From 3ddd2ef8c518efe10c811a229f056e96864d83c1 Mon Sep 17 00:00:00 2001 From: Manish Goregaokar Date: Wed, 2 Jul 2025 13:15:07 -0700 Subject: Mention that casting to *const () is a way to roundtrip with from_raw_parts --- library/core/src/ptr/metadata.rs | 3 +++ 1 file changed, 3 insertions(+) (limited to 'library/core/src') 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(ptr: *const T) -> ::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] -- cgit 1.4.1-3-g733a5 From 613080b5f1d58a2496d9f8dc8c314d5432fe3d6f Mon Sep 17 00:00:00 2001 From: Jeremy Smart Date: Thu, 31 Jul 2025 23:46:04 -0400 Subject: add Option::reduce --- library/core/src/option.rs | 36 ++++++++++++++++++++++++++++++++++++ library/coretests/tests/lib.rs | 1 + 2 files changed, 37 insertions(+) (limited to 'library/core/src') diff --git a/library/core/src/option.rs b/library/core/src/option.rs index 0e481519f76..f3ddc05bc1a 100644 --- a/library/core/src/option.rs +++ b/library/core/src/option.rs @@ -1961,6 +1961,42 @@ impl Option { _ => 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(self, other: Option, f: F) -> Option + where + T: Into, + U: Into, + 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 Option<(T, U)> { diff --git a/library/coretests/tests/lib.rs b/library/coretests/tests/lib.rs index 4cfac9ecc2a..3f817ab128b 100644 --- a/library/coretests/tests/lib.rs +++ b/library/coretests/tests/lib.rs @@ -77,6 +77,7 @@ #![feature(never_type)] #![feature(next_index)] #![feature(numfmt)] +#![feature(option_reduce)] #![feature(pattern)] #![feature(pointer_is_aligned_to)] #![feature(portable_simd)] -- cgit 1.4.1-3-g733a5 From b951b5dca116803b89380aae55ce9053a4674f31 Mon Sep 17 00:00:00 2001 From: Kivooeo Date: Fri, 15 Aug 2025 16:56:11 +0000 Subject: stabilize strict provenance atomic ptr --- library/core/src/sync/atomic.rs | 21 +++++++-------------- library/coretests/tests/lib.rs | 1 - library/std/src/lib.rs | 1 - src/tools/miri/tests/pass/atomic.rs | 1 - tests/codegen-llvm/atomicptr.rs | 1 - 5 files changed, 7 insertions(+), 18 deletions(-) (limited to 'library/core/src') 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 AtomicPtr { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let atom = AtomicPtr::::new(core::ptr::null_mut()); @@ -2209,7 +2208,7 @@ impl AtomicPtr { /// ``` #[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::()), order) @@ -2240,7 +2239,6 @@ impl AtomicPtr { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let array = [1i32, 2i32]; @@ -2254,7 +2252,7 @@ impl AtomicPtr { /// ``` #[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::()), order) @@ -2279,7 +2277,6 @@ impl AtomicPtr { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let atom = AtomicPtr::::new(core::ptr::null_mut()); @@ -2289,7 +2286,7 @@ impl AtomicPtr { /// ``` #[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 AtomicPtr { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let mut arr = [0i64, 1]; @@ -2325,7 +2321,7 @@ impl AtomicPtr { /// ``` #[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 AtomicPtr { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let pointer = &mut 3i64 as *mut i64; @@ -2376,7 +2371,7 @@ impl AtomicPtr { /// ``` #[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 AtomicPtr { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let pointer = &mut 3i64 as *mut i64; @@ -2426,7 +2420,7 @@ impl AtomicPtr { /// ``` #[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 AtomicPtr { /// # Examples /// /// ``` - /// #![feature(strict_provenance_atomic_ptr)] /// use core::sync::atomic::{AtomicPtr, Ordering}; /// /// let pointer = &mut 3i64 as *mut i64; @@ -2474,7 +2467,7 @@ impl AtomicPtr { /// ``` #[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/coretests/tests/lib.rs b/library/coretests/tests/lib.rs index 0c4d49f3c99..92774e12681 100644 --- a/library/coretests/tests/lib.rs +++ b/library/coretests/tests/lib.rs @@ -95,7 +95,6 @@ #![feature(std_internals)] #![feature(step_trait)] #![feature(str_internals)] -#![feature(strict_provenance_atomic_ptr)] #![feature(strict_provenance_lints)] #![feature(test)] #![feature(trusted_len)] diff --git a/library/std/src/lib.rs b/library/std/src/lib.rs index f111fcb4a47..77c7878ce08 100644 --- a/library/std/src/lib.rs +++ b/library/std/src/lib.rs @@ -370,7 +370,6 @@ #![feature(slice_range)] #![feature(std_internals)] #![feature(str_internals)] -#![feature(strict_provenance_atomic_ptr)] #![feature(sync_unsafe_cell)] #![feature(temporary_niche_types)] #![feature(ub_checks)] diff --git a/src/tools/miri/tests/pass/atomic.rs b/src/tools/miri/tests/pass/atomic.rs index 3de34e570c7..d8ac5114f27 100644 --- a/src/tools/miri/tests/pass/atomic.rs +++ b/src/tools/miri/tests/pass/atomic.rs @@ -2,7 +2,6 @@ //@[tree]compile-flags: -Zmiri-tree-borrows //@compile-flags: -Zmiri-strict-provenance -#![feature(strict_provenance_atomic_ptr)] // FIXME(static_mut_refs): Do not allow `static_mut_refs` lint #![allow(static_mut_refs)] diff --git a/tests/codegen-llvm/atomicptr.rs b/tests/codegen-llvm/atomicptr.rs index ce6c4aa0d2b..9d5e618fe76 100644 --- a/tests/codegen-llvm/atomicptr.rs +++ b/tests/codegen-llvm/atomicptr.rs @@ -6,7 +6,6 @@ //@ compile-flags: -Copt-level=3 -Cno-prepopulate-passes #![crate_type = "lib"] -#![feature(strict_provenance_atomic_ptr)] use std::ptr::without_provenance_mut; use std::sync::atomic::AtomicPtr; -- cgit 1.4.1-3-g733a5 From 5a2fceefd312ec027bdeaa89ebefbe4c33d94de1 Mon Sep 17 00:00:00 2001 From: ltdk Date: Wed, 13 Aug 2025 01:16:09 -0400 Subject: Copy WTF-8 code into core/alloc (for better diffs) --- library/alloc/src/wtf8/mod.rs | 1083 ++++++++++++++++++++++++++++++ library/alloc/src/wtf8/tests.rs | 751 +++++++++++++++++++++ library/core/src/wtf8.rs | 1083 ++++++++++++++++++++++++++++++ library/std/src/sys_common/wtf8.rs | 1083 ------------------------------ library/std/src/sys_common/wtf8/tests.rs | 751 --------------------- 5 files changed, 2917 insertions(+), 1834 deletions(-) create mode 100644 library/alloc/src/wtf8/mod.rs create mode 100644 library/alloc/src/wtf8/tests.rs create mode 100644 library/core/src/wtf8.rs delete mode 100644 library/std/src/sys_common/wtf8.rs delete mode 100644 library/std/src/sys_common/wtf8/tests.rs (limited to 'library/core/src') diff --git a/library/alloc/src/wtf8/mod.rs b/library/alloc/src/wtf8/mod.rs new file mode 100644 index 00000000000..50bde88b5a4 --- /dev/null +++ b/library/alloc/src/wtf8/mod.rs @@ -0,0 +1,1083 @@ +//! 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. + +// this module is imported from @SimonSapin's repo and has tons of dead code on +// unix (it's mostly used on windows), so don't worry about dead code here. +#![allow(dead_code)] + +#[cfg(test)] +mod tests; + +use core::char::{MAX_LEN_UTF8, MAX_LEN_UTF16, encode_utf8_raw, encode_utf16_raw}; +use core::clone::CloneToUninit; +use core::str::next_code_point; + +use crate::borrow::Cow; +use crate::collections::TryReserveError; +use crate::hash::{Hash, Hasher}; +use crate::iter::FusedIterator; +use crate::rc::Rc; +use crate::sync::Arc; +use crate::sys_common::AsInner; +use crate::{fmt, mem, ops, slice, str}; + +const UTF8_REPLACEMENT_CHARACTER: &str = "\u{FFFD}"; + +/// 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)] +pub struct CodePoint { + value: u32, +} + +/// 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.value) + } +} + +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 { + CodePoint { 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 { + match value { + 0..=0x10FFFF => Some(CodePoint { value }), + _ => None, + } + } + + /// 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 { + CodePoint { value: value as u32 } + } + + /// Returns the numeric value of the code point. + #[inline] + pub fn to_u32(&self) -> u32 { + self.value + } + + /// Returns the numeric value of the code point if it is a leading surrogate. + #[inline] + pub fn to_lead_surrogate(&self) -> Option { + match self.value { + 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 { + match self.value { + 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 { + match self.value { + 0xD800..=0xDFFF => None, + _ => Some(unsafe { char::from_u32_unchecked(self.value) }), + } + } + + /// 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('\u{FFFD}') + } +} + +/// An owned, growable string of well-formed WTF-8 data. +/// +/// Similar to `String`, but can additionally contain surrogate code points +/// if they’re not in a surrogate pair. +#[derive(Eq, PartialEq, Ord, PartialOrd, Clone)] +pub struct Wtf8Buf { + bytes: Vec, + + /// Do we know that `bytes` holds a valid UTF-8 encoding? We can easily + /// know this if we're constructed from a `String` or `&str`. + /// + /// It is possible for `bytes` to have valid UTF-8 without this being + /// set, such as when we're concatenating `&Wtf8`'s and surrogates become + /// paired, as we don't bother to rescan the entire string. + is_known_utf8: bool, +} + +impl ops::Deref for Wtf8Buf { + type Target = Wtf8; + + fn deref(&self) -> &Wtf8 { + self.as_slice() + } +} + +impl ops::DerefMut for Wtf8Buf { + fn deref_mut(&mut self) -> &mut Wtf8 { + self.as_mut_slice() + } +} + +/// 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. +/// +/// For example, the code units [U+0061, U+D800, U+000A] are formatted as +/// `"a\u{D800}\n"`. +impl fmt::Debug for Wtf8Buf { + #[inline] + fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt::Debug::fmt(&**self, formatter) + } +} + +/// Formats the string with unpaired surrogates substituted with the replacement +/// character, U+FFFD. +impl fmt::Display for Wtf8Buf { + fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { + if let Some(s) = self.as_known_utf8() { + fmt::Display::fmt(s, formatter) + } else { + fmt::Display::fmt(&**self, formatter) + } + } +} + +impl Wtf8Buf { + /// Creates a new, empty WTF-8 string. + #[inline] + pub fn new() -> Wtf8Buf { + Wtf8Buf { bytes: Vec::new(), is_known_utf8: true } + } + + /// Creates a new, empty WTF-8 string with pre-allocated capacity for `capacity` bytes. + #[inline] + pub fn with_capacity(capacity: usize) -> Wtf8Buf { + Wtf8Buf { bytes: Vec::with_capacity(capacity), is_known_utf8: true } + } + + /// Creates a WTF-8 string from a WTF-8 byte vec. + /// + /// Since the byte vec is not checked for valid WTF-8, this function is + /// marked unsafe. + #[inline] + pub unsafe fn from_bytes_unchecked(value: Vec) -> Wtf8Buf { + Wtf8Buf { bytes: value, is_known_utf8: false } + } + + /// Creates a WTF-8 string from a UTF-8 `String`. + /// + /// This takes ownership of the `String` and does not copy. + /// + /// Since WTF-8 is a superset of UTF-8, this always succeeds. + #[inline] + pub const fn from_string(string: String) -> Wtf8Buf { + Wtf8Buf { bytes: string.into_bytes(), is_known_utf8: true } + } + + /// Creates a WTF-8 string from a UTF-8 `&str` slice. + /// + /// This copies the content of the slice. + /// + /// Since WTF-8 is a superset of UTF-8, this always succeeds. + #[inline] + pub fn from_str(s: &str) -> Wtf8Buf { + Wtf8Buf { bytes: s.as_bytes().to_vec(), is_known_utf8: true } + } + + pub fn clear(&mut self) { + self.bytes.clear(); + self.is_known_utf8 = true; + } + + /// Creates a WTF-8 string from a potentially ill-formed UTF-16 slice of 16-bit code units. + /// + /// This is lossless: calling `.encode_wide()` on the resulting string + /// will always return the original code units. + pub fn from_wide(v: &[u16]) -> Wtf8Buf { + let mut string = Wtf8Buf::with_capacity(v.len()); + for item in char::decode_utf16(v.iter().cloned()) { + match item { + Ok(ch) => string.push_char(ch), + Err(surrogate) => { + let surrogate = surrogate.unpaired_surrogate(); + // Surrogates are known to be in the code point range. + let code_point = unsafe { CodePoint::from_u32_unchecked(surrogate as u32) }; + // The string will now contain an unpaired surrogate. + string.is_known_utf8 = false; + // Skip the WTF-8 concatenation check, + // surrogate pairs are already decoded by decode_utf16 + string.push_code_point_unchecked(code_point); + } + } + } + string + } + + /// Appends the given `char` to the end of this string. + /// This does **not** include the WTF-8 concatenation check or `is_known_utf8` check. + /// Copied from String::push. + fn push_code_point_unchecked(&mut self, code_point: CodePoint) { + let mut bytes = [0; MAX_LEN_UTF8]; + let bytes = encode_utf8_raw(code_point.value, &mut bytes); + self.bytes.extend_from_slice(bytes) + } + + #[inline] + pub fn as_slice(&self) -> &Wtf8 { + unsafe { Wtf8::from_bytes_unchecked(&self.bytes) } + } + + #[inline] + pub fn as_mut_slice(&mut self) -> &mut Wtf8 { + // Safety: `Wtf8` doesn't expose any way to mutate the bytes that would + // cause them to change from well-formed UTF-8 to ill-formed UTF-8, + // which would break the assumptions of the `is_known_utf8` field. + unsafe { Wtf8::from_mut_bytes_unchecked(&mut self.bytes) } + } + + /// Converts the string to UTF-8 without validation, if it was created from + /// valid UTF-8. + #[inline] + fn as_known_utf8(&self) -> Option<&str> { + if self.is_known_utf8 { + // SAFETY: The buffer is known to be valid UTF-8. + Some(unsafe { str::from_utf8_unchecked(self.as_bytes()) }) + } else { + None + } + } + + /// Reserves capacity for at least `additional` more bytes to be inserted + /// in the given `Wtf8Buf`. + /// The collection may reserve more space to avoid frequent reallocations. + /// + /// # Panics + /// + /// Panics if the new capacity exceeds `isize::MAX` bytes. + #[inline] + pub fn reserve(&mut self, additional: usize) { + self.bytes.reserve(additional) + } + + /// Tries to reserve capacity for at least `additional` more bytes to be + /// inserted in the given `Wtf8Buf`. The `Wtf8Buf` may reserve more space to + /// avoid frequent reallocations. After calling `try_reserve`, capacity will + /// be greater than or equal to `self.len() + additional`. Does nothing if + /// capacity is already sufficient. This method preserves the contents even + /// if an error occurs. + /// + /// # Errors + /// + /// If the capacity overflows, or the allocator reports a failure, then an error + /// is returned. + #[inline] + pub fn try_reserve(&mut self, additional: usize) -> Result<(), TryReserveError> { + self.bytes.try_reserve(additional) + } + + #[inline] + pub fn reserve_exact(&mut self, additional: usize) { + self.bytes.reserve_exact(additional) + } + + /// Tries to reserve the minimum capacity for exactly `additional` more + /// bytes to be inserted in the given `Wtf8Buf`. After calling + /// `try_reserve_exact`, capacity will be greater than or equal to + /// `self.len() + additional` if it returns `Ok(())`. + /// Does nothing if the capacity is already sufficient. + /// + /// Note that the allocator may give the `Wtf8Buf` more space than it + /// requests. Therefore, capacity can not be relied upon to be precisely + /// minimal. Prefer [`try_reserve`] if future insertions are expected. + /// + /// [`try_reserve`]: Wtf8Buf::try_reserve + /// + /// # Errors + /// + /// If the capacity overflows, or the allocator reports a failure, then an error + /// is returned. + #[inline] + pub fn try_reserve_exact(&mut self, additional: usize) -> Result<(), TryReserveError> { + self.bytes.try_reserve_exact(additional) + } + + #[inline] + pub fn shrink_to_fit(&mut self) { + self.bytes.shrink_to_fit() + } + + #[inline] + pub fn shrink_to(&mut self, min_capacity: usize) { + self.bytes.shrink_to(min_capacity) + } + + #[inline] + pub fn leak<'a>(self) -> &'a mut Wtf8 { + unsafe { Wtf8::from_mut_bytes_unchecked(self.bytes.leak()) } + } + + /// Returns the number of bytes that this string buffer can hold without reallocating. + #[inline] + pub fn capacity(&self) -> usize { + self.bytes.capacity() + } + + /// Append a UTF-8 slice at the end of the string. + #[inline] + pub fn push_str(&mut self, other: &str) { + self.bytes.extend_from_slice(other.as_bytes()) + } + + /// Append a WTF-8 slice at the end of the string. + /// + /// This replaces newly paired surrogates at the boundary + /// with a supplementary code point, + /// like concatenating ill-formed UTF-16 strings effectively would. + #[inline] + pub fn push_wtf8(&mut self, other: &Wtf8) { + match ((&*self).final_lead_surrogate(), other.initial_trail_surrogate()) { + // Replace newly paired surrogates by a supplementary code point. + (Some(lead), Some(trail)) => { + let len_without_lead_surrogate = self.len() - 3; + self.bytes.truncate(len_without_lead_surrogate); + let other_without_trail_surrogate = &other.bytes[3..]; + // 4 bytes for the supplementary code point + self.bytes.reserve(4 + other_without_trail_surrogate.len()); + self.push_char(decode_surrogate_pair(lead, trail)); + self.bytes.extend_from_slice(other_without_trail_surrogate); + } + _ => { + // If we'll be pushing a string containing a surrogate, we may + // no longer have UTF-8. + if self.is_known_utf8 && other.next_surrogate(0).is_some() { + self.is_known_utf8 = false; + } + + self.bytes.extend_from_slice(&other.bytes); + } + } + } + + /// Append a Unicode scalar value at the end of the string. + #[inline] + pub fn push_char(&mut self, c: char) { + self.push_code_point_unchecked(CodePoint::from_char(c)) + } + + /// Append a code point at the end of the string. + /// + /// This replaces newly paired surrogates at the boundary + /// with a supplementary code point, + /// like concatenating ill-formed UTF-16 strings effectively would. + #[inline] + pub fn push(&mut self, code_point: CodePoint) { + if let Some(trail) = code_point.to_trail_surrogate() { + if let Some(lead) = (&*self).final_lead_surrogate() { + let len_without_lead_surrogate = self.len() - 3; + self.bytes.truncate(len_without_lead_surrogate); + self.push_char(decode_surrogate_pair(lead, trail)); + return; + } + + // We're pushing a trailing surrogate. + self.is_known_utf8 = false; + } else if code_point.to_lead_surrogate().is_some() { + // We're pushing a leading surrogate. + self.is_known_utf8 = false; + } + + // No newly paired surrogates at the boundary. + self.push_code_point_unchecked(code_point) + } + + /// Shortens a string to the specified length. + /// + /// # Panics + /// + /// Panics if `new_len` > current length, + /// or if `new_len` is not a code point boundary. + #[inline] + pub fn truncate(&mut self, new_len: usize) { + assert!(is_code_point_boundary(self, new_len)); + self.bytes.truncate(new_len) + } + + /// Consumes the WTF-8 string and tries to convert it to a vec of bytes. + #[inline] + pub fn into_bytes(self) -> Vec { + self.bytes + } + + /// Consumes the WTF-8 string and tries to convert it to UTF-8. + /// + /// This does not copy the data. + /// + /// If the contents are not well-formed UTF-8 + /// (that is, if the string contains surrogates), + /// the original WTF-8 string is returned instead. + pub fn into_string(self) -> Result { + if self.is_known_utf8 || self.next_surrogate(0).is_none() { + Ok(unsafe { String::from_utf8_unchecked(self.bytes) }) + } else { + Err(self) + } + } + + /// Consumes the WTF-8 string and converts it lossily to UTF-8. + /// + /// This does not copy the data (but may overwrite parts of it in place). + /// + /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”) + pub fn into_string_lossy(mut self) -> String { + if !self.is_known_utf8 { + let mut pos = 0; + while let Some((surrogate_pos, _)) = self.next_surrogate(pos) { + pos = surrogate_pos + 3; + // Surrogates and the replacement character are all 3 bytes, so + // they can substituted in-place. + self.bytes[surrogate_pos..pos] + .copy_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); + } + } + unsafe { String::from_utf8_unchecked(self.bytes) } + } + + /// Converts this `Wtf8Buf` into a boxed `Wtf8`. + #[inline] + pub fn into_box(self) -> Box { + // SAFETY: relies on `Wtf8` being `repr(transparent)`. + unsafe { mem::transmute(self.bytes.into_boxed_slice()) } + } + + /// Converts a `Box` into a `Wtf8Buf`. + pub fn from_box(boxed: Box) -> Wtf8Buf { + let bytes: Box<[u8]> = unsafe { mem::transmute(boxed) }; + Wtf8Buf { bytes: bytes.into_vec(), is_known_utf8: false } + } + + /// Provides plumbing to core `Vec::extend_from_slice`. + /// More well behaving alternative to allowing outer types + /// full mutable access to the core `Vec`. + #[inline] + pub(crate) fn extend_from_slice(&mut self, other: &[u8]) { + self.bytes.extend_from_slice(other); + self.is_known_utf8 = false; + } +} + +/// Creates a new WTF-8 string from an iterator of code points. +/// +/// This replaces surrogate code point pairs with supplementary code points, +/// like concatenating ill-formed UTF-16 strings effectively would. +impl FromIterator for Wtf8Buf { + fn from_iter>(iter: T) -> Wtf8Buf { + let mut string = Wtf8Buf::new(); + string.extend(iter); + string + } +} + +/// Append code points from an iterator to the string. +/// +/// This replaces surrogate code point pairs with supplementary code points, +/// like concatenating ill-formed UTF-16 strings effectively would. +impl Extend for Wtf8Buf { + fn extend>(&mut self, iter: T) { + let iterator = iter.into_iter(); + let (low, _high) = iterator.size_hint(); + // Lower bound of one byte per code point (ASCII only) + self.bytes.reserve(low); + iterator.for_each(move |code_point| self.push(code_point)); + } + + #[inline] + fn extend_one(&mut self, code_point: CodePoint) { + self.push(code_point); + } + + #[inline] + fn extend_reserve(&mut self, additional: usize) { + // Lower bound of one byte per code point (ASCII only) + self.bytes.reserve(additional); + } +} + +/// 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)] +pub struct Wtf8 { + bytes: [u8], +} + +impl AsInner<[u8]> for Wtf8 { + #[inline] + fn as_inner(&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) { + write_str_escaped(formatter, unsafe { + str::from_utf8_unchecked(&self.bytes[pos..surrogate_pos]) + })?; + write!(formatter, "\\u{{{:x}}}", surrogate)?; + pos = surrogate_pos + 3; + } + 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, _)) => { + formatter.write_str(unsafe { + str::from_utf8_unchecked(&wtf8_bytes[pos..surrogate_pos]) + })?; + formatter.write_str(UTF8_REPLACEMENT_CHARACTER)?; + pos = surrogate_pos + 3; + } + None => { + 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. + /// + /// Since WTF-8 is a superset of UTF-8, this always succeeds. + #[inline] + pub fn from_str(value: &str) -> &Wtf8 { + 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] + 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) + } + + /// Creates an owned `Wtf8Buf` from a borrowed `Wtf8`. + pub fn to_owned(&self) -> Wtf8Buf { + Wtf8Buf { bytes: self.bytes.to_vec(), is_known_utf8: false } + } + + /// Lossily converts the string to UTF-8. + /// Returns a UTF-8 `&str` slice if the contents are well-formed in UTF-8. + /// + /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”). + /// + /// This only copies the data if necessary (if it contains any surrogate). + pub fn to_string_lossy(&self) -> Cow<'_, str> { + let Some((surrogate_pos, _)) = self.next_surrogate(0) else { + return Cow::Borrowed(unsafe { str::from_utf8_unchecked(&self.bytes) }); + }; + let wtf8_bytes = &self.bytes; + let mut utf8_bytes = Vec::with_capacity(self.len()); + utf8_bytes.extend_from_slice(&wtf8_bytes[..surrogate_pos]); + utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); + let mut pos = surrogate_pos + 3; + loop { + match self.next_surrogate(pos) { + Some((surrogate_pos, _)) => { + utf8_bytes.extend_from_slice(&wtf8_bytes[pos..surrogate_pos]); + utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); + pos = surrogate_pos + 3; + } + None => { + utf8_bytes.extend_from_slice(&wtf8_bytes[pos..]); + return Cow::Owned(unsafe { String::from_utf8_unchecked(utf8_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] + 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] + fn final_lead_surrogate(&self) -> Option { + match self.bytes { + [.., 0xED, b2 @ 0xA0..=0xAF, b3] => Some(decode_surrogate(b2, b3)), + _ => None, + } + } + + #[inline] + fn initial_trail_surrogate(&self) -> Option { + match self.bytes { + [0xED, b2 @ 0xB0..