } +} + +/// Returns a slice of the given string from its beginning to byte `end`. +/// +/// # Panics +/// +/// Panics when `end` is not at a code point boundary, +/// or is beyond the end of the string. +impl ops::Index<ops::RangeTo<usize>> for Wtf8 { + type Output = Wtf8; + + #[inline] + fn index(&self, range: ops::RangeTo<usize>) -> &Wtf8 { + if self.is_code_point_boundary(range.end) { + // SAFETY: is_code_point_boundary checks that the index is valid + unsafe { slice_unchecked(self, 0, range.end) } + } else { + slice_error_fail(self, 0, range.end) + } + } +} + +impl ops::Index<ops::RangeFull> for Wtf8 { + type Output = Wtf8; + + #[inline] + fn index(&self, _range: ops::RangeFull) -> &Wtf8 { + self + } +} + +#[inline] +fn decode_surrogate(second_byte: u8, third_byte: u8) -> u16 { + // The first byte is assumed to be 0xED + 0xD800 | (second_byte as u16 & 0x3F) << 6 | third_byte as u16 & 0x3F +} + +impl Wtf8 { + /// Copied from str::is_char_boundary + #[inline] + pub fn is_code_point_boundary(&self, index: usize) -> bool { + if index == 0 { + return true; + } + match self.bytes.get(index) { + None => index == self.len(), + Some(&b) => (b as i8) >= -0x40, + } + } + + /// Verify that `index` is at the edge of either a valid UTF-8 codepoint + /// (i.e. a codepoint that's not a surrogate) or of the whole string. + /// + /// These are the cases currently permitted by `OsStr::self_encoded_bytes`. + /// Splitting between surrogates is valid as far as WTF-8 is concerned, but + /// we do not permit it in the public API because WTF-8 is considered an + /// implementation detail. + #[track_caller] + #[inline] + pub fn check_utf8_boundary(&self, index: usize) { + if index == 0 { + return; + } + match self.bytes.get(index) { + Some(0xED) => (), // Might be a surrogate + Some(&b) if (b as i8) >= -0x40 => return, + Some(_) => panic!("byte index {index} is not a codepoint boundary"), + None if index == self.len() => return, + None => panic!("byte index {index} is out of bounds"), + } + if self.bytes[index + 1] >= 0xA0 { + // There's a surrogate after index. Now check before index. + if index >= 3 && self.bytes[index - 3] == 0xED && self.bytes[index - 2] >= 0xA0 { + panic!("byte index {index} lies between surrogate codepoints"); + } + } + } +} + +/// Copied from core::str::raw::slice_unchecked +#[inline] +unsafe fn slice_unchecked(s: &Wtf8, begin: usize, end: usize) -> &Wtf8 { + // SAFETY: memory layout of a &[u8] and &Wtf8 are the same + unsafe { + let len = end - begin; + let start = s.as_bytes().as_ptr().add(begin); + Wtf8::from_bytes_unchecked(slice::from_raw_parts(start, len)) + } +} + +/// Copied from core::str::raw::slice_error_fail +#[inline(never)] +fn slice_error_fail(s: &Wtf8, begin: usize, end: usize) -> ! { + assert!(begin <= end); + panic!("index {begin} and/or {end} in `{s:?}` do not lie on character boundary"); +} + +/// Iterator for the code points of a WTF-8 string. +/// +/// Created with the method `.code_points()`. +#[derive(Clone)] +#[doc(hidden)] +pub struct Wtf8CodePoints<'a> { + bytes: slice::Iter<'a, u8>, +} + +impl Iterator for Wtf8CodePoints<'_> { + type Item = CodePoint; + + #[inline] + fn next(&mut self) -> Option<CodePoint> { + // SAFETY: `self.bytes` has been created from a WTF-8 string + unsafe { next_code_point(&mut self.bytes).map(|c| CodePoint::from_u32_unchecked(c)) } + } + + #[inline] + fn size_hint(&self) -> (usize, Option<usize>) { + let len = self.bytes.len(); + (len.saturating_add(3) / 4, Some(len)) + } +} + +impl fmt::Debug for Wtf8CodePoints<'_> { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.debug_tuple("Wtf8CodePoints") + // SAFETY: We always leave the string in a valid state after each iteration. + .field(&unsafe { Wtf8::from_bytes_unchecked(self.bytes.as_slice()) }) + .finish() + } +} + +/// Generates a wide character sequence for potentially ill-formed UTF-16. +#[stable(feature = "rust1", since = "1.0.0")] +#[derive(Clone)] +#[doc(hidden)] +pub struct EncodeWide<'a> { + code_points: Wtf8CodePoints<'a>, + extra: u16, +} + +// Copied from libunicode/u_str.rs +#[stable(feature = "rust1", since = "1.0.0")] +impl Iterator for EncodeWide<'_> { + type Item = u16; + + #[inline] + fn next(&mut self) -> Option<u16> { + if self.extra != 0 { + let tmp = self.extra; + self.extra = 0; + return Some(tmp); + } + + let mut buf = [0; MAX_LEN_UTF16]; + self.code_points.next().map(|code_point| { + let n = encode_utf16_raw(code_point.to_u32(), &mut buf).len(); + if n == 2 { + self.extra = buf[1]; + } + buf[0] + }) + } + + #[inline] + fn size_hint(&self) -> (usize, Option<usize>) { + let (low, high) = self.code_points.size_hint(); + let ext = (self.extra != 0) as usize; + // every code point gets either one u16 or two u16, + // so this iterator is between 1 or 2 times as + // long as the underlying iterator. + (low + ext, high.and_then(|n| n.checked_mul(2)).and_then(|n| n.checked_add(ext))) + } +} + +impl fmt::Debug for EncodeWide<'_> { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.debug_struct("EncodeWide").finish_non_exhaustive() + } +} + +#[stable(feature = "encode_wide_fused_iterator", since = "1.62.0")] +impl FusedIterator for EncodeWide<'_> {} + +impl Hash for CodePoint { + #[inline] + fn hash<H: Hasher>(&self, state: &mut H) { + self.0.hash(state) + } +} + +impl Hash for Wtf8 { + #[inline] + fn hash<H: Hasher>(&self, state: &mut H) { + state.write(&self.bytes); + 0xfeu8.hash(state) + } +} + +#[unstable(feature = "clone_to_uninit", issue = "126799")] +unsafe impl CloneToUninit for Wtf8 { + #[inline] + #[cfg_attr(debug_assertions, track_caller)] + unsafe fn clone_to_uninit(&self, dst: *mut u8) { + // SAFETY: we're just a transparent wrapper around [u8] + unsafe { self.bytes.clone_to_uninit(dst) } + } +} |