=0xBF, b3, ..] => Some(decode_surrogate(b2, b3)), + _ => None, + } + } + + pub fn clone_into(&self, buf: &mut Wtf8Buf) { + buf.is_known_utf8 = false; + self.bytes.clone_into(&mut buf.bytes); + } + + /// Boxes this `Wtf8`. + #[inline] + pub fn into_box(&self) -> Box { + let boxed: Box<[u8]> = self.bytes.into(); + unsafe { mem::transmute(boxed) } + } + + /// Creates a boxed, empty `Wtf8`. + pub fn empty_box() -> Box { + let boxed: Box<[u8]> = Default::default(); + unsafe { mem::transmute(boxed) } + } + + #[inline] + pub fn into_arc(&self) -> Arc { + let arc: Arc<[u8]> = Arc::from(&self.bytes); + unsafe { Arc::from_raw(Arc::into_raw(arc) as *const Wtf8) } + } + + #[inline] + pub fn into_rc(&self) -> Rc { + let rc: Rc<[u8]> = Rc::from(&self.bytes); + unsafe { Rc::from_raw(Rc::into_raw(rc) as *const Wtf8) } + } + + #[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 to_ascii_lowercase(&self) -> Wtf8Buf { + Wtf8Buf { bytes: self.bytes.to_ascii_lowercase(), is_known_utf8: false } + } + + #[inline] + pub fn to_ascii_uppercase(&self) -> Wtf8Buf { + Wtf8Buf { bytes: self.bytes.to_ascii_uppercase(), is_known_utf8: false } + } + + #[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> for Wtf8 { + type Output = Wtf8; + + #[inline] + fn index(&self, range: ops::Range) -> &Wtf8 { + // is_code_point_boundary checks that the index is in [0, .len()] + if range.start <= range.end + && is_code_point_boundary(self, range.start) + && is_code_point_boundary(self, range.end) + { + 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> for Wtf8 { + type Output = Wtf8; + + #[inline] + fn index(&self, range: ops::RangeFrom) -> &Wtf8 { + // is_code_point_boundary checks that the index is in [0, .len()] + if is_code_point_boundary(self, range.start) { + 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> for Wtf8 { + type Output = Wtf8; + + #[inline] + fn index(&self, range: ops::RangeTo) -> &Wtf8 { + // is_code_point_boundary checks that the index is in [0, .len()] + if is_code_point_boundary(self, range.end) { + unsafe { slice_unchecked(self, 0, range.end) } + } else { + slice_error_fail(self, 0, range.end) + } + } +} + +impl ops::Index 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 +} + +#[inline] +fn decode_surrogate_pair(lead: u16, trail: u16) -> char { + let code_point = 0x10000 + ((((lead - 0xD800) as u32) << 10) | (trail - 0xDC00) as u32); + unsafe { char::from_u32_unchecked(code_point) } +} + +/// Copied from str::is_char_boundary +#[inline] +pub fn is_code_point_boundary(slice: &Wtf8, index: usize) -> bool { + if index == 0 { + return true; + } + match slice.bytes.get(index) { + None => index == slice.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::slice_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(slice: &Wtf8, index: usize) { + if index == 0 { + return; + } + match slice.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 == slice.len() => return, + None => panic!("byte index {index} is out of bounds"), + } + if slice.bytes[index + 1] >= 0xA0 { + // There's a surrogate after index. Now check before index. + if index >= 3 && slice.bytes[index - 3] == 0xED && slice.bytes[index - 2] >= 0xA0 { + panic!("byte index {index} lies between surrogate codepoints"); + } + } +} + +/// Copied from core::str::raw::slice_unchecked +#[inline] +pub 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)] +pub 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)] +pub struct Wtf8CodePoints<'a> { + bytes: slice::Iter<'a, u8>, +} + +impl Iterator for Wtf8CodePoints<'_> { + type Item = CodePoint; + + #[inline] + fn next(&mut self) -> Option { + // SAFETY: `self.bytes` has been created from a WTF-8 string + unsafe { next_code_point(&mut self.bytes).map(|c| CodePoint { value: c }) } + } + + #[inline] + fn size_hint(&self) -> (usize, Option) { + let len = self.bytes.len(); + (len.saturating_add(3) / 4, Some(len)) + } +} + +/// Generates a wide character sequence for potentially ill-formed UTF-16. +#[stable(feature = "rust1", since = "1.0.0")] +#[derive(Clone)] +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 { + 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.value, &mut buf).len(); + if n == 2 { + self.extra = buf[1]; + } + buf[0] + }) + } + + #[inline] + fn size_hint(&self) -> (usize, Option) { + 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))) + } +} + +#[stable(feature = "encode_wide_fused_iterator", since = "1.62.0")] +impl FusedIterator for EncodeWide<'_> {} + +impl Hash for CodePoint { + #[inline] + fn hash(&self, state: &mut H) { + self.value.hash(state) + } +} + +impl Hash for Wtf8Buf { + #[inline] + fn hash(&self, state: &mut H) { + state.write(&self.bytes); + 0xfeu8.hash(state) + } +} + +impl Hash for Wtf8 { + #[inline] + fn hash(&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) } + } +} diff --git a/library/alloc/src/wtf8/tests.rs b/library/alloc/src/wtf8/tests.rs new file mode 100644 index 00000000000..b57c99a8452 --- /dev/null +++ b/library/alloc/src/wtf8/tests.rs @@ -0,0 +1,751 @@ +use super::*; + +#[test] +fn code_point_from_u32() { + assert!(CodePoint::from_u32(0).is_some()); + assert!(CodePoint::from_u32(0xD800).is_some()); + assert!(CodePoint::from_u32(0x10FFFF).is_some()); + assert!(CodePoint::from_u32(0x110000).is_none()); +} + +#[test] +fn code_point_to_u32() { + fn c(value: u32) -> CodePoint { + CodePoint::from_u32(value).unwrap() + } + assert_eq!(c(0).to_u32(), 0); + assert_eq!(c(0xD800).to_u32(), 0xD800); + assert_eq!(c(0x10FFFF).to_u32(), 0x10FFFF); +} + +#[test] +fn code_point_to_lead_surrogate() { + fn c(value: u32) -> CodePoint { + CodePoint::from_u32(value).unwrap() + } + assert_eq!(c(0).to_lead_surrogate(), None); + assert_eq!(c(0xE9).to_lead_surrogate(), None); + assert_eq!(c(0xD800).to_lead_surrogate(), Some(0xD800)); + assert_eq!(c(0xDBFF).to_lead_surrogate(), Some(0xDBFF)); + assert_eq!(c(0xDC00).to_lead_surrogate(), None); + assert_eq!(c(0xDFFF).to_lead_surrogate(), None); + assert_eq!(c(0x1F4A9).to_lead_surrogate(), None); + assert_eq!(c(0x10FFFF).to_lead_surrogate(), None); +} + +#[test] +fn code_point_to_trail_surrogate() { + fn c(value: u32) -> CodePoint { + CodePoint::from_u32(value).unwrap() + } + assert_eq!(c(0).to_trail_surrogate(), None); + assert_eq!(c(0xE9).to_trail_surrogate(), None); + assert_eq!(c(0xD800).to_trail_surrogate(), None); + assert_eq!(c(0xDBFF).to_trail_surrogate(), None); + assert_eq!(c(0xDC00).to_trail_surrogate(), Some(0xDC00)); + assert_eq!(c(0xDFFF).to_trail_surrogate(), Some(0xDFFF)); + assert_eq!(c(0x1F4A9).to_trail_surrogate(), None); + assert_eq!(c(0x10FFFF).to_trail_surrogate(), None); +} + +#[test] +fn code_point_from_char() { + assert_eq!(CodePoint::from_char('a').to_u32(), 0x61); + assert_eq!(CodePoint::from_char('💩').to_u32(), 0x1F4A9); +} + +#[test] +fn code_point_to_string() { + assert_eq!(format!("{:?}", CodePoint::from_char('a')), "U+0061"); + assert_eq!(format!("{:?}", CodePoint::from_char('💩')), "U+1F4A9"); +} + +#[test] +fn code_point_to_char() { + fn c(value: u32) -> CodePoint { + CodePoint::from_u32(value).unwrap() + } + assert_eq!(c(0x61).to_char(), Some('a')); + assert_eq!(c(0x1F4A9).to_char(), Some('💩')); + assert_eq!(c(0xD800).to_char(), None); +} + +#[test] +fn code_point_to_char_lossy() { + fn c(value: u32) -> CodePoint { + CodePoint::from_u32(value).unwrap() + } + assert_eq!(c(0x61).to_char_lossy(), 'a'); + assert_eq!(c(0x1F4A9).to_char_lossy(), '💩'); + assert_eq!(c(0xD800).to_char_lossy(), '\u{FFFD}'); +} + +#[test] +fn wtf8buf_new() { + assert_eq!(Wtf8Buf::new().bytes, b""); +} + +#[test] +fn wtf8buf_from_str() { + assert_eq!(Wtf8Buf::from_str("").bytes, b""); + assert_eq!(Wtf8Buf::from_str("aé 💩").bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); +} + +#[test] +fn wtf8buf_from_string() { + assert_eq!(Wtf8Buf::from_string(String::from("")).bytes, b""); + assert_eq!(Wtf8Buf::from_string(String::from("aé 💩")).bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); +} + +#[test] +fn wtf8buf_from_wide() { + let buf = Wtf8Buf::from_wide(&[]); + assert_eq!(buf.bytes, b""); + assert!(buf.is_known_utf8); + + let buf = Wtf8Buf::from_wide(&[0x61, 0xE9, 0x20, 0xD83D, 0xDCA9]); + assert_eq!(buf.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert!(buf.is_known_utf8); + + let buf = Wtf8Buf::from_wide(&[0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]); + assert_eq!(buf.bytes, b"a\xC3\xA9 \xED\xA0\xBD\xF0\x9F\x92\xA9"); + assert!(!buf.is_known_utf8); + + let buf = Wtf8Buf::from_wide(&[0xD800]); + assert_eq!(buf.bytes, b"\xED\xA0\x80"); + assert!(!buf.is_known_utf8); + + let buf = Wtf8Buf::from_wide(&[0xDBFF]); + assert_eq!(buf.bytes, b"\xED\xAF\xBF"); + assert!(!buf.is_known_utf8); + + let buf = Wtf8Buf::from_wide(&[0xDC00]); + assert_eq!(buf.bytes, b"\xED\xB0\x80"); + assert!(!buf.is_known_utf8); + + let buf = Wtf8Buf::from_wide(&[0xDFFF]); + assert_eq!(buf.bytes, b"\xED\xBF\xBF"); + assert!(!buf.is_known_utf8); +} + +#[test] +fn wtf8buf_push_str() { + let mut string = Wtf8Buf::new(); + assert_eq!(string.bytes, b""); + assert!(string.is_known_utf8); + + string.push_str("aé 💩"); + assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert!(string.is_known_utf8); +} + +#[test] +fn wtf8buf_push_char() { + let mut string = Wtf8Buf::from_str("aé "); + assert_eq!(string.bytes, b"a\xC3\xA9 "); + assert!(string.is_known_utf8); + + string.push_char('💩'); + assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert!(string.is_known_utf8); +} + +#[test] +fn wtf8buf_push() { + let mut string = Wtf8Buf::from_str("aé "); + assert_eq!(string.bytes, b"a\xC3\xA9 "); + assert!(string.is_known_utf8); + + string.push(CodePoint::from_char('💩')); + assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert!(string.is_known_utf8); + + fn c(value: u32) -> CodePoint { + CodePoint::from_u32(value).unwrap() + } + + let mut string = Wtf8Buf::new(); + string.push(c(0xD83D)); // lead + assert!(!string.is_known_utf8); + string.push(c(0xDCA9)); // trail + assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); // Magic! + + let mut string = Wtf8Buf::new(); + string.push(c(0xD83D)); // lead + assert!(!string.is_known_utf8); + string.push(c(0x20)); // not surrogate + string.push(c(0xDCA9)); // trail + assert_eq!(string.bytes, b"\xED\xA0\xBD \xED\xB2\xA9"); + + let mut string = Wtf8Buf::new(); + string.push(c(0xD800)); // lead + assert!(!string.is_known_utf8); + string.push(c(0xDBFF)); // lead + assert_eq!(string.bytes, b"\xED\xA0\x80\xED\xAF\xBF"); + + let mut string = Wtf8Buf::new(); + string.push(c(0xD800)); // lead + assert!(!string.is_known_utf8); + string.push(c(0xE000)); // not surrogate + assert_eq!(string.bytes, b"\xED\xA0\x80\xEE\x80\x80"); + + let mut string = Wtf8Buf::new(); + string.push(c(0xD7FF)); // not surrogate + assert!(string.is_known_utf8); + string.push(c(0xDC00)); // trail + assert!(!string.is_known_utf8); + assert_eq!(string.bytes, b"\xED\x9F\xBF\xED\xB0\x80"); + + let mut string = Wtf8Buf::new(); + string.push(c(0x61)); // not surrogate, < 3 bytes + assert!(string.is_known_utf8); + string.push(c(0xDC00)); // trail + assert!(!string.is_known_utf8); + assert_eq!(string.bytes, b"\x61\xED\xB0\x80"); + + let mut string = Wtf8Buf::new(); + string.push(c(0xDC00)); // trail + assert!(!string.is_known_utf8); + assert_eq!(string.bytes, b"\xED\xB0\x80"); +} + +#[test] +fn wtf8buf_push_wtf8() { + let mut string = Wtf8Buf::from_str("aé"); + assert_eq!(string.bytes, b"a\xC3\xA9"); + string.push_wtf8(Wtf8::from_str(" 💩")); + assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert!(string.is_known_utf8); + + fn w(v: &[u8]) -> &Wtf8 { + unsafe { Wtf8::from_bytes_unchecked(v) } + } + + let mut string = Wtf8Buf::new(); + string.push_wtf8(w(b"\xED\xA0\xBD")); // lead + string.push_wtf8(w(b"\xED\xB2\xA9")); // trail + assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); // Magic! + + let mut string = Wtf8Buf::new(); + string.push_wtf8(w(b"\xED\xA0\xBD")); // lead + string.push_wtf8(w(b" ")); // not surrogate + string.push_wtf8(w(b"\xED\xB2\xA9")); // trail + assert_eq!(string.bytes, b"\xED\xA0\xBD \xED\xB2\xA9"); + assert!(!string.is_known_utf8); + + let mut string = Wtf8Buf::new(); + string.push_wtf8(w(b"\xED\xA0\x80")); // lead + string.push_wtf8(w(b"\xED\xAF\xBF")); // lead + assert_eq!(string.bytes, b"\xED\xA0\x80\xED\xAF\xBF"); + assert!(!string.is_known_utf8); + + let mut string = Wtf8Buf::new(); + string.push_wtf8(w(b"\xED\xA0\x80")); // lead + string.push_wtf8(w(b"\xEE\x80\x80")); // not surrogate + assert_eq!(string.bytes, b"\xED\xA0\x80\xEE\x80\x80"); + assert!(!string.is_known_utf8); + + let mut string = Wtf8Buf::new(); + string.push_wtf8(w(b"\xED\x9F\xBF")); // not surrogate + string.push_wtf8(w(b"\xED\xB0\x80")); // trail + assert_eq!(string.bytes, b"\xED\x9F\xBF\xED\xB0\x80"); + assert!(!string.is_known_utf8); + + let mut string = Wtf8Buf::new(); + string.push_wtf8(w(b"a")); // not surrogate, < 3 bytes + string.push_wtf8(w(b"\xED\xB0\x80")); // trail + assert_eq!(string.bytes, b"\x61\xED\xB0\x80"); + assert!(!string.is_known_utf8); + + let mut string = Wtf8Buf::new(); + string.push_wtf8(w(b"\xED\xB0\x80")); // trail + assert_eq!(string.bytes, b"\xED\xB0\x80"); + assert!(!string.is_known_utf8); +} + +#[test] +fn wtf8buf_truncate() { + let mut string = Wtf8Buf::from_str("aé"); + assert!(string.is_known_utf8); + + string.truncate(3); + assert_eq!(string.bytes, b"a\xC3\xA9"); + assert!(string.is_known_utf8); + + string.truncate(1); + assert_eq!(string.bytes, b"a"); + assert!(string.is_known_utf8); + + string.truncate(0); + assert_eq!(string.bytes, b""); + assert!(string.is_known_utf8); +} + +#[test] +fn wtf8buf_truncate_around_non_bmp() { + let mut string = Wtf8Buf::from_str("💩"); + assert!(string.is_known_utf8); + + string.truncate(4); + assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); + assert!(string.is_known_utf8); + + string.truncate(0); + assert_eq!(string.bytes, b""); + assert!(string.is_known_utf8); +} + +#[test] +#[should_panic] +fn wtf8buf_truncate_fail_code_point_boundary() { + let mut string = Wtf8Buf::from_str("aé"); + string.truncate(2); +} + +#[test] +#[should_panic] +fn wtf8buf_truncate_fail_longer() { + let mut string = Wtf8Buf::from_str("aé"); + string.truncate(4); +} + +#[test] +#[should_panic] +fn wtf8buf_truncate_splitting_non_bmp3() { + let mut string = Wtf8Buf::from_str("💩"); + assert!(string.is_known_utf8); + string.truncate(3); +} + +#[test] +#[should_panic] +fn wtf8buf_truncate_splitting_non_bmp2() { + let mut string = Wtf8Buf::from_str("💩"); + assert!(string.is_known_utf8); + string.truncate(2); +} + +#[test] +#[should_panic] +fn wtf8buf_truncate_splitting_non_bmp1() { + let mut string = Wtf8Buf::from_str("💩"); + assert!(string.is_known_utf8); + string.truncate(1); +} + +#[test] +fn wtf8buf_into_string() { + let mut string = Wtf8Buf::from_str("aé 💩"); + assert!(string.is_known_utf8); + assert_eq!(string.clone().into_string(), Ok(String::from("aé 💩"))); + string.push(CodePoint::from_u32(0xD800).unwrap()); + assert!(!string.is_known_utf8); + assert_eq!(string.clone().into_string(), Err(string)); +} + +#[test] +fn wtf8buf_into_string_lossy() { + let mut string = Wtf8Buf::from_str("aé 💩"); + assert_eq!(string.clone().into_string_lossy(), String::from("aé 💩")); + string.push(CodePoint::from_u32(0xD800).unwrap()); + assert_eq!(string.clone().into_string_lossy(), String::from("aé 💩�")); +} + +#[test] +fn wtf8buf_from_iterator() { + fn f(values: &[u32]) -> Wtf8Buf { + values.iter().map(|&c| CodePoint::from_u32(c).unwrap()).collect::() + } + assert_eq!( + f(&[0x61, 0xE9, 0x20, 0x1F4A9]), + Wtf8Buf { bytes: b"a\xC3\xA9 \xF0\x9F\x92\xA9".to_vec(), is_known_utf8: true } + ); + + assert_eq!(f(&[0xD83D, 0xDCA9]).bytes, b"\xF0\x9F\x92\xA9"); // Magic! + assert_eq!( + f(&[0xD83D, 0x20, 0xDCA9]), + Wtf8Buf { bytes: b"\xED\xA0\xBD \xED\xB2\xA9".to_vec(), is_known_utf8: false } + ); + assert_eq!( + f(&[0xD800, 0xDBFF]), + Wtf8Buf { bytes: b"\xED\xA0\x80\xED\xAF\xBF".to_vec(), is_known_utf8: false } + ); + assert_eq!( + f(&[0xD800, 0xE000]), + Wtf8Buf { bytes: b"\xED\xA0\x80\xEE\x80\x80".to_vec(), is_known_utf8: false } + ); + assert_eq!( + f(&[0xD7FF, 0xDC00]), + Wtf8Buf { bytes: b"\xED\x9F\xBF\xED\xB0\x80".to_vec(), is_known_utf8: false } + ); + assert_eq!( + f(&[0x61, 0xDC00]), + Wtf8Buf { bytes: b"\x61\xED\xB0\x80".to_vec(), is_known_utf8: false } + ); + assert_eq!(f(&[0xDC00]), Wtf8Buf { bytes: b"\xED\xB0\x80".to_vec(), is_known_utf8: false }); +} + +#[test] +fn wtf8buf_extend() { + fn e(initial: &[u32], extended: &[u32]) -> Wtf8Buf { + fn c(value: &u32) -> CodePoint { + CodePoint::from_u32(*value).unwrap() + } + let mut string = initial.iter().map(c).collect::(); + string.extend(extended.iter().map(c)); + string + } + + assert_eq!( + e(&[0x61, 0xE9], &[0x20, 0x1F4A9]), + Wtf8Buf { bytes: b"a\xC3\xA9 \xF0\x9F\x92\xA9".to_vec(), is_known_utf8: true } + ); + + assert_eq!(e(&[0xD83D], &[0xDCA9]).bytes, b"\xF0\x9F\x92\xA9"); // Magic! + assert_eq!( + e(&[0xD83D, 0x20], &[0xDCA9]), + Wtf8Buf { bytes: b"\xED\xA0\xBD \xED\xB2\xA9".to_vec(), is_known_utf8: false } + ); + assert_eq!( + e(&[0xD800], &[0xDBFF]), + Wtf8Buf { bytes: b"\xED\xA0\x80\xED\xAF\xBF".to_vec(), is_known_utf8: false } + ); + assert_eq!( + e(&[0xD800], &[0xE000]), + Wtf8Buf { bytes: b"\xED\xA0\x80\xEE\x80\x80".to_vec(), is_known_utf8: false } + ); + assert_eq!( + e(&[0xD7FF], &[0xDC00]), + Wtf8Buf { bytes: b"\xED\x9F\xBF\xED\xB0\x80".to_vec(), is_known_utf8: false } + ); + assert_eq!( + e(&[0x61], &[0xDC00]), + Wtf8Buf { bytes: b"\x61\xED\xB0\x80".to_vec(), is_known_utf8: false } + ); + assert_eq!( + e(&[], &[0xDC00]), + Wtf8Buf { bytes: b"\xED\xB0\x80".to_vec(), is_known_utf8: false } + ); +} + +#[test] +fn wtf8buf_show() { + let mut string = Wtf8Buf::from_str("a\té \u{7f}💩\r"); + string.push(CodePoint::from_u32(0xD800).unwrap()); + assert_eq!(format!("{string:?}"), "\"a\\té \\u{7f}\u{1f4a9}\\r\\u{d800}\""); +} + +#[test] +fn wtf8buf_as_slice() { + assert_eq!(Wtf8Buf::from_str("aé").as_slice(), Wtf8::from_str("aé")); +} + +#[test] +fn wtf8buf_show_str() { + let text = "a\té 💩\r"; + let string = Wtf8Buf::from_str(text); + assert_eq!(format!("{text:?}"), format!("{string:?}")); +} + +#[test] +fn wtf8_from_str() { + assert_eq!(&Wtf8::from_str("").bytes, b""); + assert_eq!(&Wtf8::from_str("aé 💩").bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); +} + +#[test] +fn wtf8_len() { + assert_eq!(Wtf8::from_str("").len(), 0); + assert_eq!(Wtf8::from_str("aé 💩").len(), 8); +} + +#[test] +fn wtf8_slice() { + assert_eq!(&Wtf8::from_str("aé 💩")[1..4].bytes, b"\xC3\xA9 "); +} + +#[test] +#[should_panic] +fn wtf8_slice_not_code_point_boundary() { + let _ = &Wtf8::from_str("aé 💩")[2..4]; +} + +#[test] +fn wtf8_slice_from() { + assert_eq!(&Wtf8::from_str("aé 💩")[1..].bytes, b"\xC3\xA9 \xF0\x9F\x92\xA9"); +} + +#[test] +#[should_panic] +fn wtf8_slice_from_not_code_point_boundary() { + let _ = &Wtf8::from_str("aé 💩")[2..]; +} + +#[test] +fn wtf8_slice_to() { + assert_eq!(&Wtf8::from_str("aé 💩")[..4].bytes, b"a\xC3\xA9 "); +} + +#[test] +#[should_panic] +fn wtf8_slice_to_not_code_point_boundary() { + let _ = &Wtf8::from_str("aé 💩")[5..]; +} + +#[test] +fn wtf8_ascii_byte_at() { + let slice = Wtf8::from_str("aé 💩"); + assert_eq!(slice.ascii_byte_at(0), b'a'); + assert_eq!(slice.ascii_byte_at(1), b'\xFF'); + assert_eq!(slice.ascii_byte_at(2), b'\xFF'); + assert_eq!(slice.ascii_byte_at(3), b' '); + assert_eq!(slice.ascii_byte_at(4), b'\xFF'); +} + +#[test] +fn wtf8_code_points() { + fn c(value: u32) -> CodePoint { + CodePoint::from_u32(value).unwrap() + } + fn cp(string: &Wtf8Buf) -> Vec> { + string.code_points().map(|c| c.to_char()).collect::>() + } + let mut string = Wtf8Buf::from_str("é "); + assert_eq!(cp(&string), [Some('é'), Some(' ')]); + string.push(c(0xD83D)); + assert_eq!(cp(&string), [Some('é'), Some(' '), None]); + string.push(c(0xDCA9)); + assert_eq!(cp(&string), [Some('é'), Some(' '), Some('💩')]); +} + +#[test] +fn wtf8_as_str() { + assert_eq!(Wtf8::from_str("").as_str(), Ok("")); + assert_eq!(Wtf8::from_str("aé 💩").as_str(), Ok("aé 💩")); + let mut string = Wtf8Buf::new(); + string.push(CodePoint::from_u32(0xD800).unwrap()); + assert!(string.as_str().is_err()); +} + +#[test] +fn wtf8_to_string_lossy() { + assert_eq!(Wtf8::from_str("").to_string_lossy(), Cow::Borrowed("")); + assert_eq!(Wtf8::from_str("aé 💩").to_string_lossy(), Cow::Borrowed("aé 💩")); + let mut string = Wtf8Buf::from_str("aé 💩"); + string.push(CodePoint::from_u32(0xD800).unwrap()); + let expected: Cow<'_, str> = Cow::Owned(String::from("aé 💩�")); + assert_eq!(string.to_string_lossy(), expected); +} + +#[test] +fn wtf8_display() { + fn d(b: &[u8]) -> String { + (&unsafe { Wtf8::from_bytes_unchecked(b) }).to_string() + } + + assert_eq!("", d("".as_bytes())); + assert_eq!("aé 💩", d("aé 💩".as_bytes())); + + let mut string = Wtf8Buf::from_str("aé 💩"); + string.push(CodePoint::from_u32(0xD800).unwrap()); + assert_eq!("aé 💩�", d(string.as_inner())); +} + +#[test] +fn wtf8_encode_wide() { + let mut string = Wtf8Buf::from_str("aé "); + string.push(CodePoint::from_u32(0xD83D).unwrap()); + string.push_char('💩'); + assert_eq!( + string.encode_wide().collect::>(), + vec![0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9] + ); +} + +#[test] +fn wtf8_encode_wide_size_hint() { + let string = Wtf8Buf::from_str("\u{12345}"); + let mut iter = string.encode_wide(); + assert_eq!((1, Some(8)), iter.size_hint()); + iter.next().unwrap(); + assert_eq!((1, Some(1)), iter.size_hint()); + iter.next().unwrap(); + assert_eq!((0, Some(0)), iter.size_hint()); + assert!(iter.next().is_none()); +} + +#[test] +fn wtf8_clone_into() { + let mut string = Wtf8Buf::new(); + Wtf8::from_str("green").clone_into(&mut string); + assert_eq!(string.bytes, b"green"); + + let mut string = Wtf8Buf::from_str("green"); + Wtf8::from_str("").clone_into(&mut string); + assert_eq!(string.bytes, b""); + + let mut string = Wtf8Buf::from_str("red"); + Wtf8::from_str("green").clone_into(&mut string); + assert_eq!(string.bytes, b"green"); + + let mut string = Wtf8Buf::from_str("green"); + Wtf8::from_str("red").clone_into(&mut string); + assert_eq!(string.bytes, b"red"); + + let mut string = Wtf8Buf::from_str("green"); + assert!(string.is_known_utf8); + unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").clone_into(&mut string) }; + assert_eq!(string.bytes, b"\xED\xA0\x80"); + assert!(!string.is_known_utf8); +} + +#[test] +fn wtf8_to_ascii_lowercase() { + let lowercase = Wtf8::from_str("").to_ascii_lowercase(); + assert_eq!(lowercase.bytes, b""); + + let lowercase = Wtf8::from_str("GrEeN gRaPeS! 🍇").to_ascii_lowercase(); + assert_eq!(lowercase.bytes, b"green grapes! \xf0\x9f\x8d\x87"); + + let lowercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_ascii_lowercase() }; + assert_eq!(lowercase.bytes, b"\xED\xA0\x80"); + assert!(!lowercase.is_known_utf8); +} + +#[test] +fn wtf8_to_ascii_uppercase() { + let uppercase = Wtf8::from_str("").to_ascii_uppercase(); + assert_eq!(uppercase.bytes, b""); + + let uppercase = Wtf8::from_str("GrEeN gRaPeS! 🍇").to_ascii_uppercase(); + assert_eq!(uppercase.bytes, b"GREEN GRAPES! \xf0\x9f\x8d\x87"); + + let uppercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_ascii_uppercase() }; + assert_eq!(uppercase.bytes, b"\xED\xA0\x80"); + assert!(!uppercase.is_known_utf8); +} + +#[test] +fn wtf8_make_ascii_lowercase() { + let mut lowercase = Wtf8Buf::from_str(""); + lowercase.make_ascii_lowercase(); + assert_eq!(lowercase.bytes, b""); + + let mut lowercase = Wtf8Buf::from_str("GrEeN gRaPeS! 🍇"); + lowercase.make_ascii_lowercase(); + assert_eq!(lowercase.bytes, b"green grapes! \xf0\x9f\x8d\x87"); + + let mut lowercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; + lowercase.make_ascii_lowercase(); + assert_eq!(lowercase.bytes, b"\xED\xA0\x80"); + assert!(!lowercase.is_known_utf8); +} + +#[test] +fn wtf8_make_ascii_uppercase() { + let mut uppercase = Wtf8Buf::from_str(""); + uppercase.make_ascii_uppercase(); + assert_eq!(uppercase.bytes, b""); + + let mut uppercase = Wtf8Buf::from_str("GrEeN gRaPeS! 🍇"); + uppercase.make_ascii_uppercase(); + assert_eq!(uppercase.bytes, b"GREEN GRAPES! \xf0\x9f\x8d\x87"); + + let mut uppercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; + uppercase.make_ascii_uppercase(); + assert_eq!(uppercase.bytes, b"\xED\xA0\x80"); + assert!(!uppercase.is_known_utf8); +} + +#[test] +fn wtf8_to_owned() { + let string = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; + assert_eq!(string.bytes, b"\xED\xA0\x80"); + assert!(!string.is_known_utf8); +} + +#[test] +fn wtf8_valid_utf8_boundaries() { + let mut string = Wtf8Buf::from_str("aé 💩"); + string.push(CodePoint::from_u32(0xD800).unwrap()); + string.push(CodePoint::from_u32(0xD800).unwrap()); + check_utf8_boundary(&string, 0); + check_utf8_boundary(&string, 1); + check_utf8_boundary(&string, 3); + check_utf8_boundary(&string, 4); + check_utf8_boundary(&string, 8); + check_utf8_boundary(&string, 14); + assert_eq!(string.len(), 14); + + string.push_char('a'); + check_utf8_boundary(&string, 14); + check_utf8_boundary(&string, 15); + + let mut string = Wtf8Buf::from_str("a"); + string.push(CodePoint::from_u32(0xD800).unwrap()); + check_utf8_boundary(&string, 1); + + let mut string = Wtf8Buf::from_str("\u{D7FF}"); + string.push(CodePoint::from_u32(0xD800).unwrap()); + check_utf8_boundary(&string, 3); + + let mut string = Wtf8Buf::new(); + string.push(CodePoint::from_u32(0xD800).unwrap()); + string.push_char('\u{D7FF}'); + check_utf8_boundary(&string, 3); +} + +#[test] +#[should_panic(expected = "byte index 4 is out of bounds")] +fn wtf8_utf8_boundary_out_of_bounds() { + let string = Wtf8::from_str("aé"); + check_utf8_boundary(&string, 4); +} + +#[test] +#[should_panic(expected = "byte index 1 is not a codepoint boundary")] +fn wtf8_utf8_boundary_inside_codepoint() { + let string = Wtf8::from_str("é"); + check_utf8_boundary(&string, 1); +} + +#[test] +#[should_panic(expected = "byte index 1 is not a codepoint boundary")] +fn wtf8_utf8_boundary_inside_surrogate() { + let mut string = Wtf8Buf::new(); + string.push(CodePoint::from_u32(0xD800).unwrap()); + check_utf8_boundary(&string, 1); +} + +#[test] +#[should_panic(expected = "byte index 3 lies between surrogate codepoints")] +fn wtf8_utf8_boundary_between_surrogates() { + let mut string = Wtf8Buf::new(); + string.push(CodePoint::from_u32(0xD800).unwrap()); + string.push(CodePoint::from_u32(0xD800).unwrap()); + check_utf8_boundary(&string, 3); +} + +#[test] +fn wobbled_wtf8_plus_bytes_isnt_utf8() { + let mut string: Wtf8Buf = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; + assert!(!string.is_known_utf8); + string.extend_from_slice(b"some utf-8"); + assert!(!string.is_known_utf8); +} + +#[test] +fn wobbled_wtf8_plus_str_isnt_utf8() { + let mut string: Wtf8Buf = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; + assert!(!string.is_known_utf8); + string.push_str("some utf-8"); + assert!(!string.is_known_utf8); +} + +#[test] +fn unwobbly_wtf8_plus_utf8_is_utf8() { + let mut string: Wtf8Buf = Wtf8Buf::from_str("hello world"); + assert!(string.is_known_utf8); + string.push_str("some utf-8"); + assert!(string.is_known_utf8); +} diff --git a/library/core/src/wtf8.rs b/library/core/src/wtf8.rs new file mode 100644 index 00000000000..50bde88b5a4 --- /dev/null +++ b/library/core/src/wtf8.rs @@ -0,0 +1,1083 @@ +//! 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. + +// this module is imported from @SimonSapin's repo and has tons of dead code on +// unix (it's mostly used on windows), so don't worry about dead code here. +#![allow(dead_code)] + +#[cfg(test)] +mod tests; + +use core::char::{MAX_LEN_UTF8, MAX_LEN_UTF16, encode_utf8_raw, encode_utf16_raw}; +use core::clone::CloneToUninit; +use core::str::next_code_point; + +use crate::borrow::Cow; +use crate::collections::TryReserveError; +use crate::hash::{Hash, Hasher}; +use crate::iter::FusedIterator; +use crate::rc::Rc; +use crate::sync::Arc; +use crate::sys_common::AsInner; +use crate::{fmt, mem, ops, slice, str}; + +const UTF8_REPLACEMENT_CHARACTER: &str = "\u{FFFD}"; + +/// 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)] +pub struct CodePoint { + value: u32, +} + +/// 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.value) + } +} + +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 { + CodePoint { 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 { + match value { + 0..=0x10FFFF => Some(CodePoint { value }), + _ => None, + } + } + + /// 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 { + CodePoint { value: value as u32 } + } + + /// Returns the numeric value of the code point. + #[inline] + pub fn to_u32(&self) -> u32 { + self.value + } + + /// Returns the numeric value of the code point if it is a leading surrogate. + #[inline] + pub fn to_lead_surrogate(&self) -> Option { + match self.value { + 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 { + match self.value { + 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 { + match self.value { + 0xD800..=0xDFFF => None, + _ => Some(unsafe { char::from_u32_unchecked(self.value) }), + } + } + + /// 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('\u{FFFD}') + } +} + +/// An owned, growable string of well-formed WTF-8 data. +/// +/// Similar to `String`, but can additionally contain surrogate code points +/// if they’re not in a surrogate pair. +#[derive(Eq, PartialEq, Ord, PartialOrd, Clone)] +pub struct Wtf8Buf { + bytes: Vec, + + /// Do we know that `bytes` holds a valid UTF-8 encoding? We can easily + /// know this if we're constructed from a `String` or `&str`. + /// + /// It is possible for `bytes` to have valid UTF-8 without this being + /// set, such as when we're concatenating `&Wtf8`'s and surrogates become + /// paired, as we don't bother to rescan the entire string. + is_known_utf8: bool, +} + +impl ops::Deref for Wtf8Buf { + type Target = Wtf8; + + fn deref(&self) -> &Wtf8 { + self.as_slice() + } +} + +impl ops::DerefMut for Wtf8Buf { + fn deref_mut(&mut self) -> &mut Wtf8 { + self.as_mut_slice() + } +} + +/// 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. +/// +/// For example, the code units [U+0061, U+D800, U+000A] are formatted as +/// `"a\u{D800}\n"`. +impl fmt::Debug for Wtf8Buf { + #[inline] + fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt::Debug::fmt(&**self, formatter) + } +} + +/// Formats the string with unpaired surrogates substituted with the replacement +/// character, U+FFFD. +impl fmt::Display for Wtf8Buf { + fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { + if let Some(s) = self.as_known_utf8() { + fmt::Display::fmt(s, formatter) + } else { + fmt::Display::fmt(&**self, formatter) + } + } +} + +impl Wtf8Buf { + /// Creates a new, empty WTF-8 string. + #[inline] + pub fn new() -> Wtf8Buf { + Wtf8Buf { bytes: Vec::new(), is_known_utf8: true } + } + + /// Creates a new, empty WTF-8 string with pre-allocated capacity for `capacity` bytes. + #[inline] + pub fn with_capacity(capacity: usize) -> Wtf8Buf { + Wtf8Buf { bytes: Vec::with_capacity(capacity), is_known_utf8: true } + } + + /// Creates a WTF-8 string from a WTF-8 byte vec. + /// + /// Since the byte vec is not checked for valid WTF-8, this function is + /// marked unsafe. + #[inline] + pub unsafe fn from_bytes_unchecked(value: Vec) -> Wtf8Buf { + Wtf8Buf { bytes: value, is_known_utf8: false } + } + + /// Creates a WTF-8 string from a UTF-8 `String`. + /// + /// This takes ownership of the `String` and does not copy. + /// + /// Since WTF-8 is a superset of UTF-8, this always succeeds. + #[inline] + pub const fn from_string(string: String) -> Wtf8Buf { + Wtf8Buf { bytes: string.into_bytes(), is_known_utf8: true } + } + + /// Creates a WTF-8 string from a UTF-8 `&str` slice. + /// + /// This copies the content of the slice. + /// + /// Since WTF-8 is a superset of UTF-8, this always succeeds. + #[inline] + pub fn from_str(s: &str) -> Wtf8Buf { + Wtf8Buf { bytes: s.as_bytes().to_vec(), is_known_utf8: true } + } + + pub fn clear(&mut self) { + self.bytes.clear(); + self.is_known_utf8 = true; + } + + /// Creates a WTF-8 string from a potentially ill-formed UTF-16 slice of 16-bit code units. + /// + /// This is lossless: calling `.encode_wide()` on the resulting string + /// will always return the original code units. + pub fn from_wide(v: &[u16]) -> Wtf8Buf { + let mut string = Wtf8Buf::with_capacity(v.len()); + for item in char::decode_utf16(v.iter().cloned()) { + match item { + Ok(ch) => string.push_char(ch), + Err(surrogate) => { + let surrogate = surrogate.unpaired_surrogate(); + // Surrogates are known to be in the code point range. + let code_point = unsafe { CodePoint::from_u32_unchecked(surrogate as u32) }; + // The string will now contain an unpaired surrogate. + string.is_known_utf8 = false; + // Skip the WTF-8 concatenation check, + // surrogate pairs are already decoded by decode_utf16 + string.push_code_point_unchecked(code_point); + } + } + } + string + } + + /// Appends the given `char` to the end of this string. + /// This does **not** include the WTF-8 concatenation check or `is_known_utf8` check. + /// Copied from String::push. + fn push_code_point_unchecked(&mut self, code_point: CodePoint) { + let mut bytes = [0; MAX_LEN_UTF8]; + let bytes = encode_utf8_raw(code_point.value, &mut bytes); + self.bytes.extend_from_slice(bytes) + } + + #[inline] + pub fn as_slice(&self) -> &Wtf8 { + unsafe { Wtf8::from_bytes_unchecked(&self.bytes) } + } + + #[inline] + pub fn as_mut_slice(&mut self) -> &mut Wtf8 { + // Safety: `Wtf8` doesn't expose any way to mutate the bytes that would + // cause them to change from well-formed UTF-8 to ill-formed UTF-8, + // which would break the assumptions of the `is_known_utf8` field. + unsafe { Wtf8::from_mut_bytes_unchecked(&mut self.bytes) } + } + + /// Converts the string to UTF-8 without validation, if it was created from + /// valid UTF-8. + #[inline] + fn as_known_utf8(&self) -> Option<&str> { + if self.is_known_utf8 { + // SAFETY: The buffer is known to be valid UTF-8. + Some(unsafe { str::from_utf8_unchecked(self.as_bytes()) }) + } else { + None + } + } + + /// Reserves capacity for at least `additional` more bytes to be inserted + /// in the given `Wtf8Buf`. + /// The collection may reserve more space to avoid frequent reallocations. + /// + /// # Panics + /// + /// Panics if the new capacity exceeds `isize::MAX` bytes. + #[inline] + pub fn reserve(&mut self, additional: usize) { + self.bytes.reserve(additional) + } + + /// Tries to reserve capacity for at least `additional` more bytes to be + /// inserted in the given `Wtf8Buf`. The `Wtf8Buf` may reserve more space to + /// avoid frequent reallocations. After calling `try_reserve`, capacity will + /// be greater than or equal to `self.len() + additional`. Does nothing if + /// capacity is already sufficient. This method preserves the contents even + /// if an error occurs. + /// + /// # Errors + /// + /// If the capacity overflows, or the allocator reports a failure, then an error + /// is returned. + #[inline] + pub fn try_reserve(&mut self, additional: usize) -> Result<(), TryReserveError> { + self.bytes.try_reserve(additional) + } + + #[inline] + pub fn reserve_exact(&mut self, additional: usize) { + self.bytes.reserve_exact(additional) + } + + /// Tries to reserve the minimum capacity for exactly `additional` more + /// bytes to be inserted in the given `Wtf8Buf`. After calling + /// `try_reserve_exact`, capacity will be greater than or equal to + /// `self.len() + additional` if it returns `Ok(())`. + /// Does nothing if the capacity is already sufficient. + /// + /// Note that the allocator may give the `Wtf8Buf` more space than it + /// requests. Therefore, capacity can not be relied upon to be precisely + /// minimal. Prefer [`try_reserve`] if future insertions are expected. + /// + /// [`try_reserve`]: Wtf8Buf::try_reserve + /// + /// # Errors + /// + /// If the capacity overflows, or the allocator reports a failure, then an error + /// is returned. + #[inline] + pub fn try_reserve_exact(&mut self, additional: usize) -> Result<(), TryReserveError> { + self.bytes.try_reserve_exact(additional) + } + + #[inline] + pub fn shrink_to_fit(&mut self) { + self.bytes.shrink_to_fit() + } + + #[inline] + pub fn shrink_to(&mut self, min_capacity: usize) { + self.bytes.shrink_to(min_capacity) + } + + #[inline] + pub fn leak<'a>(self) -> &'a mut Wtf8 { + unsafe { Wtf8::from_mut_bytes_unchecked(self.bytes.leak()) } + } + + /// Returns the number of bytes that this string buffer can hold without reallocating. + #[inline] + pub fn capacity(&self) -> usize { + self.bytes.capacity() + } + + /// Append a UTF-8 slice at the end of the string. + #[inline] + pub fn push_str(&mut self, other: &str) { + self.bytes.extend_from_slice(other.as_bytes()) + } + + /// Append a WTF-8 slice at the end of the string. + /// + /// This replaces newly paired surrogates at the boundary + /// with a supplementary code point, + /// like concatenating ill-formed UTF-16 strings effectively would. + #[inline] + pub fn push_wtf8(&mut self, other: &Wtf8) { + match ((&*self).final_lead_surrogate(), other.initial_trail_surrogate()) { + // Replace newly paired surrogates by a supplementary code point. + (Some(lead), Some(trail)) => { + let len_without_lead_surrogate = self.len() - 3; + self.bytes.truncate(len_without_lead_surrogate); + let other_without_trail_surrogate = &other.bytes[3..]; + // 4 bytes for the supplementary code point + self.bytes.reserve(4 + other_without_trail_surrogate.len()); + self.push_char(decode_surrogate_pair(lead, trail)); + self.bytes.extend_from_slice(other_without_trail_surrogate); + } + _ => { + // If we'll be pushing a string containing a surrogate, we may + // no longer have UTF-8. + if self.is_known_utf8 && other.next_surrogate(0).is_some() { + self.is_known_utf8 = false; + } + + self.bytes.extend_from_slice(&other.bytes); + } + } + } + + /// Append a Unicode scalar value at the end of the string. + #[inline] + pub fn push_char(&mut self, c: char) { + self.push_code_point_unchecked(CodePoint::from_char(c)) + } + + /// Append a code point at the end of the string. + /// + /// This replaces newly paired surrogates at the boundary + /// with a supplementary code point, + /// like concatenating ill-formed UTF-16 strings effectively would. + #[inline] + pub fn push(&mut self, code_point: CodePoint) { + if let Some(trail) = code_point.to_trail_surrogate() { + if let Some(lead) = (&*self).final_lead_surrogate() { + let len_without_lead_surrogate = self.len() - 3; + self.bytes.truncate(len_without_lead_surrogate); + self.push_char(decode_surrogate_pair(lead, trail)); + return; + } + + // We're pushing a trailing surrogate. + self.is_known_utf8 = false; + } else if code_point.to_lead_surrogate().is_some() { + // We're pushing a leading surrogate. + self.is_known_utf8 = false; + } + + // No newly paired surrogates at the boundary. + self.push_code_point_unchecked(code_point) + } + + /// Shortens a string to the specified length. + /// + /// # Panics + /// + /// Panics if `new_len` > current length, + /// or if `new_len` is not a code point boundary. + #[inline] + pub fn truncate(&mut self, new_len: usize) { + assert!(is_code_point_boundary(self, new_len)); + self.bytes.truncate(new_len) + } + + /// Consumes the WTF-8 string and tries to convert it to a vec of bytes. + #[inline] + pub fn into_bytes(self) -> Vec { + self.bytes + } + + /// Consumes the WTF-8 string and tries to convert it to UTF-8. + /// + /// This does not copy the data. + /// + /// If the contents are not well-formed UTF-8 + /// (that is, if the string contains surrogates), + /// the original WTF-8 string is returned instead. + pub fn into_string(self) -> Result { + if self.is_known_utf8 || self.next_surrogate(0).is_none() { + Ok(unsafe { String::from_utf8_unchecked(self.bytes) }) + } else { + Err(self) + } + } + + /// Consumes the WTF-8 string and converts it lossily to UTF-8. + /// + /// This does not copy the data (but may overwrite parts of it in place). + /// + /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”) + pub fn into_string_lossy(mut self) -> String { + if !self.is_known_utf8 { + let mut pos = 0; + while let Some((surrogate_pos, _)) = self.next_surrogate(pos) { + pos = surrogate_pos + 3; + // Surrogates and the replacement character are all 3 bytes, so + // they can substituted in-place. + self.bytes[surrogate_pos..pos] + .copy_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); + } + } + unsafe { String::from_utf8_unchecked(self.bytes) } + } + + /// Converts this `Wtf8Buf` into a boxed `Wtf8`. + #[inline] + pub fn into_box(self) -> Box { + // SAFETY: relies on `Wtf8` being `repr(transparent)`. + unsafe { mem::transmute(self.bytes.into_boxed_slice()) } + } + + /// Converts a `Box` into a `Wtf8Buf`. + pub fn from_box(boxed: Box) -> Wtf8Buf { + let bytes: Box<[u8]> = unsafe { mem::transmute(boxed) }; + Wtf8Buf { bytes: bytes.into_vec(), is_known_utf8: false } + } + + /// Provides plumbing to core `Vec::extend_from_slice`. + /// More well behaving alternative to allowing outer types + /// full mutable access to the core `Vec`. + #[inline] + pub(crate) fn extend_from_slice(&mut self, other: &[u8]) { + self.bytes.extend_from_slice(other); + self.is_known_utf8 = false; + } +} + +/// Creates a new WTF-8 string from an iterator of code points. +/// +/// This replaces surrogate code point pairs with supplementary code points, +/// like concatenating ill-formed UTF-16 strings effectively would. +impl FromIterator for Wtf8Buf { + fn from_iter>(iter: T) -> Wtf8Buf { + let mut string = Wtf8Buf::new(); + string.extend(iter); + string + } +} + +/// Append code points from an iterator to the string. +/// +/// This replaces surrogate code point pairs with supplementary code points, +/// like concatenating ill-formed UTF-16 strings effectively would. +impl Extend for Wtf8Buf { + fn extend>(&mut self, iter: T) { + let iterator = iter.into_iter(); + let (low, _high) = iterator.size_hint(); + // Lower bound of one byte per code point (ASCII only) + self.bytes.reserve(low); + iterator.for_each(move |code_point| self.push(code_point)); + } + + #[inline] + fn extend_one(&mut self, code_point: CodePoint) { + self.push(code_point); + } + + #[inline] + fn extend_reserve(&mut self, additional: usize) { + // Lower bound of one byte per code point (ASCII only) + self.bytes.reserve(additional); + } +} + +/// 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)] +pub struct Wtf8 { + bytes: [u8], +} + +impl AsInner<[u8]> for Wtf8 { + #[inline] + fn as_inner(&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) { + write_str_escaped(formatter, unsafe { + str::from_utf8_unchecked(&self.bytes[pos..surrogate_pos]) + })?; + write!(formatter, "\\u{{{:x}}}", surrogate)?; + pos = surrogate_pos + 3; + } + 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, _)) => { + formatter.write_str(unsafe { + str::from_utf8_unchecked(&wtf8_bytes[pos..surrogate_pos]) + })?; + formatter.write_str(UTF8_REPLACEMENT_CHARACTER)?; + pos = surrogate_pos + 3; + } + None => { + 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. + /// + /// Since WTF-8 is a superset of UTF-8, this always succeeds. + #[inline] + pub fn from_str(value: &str) -> &Wtf8 { + 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] + 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) + } + + /// Creates an owned `Wtf8Buf` from a borrowed `Wtf8`. + pub fn to_owned(&self) -> Wtf8Buf { + Wtf8Buf { bytes: self.bytes.to_vec(), is_known_utf8: false } + } + + /// Lossily converts the string to UTF-8. + /// Returns a UTF-8 `&str` slice if the contents are well-formed in UTF-8. + /// + /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”). + /// + /// This only copies the data if necessary (if it contains any surrogate). + pub fn to_string_lossy(&self) -> Cow<'_, str> { + let Some((surrogate_pos, _)) = self.next_surrogate(0) else { + return Cow::Borrowed(unsafe { str::from_utf8_unchecked(&self.bytes) }); + }; + let wtf8_bytes = &self.bytes; + let mut utf8_bytes = Vec::with_capacity(self.len()); + utf8_bytes.extend_from_slice(&wtf8_bytes[..surrogate_pos]); + utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); + let mut pos = surrogate_pos + 3; + loop { + match self.next_surrogate(pos) { + Some((surrogate_pos, _)) => { + utf8_bytes.extend_from_slice(&wtf8_bytes[pos..surrogate_pos]); + utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); + pos = surrogate_pos + 3; + } + None => { + utf8_bytes.extend_from_slice(&wtf8_bytes[pos..]); + return Cow::Owned(unsafe { String::from_utf8_unchecked(utf8_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] + 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] + fn final_lead_surrogate(&self) -> Option { + match self.bytes { + [.., 0xED, b2 @ 0xA0..=0xAF, b3] => Some(decode_surrogate(b2, b3)), + _ => None, + } + } + + #[inline] + fn initial_trail_surrogate(&self) -> Option { + match self.bytes { + [0xED, b2 @ 0xB0..=0xBF, b3, ..] => Some(decode_surrogate(b2, b3)), + _ => None, + } + } + + pub fn clone_into(&self, buf: &mut Wtf8Buf) { + buf.is_known_utf8 = false; + self.bytes.clone_into(&mut buf.bytes); + } + + /// Boxes this `Wtf8`. + #[inline] + pub fn into_box(&self) -> Box { + let boxed: Box<[u8]> = self.bytes.into(); + unsafe { mem::transmute(boxed) } + } + + /// Creates a boxed, empty `Wtf8`. + pub fn empty_box() -> Box { + let boxed: Box<[u8]> = Default::default(); + unsafe { mem::transmute(boxed) } + } + + #[inline] + pub fn into_arc(&self) -> Arc { + let arc: Arc<[u8]> = Arc::from(&self.bytes); + unsafe { Arc::from_raw(Arc::into_raw(arc) as *const Wtf8) } + } + + #[inline] + pub fn into_rc(&self) -> Rc { + let rc: Rc<[u8]> = Rc::from(&self.bytes); + unsafe { Rc::from_raw(Rc::into_raw(rc) as *const Wtf8) } + } + + #[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 to_ascii_lowercase(&self) -> Wtf8Buf { + Wtf8Buf { bytes: self.bytes.to_ascii_lowercase(), is_known_utf8: false } + } + + #[inline] + pub fn to_ascii_uppercase(&self) -> Wtf8Buf { + Wtf8Buf { bytes: self.bytes.to_ascii_uppercase(), is_known_utf8: false } + } + + #[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> for Wtf8 { + type Output = Wtf8; + + #[inline] + fn index(&self, range: ops::Range) -> &Wtf8 { + // is_code_point_boundary checks that the index is in [0, .len()] + if range.start <= range.end + && is_code_point_boundary(self, range.start) + && is_code_point_boundary(self, range.end) + { + 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> for Wtf8 { + type Output = Wtf8; + + #[inline] + fn index(&self, range: ops::RangeFrom) -> &Wtf8 { + // is_code_point_boundary checks that the index is in [0, .len()] + if is_code_point_boundary(self, range.start) { + 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> for Wtf8 { + type Output = Wtf8; + + #[inline] + fn index(&self, range: ops::RangeTo) -> &Wtf8 { + // is_code_point_boundary checks that the index is in [0, .len()] + if is_code_point_boundary(self, range.end) { + unsafe { slice_unchecked(self, 0, range.end) } + } else { + slice_error_fail(self, 0, range.end) + } + } +} + +impl ops::Index 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 +} + +#[inline] +fn decode_surrogate_pair(lead: u16, trail: u16) -> char { + let code_point = 0x10000 + ((((lead - 0xD800) as u32) << 10) | (trail - 0xDC00) as u32); + unsafe { char::from_u32_unchecked(code_point) } +} + +/// Copied from str::is_char_boundary +#[inline] +pub fn is_code_point_boundary(slice: &Wtf8, index: usize) -> bool { + if index == 0 { + return true; + } + match slice.bytes.get(index) { + None => index == slice.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::slice_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(slice: &Wtf8, index: usize) { + if index == 0 { + return; + } + match slice.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 == slice.len() => return, + None => panic!("byte index {index} is out of bounds"), + } + if slice.bytes[index + 1] >= 0xA0 { + // There's a surrogate after index. Now check before index. + if index >= 3 && slice.bytes[index - 3] == 0xED && slice.bytes[index - 2] >= 0xA0 { + panic!("byte index {index} lies between surrogate codepoints"); + } + } +} + +/// Copied from core::str::raw::slice_unchecked +#[inline] +pub 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)] +pub 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)] +pub struct Wtf8CodePoints<'a> { + bytes: slice::Iter<'a, u8>, +} + +impl Iterator for Wtf8CodePoints<'_> { + type Item = CodePoint; + + #[inline] + fn next(&mut self) -> Option { + // SAFETY: `self.bytes` has been created from a WTF-8 string + unsafe { next_code_point(&mut self.bytes).map(|c| CodePoint { value: c }) } + } + + #[inline] + fn size_hint(&self) -> (usize, Option) { + let len = self.bytes.len(); + (len.saturating_add(3) / 4, Some(len)) + } +} + +/// Generates a wide character sequence for potentially ill-formed UTF-16. +#[stable(feature = "rust1", since = "1.0.0")] +#[derive(Clone)] +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 { + 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.value, &mut buf).len(); + if n == 2 { + self.extra = buf[1]; + } + buf[0] + }) + } + + #[inline] + fn size_hint(&self) -> (usize, Option) { + 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))) + } +} + +#[stable(feature = "encode_wide_fused_iterator", since = "1.62.0")] +impl FusedIterator for EncodeWide<'_> {} + +impl Hash for CodePoint { + #[inline] + fn hash(&self, state: &mut H) { + self.value.hash(state) + } +} + +impl Hash for Wtf8Buf { + #[inline] + fn hash(&self, state: &mut H) { + state.write(&self.bytes); + 0xfeu8.hash(state) + } +} + +impl Hash for Wtf8 { + #[inline] + fn hash(&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) } + } +} diff --git a/library/std/src/sys_common/wtf8.rs b/library/std/src/sys_common/wtf8.rs deleted file mode 100644 index 50bde88b5a4..00000000000 --- a/library/std/src/sys_common/wtf8.rs +++ /dev/null @@ -1,1083 +0,0 @@ -//! 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. - -// this module is imported from @SimonSapin's repo and has tons of dead code on -// unix (it's mostly used on windows), so don't worry about dead code here. -#![allow(dead_code)] - -#[cfg(test)] -mod tests; - -use core::char::{MAX_LEN_UTF8, MAX_LEN_UTF16, encode_utf8_raw, encode_utf16_raw}; -use core::clone::CloneToUninit; -use core::str::next_code_point; - -use crate::borrow::Cow; -use crate::collections::TryReserveError; -use crate::hash::{Hash, Hasher}; -use crate::iter::FusedIterator; -use crate::rc::Rc; -use crate::sync::Arc; -use crate::sys_common::AsInner; -use crate::{fmt, mem, ops, slice, str}; - -const UTF8_REPLACEMENT_CHARACTER: &str = "\u{FFFD}"; - -/// 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)] -pub struct CodePoint { - value: u32, -} - -/// 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.value) - } -} - -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 { - CodePoint { 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 { - match value { - 0..=0x10FFFF => Some(CodePoint { value }), - _ => None, - } - } - - /// 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 { - CodePoint { value: value as u32 } - } - - /// Returns the numeric value of the code point. - #[inline] - pub fn to_u32(&self) -> u32 { - self.value - } - - /// Returns the numeric value of the code point if it is a leading surrogate. - #[inline] - pub fn to_lead_surrogate(&self) -> Option { - match self.value { - 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 { - match self.value { - 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 { - match self.value { - 0xD800..=0xDFFF => None, - _ => Some(unsafe { char::from_u32_unchecked(self.value) }), - } - } - - /// 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('\u{FFFD}') - } -} - -/// An owned, growable string of well-formed WTF-8 data. -/// -/// Similar to `String`, but can additionally contain surrogate code points -/// if they’re not in a surrogate pair. -#[derive(Eq, PartialEq, Ord, PartialOrd, Clone)] -pub struct Wtf8Buf { - bytes: Vec, - - /// Do we know that `bytes` holds a valid UTF-8 encoding? We can easily - /// know this if we're constructed from a `String` or `&str`. - /// - /// It is possible for `bytes` to have valid UTF-8 without this being - /// set, such as when we're concatenating `&Wtf8`'s and surrogates become - /// paired, as we don't bother to rescan the entire string. - is_known_utf8: bool, -} - -impl ops::Deref for Wtf8Buf { - type Target = Wtf8; - - fn deref(&self) -> &Wtf8 { - self.as_slice() - } -} - -impl ops::DerefMut for Wtf8Buf { - fn deref_mut(&mut self) -> &mut Wtf8 { - self.as_mut_slice() - } -} - -/// 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. -/// -/// For example, the code units [U+0061, U+D800, U+000A] are formatted as -/// `"a\u{D800}\n"`. -impl fmt::Debug for Wtf8Buf { - #[inline] - fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { - fmt::Debug::fmt(&**self, formatter) - } -} - -/// Formats the string with unpaired surrogates substituted with the replacement -/// character, U+FFFD. -impl fmt::Display for Wtf8Buf { - fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { - if let Some(s) = self.as_known_utf8() { - fmt::Display::fmt(s, formatter) - } else { - fmt::Display::fmt(&**self, formatter) - } - } -} - -impl Wtf8Buf { - /// Creates a new, empty WTF-8 string. - #[inline] - pub fn new() -> Wtf8Buf { - Wtf8Buf { bytes: Vec::new(), is_known_utf8: true } - } - - /// Creates a new, empty WTF-8 string with pre-allocated capacity for `capacity` bytes. - #[inline] - pub fn with_capacity(capacity: usize) -> Wtf8Buf { - Wtf8Buf { bytes: Vec::with_capacity(capacity), is_known_utf8: true } - } - - /// Creates a WTF-8 string from a WTF-8 byte vec. - /// - /// Since the byte vec is not checked for valid WTF-8, this function is - /// marked unsafe. - #[inline] - pub unsafe fn from_bytes_unchecked(value: Vec) -> Wtf8Buf { - Wtf8Buf { bytes: value, is_known_utf8: false } - } - - /// Creates a WTF-8 string from a UTF-8 `String`. - /// - /// This takes ownership of the `String` and does not copy. - /// - /// Since WTF-8 is a superset of UTF-8, this always succeeds. - #[inline] - pub const fn from_string(string: String) -> Wtf8Buf { - Wtf8Buf { bytes: string.into_bytes(), is_known_utf8: true } - } - - /// Creates a WTF-8 string from a UTF-8 `&str` slice. - /// - /// This copies the content of the slice. - /// - /// Since WTF-8 is a superset of UTF-8, this always succeeds. - #[inline] - pub fn from_str(s: &str) -> Wtf8Buf { - Wtf8Buf { bytes: s.as_bytes().to_vec(), is_known_utf8: true } - } - - pub fn clear(&mut self) { - self.bytes.clear(); - self.is_known_utf8 = true; - } - - /// Creates a WTF-8 string from a potentially ill-formed UTF-16 slice of 16-bit code units. - /// - /// This is lossless: calling `.encode_wide()` on the resulting string - /// will always return the original code units. - pub fn from_wide(v: &[u16]) -> Wtf8Buf { - let mut string = Wtf8Buf::with_capacity(v.len()); - for item in char::decode_utf16(v.iter().cloned()) { - match item { - Ok(ch) => string.push_char(ch), - Err(surrogate) => { - let surrogate = surrogate.unpaired_surrogate(); - // Surrogates are known to be in the code point range. - let code_point = unsafe { CodePoint::from_u32_unchecked(surrogate as u32) }; - // The string will now contain an unpaired surrogate. - string.is_known_utf8 = false; - // Skip the WTF-8 concatenation check, - // surrogate pairs are already decoded by decode_utf16 - string.push_code_point_unchecked(code_point); - } - } - } - string - } - - /// Appends the given `char` to the end of this string. - /// This does **not** include the WTF-8 concatenation check or `is_known_utf8` check. - /// Copied from String::push. - fn push_code_point_unchecked(&mut self, code_point: CodePoint) { - let mut bytes = [0; MAX_LEN_UTF8]; - let bytes = encode_utf8_raw(code_point.value, &mut bytes); - self.bytes.extend_from_slice(bytes) - } - - #[inline] - pub fn as_slice(&self) -> &Wtf8 { - unsafe { Wtf8::from_bytes_unchecked(&self.bytes) } - } - - #[inline] - pub fn as_mut_slice(&mut self) -> &mut Wtf8 { - // Safety: `Wtf8` doesn't expose any way to mutate the bytes that would - // cause them to change from well-formed UTF-8 to ill-formed UTF-8, - // which would break the assumptions of the `is_known_utf8` field. - unsafe { Wtf8::from_mut_bytes_unchecked(&mut self.bytes) } - } - - /// Converts the string to UTF-8 without validation, if it was created from - /// valid UTF-8. - #[inline] - fn as_known_utf8(&self) -> Option<&str> { - if self.is_known_utf8 { - // SAFETY: The buffer is known to be valid UTF-8. - Some(unsafe { str::from_utf8_unchecked(self.as_bytes()) }) - } else { - None - } - } - - /// Reserves capacity for at least `additional` more bytes to be inserted - /// in the given `Wtf8Buf`. - /// The collection may reserve more space to avoid frequent reallocations. - /// - /// # Panics - /// - /// Panics if the new capacity exceeds `isize::MAX` bytes. - #[inline] - pub fn reserve(&mut self, additional: usize) { - self.bytes.reserve(additional) - } - - /// Tries to reserve capacity for at least `additional` more bytes to be - /// inserted in the given `Wtf8Buf`. The `Wtf8Buf` may reserve more space to - /// avoid frequent reallocations. After calling `try_reserve`, capacity will - /// be greater than or equal to `self.len() + additional`. Does nothing if - /// capacity is already sufficient. This method preserves the contents even - /// if an error occurs. - /// - /// # Errors - /// - /// If the capacity overflows, or the allocator reports a failure, then an error - /// is returned. - #[inline] - pub fn try_reserve(&mut self, additional: usize) -> Result<(), TryReserveError> { - self.bytes.try_reserve(additional) - } - - #[inline] - pub fn reserve_exact(&mut self, additional: usize) { - self.bytes.reserve_exact(additional) - } - - /// Tries to reserve the minimum capacity for exactly `additional` more - /// bytes to be inserted in the given `Wtf8Buf`. After calling - /// `try_reserve_exact`, capacity will be greater than or equal to - /// `self.len() + additional` if it returns `Ok(())`. - /// Does nothing if the capacity is already sufficient. - /// - /// Note that the allocator may give the `Wtf8Buf` more space than it - /// requests. Therefore, capacity can not be relied upon to be precisely - /// minimal. Prefer [`try_reserve`] if future insertions are expected. - /// - /// [`try_reserve`]: Wtf8Buf::try_reserve - /// - /// # Errors - /// - /// If the capacity overflows, or the allocator reports a failure, then an error - /// is returned. - #[inline] - pub fn try_reserve_exact(&mut self, additional: usize) -> Result<(), TryReserveError> { - self.bytes.try_reserve_exact(additional) - } - - #[inline] - pub fn shrink_to_fit(&mut self) { - self.bytes.shrink_to_fit() - } - - #[inline] - pub fn shrink_to(&mut self, min_capacity: usize) { - self.bytes.shrink_to(min_capacity) - } - - #[inline] - pub fn leak<'a>(self) -> &'a mut Wtf8 { - unsafe { Wtf8::from_mut_bytes_unchecked(self.bytes.leak()) } - } - - /// Returns the number of bytes that this string buffer can hold without reallocating. - #[inline] - pub fn capacity(&self) -> usize { - self.bytes.capacity() - } - - /// Append a UTF-8 slice at the end of the string. - #[inline] - pub fn push_str(&mut self, other: &str) { - self.bytes.extend_from_slice(other.as_bytes()) - } - - /// Append a WTF-8 slice at the end of the string. - /// - /// This replaces newly paired surrogates at the boundary - /// with a supplementary code point, - /// like concatenating ill-formed UTF-16 strings effectively would. - #[inline] - pub fn push_wtf8(&mut self, other: &Wtf8) { - match ((&*self).final_lead_surrogate(), other.initial_trail_surrogate()) { - // Replace newly paired surrogates by a supplementary code point. - (Some(lead), Some(trail)) => { - let len_without_lead_surrogate = self.len() - 3; - self.bytes.truncate(len_without_lead_surrogate); - let other_without_trail_surrogate = &other.bytes[3..]; - // 4 bytes for the supplementary code point - self.bytes.reserve(4 + other_without_trail_surrogate.len()); - self.push_char(decode_surrogate_pair(lead, trail)); - self.bytes.extend_from_slice(other_without_trail_surrogate); - } - _ => { - // If we'll be pushing a string containing a surrogate, we may - // no longer have UTF-8. - if self.is_known_utf8 && other.next_surrogate(0).is_some() { - self.is_known_utf8 = false; - } - - self.bytes.extend_from_slice(&other.bytes); - } - } - } - - /// Append a Unicode scalar value at the end of the string. - #[inline] - pub fn push_char(&mut self, c: char) { - self.push_code_point_unchecked(CodePoint::from_char(c)) - } - - /// Append a code point at the end of the string. - /// - /// This replaces newly paired surrogates at the boundary - /// with a supplementary code point, - /// like concatenating ill-formed UTF-16 strings effectively would. - #[inline] - pub fn push(&mut self, code_point: CodePoint) { - if let Some(trail) = code_point.to_trail_surrogate() { - if let Some(lead) = (&*self).final_lead_surrogate() { - let len_without_lead_surrogate = self.len() - 3; - self.bytes.truncate(len_without_lead_surrogate); - self.push_char(decode_surrogate_pair(lead, trail)); - return; - } - - // We're pushing a trailing surrogate. - self.is_known_utf8 = false; - } else if code_point.to_lead_surrogate().is_some() { - // We're pushing a leading surrogate. - self.is_known_utf8 = false; - } - - // No newly paired surrogates at the boundary. - self.push_code_point_unchecked(code_point) - } - - /// Shortens a string to the specified length. - /// - /// # Panics - /// - /// Panics if `new_len` > current length, - /// or if `new_len` is not a code point boundary. - #[inline] - pub fn truncate(&mut self, new_len: usize) { - assert!(is_code_point_boundary(self, new_len)); - self.bytes.truncate(new_len) - } - - /// Consumes the WTF-8 string and tries to convert it to a vec of bytes. - #[inline] - pub fn into_bytes(self) -> Vec { - self.bytes - } - - /// Consumes the WTF-8 string and tries to convert it to UTF-8. - /// - /// This does not copy the data. - /// - /// If the contents are not well-formed UTF-8 - /// (that is, if the string contains surrogates), - /// the original WTF-8 string is returned instead. - pub fn into_string(self) -> Result { - if self.is_known_utf8 || self.next_surrogate(0).is_none() { - Ok(unsafe { String::from_utf8_unchecked(self.bytes) }) - } else { - Err(self) - } - } - - /// Consumes the WTF-8 string and converts it lossily to UTF-8. - /// - /// This does not copy the data (but may overwrite parts of it in place). - /// - /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”) - pub fn into_string_lossy(mut self) -> String { - if !self.is_known_utf8 { - let mut pos = 0; - while let Some((surrogate_pos, _)) = self.next_surrogate(pos) { - pos = surrogate_pos + 3; - // Surrogates and the replacement character are all 3 bytes, so - // they can substituted in-place. - self.bytes[surrogate_pos..pos] - .copy_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); - } - } - unsafe { String::from_utf8_unchecked(self.bytes) } - } - - /// Converts this `Wtf8Buf` into a boxed `Wtf8`. - #[inline] - pub fn into_box(self) -> Box { - // SAFETY: relies on `Wtf8` being `repr(transparent)`. - unsafe { mem::transmute(self.bytes.into_boxed_slice()) } - } - - /// Converts a `Box` into a `Wtf8Buf`. - pub fn from_box(boxed: Box) -> Wtf8Buf { - let bytes: Box<[u8]> = unsafe { mem::transmute(boxed) }; - Wtf8Buf { bytes: bytes.into_vec(), is_known_utf8: false } - } - - /// Provides plumbing to core `Vec::extend_from_slice`. - /// More well behaving alternative to allowing outer types - /// full mutable access to the core `Vec`. - #[inline] - pub(crate) fn extend_from_slice(&mut self, other: &[u8]) { - self.bytes.extend_from_slice(other); - self.is_known_utf8 = false; - } -} - -/// Creates a new WTF-8 string from an iterator of code points. -/// -/// This replaces surrogate code point pairs with supplementary code points, -/// like concatenating ill-formed UTF-16 strings effectively would. -impl FromIterator for Wtf8Buf { - fn from_iter>(iter: T) -> Wtf8Buf { - let mut string = Wtf8Buf::new(); - string.extend(iter); - string - } -} - -/// Append code points from an iterator to the string. -/// -/// This replaces surrogate code point pairs with supplementary code points, -/// like concatenating ill-formed UTF-16 strings effectively would. -impl Extend for Wtf8Buf { - fn extend>(&mut self, iter: T) { - let iterator = iter.into_iter(); - let (low, _high) = iterator.size_hint(); - // Lower bound of one byte per code point (ASCII only) - self.bytes.reserve(low); - iterator.for_each(move |code_point| self.push(code_point)); - } - - #[inline] - fn extend_one(&mut self, code_point: CodePoint) { - self.push(code_point); - } - - #[inline] - fn extend_reserve(&mut self, additional: usize) { - // Lower bound of one byte per code point (ASCII only) - self.bytes.reserve(additional); - } -} - -/// 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)] -pub struct Wtf8 { - bytes: [u8], -} - -impl AsInner<[u8]> for Wtf8 { - #[inline] - fn as_inner(&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) { - write_str_escaped(formatter, unsafe { - str::from_utf8_unchecked(&self.bytes[pos..surrogate_pos]) - })?; - write!(formatter, "\\u{{{:x}}}", surrogate)?; - pos = surrogate_pos + 3; - } - 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, _)) => { - formatter.write_str(unsafe { - str::from_utf8_unchecked(&wtf8_bytes[pos..surrogate_pos]) - })?; - formatter.write_str(UTF8_REPLACEMENT_CHARACTER)?; - pos = surrogate_pos + 3; - } - None => { - 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. - /// - /// Since WTF-8 is a superset of UTF-8, this always succeeds. - #[inline] - pub fn from_str(value: &str) -> &Wtf8 { - 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] - 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) - } - - /// Creates an owned `Wtf8Buf` from a borrowed `Wtf8`. - pub fn to_owned(&self) -> Wtf8Buf { - Wtf8Buf { bytes: self.bytes.to_vec(), is_known_utf8: false } - } - - /// Lossily converts the string to UTF-8. - /// Returns a UTF-8 `&str` slice if the contents are well-formed in UTF-8. - /// - /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”). - /// - /// This only copies the data if necessary (if it contains any surrogate). - pub fn to_string_lossy(&self) -> Cow<'_, str> { - let Some((surrogate_pos, _)) = self.next_surrogate(0) else { - return Cow::Borrowed(unsafe { str::from_utf8_unchecked(&self.bytes) }); - }; - let wtf8_bytes = &self.bytes; - let mut utf8_bytes = Vec::with_capacity(self.len()); - utf8_bytes.extend_from_slice(&wtf8_bytes[..surrogate_pos]); - utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); - let mut pos = surrogate_pos + 3; - loop { - match self.next_surrogate(pos) { - Some((surrogate_pos, _)) => { - utf8_bytes.extend_from_slice(&wtf8_bytes[pos..surrogate_pos]); - utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); - pos = surrogate_pos + 3; - } - None => { - utf8_bytes.extend_from_slice(&wtf8_bytes[pos..]); - return Cow::Owned(unsafe { String::from_utf8_unchecked(utf8_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] - 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] - fn final_lead_surrogate(&self) -> Option { - match self.bytes { - [.., 0xED, b2 @ 0xA0..=0xAF, b3] => Some(decode_surrogate(b2, b3)), - _ => None, - } - } - - #[inline] - fn initial_trail_surrogate(&self) -> Option { - match self.bytes { - [0xED, b2 @ 0xB0..=0xBF, b3, ..] => Some(decode_surrogate(b2, b3)), - _ => None, - } - } - - pub fn clone_into(&self, buf: &mut Wtf8Buf) { - buf.is_known_utf8 = false; - self.bytes.clone_into(&mut buf.bytes); - } - - /// Boxes this `Wtf8`. - #[inline] - pub fn into_box(&self) -> Box { - let boxed: Box<[u8]> = self.bytes.into(); - unsafe { mem::transmute(boxed) } - } - - /// Creates a boxed, empty `Wtf8`. - pub fn empty_box() -> Box { - let boxed: Box<[u8]> = Default::default(); - unsafe { mem::transmute(boxed) } - } - - #[inline] - pub fn into_arc(&self) -> Arc { - let arc: Arc<[u8]> = Arc::from(&self.bytes); - unsafe { Arc::from_raw(Arc::into_raw(arc) as *const Wtf8) } - } - - #[inline] - pub fn into_rc(&self) -> Rc { - let rc: Rc<[u8]> = Rc::from(&self.bytes); - unsafe { Rc::from_raw(Rc::into_raw(rc) as *const Wtf8) } - } - - #[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 to_ascii_lowercase(&self) -> Wtf8Buf { - Wtf8Buf { bytes: self.bytes.to_ascii_lowercase(), is_known_utf8: false } - } - - #[inline] - pub fn to_ascii_uppercase(&self) -> Wtf8Buf { - Wtf8Buf { bytes: self.bytes.to_ascii_uppercase(), is_known_utf8: false } - } - - #[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> for Wtf8 { - type Output = Wtf8; - - #[inline] - fn index(&self, range: ops::Range) -> &Wtf8 { - // is_code_point_boundary checks that the index is in [0, .len()] - if range.start <= range.end - && is_code_point_boundary(self, range.start) - && is_code_point_boundary(self, range.end) - { - 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> for Wtf8 { - type Output = Wtf8; - - #[inline] - fn index(&self, range: ops::RangeFrom) -> &Wtf8 { - // is_code_point_boundary checks that the index is in [0, .len()] - if is_code_point_boundary(self, range.start) { - 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> for Wtf8 { - type Output = Wtf8; - - #[inline] - fn index(&self, range: ops::RangeTo) -> &Wtf8 { - // is_code_point_boundary checks that the index is in [0, .len()] - if is_code_point_boundary(self, range.end) { - unsafe { slice_unchecked(self, 0, range.end) } - } else { - slice_error_fail(self, 0, range.end) - } - } -} - -impl ops::Index 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 -} - -#[inline] -fn decode_surrogate_pair(lead: u16, trail: u16) -> char { - let code_point = 0x10000 + ((((lead - 0xD800) as u32) << 10) | (trail - 0xDC00) as u32); - unsafe { char::from_u32_unchecked(code_point) } -} - -/// Copied from str::is_char_boundary -#[inline] -pub fn is_code_point_boundary(slice: &Wtf8, index: usize) -> bool { - if index == 0 { - return true; - } - match slice.bytes.get(index) { - None => index == slice.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::slice_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(slice: &Wtf8, index: usize) { - if index == 0 { - return; - } - match slice.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 == slice.len() => return, - None => panic!("byte index {index} is out of bounds"), - } - if slice.bytes[index + 1] >= 0xA0 { - // There's a surrogate after index. Now check before index. - if index >= 3 && slice.bytes[index - 3] == 0xED && slice.bytes[index - 2] >= 0xA0 { - panic!("byte index {index} lies between surrogate codepoints"); - } - } -} - -/// Copied from core::str::raw::slice_unchecked -#[inline] -pub 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)] -pub 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)] -pub struct Wtf8CodePoints<'a> { - bytes: slice::Iter<'a, u8>, -} - -impl Iterator for Wtf8CodePoints<'_> { - type Item = CodePoint; - - #[inline] - fn next(&mut self) -> Option { - // SAFETY: `self.bytes` has been created from a WTF-8 string - unsafe { next_code_point(&mut self.bytes).map(|c| CodePoint { value: c }) } - } - - #[inline] - fn size_hint(&self) -> (usize, Option) { - let len = self.bytes.len(); - (len.saturating_add(3) / 4, Some(len)) - } -} - -/// Generates a wide character sequence for potentially ill-formed UTF-16. -#[stable(feature = "rust1", since = "1.0.0")] -#[derive(Clone)] -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 { - 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.value, &mut buf).len(); - if n == 2 { - self.extra = buf[1]; - } - buf[0] - }) - } - - #[inline] - fn size_hint(&self) -> (usize, Option) { - 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))) - } -} - -#[stable(feature = "encode_wide_fused_iterator", since = "1.62.0")] -impl FusedIterator for EncodeWide<'_> {} - -impl Hash for CodePoint { - #[inline] - fn hash(&self, state: &mut H) { - self.value.hash(state) - } -} - -impl Hash for Wtf8Buf { - #[inline] - fn hash(&self, state: &mut H) { - state.write(&self.bytes); - 0xfeu8.hash(state) - } -} - -impl Hash for Wtf8 { - #[inline] - fn hash(&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) } - } -} diff --git a/library/std/src/sys_common/wtf8/tests.rs b/library/std/src/sys_common/wtf8/tests.rs deleted file mode 100644 index b57c99a8452..00000000000 --- a/library/std/src/sys_common/wtf8/tests.rs +++ /dev/null @@ -1,751 +0,0 @@ -use super::*; - -#[test] -fn code_point_from_u32() { - assert!(CodePoint::from_u32(0).is_some()); - assert!(CodePoint::from_u32(0xD800).is_some()); - assert!(CodePoint::from_u32(0x10FFFF).is_some()); - assert!(CodePoint::from_u32(0x110000).is_none()); -} - -#[test] -fn code_point_to_u32() { - fn c(value: u32) -> CodePoint { - CodePoint::from_u32(value).unwrap() - } - assert_eq!(c(0).to_u32(), 0); - assert_eq!(c(0xD800).to_u32(), 0xD800); - assert_eq!(c(0x10FFFF).to_u32(), 0x10FFFF); -} - -#[test] -fn code_point_to_lead_surrogate() { - fn c(value: u32) -> CodePoint { - CodePoint::from_u32(value).unwrap() - } - assert_eq!(c(0).to_lead_surrogate(), None); - assert_eq!(c(0xE9).to_lead_surrogate(), None); - assert_eq!(c(0xD800).to_lead_surrogate(), Some(0xD800)); - assert_eq!(c(0xDBFF).to_lead_surrogate(), Some(0xDBFF)); - assert_eq!(c(0xDC00).to_lead_surrogate(), None); - assert_eq!(c(0xDFFF).to_lead_surrogate(), None); - assert_eq!(c(0x1F4A9).to_lead_surrogate(), None); - assert_eq!(c(0x10FFFF).to_lead_surrogate(), None); -} - -#[test] -fn code_point_to_trail_surrogate() { - fn c(value: u32) -> CodePoint { - CodePoint::from_u32(value).unwrap() - } - assert_eq!(c(0).to_trail_surrogate(), None); - assert_eq!(c(0xE9).to_trail_surrogate(), None); - assert_eq!(c(0xD800).to_trail_surrogate(), None); - assert_eq!(c(0xDBFF).to_trail_surrogate(), None); - assert_eq!(c(0xDC00).to_trail_surrogate(), Some(0xDC00)); - assert_eq!(c(0xDFFF).to_trail_surrogate(), Some(0xDFFF)); - assert_eq!(c(0x1F4A9).to_trail_surrogate(), None); - assert_eq!(c(0x10FFFF).to_trail_surrogate(), None); -} - -#[test] -fn code_point_from_char() { - assert_eq!(CodePoint::from_char('a').to_u32(), 0x61); - assert_eq!(CodePoint::from_char('💩').to_u32(), 0x1F4A9); -} - -#[test] -fn code_point_to_string() { - assert_eq!(format!("{:?}", CodePoint::from_char('a')), "U+0061"); - assert_eq!(format!("{:?}", CodePoint::from_char('💩')), "U+1F4A9"); -} - -#[test] -fn code_point_to_char() { - fn c(value: u32) -> CodePoint { - CodePoint::from_u32(value).unwrap() - } - assert_eq!(c(0x61).to_char(), Some('a')); - assert_eq!(c(0x1F4A9).to_char(), Some('💩')); - assert_eq!(c(0xD800).to_char(), None); -} - -#[test] -fn code_point_to_char_lossy() { - fn c(value: u32) -> CodePoint { - CodePoint::from_u32(value).unwrap() - } - assert_eq!(c(0x61).to_char_lossy(), 'a'); - assert_eq!(c(0x1F4A9).to_char_lossy(), '💩'); - assert_eq!(c(0xD800).to_char_lossy(), '\u{FFFD}'); -} - -#[test] -fn wtf8buf_new() { - assert_eq!(Wtf8Buf::new().bytes, b""); -} - -#[test] -fn wtf8buf_from_str() { - assert_eq!(Wtf8Buf::from_str("").bytes, b""); - assert_eq!(Wtf8Buf::from_str("aé 💩").bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); -} - -#[test] -fn wtf8buf_from_string() { - assert_eq!(Wtf8Buf::from_string(String::from("")).bytes, b""); - assert_eq!(Wtf8Buf::from_string(String::from("aé 💩")).bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); -} - -#[test] -fn wtf8buf_from_wide() { - let buf = Wtf8Buf::from_wide(&[]); - assert_eq!(buf.bytes, b""); - assert!(buf.is_known_utf8); - - let buf = Wtf8Buf::from_wide(&[0x61, 0xE9, 0x20, 0xD83D, 0xDCA9]); - assert_eq!(buf.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); - assert!(buf.is_known_utf8); - - let buf = Wtf8Buf::from_wide(&[0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]); - assert_eq!(buf.bytes, b"a\xC3\xA9 \xED\xA0\xBD\xF0\x9F\x92\xA9"); - assert!(!buf.is_known_utf8); - - let buf = Wtf8Buf::from_wide(&[0xD800]); - assert_eq!(buf.bytes, b"\xED\xA0\x80"); - assert!(!buf.is_known_utf8); - - let buf = Wtf8Buf::from_wide(&[0xDBFF]); - assert_eq!(buf.bytes, b"\xED\xAF\xBF"); - assert!(!buf.is_known_utf8); - - let buf = Wtf8Buf::from_wide(&[0xDC00]); - assert_eq!(buf.bytes, b"\xED\xB0\x80"); - assert!(!buf.is_known_utf8); - - let buf = Wtf8Buf::from_wide(&[0xDFFF]); - assert_eq!(buf.bytes, b"\xED\xBF\xBF"); - assert!(!buf.is_known_utf8); -} - -#[test] -fn wtf8buf_push_str() { - let mut string = Wtf8Buf::new(); - assert_eq!(string.bytes, b""); - assert!(string.is_known_utf8); - - string.push_str("aé 💩"); - assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); - assert!(string.is_known_utf8); -} - -#[test] -fn wtf8buf_push_char() { - let mut string = Wtf8Buf::from_str("aé "); - assert_eq!(string.bytes, b"a\xC3\xA9 "); - assert!(string.is_known_utf8); - - string.push_char('💩'); - assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); - assert!(string.is_known_utf8); -} - -#[test] -fn wtf8buf_push() { - let mut string = Wtf8Buf::from_str("aé "); - assert_eq!(string.bytes, b"a\xC3\xA9 "); - assert!(string.is_known_utf8); - - string.push(CodePoint::from_char('💩')); - assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); - assert!(string.is_known_utf8); - - fn c(value: u32) -> CodePoint { - CodePoint::from_u32(value).unwrap() - } - - let mut string = Wtf8Buf::new(); - string.push(c(0xD83D)); // lead - assert!(!string.is_known_utf8); - string.push(c(0xDCA9)); // trail - assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); // Magic! - - let mut string = Wtf8Buf::new(); - string.push(c(0xD83D)); // lead - assert!(!string.is_known_utf8); - string.push(c(0x20)); // not surrogate - string.push(c(0xDCA9)); // trail - assert_eq!(string.bytes, b"\xED\xA0\xBD \xED\xB2\xA9"); - - let mut string = Wtf8Buf::new(); - string.push(c(0xD800)); // lead - assert!(!string.is_known_utf8); - string.push(c(0xDBFF)); // lead - assert_eq!(string.bytes, b"\xED\xA0\x80\xED\xAF\xBF"); - - let mut string = Wtf8Buf::new(); - string.push(c(0xD800)); // lead - assert!(!string.is_known_utf8); - string.push(c(0xE000)); // not surrogate - assert_eq!(string.bytes, b"\xED\xA0\x80\xEE\x80\x80"); - - let mut string = Wtf8Buf::new(); - string.push(c(0xD7FF)); // not surrogate - assert!(string.is_known_utf8); - string.push(c(0xDC00)); // trail - assert!(!string.is_known_utf8); - assert_eq!(string.bytes, b"\xED\x9F\xBF\xED\xB0\x80"); - - let mut string = Wtf8Buf::new(); - string.push(c(0x61)); // not surrogate, < 3 bytes - assert!(string.is_known_utf8); - string.push(c(0xDC00)); // trail - assert!(!string.is_known_utf8); - assert_eq!(string.bytes, b"\x61\xED\xB0\x80"); - - let mut string = Wtf8Buf::new(); - string.push(c(0xDC00)); // trail - assert!(!string.is_known_utf8); - assert_eq!(string.bytes, b"\xED\xB0\x80"); -} - -#[test] -fn wtf8buf_push_wtf8() { - let mut string = Wtf8Buf::from_str("aé"); - assert_eq!(string.bytes, b"a\xC3\xA9"); - string.push_wtf8(Wtf8::from_str(" 💩")); - assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); - assert!(string.is_known_utf8); - - fn w(v: &[u8]) -> &Wtf8 { - unsafe { Wtf8::from_bytes_unchecked(v) } - } - - let mut string = Wtf8Buf::new(); - string.push_wtf8(w(b"\xED\xA0\xBD")); // lead - string.push_wtf8(w(b"\xED\xB2\xA9")); // trail - assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); // Magic! - - let mut string = Wtf8Buf::new(); - string.push_wtf8(w(b"\xED\xA0\xBD")); // lead - string.push_wtf8(w(b" ")); // not surrogate - string.push_wtf8(w(b"\xED\xB2\xA9")); // trail - assert_eq!(string.bytes, b"\xED\xA0\xBD \xED\xB2\xA9"); - assert!(!string.is_known_utf8); - - let mut string = Wtf8Buf::new(); - string.push_wtf8(w(b"\xED\xA0\x80")); // lead - string.push_wtf8(w(b"\xED\xAF\xBF")); // lead - assert_eq!(string.bytes, b"\xED\xA0\x80\xED\xAF\xBF"); - assert!(!string.is_known_utf8); - - let mut string = Wtf8Buf::new(); - string.push_wtf8(w(b"\xED\xA0\x80")); // lead - string.push_wtf8(w(b"\xEE\x80\x80")); // not surrogate - assert_eq!(string.bytes, b"\xED\xA0\x80\xEE\x80\x80"); - assert!(!string.is_known_utf8); - - let mut string = Wtf8Buf::new(); - string.push_wtf8(w(b"\xED\x9F\xBF")); // not surrogate - string.push_wtf8(w(b"\xED\xB0\x80")); // trail - assert_eq!(string.bytes, b"\xED\x9F\xBF\xED\xB0\x80"); - assert!(!string.is_known_utf8); - - let mut string = Wtf8Buf::new(); - string.push_wtf8(w(b"a")); // not surrogate, < 3 bytes - string.push_wtf8(w(b"\xED\xB0\x80")); // trail - assert_eq!(string.bytes, b"\x61\xED\xB0\x80"); - assert!(!string.is_known_utf8); - - let mut string = Wtf8Buf::new(); - string.push_wtf8(w(b"\xED\xB0\x80")); // trail - assert_eq!(string.bytes, b"\xED\xB0\x80"); - assert!(!string.is_known_utf8); -} - -#[test] -fn wtf8buf_truncate() { - let mut string = Wtf8Buf::from_str("aé"); - assert!(string.is_known_utf8); - - string.truncate(3); - assert_eq!(string.bytes, b"a\xC3\xA9"); - assert!(string.is_known_utf8); - - string.truncate(1); - assert_eq!(string.bytes, b"a"); - assert!(string.is_known_utf8); - - string.truncate(0); - assert_eq!(string.bytes, b""); - assert!(string.is_known_utf8); -} - -#[test] -fn wtf8buf_truncate_around_non_bmp() { - let mut string = Wtf8Buf::from_str("💩"); - assert!(string.is_known_utf8); - - string.truncate(4); - assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); - assert!(string.is_known_utf8); - - string.truncate(0); - assert_eq!(string.bytes, b""); - assert!(string.is_known_utf8); -} - -#[test] -#[should_panic] -fn wtf8buf_truncate_fail_code_point_boundary() { - let mut string = Wtf8Buf::from_str("aé"); - string.truncate(2); -} - -#[test] -#[should_panic] -fn wtf8buf_truncate_fail_longer() { - let mut string = Wtf8Buf::from_str("aé"); - string.truncate(4); -} - -#[test] -#[should_panic] -fn wtf8buf_truncate_splitting_non_bmp3() { - let mut string = Wtf8Buf::from_str("💩"); - assert!(string.is_known_utf8); - string.truncate(3); -} - -#[test] -#[should_panic] -fn wtf8buf_truncate_splitting_non_bmp2() { - let mut string = Wtf8Buf::from_str("💩"); - assert!(string.is_known_utf8); - string.truncate(2); -} - -#[test] -#[should_panic] -fn wtf8buf_truncate_splitting_non_bmp1() { - let mut string = Wtf8Buf::from_str("💩"); - assert!(string.is_known_utf8); - string.truncate(1); -} - -#[test] -fn wtf8buf_into_string() { - let mut string = Wtf8Buf::from_str("aé 💩"); - assert!(string.is_known_utf8); - assert_eq!(string.clone().into_string(), Ok(String::from("aé 💩"))); - string.push(CodePoint::from_u32(0xD800).unwrap()); - assert!(!string.is_known_utf8); - assert_eq!(string.clone().into_string(), Err(string)); -} - -#[test] -fn wtf8buf_into_string_lossy() { - let mut string = Wtf8Buf::from_str("aé 💩"); - assert_eq!(string.clone().into_string_lossy(), String::from("aé 💩")); - string.push(CodePoint::from_u32(0xD800).unwrap()); - assert_eq!(string.clone().into_string_lossy(), String::from("aé 💩�")); -} - -#[test] -fn wtf8buf_from_iterator() { - fn f(values: &[u32]) -> Wtf8Buf { - values.iter().map(|&c| CodePoint::from_u32(c).unwrap()).collect::() - } - assert_eq!( - f(&[0x61, 0xE9, 0x20, 0x1F4A9]), - Wtf8Buf { bytes: b"a\xC3\xA9 \xF0\x9F\x92\xA9".to_vec(), is_known_utf8: true } - ); - - assert_eq!(f(&[0xD83D, 0xDCA9]).bytes, b"\xF0\x9F\x92\xA9"); // Magic! - assert_eq!( - f(&[0xD83D, 0x20, 0xDCA9]), - Wtf8Buf { bytes: b"\xED\xA0\xBD \xED\xB2\xA9".to_vec(), is_known_utf8: false } - ); - assert_eq!( - f(&[0xD800, 0xDBFF]), - Wtf8Buf { bytes: b"\xED\xA0\x80\xED\xAF\xBF".to_vec(), is_known_utf8: false } - ); - assert_eq!( - f(&[0xD800, 0xE000]), - Wtf8Buf { bytes: b"\xED\xA0\x80\xEE\x80\x80".to_vec(), is_known_utf8: false } - ); - assert_eq!( - f(&[0xD7FF, 0xDC00]), - Wtf8Buf { bytes: b"\xED\x9F\xBF\xED\xB0\x80".to_vec(), is_known_utf8: false } - ); - assert_eq!( - f(&[0x61, 0xDC00]), - Wtf8Buf { bytes: b"\x61\xED\xB0\x80".to_vec(), is_known_utf8: false } - ); - assert_eq!(f(&[0xDC00]), Wtf8Buf { bytes: b"\xED\xB0\x80".to_vec(), is_known_utf8: false }); -} - -#[test] -fn wtf8buf_extend() { - fn e(initial: &[u32], extended: &[u32]) -> Wtf8Buf { - fn c(value: &u32) -> CodePoint { - CodePoint::from_u32(*value).unwrap() - } - let mut string = initial.iter().map(c).collect::(); - string.extend(extended.iter().map(c)); - string - } - - assert_eq!( - e(&[0x61, 0xE9], &[0x20, 0x1F4A9]), - Wtf8Buf { bytes: b"a\xC3\xA9 \xF0\x9F\x92\xA9".to_vec(), is_known_utf8: true } - ); - - assert_eq!(e(&[0xD83D], &[0xDCA9]).bytes, b"\xF0\x9F\x92\xA9"); // Magic! - assert_eq!( - e(&[0xD83D, 0x20], &[0xDCA9]), - Wtf8Buf { bytes: b"\xED\xA0\xBD \xED\xB2\xA9".to_vec(), is_known_utf8: false } - ); - assert_eq!( - e(&[0xD800], &[0xDBFF]), - Wtf8Buf { bytes: b"\xED\xA0\x80\xED\xAF\xBF".to_vec(), is_known_utf8: false } - ); - assert_eq!( - e(&[0xD800], &[0xE000]), - Wtf8Buf { bytes: b"\xED\xA0\x80\xEE\x80\x80".to_vec(), is_known_utf8: false } - ); - assert_eq!( - e(&[0xD7FF], &[0xDC00]), - Wtf8Buf { bytes: b"\xED\x9F\xBF\xED\xB0\x80".to_vec(), is_known_utf8: false } - ); - assert_eq!( - e(&[0x61], &[0xDC00]), - Wtf8Buf { bytes: b"\x61\xED\xB0\x80".to_vec(), is_known_utf8: false } - ); - assert_eq!( - e(&[], &[0xDC00]), - Wtf8Buf { bytes: b"\xED\xB0\x80".to_vec(), is_known_utf8: false } - ); -} - -#[test] -fn wtf8buf_show() { - let mut string = Wtf8Buf::from_str("a\té \u{7f}💩\r"); - string.push(CodePoint::from_u32(0xD800).unwrap()); - assert_eq!(format!("{string:?}"), "\"a\\té \\u{7f}\u{1f4a9}\\r\\u{d800}\""); -} - -#[test] -fn wtf8buf_as_slice() { - assert_eq!(Wtf8Buf::from_str("aé").as_slice(), Wtf8::from_str("aé")); -} - -#[test] -fn wtf8buf_show_str() { - let text = "a\té 💩\r"; - let string = Wtf8Buf::from_str(text); - assert_eq!(format!("{text:?}"), format!("{string:?}")); -} - -#[test] -fn wtf8_from_str() { - assert_eq!(&Wtf8::from_str("").bytes, b""); - assert_eq!(&Wtf8::from_str("aé 💩").bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); -} - -#[test] -fn wtf8_len() { - assert_eq!(Wtf8::from_str("").len(), 0); - assert_eq!(Wtf8::from_str("aé 💩").len(), 8); -} - -#[test] -fn wtf8_slice() { - assert_eq!(&Wtf8::from_str("aé 💩")[1..4].bytes, b"\xC3\xA9 "); -} - -#[test] -#[should_panic] -fn wtf8_slice_not_code_point_boundary() { - let _ = &Wtf8::from_str("aé 💩")[2..4]; -} - -#[test] -fn wtf8_slice_from() { - assert_eq!(&Wtf8::from_str("aé 💩")[1..].bytes, b"\xC3\xA9 \xF0\x9F\x92\xA9"); -} - -#[test] -#[should_panic] -fn wtf8_slice_from_not_code_point_boundary() { - let _ = &Wtf8::from_str("aé 💩")[2..]; -} - -#[test] -fn wtf8_slice_to() { - assert_eq!(&Wtf8::from_str("aé 💩")[..4].bytes, b"a\xC3\xA9 "); -} - -#[test] -#[should_panic] -fn wtf8_slice_to_not_code_point_boundary() { - let _ = &Wtf8::from_str("aé 💩")[5..]; -} - -#[test] -fn wtf8_ascii_byte_at() { - let slice = Wtf8::from_str("aé 💩"); - assert_eq!(slice.ascii_byte_at(0), b'a'); - assert_eq!(slice.ascii_byte_at(1), b'\xFF'); - assert_eq!(slice.ascii_byte_at(2), b'\xFF'); - assert_eq!(slice.ascii_byte_at(3), b' '); - assert_eq!(slice.ascii_byte_at(4), b'\xFF'); -} - -#[test] -fn wtf8_code_points() { - fn c(value: u32) -> CodePoint { - CodePoint::from_u32(value).unwrap() - } - fn cp(string: &Wtf8Buf) -> Vec> { - string.code_points().map(|c| c.to_char()).collect::>() - } - let mut string = Wtf8Buf::from_str("é "); - assert_eq!(cp(&string), [Some('é'), Some(' ')]); - string.push(c(0xD83D)); - assert_eq!(cp(&string), [Some('é'), Some(' '), None]); - string.push(c(0xDCA9)); - assert_eq!(cp(&string), [Some('é'), Some(' '), Some('💩')]); -} - -#[test] -fn wtf8_as_str() { - assert_eq!(Wtf8::from_str("").as_str(), Ok("")); - assert_eq!(Wtf8::from_str("aé 💩").as_str(), Ok("aé 💩")); - let mut string = Wtf8Buf::new(); - string.push(CodePoint::from_u32(0xD800).unwrap()); - assert!(string.as_str().is_err()); -} - -#[test] -fn wtf8_to_string_lossy() { - assert_eq!(Wtf8::from_str("").to_string_lossy(), Cow::Borrowed("")); - assert_eq!(Wtf8::from_str("aé 💩").to_string_lossy(), Cow::Borrowed("aé 💩")); - let mut string = Wtf8Buf::from_str("aé 💩"); - string.push(CodePoint::from_u32(0xD800).unwrap()); - let expected: Cow<'_, str> = Cow::Owned(String::from("aé 💩�")); - assert_eq!(string.to_string_lossy(), expected); -} - -#[test] -fn wtf8_display() { - fn d(b: &[u8]) -> String { - (&unsafe { Wtf8::from_bytes_unchecked(b) }).to_string() - } - - assert_eq!("", d("".as_bytes())); - assert_eq!("aé 💩", d("aé 💩".as_bytes())); - - let mut string = Wtf8Buf::from_str("aé 💩"); - string.push(CodePoint::from_u32(0xD800).unwrap()); - assert_eq!("aé 💩�", d(string.as_inner())); -} - -#[test] -fn wtf8_encode_wide() { - let mut string = Wtf8Buf::from_str("aé "); - string.push(CodePoint::from_u32(0xD83D).unwrap()); - string.push_char('💩'); - assert_eq!( - string.encode_wide().collect::>(), - vec![0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9] - ); -} - -#[test] -fn wtf8_encode_wide_size_hint() { - let string = Wtf8Buf::from_str("\u{12345}"); - let mut iter = string.encode_wide(); - assert_eq!((1, Some(8)), iter.size_hint()); - iter.next().unwrap(); - assert_eq!((1, Some(1)), iter.size_hint()); - iter.next().unwrap(); - assert_eq!((0, Some(0)), iter.size_hint()); - assert!(iter.next().is_none()); -} - -#[test] -fn wtf8_clone_into() { - let mut string = Wtf8Buf::new(); - Wtf8::from_str("green").clone_into(&mut string); - assert_eq!(string.bytes, b"green"); - - let mut string = Wtf8Buf::from_str("green"); - Wtf8::from_str("").clone_into(&mut string); - assert_eq!(string.bytes, b""); - - let mut string = Wtf8Buf::from_str("red"); - Wtf8::from_str("green").clone_into(&mut string); - assert_eq!(string.bytes, b"green"); - - let mut string = Wtf8Buf::from_str("green"); - Wtf8::from_str("red").clone_into(&mut string); - assert_eq!(string.bytes, b"red"); - - let mut string = Wtf8Buf::from_str("green"); - assert!(string.is_known_utf8); - unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").clone_into(&mut string) }; - assert_eq!(string.bytes, b"\xED\xA0\x80"); - assert!(!string.is_known_utf8); -} - -#[test] -fn wtf8_to_ascii_lowercase() { - let lowercase = Wtf8::from_str("").to_ascii_lowercase(); - assert_eq!(lowercase.bytes, b""); - - let lowercase = Wtf8::from_str("GrEeN gRaPeS! 🍇").to_ascii_lowercase(); - assert_eq!(lowercase.bytes, b"green grapes! \xf0\x9f\x8d\x87"); - - let lowercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_ascii_lowercase() }; - assert_eq!(lowercase.bytes, b"\xED\xA0\x80"); - assert!(!lowercase.is_known_utf8); -} - -#[test] -fn wtf8_to_ascii_uppercase() { - let uppercase = Wtf8::from_str("").to_ascii_uppercase(); - assert_eq!(uppercase.bytes, b""); - - let uppercase = Wtf8::from_str("GrEeN gRaPeS! 🍇").to_ascii_uppercase(); - assert_eq!(uppercase.bytes, b"GREEN GRAPES! \xf0\x9f\x8d\x87"); - - let uppercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_ascii_uppercase() }; - assert_eq!(uppercase.bytes, b"\xED\xA0\x80"); - assert!(!uppercase.is_known_utf8); -} - -#[test] -fn wtf8_make_ascii_lowercase() { - let mut lowercase = Wtf8Buf::from_str(""); - lowercase.make_ascii_lowercase(); - assert_eq!(lowercase.bytes, b""); - - let mut lowercase = Wtf8Buf::from_str("GrEeN gRaPeS! 🍇"); - lowercase.make_ascii_lowercase(); - assert_eq!(lowercase.bytes, b"green grapes! \xf0\x9f\x8d\x87"); - - let mut lowercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; - lowercase.make_ascii_lowercase(); - assert_eq!(lowercase.bytes, b"\xED\xA0\x80"); - assert!(!lowercase.is_known_utf8); -} - -#[test] -fn wtf8_make_ascii_uppercase() { - let mut uppercase = Wtf8Buf::from_str(""); - uppercase.make_ascii_uppercase(); - assert_eq!(uppercase.bytes, b""); - - let mut uppercase = Wtf8Buf::from_str("GrEeN gRaPeS! 🍇"); - uppercase.make_ascii_uppercase(); - assert_eq!(uppercase.bytes, b"GREEN GRAPES! \xf0\x9f\x8d\x87"); - - let mut uppercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; - uppercase.make_ascii_uppercase(); - assert_eq!(uppercase.bytes, b"\xED\xA0\x80"); - assert!(!uppercase.is_known_utf8); -} - -#[test] -fn wtf8_to_owned() { - let string = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; - assert_eq!(string.bytes, b"\xED\xA0\x80"); - assert!(!string.is_known_utf8); -} - -#[test] -fn wtf8_valid_utf8_boundaries() { - let mut string = Wtf8Buf::from_str("aé 💩"); - string.push(CodePoint::from_u32(0xD800).unwrap()); - string.push(CodePoint::from_u32(0xD800).unwrap()); - check_utf8_boundary(&string, 0); - check_utf8_boundary(&string, 1); - check_utf8_boundary(&string, 3); - check_utf8_boundary(&string, 4); - check_utf8_boundary(&string, 8); - check_utf8_boundary(&string, 14); - assert_eq!(string.len(), 14); - - string.push_char('a'); - check_utf8_boundary(&string, 14); - check_utf8_boundary(&string, 15); - - let mut string = Wtf8Buf::from_str("a"); - string.push(CodePoint::from_u32(0xD800).unwrap()); - check_utf8_boundary(&string, 1); - - let mut string = Wtf8Buf::from_str("\u{D7FF}"); - string.push(CodePoint::from_u32(0xD800).unwrap()); - check_utf8_boundary(&string, 3); - - let mut string = Wtf8Buf::new(); - string.push(CodePoint::from_u32(0xD800).unwrap()); - string.push_char('\u{D7FF}'); - check_utf8_boundary(&string, 3); -} - -#[test] -#[should_panic(expected = "byte index 4 is out of bounds")] -fn wtf8_utf8_boundary_out_of_bounds() { - let string = Wtf8::from_str("aé"); - check_utf8_boundary(&string, 4); -} - -#[test] -#[should_panic(expected = "byte index 1 is not a codepoint boundary")] -fn wtf8_utf8_boundary_inside_codepoint() { - let string = Wtf8::from_str("é"); - check_utf8_boundary(&string, 1); -} - -#[test] -#[should_panic(expected = "byte index 1 is not a codepoint boundary")] -fn wtf8_utf8_boundary_inside_surrogate() { - let mut string = Wtf8Buf::new(); - string.push(CodePoint::from_u32(0xD800).unwrap()); - check_utf8_boundary(&string, 1); -} - -#[test] -#[should_panic(expected = "byte index 3 lies between surrogate codepoints")] -fn wtf8_utf8_boundary_between_surrogates() { - let mut string = Wtf8Buf::new(); - string.push(CodePoint::from_u32(0xD800).unwrap()); - string.push(CodePoint::from_u32(0xD800).unwrap()); - check_utf8_boundary(&string, 3); -} - -#[test] -fn wobbled_wtf8_plus_bytes_isnt_utf8() { - let mut string: Wtf8Buf = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; - assert!(!string.is_known_utf8); - string.extend_from_slice(b"some utf-8"); - assert!(!string.is_known_utf8); -} - -#[test] -fn wobbled_wtf8_plus_str_isnt_utf8() { - let mut string: Wtf8Buf = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; - assert!(!string.is_known_utf8); - string.push_str("some utf-8"); - assert!(!string.is_known_utf8); -} - -#[test] -fn unwobbly_wtf8_plus_utf8_is_utf8() { - let mut string: Wtf8Buf = Wtf8Buf::from_str("hello world"); - assert!(string.is_known_utf8); - string.push_str("some utf-8"); - assert!(string.is_known_utf8); -} -- cgit 1.4.1-3-g733a5 From 2914291e09cb13aab64207f9e11f2aaf74de3904 Mon Sep 17 00:00:00 2001 From: ltdk Date: Wed, 13 Aug 2025 01:16:42 -0400 Subject: Move WTF-8 code from std to core/alloc --- library/alloc/src/lib.rs | 3 + library/alloc/src/wtf8/mod.rs | 674 +++++------------------------------- library/alloc/src/wtf8/tests.rs | 209 +++++------ library/alloctests/lib.rs | 12 +- library/alloctests/tests/lib.rs | 1 + library/core/src/lib.rs | 2 + library/core/src/num/niche_types.rs | 15 + library/core/src/wtf8.rs | 642 +++++----------------------------- library/coretests/tests/lib.rs | 1 + library/coretests/tests/wtf8.rs | 1 + library/std/src/lib.rs | 1 + library/std/src/os/windows/ffi.rs | 38 +- library/std/src/sys/os_str/wtf8.rs | 8 +- library/std/src/sys_common/mod.rs | 1 - src/tools/tidy/src/unit_tests.rs | 1 + 15 files changed, 331 insertions(+), 1278 deletions(-) create mode 100644 library/coretests/tests/wtf8.rs (limited to 'library/core/src') diff --git a/library/alloc/src/lib.rs b/library/alloc/src/lib.rs index 639c5d4c930..711092ae8eb 100644 --- a/library/alloc/src/lib.rs +++ b/library/alloc/src/lib.rs @@ -159,6 +159,7 @@ #![feature(unicode_internals)] #![feature(unsize)] #![feature(unwrap_infallible)] +#![feature(wtf8_internals)] // tidy-alphabetical-end // // Language features: @@ -232,6 +233,8 @@ pub mod sync; #[cfg(all(not(no_global_oom_handling), not(no_rc), not(no_sync)))] pub mod task; pub mod vec; +#[cfg(all(not(no_rc), not(no_sync), not(no_global_oom_handling)))] +pub mod wtf8; #[doc(hidden)] #[unstable(feature = "liballoc_internals", issue = "none", reason = "implementation detail")] diff --git a/library/alloc/src/wtf8/mod.rs b/library/alloc/src/wtf8/mod.rs index 50bde88b5a4..95d317a5efb 100644 --- a/library/alloc/src/wtf8/mod.rs +++ b/library/alloc/src/wtf8/mod.rs @@ -1,135 +1,42 @@ -//! 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. - -// this module is imported from @SimonSapin's repo and has tons of dead code on -// unix (it's mostly used on windows), so don't worry about dead code here. -#![allow(dead_code)] +//! Heap-allocated counterpart to core `wtf8` module. +#![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)] + +// Note: This module is also included in the alloctests crate using #[path] to +// run the tests. See the comment there for an explanation why this is the case. #[cfg(test)] mod tests; -use core::char::{MAX_LEN_UTF8, MAX_LEN_UTF16, encode_utf8_raw, encode_utf16_raw}; -use core::clone::CloneToUninit; -use core::str::next_code_point; +use core::char::{MAX_LEN_UTF8, encode_utf8_raw}; +use core::hash::{Hash, Hasher}; +pub use core::wtf8::{CodePoint, Wtf8}; +#[cfg(not(test))] +pub use core::wtf8::{EncodeWide, Wtf8CodePoints}; +use core::{fmt, mem, ops, str}; -use crate::borrow::Cow; +use crate::borrow::{Cow, ToOwned}; +use crate::boxed::Box; use crate::collections::TryReserveError; -use crate::hash::{Hash, Hasher}; -use crate::iter::FusedIterator; +#[cfg(not(test))] use crate::rc::Rc; +use crate::string::String; +#[cfg(all(not(test), target_has_atomic = "ptr"))] use crate::sync::Arc; -use crate::sys_common::AsInner; -use crate::{fmt, mem, ops, slice, str}; - -const UTF8_REPLACEMENT_CHARACTER: &str = "\u{FFFD}"; - -/// 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)] -pub struct CodePoint { - value: u32, -} - -/// 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.value) - } -} - -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 { - CodePoint { 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 { - match value { - 0..=0x10FFFF => Some(CodePoint { value }), - _ => None, - } - } - - /// 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 { - CodePoint { value: value as u32 } - } - - /// Returns the numeric value of the code point. - #[inline] - pub fn to_u32(&self) -> u32 { - self.value - } - - /// Returns the numeric value of the code point if it is a leading surrogate. - #[inline] - pub fn to_lead_surrogate(&self) -> Option { - match self.value { - 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 { - match self.value { - 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 { - match self.value { - 0xD800..=0xDFFF => None, - _ => Some(unsafe { char::from_u32_unchecked(self.value) }), - } - } - - /// 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('\u{FFFD}') - } -} +use crate::vec::Vec; /// An owned, growable string of well-formed WTF-8 data. /// /// Similar to `String`, but can additionally contain surrogate code points /// if they’re not in a surrogate pair. #[derive(Eq, PartialEq, Ord, PartialOrd, Clone)] +#[doc(hidden)] pub struct Wtf8Buf { bytes: Vec, @@ -181,6 +88,7 @@ impl fmt::Display for Wtf8Buf { } } +#[cfg_attr(test, allow(dead_code))] impl Wtf8Buf { /// Creates a new, empty WTF-8 string. #[inline] @@ -245,7 +153,9 @@ impl Wtf8Buf { string.is_known_utf8 = false; // Skip the WTF-8 concatenation check, // surrogate pairs are already decoded by decode_utf16 - string.push_code_point_unchecked(code_point); + unsafe { + string.push_code_point_unchecked(code_point); + } } } } @@ -255,9 +165,9 @@ impl Wtf8Buf { /// Appends the given `char` to the end of this string. /// This does **not** include the WTF-8 concatenation check or `is_known_utf8` check. /// Copied from String::push. - fn push_code_point_unchecked(&mut self, code_point: CodePoint) { + unsafe fn push_code_point_unchecked(&mut self, code_point: CodePoint) { let mut bytes = [0; MAX_LEN_UTF8]; - let bytes = encode_utf8_raw(code_point.value, &mut bytes); + let bytes = encode_utf8_raw(code_point.to_u32(), &mut bytes); self.bytes.extend_from_slice(bytes) } @@ -379,7 +289,7 @@ impl Wtf8Buf { (Some(lead), Some(trail)) => { let len_without_lead_surrogate = self.len() - 3; self.bytes.truncate(len_without_lead_surrogate); - let other_without_trail_surrogate = &other.bytes[3..]; + let other_without_trail_surrogate = &other.as_bytes()[3..]; // 4 bytes for the supplementary code point self.bytes.reserve(4 + other_without_trail_surrogate.len()); self.push_char(decode_surrogate_pair(lead, trail)); @@ -392,7 +302,7 @@ impl Wtf8Buf { self.is_known_utf8 = false; } - self.bytes.extend_from_slice(&other.bytes); + self.bytes.extend_from_slice(other.as_bytes()); } } } @@ -400,7 +310,8 @@ impl Wtf8Buf { /// Append a Unicode scalar value at the end of the string. #[inline] pub fn push_char(&mut self, c: char) { - self.push_code_point_unchecked(CodePoint::from_char(c)) + // SAFETY: It's always safe to push a char. + unsafe { self.push_code_point_unchecked(CodePoint::from_char(c)) } } /// Append a code point at the end of the string. @@ -426,7 +337,7 @@ impl Wtf8Buf { } // No newly paired surrogates at the boundary. - self.push_code_point_unchecked(code_point) + unsafe { self.push_code_point_unchecked(code_point) } } /// Shortens a string to the specified length. @@ -437,7 +348,7 @@ impl Wtf8Buf { /// or if `new_len` is not a code point boundary. #[inline] pub fn truncate(&mut self, new_len: usize) { - assert!(is_code_point_boundary(self, new_len)); + assert!(self.is_code_point_boundary(new_len)); self.bytes.truncate(new_len) } @@ -474,8 +385,7 @@ impl Wtf8Buf { pos = surrogate_pos + 3; // Surrogates and the replacement character are all 3 bytes, so // they can substituted in-place. - self.bytes[surrogate_pos..pos] - .copy_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); + self.bytes[surrogate_pos..pos].copy_from_slice("\u{FFFD}".as_bytes()); } } unsafe { String::from_utf8_unchecked(self.bytes) } @@ -498,7 +408,7 @@ impl Wtf8Buf { /// More well behaving alternative to allowing outer types /// full mutable access to the core `Vec`. #[inline] - pub(crate) fn extend_from_slice(&mut self, other: &[u8]) { + pub unsafe fn extend_from_slice_unchecked(&mut self, other: &[u8]) { self.bytes.extend_from_slice(other); self.is_known_utf8 = false; } @@ -541,153 +451,13 @@ impl Extend for Wtf8Buf { } } -/// 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)] -pub struct Wtf8 { - bytes: [u8], -} - -impl AsInner<[u8]> for Wtf8 { - #[inline] - fn as_inner(&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) { - write_str_escaped(formatter, unsafe { - str::from_utf8_unchecked(&self.bytes[pos..surrogate_pos]) - })?; - write!(formatter, "\\u{{{:x}}}", surrogate)?; - pos = surrogate_pos + 3; - } - 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, _)) => { - formatter.write_str(unsafe { - str::from_utf8_unchecked(&wtf8_bytes[pos..surrogate_pos]) - })?; - formatter.write_str(UTF8_REPLACEMENT_CHARACTER)?; - pos = surrogate_pos + 3; - } - None => { - 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. - /// - /// Since WTF-8 is a superset of UTF-8, this always succeeds. - #[inline] - pub fn from_str(value: &str) -> &Wtf8 { - 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] - 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) - } +// helps diff +mod wtf8 { + use super::*; /// Creates an owned `Wtf8Buf` from a borrowed `Wtf8`. - pub fn to_owned(&self) -> Wtf8Buf { - Wtf8Buf { bytes: self.bytes.to_vec(), is_known_utf8: false } + pub(super) fn to_owned(slice: &Wtf8) -> Wtf8Buf { + Wtf8Buf { bytes: slice.as_bytes().to_vec(), is_known_utf8: false } } /// Lossily converts the string to UTF-8. @@ -696,20 +466,20 @@ impl Wtf8 { /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”). /// /// This only copies the data if necessary (if it contains any surrogate). - pub fn to_string_lossy(&self) -> Cow<'_, str> { - let Some((surrogate_pos, _)) = self.next_surrogate(0) else { - return Cow::Borrowed(unsafe { str::from_utf8_unchecked(&self.bytes) }); + pub(super) fn to_string_lossy(slice: &Wtf8) -> Cow<'_, str> { + let Some((surrogate_pos, _)) = slice.next_surrogate(0) else { + return Cow::Borrowed(unsafe { str::from_utf8_unchecked(slice.as_bytes()) }); }; - let wtf8_bytes = &self.bytes; - let mut utf8_bytes = Vec::with_capacity(self.len()); + let wtf8_bytes = slice.as_bytes(); + let mut utf8_bytes = Vec::with_capacity(slice.len()); utf8_bytes.extend_from_slice(&wtf8_bytes[..surrogate_pos]); - utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); + utf8_bytes.extend_from_slice("\u{FFFD}".as_bytes()); let mut pos = surrogate_pos + 3; loop { - match self.next_surrogate(pos) { + match slice.next_surrogate(pos) { Some((surrogate_pos, _)) => { utf8_bytes.extend_from_slice(&wtf8_bytes[pos..surrogate_pos]); - utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); + utf8_bytes.extend_from_slice("\u{FFFD}".as_bytes()); pos = surrogate_pos + 3; } None => { @@ -720,342 +490,76 @@ impl Wtf8 { } } - /// 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 } + pub(super) fn clone_into(slice: &Wtf8, buf: &mut Wtf8Buf) { + buf.is_known_utf8 = false; + slice.as_bytes().clone_into(&mut buf.bytes); } +} - #[inline] - 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; - } - } - } +use self::wtf8::{to_owned, to_string_lossy, clone_into}; - #[inline] - fn final_lead_surrogate(&self) -> Option { - match self.bytes { - [.., 0xED, b2 @ 0xA0..=0xAF, b3] => Some(decode_surrogate(b2, b3)), - _ => None, - } +#[cfg(not(test))] +impl Wtf8 { + #[rustc_allow_incoherent_impl] + pub fn to_owned(&self) -> Wtf8Buf { + to_owned(self) } - #[inline] - fn initial_trail_surrogate(&self) -> Option { - match self.bytes { - [0xED, b2 @ 0xB0..=0xBF, b3, ..] => Some(decode_surrogate(b2, b3)), - _ => None, - } + #[rustc_allow_incoherent_impl] + pub fn clone_into(&self, buf: &mut Wtf8Buf) { + clone_into(self, buf) } - pub fn clone_into(&self, buf: &mut Wtf8Buf) { - buf.is_known_utf8 = false; - self.bytes.clone_into(&mut buf.bytes); + #[rustc_allow_incoherent_impl] + pub fn to_string_lossy(&self) -> Cow<'_, str> { + to_string_lossy(self) } - /// Boxes this `Wtf8`. - #[inline] + #[rustc_allow_incoherent_impl] pub fn into_box(&self) -> Box { - let boxed: Box<[u8]> = self.bytes.into(); + let boxed: Box<[u8]> = self.as_bytes().into(); unsafe { mem::transmute(boxed) } } - /// Creates a boxed, empty `Wtf8`. + #[rustc_allow_incoherent_impl] pub fn empty_box() -> Box { let boxed: Box<[u8]> = Default::default(); unsafe { mem::transmute(boxed) } } - #[inline] + #[cfg(target_has_atomic = "ptr")] + #[rustc_allow_incoherent_impl] pub fn into_arc(&self) -> Arc { - let arc: Arc<[u8]> = Arc::from(&self.bytes); + let arc: Arc<[u8]> = Arc::from(self.as_bytes()); unsafe { Arc::from_raw(Arc::into_raw(arc) as *const Wtf8) } } - #[inline] + #[rustc_allow_incoherent_impl] pub fn into_rc(&self) -> Rc { - let rc: Rc<[u8]> = Rc::from(&self.bytes); + let rc: Rc<[u8]> = Rc::from(self.as_bytes()); unsafe { Rc::from_raw(Rc::into_raw(rc) as *const Wtf8) } } #[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] + #[rustc_allow_incoherent_impl] pub fn to_ascii_lowercase(&self) -> Wtf8Buf { - Wtf8Buf { bytes: self.bytes.to_ascii_lowercase(), is_known_utf8: false } + Wtf8Buf { bytes: self.as_bytes().to_ascii_lowercase(), is_known_utf8: false } } #[inline] + #[rustc_allow_incoherent_impl] pub fn to_ascii_uppercase(&self) -> Wtf8Buf { - Wtf8Buf { bytes: self.bytes.to_ascii_uppercase(), is_known_utf8: false } - } - - #[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> for Wtf8 { - type Output = Wtf8; - - #[inline] - fn index(&self, range: ops::Range) -> &Wtf8 { - // is_code_point_boundary checks that the index is in [0, .len()] - if range.start <= range.end - && is_code_point_boundary(self, range.start) - && is_code_point_boundary(self, range.end) - { - 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> for Wtf8 { - type Output = Wtf8; - - #[inline] - fn index(&self, range: ops::RangeFrom) -> &Wtf8 { - // is_code_point_boundary checks that the index is in [0, .len()] - if is_code_point_boundary(self, range.start) { - unsafe { slice_unchecked(self, range.start, self.len()) } - } else { - slice_error_fail(self, range.start, self.len()) - } + Wtf8Buf { bytes: self.as_bytes().to_ascii_uppercase(), is_known_utf8: false } } } -/// 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> for Wtf8 { - type Output = Wtf8; - - #[inline] - fn index(&self, range: ops::RangeTo) -> &Wtf8 { - // is_code_point_boundary checks that the index is in [0, .len()] - if is_code_point_boundary(self, range.end) { - unsafe { slice_unchecked(self, 0, range.end) } - } else { - slice_error_fail(self, 0, range.end) - } - } -} - -impl ops::Index 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 -} - #[inline] fn decode_surrogate_pair(lead: u16, trail: u16) -> char { let code_point = 0x10000 + ((((lead - 0xD800) as u32) << 10) | (trail - 0xDC00) as u32); unsafe { char::from_u32_unchecked(code_point) } } -/// Copied from str::is_char_boundary -#[inline] -pub fn is_code_point_boundary(slice: &Wtf8, index: usize) -> bool { - if index == 0 { - return true; - } - match slice.bytes.get(index) { - None => index == slice.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::slice_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(slice: &Wtf8, index: usize) { - if index == 0 { - return; - } - match slice.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 == slice.len() => return, - None => panic!("byte index {index} is out of bounds"), - } - if slice.bytes[index + 1] >= 0xA0 { - // There's a surrogate after index. Now check before index. - if index >= 3 && slice.bytes[index - 3] == 0xED && slice.bytes[index - 2] >= 0xA0 { - panic!("byte index {index} lies between surrogate codepoints"); - } - } -} - -/// Copied from core::str::raw::slice_unchecked -#[inline] -pub 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)] -pub 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)] -pub struct Wtf8CodePoints<'a> { - bytes: slice::Iter<'a, u8>, -} - -impl Iterator for Wtf8CodePoints<'_> { - type Item = CodePoint; - - #[inline] - fn next(&mut self) -> Option { - // SAFETY: `self.bytes` has been created from a WTF-8 string - unsafe { next_code_point(&mut self.bytes).map(|c| CodePoint { value: c }) } - } - - #[inline] - fn size_hint(&self) -> (usize, Option) { - let len = self.bytes.len(); - (len.saturating_add(3) / 4, Some(len)) - } -} - -/// Generates a wide character sequence for potentially ill-formed UTF-16. -#[stable(feature = "rust1", since = "1.0.0")] -#[derive(Clone)] -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 { - 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.value, &mut buf).len(); - if n == 2 { - self.extra = buf[1]; - } - buf[0] - }) - } - - #[inline] - fn size_hint(&self) -> (usize, Option) { - 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))) - } -} - -#[stable(feature = "encode_wide_fused_iterator", since = "1.62.0")] -impl FusedIterator for EncodeWide<'_> {} - -impl Hash for CodePoint { - #[inline] - fn hash(&self, state: &mut H) { - self.value.hash(state) - } -} - impl Hash for Wtf8Buf { #[inline] fn hash(&self, state: &mut H) { @@ -1063,21 +567,3 @@ impl Hash for Wtf8Buf { 0xfeu8.hash(state) } } - -impl Hash for Wtf8 { - #[inline] - fn hash(&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) } - } -} diff --git a/library/alloc/src/wtf8/tests.rs b/library/alloc/src/wtf8/tests.rs index b57c99a8452..291f63f9f9e 100644 --- a/library/alloc/src/wtf8/tests.rs +++ b/library/alloc/src/wtf8/tests.rs @@ -1,3 +1,5 @@ +use realalloc::string::ToString; + use super::*; #[test] @@ -82,82 +84,85 @@ fn code_point_to_char_lossy() { #[test] fn wtf8buf_new() { - assert_eq!(Wtf8Buf::new().bytes, b""); + assert_eq!(Wtf8Buf::new().as_bytes(), b""); } #[test] fn wtf8buf_from_str() { - assert_eq!(Wtf8Buf::from_str("").bytes, b""); - assert_eq!(Wtf8Buf::from_str("aé 💩").bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert_eq!(Wtf8Buf::from_str("").as_bytes(), b""); + assert_eq!(Wtf8Buf::from_str("aé 💩").as_bytes(), b"a\xC3\xA9 \xF0\x9F\x92\xA9"); } #[test] fn wtf8buf_from_string() { - assert_eq!(Wtf8Buf::from_string(String::from("")).bytes, b""); - assert_eq!(Wtf8Buf::from_string(String::from("aé 💩")).bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert_eq!(Wtf8Buf::from_string(String::from("")).as_bytes(), b""); + assert_eq!( + Wtf8Buf::from_string(String::from("aé 💩")).as_bytes(), + b"a\xC3\xA9 \xF0\x9F\x92\xA9" + ); } #[test] fn wtf8buf_from_wide() { let buf = Wtf8Buf::from_wide(&[]); - assert_eq!(buf.bytes, b""); + assert_eq!(buf.as_bytes(), b""); assert!(buf.is_known_utf8); let buf = Wtf8Buf::from_wide(&[0x61, 0xE9, 0x20, 0xD83D, 0xDCA9]); - assert_eq!(buf.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert_eq!(buf.as_bytes(), b"a\xC3\xA9 \xF0\x9F\x92\xA9"); assert!(buf.is_known_utf8); let buf = Wtf8Buf::from_wide(&[0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]); - assert_eq!(buf.bytes, b"a\xC3\xA9 \xED\xA0\xBD\xF0\x9F\x92\xA9"); + assert_eq!(buf.as_bytes(), b"a\xC3\xA9 \xED\xA0\xBD\xF0\x9F\x92\xA9"); assert!(!buf.is_known_utf8); let buf = Wtf8Buf::from_wide(&[0xD800]); - assert_eq!(buf.bytes, b"\xED\xA0\x80"); + assert_eq!(buf.as_bytes(), b"\xED\xA0\x80"); assert!(!buf.is_known_utf8); let buf = Wtf8Buf::from_wide(&[0xDBFF]); - assert_eq!(buf.bytes, b"\xED\xAF\xBF"); + assert_eq!(buf.as_bytes(), b"\xED\xAF\xBF"); assert!(!buf.is_known_utf8); let buf = Wtf8Buf::from_wide(&[0xDC00]); - assert_eq!(buf.bytes, b"\xED\xB0\x80"); + assert_eq!(buf.as_bytes(), b"\xED\xB0\x80"); assert!(!buf.is_known_utf8); let buf = Wtf8Buf::from_wide(&[0xDFFF]); - assert_eq!(buf.bytes, b"\xED\xBF\xBF"); + assert_eq!(buf.as_bytes(), b"\xED\xBF\xBF"); assert!(!buf.is_known_utf8); } #[test] fn wtf8buf_push_str() { let mut string = Wtf8Buf::new(); - assert_eq!(string.bytes, b""); + assert_eq!(string.as_bytes(), b""); assert!(string.is_known_utf8); string.push_str("aé 💩"); - assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert_eq!(string.as_bytes(), b"a\xC3\xA9 \xF0\x9F\x92\xA9"); assert!(string.is_known_utf8); } #[test] fn wtf8buf_push_char() { let mut string = Wtf8Buf::from_str("aé "); - assert_eq!(string.bytes, b"a\xC3\xA9 "); + assert_eq!(string.as_bytes(), b"a\xC3\xA9 "); assert!(string.is_known_utf8); string.push_char('💩'); - assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert_eq!(string.as_bytes(), b"a\xC3\xA9 \xF0\x9F\x92\xA9"); assert!(string.is_known_utf8); } #[test] fn wtf8buf_push() { let mut string = Wtf8Buf::from_str("aé "); - assert_eq!(string.bytes, b"a\xC3\xA9 "); + assert_eq!(string.as_bytes(), b"a\xC3\xA9 "); assert!(string.is_known_utf8); string.push(CodePoint::from_char('💩')); - assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert_eq!(string.as_bytes(), b"a\xC3\xA9 \xF0\x9F\x92\xA9"); assert!(string.is_known_utf8); fn c(value: u32) -> CodePoint { @@ -168,53 +173,53 @@ fn wtf8buf_push() { string.push(c(0xD83D)); // lead assert!(!string.is_known_utf8); string.push(c(0xDCA9)); // trail - assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); // Magic! + assert_eq!(string.as_bytes(), b"\xF0\x9F\x92\xA9"); // Magic! let mut string = Wtf8Buf::new(); string.push(c(0xD83D)); // lead assert!(!string.is_known_utf8); string.push(c(0x20)); // not surrogate string.push(c(0xDCA9)); // trail - assert_eq!(string.bytes, b"\xED\xA0\xBD \xED\xB2\xA9"); + assert_eq!(string.as_bytes(), b"\xED\xA0\xBD \xED\xB2\xA9"); let mut string = Wtf8Buf::new(); string.push(c(0xD800)); // lead assert!(!string.is_known_utf8); string.push(c(0xDBFF)); // lead - assert_eq!(string.bytes, b"\xED\xA0\x80\xED\xAF\xBF"); + assert_eq!(string.as_bytes(), b"\xED\xA0\x80\xED\xAF\xBF"); let mut string = Wtf8Buf::new(); string.push(c(0xD800)); // lead assert!(!string.is_known_utf8); string.push(c(0xE000)); // not surrogate - assert_eq!(string.bytes, b"\xED\xA0\x80\xEE\x80\x80"); + assert_eq!(string.as_bytes(), b"\xED\xA0\x80\xEE\x80\x80"); let mut string = Wtf8Buf::new(); string.push(c(0xD7FF)); // not surrogate assert!(string.is_known_utf8); string.push(c(0xDC00)); // trail assert!(!string.is_known_utf8); - assert_eq!(string.bytes, b"\xED\x9F\xBF\xED\xB0\x80"); + assert_eq!(string.as_bytes(), b"\xED\x9F\xBF\xED\xB0\x80"); let mut string = Wtf8Buf::new(); string.push(c(0x61)); // not surrogate, < 3 bytes assert!(string.is_known_utf8); string.push(c(0xDC00)); // trail assert!(!string.is_known_utf8); - assert_eq!(string.bytes, b"\x61\xED\xB0\x80"); + assert_eq!(string.as_bytes(), b"\x61\xED\xB0\x80"); let mut string = Wtf8Buf::new(); string.push(c(0xDC00)); // trail assert!(!string.is_known_utf8); - assert_eq!(string.bytes, b"\xED\xB0\x80"); + assert_eq!(string.as_bytes(), b"\xED\xB0\x80"); } #[test] fn wtf8buf_push_wtf8() { let mut string = Wtf8Buf::from_str("aé"); - assert_eq!(string.bytes, b"a\xC3\xA9"); + assert_eq!(string.as_bytes(), b"a\xC3\xA9"); string.push_wtf8(Wtf8::from_str(" 💩")); - assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert_eq!(string.as_bytes(), b"a\xC3\xA9 \xF0\x9F\x92\xA9"); assert!(string.is_known_utf8); fn w(v: &[u8]) -> &Wtf8 { @@ -224,42 +229,42 @@ fn wtf8buf_push_wtf8() { let mut string = Wtf8Buf::new(); string.push_wtf8(w(b"\xED\xA0\xBD")); // lead string.push_wtf8(w(b"\xED\xB2\xA9")); // trail - assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); // Magic! + assert_eq!(string.as_bytes(), b"\xF0\x9F\x92\xA9"); // Magic! let mut string = Wtf8Buf::new(); string.push_wtf8(w(b"\xED\xA0\xBD")); // lead string.push_wtf8(w(b" ")); // not surrogate string.push_wtf8(w(b"\xED\xB2\xA9")); // trail - assert_eq!(string.bytes, b"\xED\xA0\xBD \xED\xB2\xA9"); + assert_eq!(string.as_bytes(), b"\xED\xA0\xBD \xED\xB2\xA9"); assert!(!string.is_known_utf8); let mut string = Wtf8Buf::new(); string.push_wtf8(w(b"\xED\xA0\x80")); // lead string.push_wtf8(w(b"\xED\xAF\xBF")); // lead - assert_eq!(string.bytes, b"\xED\xA0\x80\xED\xAF\xBF"); + assert_eq!(string.as_bytes(), b"\xED\xA0\x80\xED\xAF\xBF"); assert!(!string.is_known_utf8); let mut string = Wtf8Buf::new(); string.push_wtf8(w(b"\xED\xA0\x80")); // lead string.push_wtf8(w(b"\xEE\x80\x80")); // not surrogate - assert_eq!(string.bytes, b"\xED\xA0\x80\xEE\x80\x80"); + assert_eq!(string.as_bytes(), b"\xED\xA0\x80\xEE\x80\x80"); assert!(!string.is_known_utf8); let mut string = Wtf8Buf::new(); string.push_wtf8(w(b"\xED\x9F\xBF")); // not surrogate string.push_wtf8(w(b"\xED\xB0\x80")); // trail - assert_eq!(string.bytes, b"\xED\x9F\xBF\xED\xB0\x80"); + assert_eq!(string.as_bytes(), b"\xED\x9F\xBF\xED\xB0\x80"); assert!(!string.is_known_utf8); let mut string = Wtf8Buf::new(); string.push_wtf8(w(b"a")); // not surrogate, < 3 bytes string.push_wtf8(w(b"\xED\xB0\x80")); // trail - assert_eq!(string.bytes, b"\x61\xED\xB0\x80"); + assert_eq!(string.as_bytes(), b"\x61\xED\xB0\x80"); assert!(!string.is_known_utf8); let mut string = Wtf8Buf::new(); string.push_wtf8(w(b"\xED\xB0\x80")); // trail - assert_eq!(string.bytes, b"\xED\xB0\x80"); + assert_eq!(string.as_bytes(), b"\xED\xB0\x80"); assert!(!string.is_known_utf8); } @@ -269,15 +274,15 @@ fn wtf8buf_truncate() { assert!(string.is_known_utf8); string.truncate(3); - assert_eq!(string.bytes, b"a\xC3\xA9"); + assert_eq!(string.as_bytes(), b"a\xC3\xA9"); assert!(string.is_known_utf8); string.truncate(1); - assert_eq!(string.bytes, b"a"); + assert_eq!(string.as_bytes(), b"a"); assert!(string.is_known_utf8); string.truncate(0); - assert_eq!(string.bytes, b""); + assert_eq!(string.as_bytes(), b""); assert!(string.is_known_utf8); } @@ -287,11 +292,11 @@ fn wtf8buf_truncate_around_non_bmp() { assert!(string.is_known_utf8); string.truncate(4); - assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); + assert_eq!(string.as_bytes(), b"\xF0\x9F\x92\xA9"); assert!(string.is_known_utf8); string.truncate(0); - assert_eq!(string.bytes, b""); + assert_eq!(string.as_bytes(), b""); assert!(string.is_known_utf8); } @@ -361,7 +366,7 @@ fn wtf8buf_from_iterator() { Wtf8Buf { bytes: b"a\xC3\xA9 \xF0\x9F\x92\xA9".to_vec(), is_known_utf8: true } ); - assert_eq!(f(&[0xD83D, 0xDCA9]).bytes, b"\xF0\x9F\x92\xA9"); // Magic! + assert_eq!(f(&[0xD83D, 0xDCA9]).as_bytes(), b"\xF0\x9F\x92\xA9"); // Magic! assert_eq!( f(&[0xD83D, 0x20, 0xDCA9]), Wtf8Buf { bytes: b"\xED\xA0\xBD \xED\xB2\xA9".to_vec(), is_known_utf8: false } @@ -401,7 +406,7 @@ fn wtf8buf_extend() { Wtf8Buf { bytes: b"a\xC3\xA9 \xF0\x9F\x92\xA9".to_vec(), is_known_utf8: true } ); - assert_eq!(e(&[0xD83D], &[0xDCA9]).bytes, b"\xF0\x9F\x92\xA9"); // Magic! + assert_eq!(e(&[0xD83D], &[0xDCA9]).as_bytes(), b"\xF0\x9F\x92\xA9"); // Magic! assert_eq!( e(&[0xD83D, 0x20], &[0xDCA9]), Wtf8Buf { bytes: b"\xED\xA0\xBD \xED\xB2\xA9".to_vec(), is_known_utf8: false } @@ -449,8 +454,8 @@ fn wtf8buf_show_str() { #[test] fn wtf8_from_str() { - assert_eq!(&Wtf8::from_str("").bytes, b""); - assert_eq!(&Wtf8::from_str("aé 💩").bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert_eq!(&Wtf8::from_str("").as_bytes(), b""); + assert_eq!(&Wtf8::from_str("aé 💩").as_bytes(), b"a\xC3\xA9 \xF0\x9F\x92\xA9"); } #[test] @@ -461,7 +466,7 @@ fn wtf8_len() { #[test] fn wtf8_slice() { - assert_eq!(&Wtf8::from_str("aé 💩")[1..4].bytes, b"\xC3\xA9 "); + assert_eq!(&Wtf8::from_str("aé 💩")[1..4].as_bytes(), b"\xC3\xA9 "); } #[test] @@ -472,7 +477,7 @@ fn wtf8_slice_not_code_point_boundary() { #[test] fn wtf8_slice_from() { - assert_eq!(&Wtf8::from_str("aé 💩")[1..].bytes, b"\xC3\xA9 \xF0\x9F\x92\xA9"); + assert_eq!(&Wtf8::from_str("aé 💩")[1..].as_bytes(), b"\xC3\xA9 \xF0\x9F\x92\xA9"); } #[test] @@ -483,7 +488,7 @@ fn wtf8_slice_from_not_code_point_boundary() { #[test] fn wtf8_slice_to() { - assert_eq!(&Wtf8::from_str("aé 💩")[..4].bytes, b"a\xC3\xA9 "); + assert_eq!(&Wtf8::from_str("aé 💩")[..4].as_bytes(), b"a\xC3\xA9 "); } #[test] @@ -529,12 +534,12 @@ fn wtf8_as_str() { #[test] fn wtf8_to_string_lossy() { - assert_eq!(Wtf8::from_str("").to_string_lossy(), Cow::Borrowed("")); - assert_eq!(Wtf8::from_str("aé 💩").to_string_lossy(), Cow::Borrowed("aé 💩")); + assert_eq!(to_string_lossy(Wtf8::from_str("")), Cow::Borrowed("")); + assert_eq!(to_string_lossy(Wtf8::from_str("aé 💩")), Cow::Borrowed("aé 💩")); let mut string = Wtf8Buf::from_str("aé 💩"); string.push(CodePoint::from_u32(0xD800).unwrap()); let expected: Cow<'_, str> = Cow::Owned(String::from("aé 💩�")); - assert_eq!(string.to_string_lossy(), expected); + assert_eq!(to_string_lossy(&string), expected); } #[test] @@ -548,7 +553,7 @@ fn wtf8_display() { let mut string = Wtf8Buf::from_str("aé 💩"); string.push(CodePoint::from_u32(0xD800).unwrap()); - assert_eq!("aé 💩�", d(string.as_inner())); + assert_eq!("aé 💩�", d(string.as_ref())); } #[test] @@ -577,67 +582,41 @@ fn wtf8_encode_wide_size_hint() { #[test] fn wtf8_clone_into() { let mut string = Wtf8Buf::new(); - Wtf8::from_str("green").clone_into(&mut string); - assert_eq!(string.bytes, b"green"); + clone_into(Wtf8::from_str("green"), &mut string); + assert_eq!(string.as_bytes(), b"green"); let mut string = Wtf8Buf::from_str("green"); - Wtf8::from_str("").clone_into(&mut string); - assert_eq!(string.bytes, b""); + clone_into(Wtf8::from_str(""), &mut string); + assert_eq!(string.as_bytes(), b""); let mut string = Wtf8Buf::from_str("red"); - Wtf8::from_str("green").clone_into(&mut string); - assert_eq!(string.bytes, b"green"); + clone_into(Wtf8::from_str("green"), &mut string); + assert_eq!(string.as_bytes(), b"green"); let mut string = Wtf8Buf::from_str("green"); - Wtf8::from_str("red").clone_into(&mut string); - assert_eq!(string.bytes, b"red"); + clone_into(Wtf8::from_str("red"), &mut string); + assert_eq!(string.as_bytes(), b"red"); let mut string = Wtf8Buf::from_str("green"); assert!(string.is_known_utf8); - unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").clone_into(&mut string) }; - assert_eq!(string.bytes, b"\xED\xA0\x80"); + clone_into(unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80") }, &mut string); + assert_eq!(string.as_bytes(), b"\xED\xA0\x80"); assert!(!string.is_known_utf8); } -#[test] -fn wtf8_to_ascii_lowercase() { - let lowercase = Wtf8::from_str("").to_ascii_lowercase(); - assert_eq!(lowercase.bytes, b""); - - let lowercase = Wtf8::from_str("GrEeN gRaPeS! 🍇").to_ascii_lowercase(); - assert_eq!(lowercase.bytes, b"green grapes! \xf0\x9f\x8d\x87"); - - let lowercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_ascii_lowercase() }; - assert_eq!(lowercase.bytes, b"\xED\xA0\x80"); - assert!(!lowercase.is_known_utf8); -} - -#[test] -fn wtf8_to_ascii_uppercase() { - let uppercase = Wtf8::from_str("").to_ascii_uppercase(); - assert_eq!(uppercase.bytes, b""); - - let uppercase = Wtf8::from_str("GrEeN gRaPeS! 🍇").to_ascii_uppercase(); - assert_eq!(uppercase.bytes, b"GREEN GRAPES! \xf0\x9f\x8d\x87"); - - let uppercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_ascii_uppercase() }; - assert_eq!(uppercase.bytes, b"\xED\xA0\x80"); - assert!(!uppercase.is_known_utf8); -} - #[test] fn wtf8_make_ascii_lowercase() { let mut lowercase = Wtf8Buf::from_str(""); lowercase.make_ascii_lowercase(); - assert_eq!(lowercase.bytes, b""); + assert_eq!(lowercase.as_bytes(), b""); let mut lowercase = Wtf8Buf::from_str("GrEeN gRaPeS! 🍇"); lowercase.make_ascii_lowercase(); - assert_eq!(lowercase.bytes, b"green grapes! \xf0\x9f\x8d\x87"); + assert_eq!(lowercase.as_bytes(), b"green grapes! \xf0\x9f\x8d\x87"); - let mut lowercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; + let mut lowercase = to_owned(unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80") }); lowercase.make_ascii_lowercase(); - assert_eq!(lowercase.bytes, b"\xED\xA0\x80"); + assert_eq!(lowercase.as_bytes(), b"\xED\xA0\x80"); assert!(!lowercase.is_known_utf8); } @@ -645,22 +624,22 @@ fn wtf8_make_ascii_lowercase() { fn wtf8_make_ascii_uppercase() { let mut uppercase = Wtf8Buf::from_str(""); uppercase.make_ascii_uppercase(); - assert_eq!(uppercase.bytes, b""); + assert_eq!(uppercase.as_bytes(), b""); let mut uppercase = Wtf8Buf::from_str("GrEeN gRaPeS! 🍇"); uppercase.make_ascii_uppercase(); - assert_eq!(uppercase.bytes, b"GREEN GRAPES! \xf0\x9f\x8d\x87"); + assert_eq!(uppercase.as_bytes(), b"GREEN GRAPES! \xf0\x9f\x8d\x87"); - let mut uppercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; + let mut uppercase = to_owned(unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80") }); uppercase.make_ascii_uppercase(); - assert_eq!(uppercase.bytes, b"\xED\xA0\x80"); + assert_eq!(uppercase.as_bytes(), b"\xED\xA0\x80"); assert!(!uppercase.is_known_utf8); } #[test] fn wtf8_to_owned() { - let string = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; - assert_eq!(string.bytes, b"\xED\xA0\x80"); + let string = to_owned(unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80") }); + assert_eq!(string.as_bytes(), b"\xED\xA0\x80"); assert!(!string.is_known_utf8); } @@ -669,44 +648,44 @@ fn wtf8_valid_utf8_boundaries() { let mut string = Wtf8Buf::from_str("aé 💩"); string.push(CodePoint::from_u32(0xD800).unwrap()); string.push(CodePoint::from_u32(0xD800).unwrap()); - check_utf8_boundary(&string, 0); - check_utf8_boundary(&string, 1); - check_utf8_boundary(&string, 3); - check_utf8_boundary(&string, 4); - check_utf8_boundary(&string, 8); - check_utf8_boundary(&string, 14); + string.check_utf8_boundary(0); + string.check_utf8_boundary(1); + string.check_utf8_boundary(3); + string.check_utf8_boundary(4); + string.check_utf8_boundary(8); + string.check_utf8_boundary(14); assert_eq!(string.len(), 14); string.push_char('a'); - check_utf8_boundary(&string, 14); - check_utf8_boundary(&string, 15); + string.check_utf8_boundary(14); + string.check_utf8_boundary(15); let mut string = Wtf8Buf::from_str("a"); string.push(CodePoint::from_u32(0xD800).unwrap()); - check_utf8_boundary(&string, 1); + string.check_utf8_boundary(1); let mut string = Wtf8Buf::from_str("\u{D7FF}"); string.push(CodePoint::from_u32(0xD800).unwrap()); - check_utf8_boundary(&string, 3); + string.check_utf8_boundary(3); let mut string = Wtf8Buf::new(); string.push(CodePoint::from_u32(0xD800).unwrap()); string.push_char('\u{D7FF}'); - check_utf8_boundary(&string, 3); + string.check_utf8_boundary(3); } #[test] #[should_panic(expected = "byte index 4 is out of bounds")] fn wtf8_utf8_boundary_out_of_bounds() { let string = Wtf8::from_str("aé"); - check_utf8_boundary(&string, 4); + string.check_utf8_boundary(4); } #[test] #[should_panic(expected = "byte index 1 is not a codepoint boundary")] fn wtf8_utf8_boundary_inside_codepoint() { let string = Wtf8::from_str("é"); - check_utf8_boundary(&string, 1); + string.check_utf8_boundary(1); } #[test] @@ -714,7 +693,7 @@ fn wtf8_utf8_boundary_inside_codepoint() { fn wtf8_utf8_boundary_inside_surrogate() { let mut string = Wtf8Buf::new(); string.push(CodePoint::from_u32(0xD800).unwrap()); - check_utf8_boundary(&string, 1); + string.check_utf8_boundary(1); } #[test] @@ -723,20 +702,22 @@ fn wtf8_utf8_boundary_between_surrogates() { let mut string = Wtf8Buf::new(); string.push(CodePoint::from_u32(0xD800).unwrap()); string.push(CodePoint::from_u32(0xD800).unwrap()); - check_utf8_boundary(&string, 3); + string.check_utf8_boundary(3); } #[test] fn wobbled_wtf8_plus_bytes_isnt_utf8() { - let mut string: Wtf8Buf = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; + let mut string: Wtf8Buf = to_owned(unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80") }); assert!(!string.is_known_utf8); - string.extend_from_slice(b"some utf-8"); + unsafe { + string.extend_from_slice_unchecked(b"some utf-8"); + } assert!(!string.is_known_utf8); } #[test] fn wobbled_wtf8_plus_str_isnt_utf8() { - let mut string: Wtf8Buf = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; + let mut string: Wtf8Buf = to_owned(unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80") }); assert!(!string.is_known_utf8); string.push_str("some utf-8"); assert!(!string.is_known_utf8); diff --git a/library/alloctests/lib.rs b/library/alloctests/lib.rs index 3241b4b0045..97de7d6a701 100644 --- a/library/alloctests/lib.rs +++ b/library/alloctests/lib.rs @@ -18,6 +18,8 @@ #![feature(allocator_api)] #![feature(array_into_iter_constructors)] #![feature(assert_matches)] +#![feature(char_internals)] +#![feature(char_max_len)] #![feature(core_intrinsics)] #![feature(exact_size_is_empty)] #![feature(extend_one)] @@ -41,6 +43,7 @@ #![feature(trusted_random_access)] #![feature(try_reserve_kind)] #![feature(try_trait_v2)] +#![feature(wtf8_internals)] // tidy-alphabetical-end // // Language features: @@ -68,15 +71,18 @@ extern crate test; mod testing; use realalloc::*; -// We are directly including collections and raw_vec here as both use non-public -// methods and fields in tests and as such need to have the types to test in the -// same crate as the tests themself. +// We are directly including collections, raw_vec, and wtf8 here as they use non-public +// methods and fields in tests and as such need to have the types to test in the same +// crate as the tests themself. #[path = "../alloc/src/collections/mod.rs"] mod collections; #[path = "../alloc/src/raw_vec/mod.rs"] mod raw_vec; +#[path = "../alloc/src/wtf8/mod.rs"] +mod wtf8; + #[allow(dead_code)] // Not used in all configurations pub(crate) mod test_helpers { /// Copied from `std::test_helpers::test_rng`, since these tests rely on the diff --git a/library/alloctests/tests/lib.rs b/library/alloctests/tests/lib.rs index fcfc7f8dd29..447af240a4b 100644 --- a/library/alloctests/tests/lib.rs +++ b/library/alloctests/tests/lib.rs @@ -3,6 +3,7 @@ #![feature(iter_array_chunks)] #![feature(assert_matches)] #![feature(btree_extract_if)] +#![feature(wtf8_internals)] #![feature(char_max_len)] #![feature(cow_is_borrowed)] #![feature(core_intrinsics)] diff --git a/library/core/src/lib.rs b/library/core/src/lib.rs index 71abd707374..2b6ea7d9afc 100644 --- a/library/core/src/lib.rs +++ b/library/core/src/lib.rs @@ -362,6 +362,8 @@ pub mod slice; pub mod str; pub mod time; +pub mod wtf8; + pub mod unicode; /* Async */ 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/wtf8.rs b/library/core/src/wtf8.rs index 50bde88b5a4..5631993dea2 100644 --- a/library/core/src/wtf8.rs +++ b/library/core/src/wtf8.rs @@ -10,28 +10,23 @@ //! 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. - -// this module is imported from @SimonSapin's repo and has tons of dead code on -// unix (it's mostly used on windows), so don't worry about dead code here. -#![allow(dead_code)] - -#[cfg(test)] -mod tests; - -use core::char::{MAX_LEN_UTF8, MAX_LEN_UTF16, encode_utf8_raw, encode_utf16_raw}; -use core::clone::CloneToUninit; -use core::str::next_code_point; - -use crate::borrow::Cow; -use crate::collections::TryReserveError; +#![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::rc::Rc; -use crate::sync::Arc; -use crate::sys_common::AsInner; -use crate::{fmt, mem, ops, slice, str}; - -const UTF8_REPLACEMENT_CHARACTER: &str = "\u{FFFD}"; +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. /// @@ -39,16 +34,15 @@ const UTF8_REPLACEMENT_CHARACTER: &str = "\u{FFFD}"; /// 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)] -pub struct CodePoint { - value: u32, -} +#[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.value) + write!(formatter, "U+{:04X}", self.0.as_inner()) } } @@ -58,7 +52,8 @@ impl CodePoint { /// Only use when `value` is known to be less than or equal to 0x10FFFF. #[inline] pub unsafe fn from_u32_unchecked(value: u32) -> CodePoint { - CodePoint { value } + // SAFETY: Guaranteed by caller. + CodePoint(unsafe { CodePointInner::new_unchecked(value) }) } /// Creates a new `CodePoint` if the value is a valid code point. @@ -66,10 +61,7 @@ impl CodePoint { /// Returns `None` if `value` is above 0x10FFFF. #[inline] pub fn from_u32(value: u32) -> Option { - match value { - 0..=0x10FFFF => Some(CodePoint { value }), - _ => None, - } + Some(CodePoint(CodePointInner::new(value)?)) } /// Creates a new `CodePoint` from a `char`. @@ -77,19 +69,20 @@ impl CodePoint { /// Since all Unicode scalar values are code points, this always succeeds. #[inline] pub fn from_char(value: char) -> CodePoint { - CodePoint { value: value as u32 } + // 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.value + 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 { - match self.value { + match self.to_u32() { lead @ 0xD800..=0xDBFF => Some(lead as u16), _ => None, } @@ -98,7 +91,7 @@ impl CodePoint { /// Returns the numeric value of the code point if it is a trailing surrogate. #[inline] pub fn to_trail_surrogate(&self) -> Option { - match self.value { + match self.to_u32() { trail @ 0xDC00..=0xDFFF => Some(trail as u16), _ => None, } @@ -109,9 +102,10 @@ impl CodePoint { /// Returns `None` if the code point is a surrogate (from U+D800 to U+DFFF). #[inline] pub fn to_char(&self) -> Option { - match self.value { + match self.to_u32() { 0xD800..=0xDFFF => None, - _ => Some(unsafe { char::from_u32_unchecked(self.value) }), + // SAFETY: We explicitly check that the char is valid. + valid => Some(unsafe { char::from_u32_unchecked(valid) }), } } @@ -121,423 +115,7 @@ impl CodePoint { /// 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('\u{FFFD}') - } -} - -/// An owned, growable string of well-formed WTF-8 data. -/// -/// Similar to `String`, but can additionally contain surrogate code points -/// if they’re not in a surrogate pair. -#[derive(Eq, PartialEq, Ord, PartialOrd, Clone)] -pub struct Wtf8Buf { - bytes: Vec, - - /// Do we know that `bytes` holds a valid UTF-8 encoding? We can easily - /// know this if we're constructed from a `String` or `&str`. - /// - /// It is possible for `bytes` to have valid UTF-8 without this being - /// set, such as when we're concatenating `&Wtf8`'s and surrogates become - /// paired, as we don't bother to rescan the entire string. - is_known_utf8: bool, -} - -impl ops::Deref for Wtf8Buf { - type Target = Wtf8; - - fn deref(&self) -> &Wtf8 { - self.as_slice() - } -} - -impl ops::DerefMut for Wtf8Buf { - fn deref_mut(&mut self) -> &mut Wtf8 { - self.as_mut_slice() - } -} - -/// 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. -/// -/// For example, the code units [U+0061, U+D800, U+000A] are formatted as -/// `"a\u{D800}\n"`. -impl fmt::Debug for Wtf8Buf { - #[inline] - fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { - fmt::Debug::fmt(&**self, formatter) - } -} - -/// Formats the string with unpaired surrogates substituted with the replacement -/// character, U+FFFD. -impl fmt::Display for Wtf8Buf { - fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { - if let Some(s) = self.as_known_utf8() { - fmt::Display::fmt(s, formatter) - } else { - fmt::Display::fmt(&**self, formatter) - } - } -} - -impl Wtf8Buf { - /// Creates a new, empty WTF-8 string. - #[inline] - pub fn new() -> Wtf8Buf { - Wtf8Buf { bytes: Vec::new(), is_known_utf8: true } - } - - /// Creates a new, empty WTF-8 string with pre-allocated capacity for `capacity` bytes. - #[inline] - pub fn with_capacity(capacity: usize) -> Wtf8Buf { - Wtf8Buf { bytes: Vec::with_capacity(capacity), is_known_utf8: true } - } - - /// Creates a WTF-8 string from a WTF-8 byte vec. - /// - /// Since the byte vec is not checked for valid WTF-8, this function is - /// marked unsafe. - #[inline] - pub unsafe fn from_bytes_unchecked(value: Vec) -> Wtf8Buf { - Wtf8Buf { bytes: value, is_known_utf8: false } - } - - /// Creates a WTF-8 string from a UTF-8 `String`. - /// - /// This takes ownership of the `String` and does not copy. - /// - /// Since WTF-8 is a superset of UTF-8, this always succeeds. - #[inline] - pub const fn from_string(string: String) -> Wtf8Buf { - Wtf8Buf { bytes: string.into_bytes(), is_known_utf8: true } - } - - /// Creates a WTF-8 string from a UTF-8 `&str` slice. - /// - /// This copies the content of the slice. - /// - /// Since WTF-8 is a superset of UTF-8, this always succeeds. - #[inline] - pub fn from_str(s: &str) -> Wtf8Buf { - Wtf8Buf { bytes: s.as_bytes().to_vec(), is_known_utf8: true } - } - - pub fn clear(&mut self) { - self.bytes.clear(); - self.is_known_utf8 = true; - } - - /// Creates a WTF-8 string from a potentially ill-formed UTF-16 slice of 16-bit code units. - /// - /// This is lossless: calling `.encode_wide()` on the resulting string - /// will always return the original code units. - pub fn from_wide(v: &[u16]) -> Wtf8Buf { - let mut string = Wtf8Buf::with_capacity(v.len()); - for item in char::decode_utf16(v.iter().cloned()) { - match item { - Ok(ch) => string.push_char(ch), - Err(surrogate) => { - let surrogate = surrogate.unpaired_surrogate(); - // Surrogates are known to be in the code point range. - let code_point = unsafe { CodePoint::from_u32_unchecked(surrogate as u32) }; - // The string will now contain an unpaired surrogate. - string.is_known_utf8 = false; - // Skip the WTF-8 concatenation check, - // surrogate pairs are already decoded by decode_utf16 - string.push_code_point_unchecked(code_point); - } - } - } - string - } - - /// Appends the given `char` to the end of this string. - /// This does **not** include the WTF-8 concatenation check or `is_known_utf8` check. - /// Copied from String::push. - fn push_code_point_unchecked(&mut self, code_point: CodePoint) { - let mut bytes = [0; MAX_LEN_UTF8]; - let bytes = encode_utf8_raw(code_point.value, &mut bytes); - self.bytes.extend_from_slice(bytes) - } - - #[inline] - pub fn as_slice(&self) -> &Wtf8 { - unsafe { Wtf8::from_bytes_unchecked(&self.bytes) } - } - - #[inline] - pub fn as_mut_slice(&mut self) -> &mut Wtf8 { - // Safety: `Wtf8` doesn't expose any way to mutate the bytes that would - // cause them to change from well-formed UTF-8 to ill-formed UTF-8, - // which would break the assumptions of the `is_known_utf8` field. - unsafe { Wtf8::from_mut_bytes_unchecked(&mut self.bytes) } - } - - /// Converts the string to UTF-8 without validation, if it was created from - /// valid UTF-8. - #[inline] - fn as_known_utf8(&self) -> Option<&str> { - if self.is_known_utf8 { - // SAFETY: The buffer is known to be valid UTF-8. - Some(unsafe { str::from_utf8_unchecked(self.as_bytes()) }) - } else { - None - } - } - - /// Reserves capacity for at least `additional` more bytes to be inserted - /// in the given `Wtf8Buf`. - /// The collection may reserve more space to avoid frequent reallocations. - /// - /// # Panics - /// - /// Panics if the new capacity exceeds `isize::MAX` bytes. - #[inline] - pub fn reserve(&mut self, additional: usize) { - self.bytes.reserve(additional) - } - - /// Tries to reserve capacity for at least `additional` more bytes to be - /// inserted in the given `Wtf8Buf`. The `Wtf8Buf` may reserve more space to - /// avoid frequent reallocations. After calling `try_reserve`, capacity will - /// be greater than or equal to `self.len() + additional`. Does nothing if - /// capacity is already sufficient. This method preserves the contents even - /// if an error occurs. - /// - /// # Errors - /// - /// If the capacity overflows, or the allocator reports a failure, then an error - /// is returned. - #[inline] - pub fn try_reserve(&mut self, additional: usize) -> Result<(), TryReserveError> { - self.bytes.try_reserve(additional) - } - - #[inline] - pub fn reserve_exact(&mut self, additional: usize) { - self.bytes.reserve_exact(additional) - } - - /// Tries to reserve the minimum capacity for exactly `additional` more - /// bytes to be inserted in the given `Wtf8Buf`. After calling - /// `try_reserve_exact`, capacity will be greater than or equal to - /// `self.len() + additional` if it returns `Ok(())`. - /// Does nothing if the capacity is already sufficient. - /// - /// Note that the allocator may give the `Wtf8Buf` more space than it - /// requests. Therefore, capacity can not be relied upon to be precisely - /// minimal. Prefer [`try_reserve`] if future insertions are expected. - /// - /// [`try_reserve`]: Wtf8Buf::try_reserve - /// - /// # Errors - /// - /// If the capacity overflows, or the allocator reports a failure, then an error - /// is returned. - #[inline] - pub fn try_reserve_exact(&mut self, additional: usize) -> Result<(), TryReserveError> { - self.bytes.try_reserve_exact(additional) - } - - #[inline] - pub fn shrink_to_fit(&mut self) { - self.bytes.shrink_to_fit() - } - - #[inline] - pub fn shrink_to(&mut self, min_capacity: usize) { - self.bytes.shrink_to(min_capacity) - } - - #[inline] - pub fn leak<'a>(self) -> &'a mut Wtf8 { - unsafe { Wtf8::from_mut_bytes_unchecked(self.bytes.leak()) } - } - - /// Returns the number of bytes that this string buffer can hold without reallocating. - #[inline] - pub fn capacity(&self) -> usize { - self.bytes.capacity() - } - - /// Append a UTF-8 slice at the end of the string. - #[inline] - pub fn push_str(&mut self, other: &str) { - self.bytes.extend_from_slice(other.as_bytes()) - } - - /// Append a WTF-8 slice at the end of the string. - /// - /// This replaces newly paired surrogates at the boundary - /// with a supplementary code point, - /// like concatenating ill-formed UTF-16 strings effectively would. - #[inline] - pub fn push_wtf8(&mut self, other: &Wtf8) { - match ((&*self).final_lead_surrogate(), other.initial_trail_surrogate()) { - // Replace newly paired surrogates by a supplementary code point. - (Some(lead), Some(trail)) => { - let len_without_lead_surrogate = self.len() - 3; - self.bytes.truncate(len_without_lead_surrogate); - let other_without_trail_surrogate = &other.bytes[3..]; - // 4 bytes for the supplementary code point - self.bytes.reserve(4 + other_without_trail_surrogate.len()); - self.push_char(decode_surrogate_pair(lead, trail)); - self.bytes.extend_from_slice(other_without_trail_surrogate); - } - _ => { - // If we'll be pushing a string containing a surrogate, we may - // no longer have UTF-8. - if self.is_known_utf8 && other.next_surrogate(0).is_some() { - self.is_known_utf8 = false; - } - - self.bytes.extend_from_slice(&other.bytes); - } - } - } - - /// Append a Unicode scalar value at the end of the string. - #[inline] - pub fn push_char(&mut self, c: char) { - self.push_code_point_unchecked(CodePoint::from_char(c)) - } - - /// Append a code point at the end of the string. - /// - /// This replaces newly paired surrogates at the boundary - /// with a supplementary code point, - /// like concatenating ill-formed UTF-16 strings effectively would. - #[inline] - pub fn push(&mut self, code_point: CodePoint) { - if let Some(trail) = code_point.to_trail_surrogate() { - if let Some(lead) = (&*self).final_lead_surrogate() { - let len_without_lead_surrogate = self.len() - 3; - self.bytes.truncate(len_without_lead_surrogate); - self.push_char(decode_surrogate_pair(lead, trail)); - return; - } - - // We're pushing a trailing surrogate. - self.is_known_utf8 = false; - } else if code_point.to_lead_surrogate().is_some() { - // We're pushing a leading surrogate. - self.is_known_utf8 = false; - } - - // No newly paired surrogates at the boundary. - self.push_code_point_unchecked(code_point) - } - - /// Shortens a string to the specified length. - /// - /// # Panics - /// - /// Panics if `new_len` > current length, - /// or if `new_len` is not a code point boundary. - #[inline] - pub fn truncate(&mut self, new_len: usize) { - assert!(is_code_point_boundary(self, new_len)); - self.bytes.truncate(new_len) - } - - /// Consumes the WTF-8 string and tries to convert it to a vec of bytes. - #[inline] - pub fn into_bytes(self) -> Vec { - self.bytes - } - - /// Consumes the WTF-8 string and tries to convert it to UTF-8. - /// - /// This does not copy the data. - /// - /// If the contents are not well-formed UTF-8 - /// (that is, if the string contains surrogates), - /// the original WTF-8 string is returned instead. - pub fn into_string(self) -> Result { - if self.is_known_utf8 || self.next_surrogate(0).is_none() { - Ok(unsafe { String::from_utf8_unchecked(self.bytes) }) - } else { - Err(self) - } - } - - /// Consumes the WTF-8 string and converts it lossily to UTF-8. - /// - /// This does not copy the data (but may overwrite parts of it in place). - /// - /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”) - pub fn into_string_lossy(mut self) -> String { - if !self.is_known_utf8 { - let mut pos = 0; - while let Some((surrogate_pos, _)) = self.next_surrogate(pos) { - pos = surrogate_pos + 3; - // Surrogates and the replacement character are all 3 bytes, so - // they can substituted in-place. - self.bytes[surrogate_pos..pos] - .copy_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); - } - } - unsafe { String::from_utf8_unchecked(self.bytes) } - } - - /// Converts this `Wtf8Buf` into a boxed `Wtf8`. - #[inline] - pub fn into_box(self) -> Box { - // SAFETY: relies on `Wtf8` being `repr(transparent)`. - unsafe { mem::transmute(self.bytes.into_boxed_slice()) } - } - - /// Converts a `Box` into a `Wtf8Buf`. - pub fn from_box(boxed: Box) -> Wtf8Buf { - let bytes: Box<[u8]> = unsafe { mem::transmute(boxed) }; - Wtf8Buf { bytes: bytes.into_vec(), is_known_utf8: false } - } - - /// Provides plumbing to core `Vec::extend_from_slice`. - /// More well behaving alternative to allowing outer types - /// full mutable access to the core `Vec`. - #[inline] - pub(crate) fn extend_from_slice(&mut self, other: &[u8]) { - self.bytes.extend_from_slice(other); - self.is_known_utf8 = false; - } -} - -/// Creates a new WTF-8 string from an iterator of code points. -/// -/// This replaces surrogate code point pairs with supplementary code points, -/// like concatenating ill-formed UTF-16 strings effectively would. -impl FromIterator for Wtf8Buf { - fn from_iter>(iter: T) -> Wtf8Buf { - let mut string = Wtf8Buf::new(); - string.extend(iter); - string - } -} - -/// Append code points from an iterator to the string. -/// -/// This replaces surrogate code point pairs with supplementary code points, -/// like concatenating ill-formed UTF-16 strings effectively would. -impl Extend for Wtf8Buf { - fn extend>(&mut self, iter: T) { - let iterator = iter.into_iter(); - let (low, _high) = iterator.size_hint(); - // Lower bound of one byte per code point (ASCII only) - self.bytes.reserve(low); - iterator.for_each(move |code_point| self.push(code_point)); - } - - #[inline] - fn extend_one(&mut self, code_point: CodePoint) { - self.push(code_point); - } - - #[inline] - fn extend_reserve(&mut self, additional: usize) { - // Lower bound of one byte per code point (ASCII only) - self.bytes.reserve(additional); + self.to_char().unwrap_or(char::REPLACEMENT_CHARACTER) } } @@ -547,13 +125,15 @@ impl Extend for Wtf8Buf { /// 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 AsInner<[u8]> for Wtf8 { +impl AsRef<[u8]> for Wtf8 { #[inline] - fn as_inner(&self) -> &[u8] { + fn as_ref(&self) -> &[u8] { &self.bytes } } @@ -574,12 +154,15 @@ impl fmt::Debug for Wtf8 { 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("\"") } @@ -594,13 +177,15 @@ impl fmt::Display for Wtf8 { 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_str(UTF8_REPLACEMENT_CHARACTER)?; + 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) } } @@ -611,10 +196,9 @@ impl fmt::Display for Wtf8 { impl Wtf8 { /// Creates a WTF-8 slice from a UTF-8 `&str` slice. - /// - /// Since WTF-8 is a superset of UTF-8, this always succeeds. #[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()) } } @@ -633,7 +217,7 @@ impl Wtf8 { /// Since the byte slice is not checked for valid WTF-8, this functions is /// marked unsafe. #[inline] - unsafe fn from_mut_bytes_unchecked(value: &mut [u8]) -> &mut Wtf8 { + 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) } } @@ -685,41 +269,6 @@ impl Wtf8 { str::from_utf8(&self.bytes) } - /// Creates an owned `Wtf8Buf` from a borrowed `Wtf8`. - pub fn to_owned(&self) -> Wtf8Buf { - Wtf8Buf { bytes: self.bytes.to_vec(), is_known_utf8: false } - } - - /// Lossily converts the string to UTF-8. - /// Returns a UTF-8 `&str` slice if the contents are well-formed in UTF-8. - /// - /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”). - /// - /// This only copies the data if necessary (if it contains any surrogate). - pub fn to_string_lossy(&self) -> Cow<'_, str> { - let Some((surrogate_pos, _)) = self.next_surrogate(0) else { - return Cow::Borrowed(unsafe { str::from_utf8_unchecked(&self.bytes) }); - }; - let wtf8_bytes = &self.bytes; - let mut utf8_bytes = Vec::with_capacity(self.len()); - utf8_bytes.extend_from_slice(&wtf8_bytes[..surrogate_pos]); - utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); - let mut pos = surrogate_pos + 3; - loop { - match self.next_surrogate(pos) { - Some((surrogate_pos, _)) => { - utf8_bytes.extend_from_slice(&wtf8_bytes[pos..surrogate_pos]); - utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes()); - pos = surrogate_pos + 3; - } - None => { - utf8_bytes.extend_from_slice(&wtf8_bytes[pos..]); - return Cow::Owned(unsafe { String::from_utf8_unchecked(utf8_bytes) }); - } - } - } - } - /// Converts the WTF-8 string to potentially ill-formed UTF-16 /// and return an iterator of 16-bit code units. /// @@ -732,7 +281,7 @@ impl Wtf8 { } #[inline] - fn next_surrogate(&self, mut pos: usize) -> Option<(usize, u16)> { + pub fn next_surrogate(&self, mut pos: usize) -> Option<(usize, u16)> { let mut iter = self.bytes[pos..].iter(); loop { let b = *iter.next()?; @@ -762,7 +311,7 @@ impl Wtf8 { } #[inline] - fn final_lead_surrogate(&self) -> Option { + pub fn final_lead_surrogate(&self) -> Option { match self.bytes { [.., 0xED, b2 @ 0xA0..=0xAF, b3] => Some(decode_surrogate(b2, b3)), _ => None, @@ -770,43 +319,13 @@ impl Wtf8 { } #[inline] - fn initial_trail_surrogate(&self) -> Option { + pub fn initial_trail_surrogate(&self) -> Option { match self.bytes { [0xED, b2 @ 0xB0..=0xBF, b3, ..] => Some(decode_surrogate(b2, b3)), _ => None, } } - pub fn clone_into(&self, buf: &mut Wtf8Buf) { - buf.is_known_utf8 = false; - self.bytes.clone_into(&mut buf.bytes); - } - - /// Boxes this `Wtf8`. - #[inline] - pub fn into_box(&self) -> Box { - let boxed: Box<[u8]> = self.bytes.into(); - unsafe { mem::transmute(boxed) } - } - - /// Creates a boxed, empty `Wtf8`. - pub fn empty_box() -> Box { - let boxed: Box<[u8]> = Default::default(); - unsafe { mem::transmute(boxed) } - } - - #[inline] - pub fn into_arc(&self) -> Arc { - let arc: Arc<[u8]> = Arc::from(&self.bytes); - unsafe { Arc::from_raw(Arc::into_raw(arc) as *const Wtf8) } - } - - #[inline] - pub fn into_rc(&self) -> Rc { - let rc: Rc<[u8]> = Rc::from(&self.bytes); - unsafe { Rc::from_raw(Rc::into_raw(rc) as *const Wtf8) } - } - #[inline] pub fn make_ascii_lowercase(&mut self) { self.bytes.make_ascii_lowercase() @@ -818,23 +337,23 @@ impl Wtf8 { } #[inline] - pub fn to_ascii_lowercase(&self) -> Wtf8Buf { - Wtf8Buf { bytes: self.bytes.to_ascii_lowercase(), is_known_utf8: false } + pub fn is_ascii(&self) -> bool { + self.bytes.is_ascii() } #[inline] - pub fn to_ascii_uppercase(&self) -> Wtf8Buf { - Wtf8Buf { bytes: self.bytes.to_ascii_uppercase(), is_known_utf8: false } + pub fn eq_ignore_ascii_case(&self, other: &Self) -> bool { + self.bytes.eq_ignore_ascii_case(&other.bytes) } #[inline] - pub fn is_ascii(&self) -> bool { - self.bytes.is_ascii() + pub fn is_code_point_boundary(&self, index: usize) -> bool { + is_code_point_boundary(self, index) } #[inline] - pub fn eq_ignore_ascii_case(&self, other: &Self) -> bool { - self.bytes.eq_ignore_ascii_case(&other.bytes) + pub fn check_utf8_boundary(&self, index: usize) { + check_utf8_boundary(self, index) } } @@ -849,11 +368,11 @@ impl ops::Index> for Wtf8 { #[inline] fn index(&self, range: ops::Range) -> &Wtf8 { - // is_code_point_boundary checks that the index is in [0, .len()] if range.start <= range.end - && is_code_point_boundary(self, range.start) - && is_code_point_boundary(self, 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) @@ -872,8 +391,8 @@ impl ops::Index> for Wtf8 { #[inline] fn index(&self, range: ops::RangeFrom) -> &Wtf8 { - // is_code_point_boundary checks that the index is in [0, .len()] - if is_code_point_boundary(self, range.start) { + 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()) @@ -892,8 +411,8 @@ impl ops::Index> for Wtf8 { #[inline] fn index(&self, range: ops::RangeTo) -> &Wtf8 { - // is_code_point_boundary checks that the index is in [0, .len()] - if is_code_point_boundary(self, range.end) { + 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) @@ -916,11 +435,7 @@ fn decode_surrogate(second_byte: u8, third_byte: u8) -> u16 { 0xD800 | (second_byte as u16 & 0x3F) << 6 | third_byte as u16 & 0x3F } -#[inline] -fn decode_surrogate_pair(lead: u16, trail: u16) -> char { - let code_point = 0x10000 + ((((lead - 0xD800) as u32) << 10) | (trail - 0xDC00) as u32); - unsafe { char::from_u32_unchecked(code_point) } -} +// helps diff to be unindented /// Copied from str::is_char_boundary #[inline] @@ -964,7 +479,7 @@ pub fn check_utf8_boundary(slice: &Wtf8, index: usize) { /// Copied from core::str::raw::slice_unchecked #[inline] -pub unsafe fn slice_unchecked(s: &Wtf8, begin: usize, end: usize) -> &Wtf8 { +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; @@ -975,7 +490,7 @@ pub unsafe fn slice_unchecked(s: &Wtf8, begin: usize, end: usize) -> &Wtf8 { /// Copied from core::str::raw::slice_error_fail #[inline(never)] -pub fn slice_error_fail(s: &Wtf8, begin: usize, end: usize) -> ! { +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"); } @@ -984,6 +499,7 @@ pub fn slice_error_fail(s: &Wtf8, begin: usize, end: usize) -> ! { /// /// Created with the method `.code_points()`. #[derive(Clone)] +#[doc(hidden)] pub struct Wtf8CodePoints<'a> { bytes: slice::Iter<'a, u8>, } @@ -994,7 +510,7 @@ impl Iterator for Wtf8CodePoints<'_> { #[inline] fn next(&mut self) -> Option { // SAFETY: `self.bytes` has been created from a WTF-8 string - unsafe { next_code_point(&mut self.bytes).map(|c| CodePoint { value: c }) } + unsafe { next_code_point(&mut self.bytes).map(|c| CodePoint::from_u32_unchecked(c)) } } #[inline] @@ -1004,9 +520,19 @@ impl Iterator for Wtf8CodePoints<'_> { } } +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, @@ -1027,7 +553,7 @@ impl Iterator for EncodeWide<'_> { let mut buf = [0; MAX_LEN_UTF16]; self.code_points.next().map(|code_point| { - let n = encode_utf16_raw(code_point.value, &mut buf).len(); + let n = encode_utf16_raw(code_point.to_u32(), &mut buf).len(); if n == 2 { self.extra = buf[1]; } @@ -1046,21 +572,19 @@ impl Iterator for EncodeWide<'_> { } } +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(&self, state: &mut H) { - self.value.hash(state) - } -} - -impl Hash for Wtf8Buf { - #[inline] - fn hash(&self, state: &mut H) { - state.write(&self.bytes); - 0xfeu8.hash(state) + self.0.hash(state) } } diff --git a/library/coretests/tests/lib.rs b/library/coretests/tests/lib.rs index d2281b1df2f..4415b614817 100644 --- a/library/coretests/tests/lib.rs +++ b/library/coretests/tests/lib.rs @@ -194,6 +194,7 @@ mod time; mod tuple; mod unicode; mod waker; +mod wtf8; /// Copied from `std::test_helpers::test_rng`, see that function for rationale. #[track_caller] diff --git a/library/coretests/tests/wtf8.rs b/library/coretests/tests/wtf8.rs new file mode 100644 index 00000000000..9f187e70630 --- /dev/null +++ b/library/coretests/tests/wtf8.rs @@ -0,0 +1 @@ +// All `wtf8` tests live in library/alloctests/tests/wtf8.rs diff --git a/library/std/src/lib.rs b/library/std/src/lib.rs index ab417b6c72f..ca76dcc5147 100644 --- a/library/std/src/lib.rs +++ b/library/std/src/lib.rs @@ -391,6 +391,7 @@ #![feature(try_with_capacity)] #![feature(unique_rc_arc)] #![feature(vec_into_raw_parts)] +#![feature(wtf8_internals)] // tidy-alphabetical-end // // Library features (unwind): diff --git a/library/std/src/os/windows/ffi.rs b/library/std/src/os/windows/ffi.rs index 496443dbbc3..345d5b74285 100644 --- a/library/std/src/os/windows/ffi.rs +++ b/library/std/src/os/windows/ffi.rs @@ -53,12 +53,13 @@ #![stable(feature = "rust1", since = "1.0.0")] +use alloc::wtf8::Wtf8Buf; + use crate::ffi::{OsStr, OsString}; +use crate::fmt; +use crate::iter::FusedIterator; use crate::sealed::Sealed; use crate::sys::os_str::Buf; -#[stable(feature = "rust1", since = "1.0.0")] -pub use crate::sys_common::wtf8::EncodeWide; -use crate::sys_common::wtf8::Wtf8Buf; use crate::sys_common::{AsInner, FromInner}; /// Windows-specific extensions to [`OsString`]. @@ -130,6 +131,35 @@ pub trait OsStrExt: Sealed { impl OsStrExt for OsStr { #[inline] fn encode_wide(&self) -> EncodeWide<'_> { - self.as_inner().inner.encode_wide() + EncodeWide { inner: self.as_inner().inner.encode_wide() } + } +} + +/// Iterator returned by [`OsStrExt::encode_wide`]. +#[stable(feature = "rust1", since = "1.0.0")] +#[derive(Clone)] +pub struct EncodeWide<'a> { + inner: alloc::wtf8::EncodeWide<'a>, +} +#[stable(feature = "encode_wide_debug", since = "CURRENT_RUSTC_VERSION")] +impl fmt::Debug for EncodeWide<'_> { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt::Debug::fmt(&self.inner, f) + } +} +#[stable(feature = "rust1", since = "1.0.0")] +impl Iterator for EncodeWide<'_> { + type Item = u16; + + #[inline] + fn next(&mut self) -> Option { + self.inner.next() + } + + #[inline] + fn size_hint(&self) -> (usize, Option) { + self.inner.size_hint() } } +#[stable(feature = "encode_wide_fused_iterator", since = "1.62.0")] +impl FusedIterator for EncodeWide<'_> {} diff --git a/library/std/src/sys/os_str/wtf8.rs b/library/std/src/sys/os_str/wtf8.rs index bbc704ebf86..96da891874e 100644 --- a/library/std/src/sys/os_str/wtf8.rs +++ b/library/std/src/sys/os_str/wtf8.rs @@ -1,12 +1,12 @@ //! The underlying OsString/OsStr implementation on Windows is a //! wrapper around the "WTF-8" encoding; see the `wtf8` module for more. +use alloc::wtf8::{Wtf8, Wtf8Buf}; use core::clone::CloneToUninit; use crate::borrow::Cow; use crate::collections::TryReserveError; use crate::rc::Rc; use crate::sync::Arc; -use crate::sys_common::wtf8::{Wtf8, Wtf8Buf, check_utf8_boundary}; use crate::sys_common::{AsInner, FromInner, IntoInner}; use crate::{fmt, mem}; @@ -220,7 +220,9 @@ impl Buf { /// trailing surrogate half. #[inline] pub unsafe fn extend_from_slice_unchecked(&mut self, other: &[u8]) { - self.inner.extend_from_slice(other); + unsafe { + self.inner.extend_from_slice_unchecked(other); + } } } @@ -238,7 +240,7 @@ impl Slice { #[track_caller] #[inline] pub fn check_public_boundary(&self, index: usize) { - check_utf8_boundary(&self.inner, index); + self.inner.check_utf8_boundary(index); } #[inline] diff --git a/library/std/src/sys_common/mod.rs b/library/std/src/sys_common/mod.rs index 24b6cff1309..ec45c723e0d 100644 --- a/library/std/src/sys_common/mod.rs +++ b/library/std/src/sys_common/mod.rs @@ -21,7 +21,6 @@ mod tests; pub mod wstr; -pub mod wtf8; // common error constructors diff --git a/src/tools/tidy/src/unit_tests.rs b/src/tools/tidy/src/unit_tests.rs index 3d14a467319..7396310ed37 100644 --- a/src/tools/tidy/src/unit_tests.rs +++ b/src/tools/tidy/src/unit_tests.rs @@ -61,6 +61,7 @@ pub fn check(root_path: &Path, stdlib: bool, bad: &mut bool) { || path.ends_with("library/alloc/src/collections/linked_list/tests.rs") || path.ends_with("library/alloc/src/collections/vec_deque/tests.rs") || path.ends_with("library/alloc/src/raw_vec/tests.rs") + || path.ends_with("library/alloc/src/wtf8/tests.rs") } }; -- cgit 1.4.1-3-g733a5 From 7c81a067ea4cfd289d30c3903ac60b113f481c87 Mon Sep 17 00:00:00 2001 From: ltdk Date: Wed, 13 Aug 2025 18:54:48 -0400 Subject: Diff-massaging commit --- library/alloc/src/wtf8/mod.rs | 75 ++++++++++++++++++--------------------- library/core/src/wtf8.rs | 82 +++++++++++++++++++------------------------ 2 files changed, 70 insertions(+), 87 deletions(-) (limited to 'library/core/src') diff --git a/library/alloc/src/wtf8/mod.rs b/library/alloc/src/wtf8/mod.rs index 95d317a5efb..047994adc44 100644 --- a/library/alloc/src/wtf8/mod.rs +++ b/library/alloc/src/wtf8/mod.rs @@ -451,53 +451,46 @@ impl Extend for Wtf8Buf { } } -// helps diff -mod wtf8 { - use super::*; - - /// Creates an owned `Wtf8Buf` from a borrowed `Wtf8`. - pub(super) fn to_owned(slice: &Wtf8) -> Wtf8Buf { - Wtf8Buf { bytes: slice.as_bytes().to_vec(), is_known_utf8: false } - } +/// Creates an owned `Wtf8Buf` from a borrowed `Wtf8`. +pub(super) fn to_owned(slice: &Wtf8) -> Wtf8Buf { + Wtf8Buf { bytes: slice.as_bytes().to_vec(), is_known_utf8: false } +} - /// Lossily converts the string to UTF-8. - /// Returns a UTF-8 `&str` slice if the contents are well-formed in UTF-8. - /// - /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”). - /// - /// This only copies the data if necessary (if it contains any surrogate). - pub(super) fn to_string_lossy(slice: &Wtf8) -> Cow<'_, str> { - let Some((surrogate_pos, _)) = slice.next_surrogate(0) else { - return Cow::Borrowed(unsafe { str::from_utf8_unchecked(slice.as_bytes()) }); - }; - let wtf8_bytes = slice.as_bytes(); - let mut utf8_bytes = Vec::with_capacity(slice.len()); - utf8_bytes.extend_from_slice(&wtf8_bytes[..surrogate_pos]); - utf8_bytes.extend_from_slice("\u{FFFD}".as_bytes()); - let mut pos = surrogate_pos + 3; - loop { - match slice.next_surrogate(pos) { - Some((surrogate_pos, _)) => { - utf8_bytes.extend_from_slice(&wtf8_bytes[pos..surrogate_pos]); - utf8_bytes.extend_from_slice("\u{FFFD}".as_bytes()); - pos = surrogate_pos + 3; - } - None => { - utf8_bytes.extend_from_slice(&wtf8_bytes[pos..]); - return Cow::Owned(unsafe { String::from_utf8_unchecked(utf8_bytes) }); - } +/// Lossily converts the string to UTF-8. +/// Returns a UTF-8 `&str` slice if the contents are well-formed in UTF-8. +/// +/// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”). +/// +/// This only copies the data if necessary (if it contains any surrogate). +pub(super) fn to_string_lossy(slice: &Wtf8) -> Cow<'_, str> { + let Some((surrogate_pos, _)) = slice.next_surrogate(0) else { + return Cow::Borrowed(unsafe { str::from_utf8_unchecked(slice.as_bytes()) }); + }; + let wtf8_bytes = slice.as_bytes(); + let mut utf8_bytes = Vec::with_capacity(slice.len()); + utf8_bytes.extend_from_slice(&wtf8_bytes[..surrogate_pos]); + utf8_bytes.extend_from_slice("\u{FFFD}".as_bytes()); + let mut pos = surrogate_pos + 3; + loop { + match slice.next_surrogate(pos) { + Some((surrogate_pos, _)) => { + utf8_bytes.extend_from_slice(&wtf8_bytes[pos..surrogate_pos]); + utf8_bytes.extend_from_slice("\u{FFFD}".as_bytes()); + pos = surrogate_pos + 3; + } + None => { + utf8_bytes.extend_from_slice(&wtf8_bytes[pos..]); + return Cow::Owned(unsafe { String::from_utf8_unchecked(utf8_bytes) }); } } } - - #[inline] - pub(super) fn clone_into(slice: &Wtf8, buf: &mut Wtf8Buf) { - buf.is_known_utf8 = false; - slice.as_bytes().clone_into(&mut buf.bytes); - } } -use self::wtf8::{to_owned, to_string_lossy, clone_into}; +#[inline] +pub(super) fn clone_into(slice: &Wtf8, buf: &mut Wtf8Buf) { + buf.is_known_utf8 = false; + slice.as_bytes().clone_into(&mut buf.bytes); +} #[cfg(not(test))] impl Wtf8 { diff --git a/library/core/src/wtf8.rs b/library/core/src/wtf8.rs index 5631993dea2..de0dfa560a3 100644 --- a/library/core/src/wtf8.rs +++ b/library/core/src/wtf8.rs @@ -345,16 +345,6 @@ impl Wtf8 { pub fn eq_ignore_ascii_case(&self, other: &Self) -> bool { self.bytes.eq_ignore_ascii_case(&other.bytes) } - - #[inline] - pub fn is_code_point_boundary(&self, index: usize) -> bool { - is_code_point_boundary(self, index) - } - - #[inline] - pub fn check_utf8_boundary(&self, index: usize) { - check_utf8_boundary(self, index) - } } /// Returns a slice of the given string for the byte range \[`begin`..`end`). @@ -435,44 +425,44 @@ fn decode_surrogate(second_byte: u8, third_byte: u8) -> u16 { 0xD800 | (second_byte as u16 & 0x3F) << 6 | third_byte as u16 & 0x3F } -// helps diff to be unindented - -/// Copied from str::is_char_boundary -#[inline] -pub fn is_code_point_boundary(slice: &Wtf8, index: usize) -> bool { - if index == 0 { - return true; - } - match slice.bytes.get(index) { - None => index == slice.len(), - Some(&b) => (b as i8) >= -0x40, +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::slice_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(slice: &Wtf8, index: usize) { - if index == 0 { - return; - } - match slice.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 == slice.len() => return, - None => panic!("byte index {index} is out of bounds"), - } - if slice.bytes[index + 1] >= 0xA0 { - // There's a surrogate after index. Now check before index. - if index >= 3 && slice.bytes[index - 3] == 0xED && slice.bytes[index - 2] >= 0xA0 { - panic!("byte index {index} lies between surrogate codepoints"); + /// 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"); + } } } } -- cgit 1.4.1-3-g733a5 From 698db13cd068b1807780f97344bb764efc87703f Mon Sep 17 00:00:00 2001 From: Karol Zwolak Date: Wed, 20 Aug 2025 20:23:08 +0200 Subject: improve float to_degrees/to_radians rounding comments and impl * revise comments explaining why we are using literal or expression * add unspecified precision comments as we don't guarantee precision * use expression in `f128::to_degrees()` * make `f64::to_degrees()` impl consistent with other functions --- library/core/src/num/f128.rs | 23 ++++++++++++++++++++--- library/core/src/num/f16.rs | 20 ++++++++++++++++++-- library/core/src/num/f32.rs | 20 +++++++++++++++++++- library/core/src/num/f64.rs | 26 ++++++++++++++++++++++---- 4 files changed, 79 insertions(+), 10 deletions(-) (limited to 'library/core/src') 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 } -- cgit 1.4.1-3-g733a5 From e42c1b12961385d6a5e9e442344fcaef8678ab58 Mon Sep 17 00:00:00 2001 From: okaneco <47607823+okaneco@users.noreply.github.com> Date: Fri, 22 Aug 2025 11:58:36 -0400 Subject: Stabilize `round_char_boundary` feature --- compiler/rustc_middle/src/lib.rs | 2 +- compiler/rustc_span/src/lib.rs | 2 +- library/alloctests/tests/lib.rs | 1 - library/core/src/str/mod.rs | 8 +++---- library/std/src/lib.rs | 1 - src/librustdoc/lib.rs | 2 +- src/tools/clippy/clippy_lints/src/lib.rs | 2 +- .../tests/ui/char_indices_as_byte_indices.fixed | 1 - .../tests/ui/char_indices_as_byte_indices.rs | 1 - .../tests/ui/char_indices_as_byte_indices.stderr | 28 +++++++++++----------- 10 files changed, 22 insertions(+), 26 deletions(-) (limited to 'library/core/src') diff --git a/compiler/rustc_middle/src/lib.rs b/compiler/rustc_middle/src/lib.rs index e5cc23c213d..5d0723a0090 100644 --- a/compiler/rustc_middle/src/lib.rs +++ b/compiler/rustc_middle/src/lib.rs @@ -29,6 +29,7 @@ #![allow(rustc::diagnostic_outside_of_impl)] #![allow(rustc::direct_use_of_rustc_type_ir)] #![allow(rustc::untranslatable_diagnostic)] +#![cfg_attr(bootstrap, feature(round_char_boundary))] #![doc(html_root_url = "https://doc.rust-lang.org/nightly/nightly-rustc/")] #![doc(rust_logo)] #![feature(allocator_api)] @@ -51,7 +52,6 @@ #![feature(negative_impls)] #![feature(never_type)] #![feature(ptr_alignment_type)] -#![feature(round_char_boundary)] #![feature(rustc_attrs)] #![feature(rustdoc_internals)] #![feature(sized_hierarchy)] diff --git a/compiler/rustc_span/src/lib.rs b/compiler/rustc_span/src/lib.rs index d647ec28aae..ae6755f0764 100644 --- a/compiler/rustc_span/src/lib.rs +++ b/compiler/rustc_span/src/lib.rs @@ -17,6 +17,7 @@ // tidy-alphabetical-start #![allow(internal_features)] +#![cfg_attr(bootstrap, feature(round_char_boundary))] #![doc(html_root_url = "https://doc.rust-lang.org/nightly/nightly-rustc/")] #![doc(rust_logo)] #![feature(array_windows)] @@ -26,7 +27,6 @@ #![feature(map_try_insert)] #![feature(negative_impls)] #![feature(read_buf)] -#![feature(round_char_boundary)] #![feature(rustc_attrs)] #![feature(rustdoc_internals)] // tidy-alphabetical-end diff --git a/library/alloctests/tests/lib.rs b/library/alloctests/tests/lib.rs index fcfc7f8dd29..cb9479abee3 100644 --- a/library/alloctests/tests/lib.rs +++ b/library/alloctests/tests/lib.rs @@ -23,7 +23,6 @@ #![feature(inplace_iteration)] #![feature(iter_advance_by)] #![feature(iter_next_chunk)] -#![feature(round_char_boundary)] #![feature(slice_partition_dedup)] #![feature(string_from_utf8_lossy_owned)] #![feature(string_remove_matches)] diff --git a/library/core/src/str/mod.rs b/library/core/src/str/mod.rs index 1b6e84175b9..89115b254b1 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() { diff --git a/library/std/src/lib.rs b/library/std/src/lib.rs index ab417b6c72f..74ac1230ff5 100644 --- a/library/std/src/lib.rs +++ b/library/std/src/lib.rs @@ -263,7 +263,6 @@ all(target_vendor = "fortanix", target_env = "sgx"), feature(slice_index_methods, coerce_unsized, sgx_platform) )] -#![cfg_attr(any(windows, target_os = "uefi"), feature(round_char_boundary))] #![cfg_attr(target_family = "wasm", feature(stdarch_wasm_atomic_wait))] #![cfg_attr(target_arch = "wasm64", feature(simd_wasm64))] // diff --git a/src/librustdoc/lib.rs b/src/librustdoc/lib.rs index 28dbd8ba7d3..a59272d49c7 100644 --- a/src/librustdoc/lib.rs +++ b/src/librustdoc/lib.rs @@ -1,4 +1,5 @@ // tidy-alphabetical-start +#![cfg_attr(bootstrap, feature(round_char_boundary))] #![doc( html_root_url = "https://doc.rust-lang.org/nightly/", html_playground_url = "https://play.rust-lang.org/" @@ -12,7 +13,6 @@ #![feature(format_args_nl)] #![feature(if_let_guard)] #![feature(iter_intersperse)] -#![feature(round_char_boundary)] #![feature(rustc_private)] #![feature(test)] #![warn(rustc::internal)] diff --git a/src/tools/clippy/clippy_lints/src/lib.rs b/src/tools/clippy/clippy_lints/src/lib.rs index 844bc1b0e39..edc91718263 100644 --- a/src/tools/clippy/clippy_lints/src/lib.rs +++ b/src/tools/clippy/clippy_lints/src/lib.rs @@ -7,7 +7,7 @@ #![feature(iter_intersperse)] #![feature(iter_partition_in_place)] #![feature(never_type)] -#![feature(round_char_boundary)] +#![cfg_attr(bootstrap, feature(round_char_boundary))] #![feature(rustc_private)] #![feature(stmt_expr_attributes)] #![feature(unwrap_infallible)] diff --git a/src/tools/clippy/tests/ui/char_indices_as_byte_indices.fixed b/src/tools/clippy/tests/ui/char_indices_as_byte_indices.fixed index 04c8f6782c5..375a101c2e3 100644 --- a/src/tools/clippy/tests/ui/char_indices_as_byte_indices.fixed +++ b/src/tools/clippy/tests/ui/char_indices_as_byte_indices.fixed @@ -1,4 +1,3 @@ -#![feature(round_char_boundary)] #![warn(clippy::char_indices_as_byte_indices)] trait StrExt { diff --git a/src/tools/clippy/tests/ui/char_indices_as_byte_indices.rs b/src/tools/clippy/tests/ui/char_indices_as_byte_indices.rs index 773a4fc65f1..eebc39962a2 100644 --- a/src/tools/clippy/tests/ui/char_indices_as_byte_indices.rs +++ b/src/tools/clippy/tests/ui/char_indices_as_byte_indices.rs @@ -1,4 +1,3 @@ -#![feature(round_char_boundary)] #![warn(clippy::char_indices_as_byte_indices)] trait StrExt { diff --git a/src/tools/clippy/tests/ui/char_indices_as_byte_indices.stderr b/src/tools/clippy/tests/ui/char_indices_as_byte_indices.stderr index e2b4c1db78c..fae81fd772d 100644 --- a/src/tools/clippy/tests/ui/char_indices_as_byte_indices.stderr +++ b/src/tools/clippy/tests/ui/char_indices_as_byte_indices.stderr @@ -1,12 +1,12 @@ error: indexing into a string with a character position where a byte index is expected - --> tests/ui/char_indices_as_byte_indices.rs:13:24 + --> tests/ui/char_indices_as_byte_indices.rs:12:24 | LL | let _ = prim[..idx]; | ^^^ | = note: a character can take up more than one byte, so they are not interchangeable note: position comes from the enumerate iterator - --> tests/ui/char_indices_as_byte_indices.rs:12:10 + --> tests/ui/char_indices_as_byte_indices.rs:11:10 | LL | for (idx, _) in prim.chars().enumerate() { | ^^^ ^^^^^^^^^^^ @@ -19,14 +19,14 @@ LL + for (idx, _) in prim.char_indices() { | error: passing a character position to a method that expects a byte index - --> tests/ui/char_indices_as_byte_indices.rs:15:23 + --> tests/ui/char_indices_as_byte_indices.rs:14:23 | LL | prim.split_at(idx); | ^^^ | = note: a character can take up more than one byte, so they are not interchangeable note: position comes from the enumerate iterator - --> tests/ui/char_indices_as_byte_indices.rs:12:10 + --> tests/ui/char_indices_as_byte_indices.rs:11:10 | LL | for (idx, _) in prim.chars().enumerate() { | ^^^ ^^^^^^^^^^^ @@ -37,14 +37,14 @@ LL + for (idx, _) in prim.char_indices() { | error: passing a character position to a method that expects a byte index - --> tests/ui/char_indices_as_byte_indices.rs:19:49 + --> tests/ui/char_indices_as_byte_indices.rs:18:49 | LL | let _ = prim[..prim.floor_char_boundary(idx)]; | ^^^ | = note: a character can take up more than one byte, so they are not interchangeable note: position comes from the enumerate iterator - --> tests/ui/char_indices_as_byte_indices.rs:12:10 + --> tests/ui/char_indices_as_byte_indices.rs:11:10 | LL | for (idx, _) in prim.chars().enumerate() { | ^^^ ^^^^^^^^^^^ @@ -55,14 +55,14 @@ LL + for (idx, _) in prim.char_indices() { | error: indexing into a string with a character position where a byte index is expected - --> tests/ui/char_indices_as_byte_indices.rs:29:24 + --> tests/ui/char_indices_as_byte_indices.rs:28:24 | LL | let _ = prim[..c.0]; | ^^^ | = note: a character can take up more than one byte, so they are not interchangeable note: position comes from the enumerate iterator - --> tests/ui/char_indices_as_byte_indices.rs:28:9 + --> tests/ui/char_indices_as_byte_indices.rs:27:9 | LL | for c in prim.chars().enumerate() { | ^ ^^^^^^^^^^^ @@ -73,14 +73,14 @@ LL + for c in prim.char_indices() { | error: passing a character position to a method that expects a byte index - --> tests/ui/char_indices_as_byte_indices.rs:31:23 + --> tests/ui/char_indices_as_byte_indices.rs:30:23 | LL | prim.split_at(c.0); | ^^^ | = note: a character can take up more than one byte, so they are not interchangeable note: position comes from the enumerate iterator - --> tests/ui/char_indices_as_byte_indices.rs:28:9 + --> tests/ui/char_indices_as_byte_indices.rs:27:9 | LL | for c in prim.chars().enumerate() { | ^ ^^^^^^^^^^^ @@ -91,14 +91,14 @@ LL + for c in prim.char_indices() { | error: indexing into a string with a character position where a byte index is expected - --> tests/ui/char_indices_as_byte_indices.rs:36:26 + --> tests/ui/char_indices_as_byte_indices.rs:35:26 | LL | let _ = string[..idx]; | ^^^ | = note: a character can take up more than one byte, so they are not interchangeable note: position comes from the enumerate iterator - --> tests/ui/char_indices_as_byte_indices.rs:35:10 + --> tests/ui/char_indices_as_byte_indices.rs:34:10 | LL | for (idx, _) in string.chars().enumerate() { | ^^^ ^^^^^^^^^^^ @@ -109,14 +109,14 @@ LL + for (idx, _) in string.char_indices() { | error: passing a character position to a method that expects a byte index - --> tests/ui/char_indices_as_byte_indices.rs:38:25 + --> tests/ui/char_indices_as_byte_indices.rs:37:25 | LL | string.split_at(idx); | ^^^ | = note: a character can take up more than one byte, so they are not interchangeable note: position comes from the enumerate iterator - --> tests/ui/char_indices_as_byte_indices.rs:35:10 + --> tests/ui/char_indices_as_byte_indices.rs:34:10 | LL | for (idx, _) in string.chars().enumerate() { | ^^^ ^^^^^^^^^^^ -- cgit 1.4.1-3-g733a5 From 0da328b2c61d39da1e0458ac528b667ae9055eb6 Mon Sep 17 00:00:00 2001 From: WANG Rui Date: Wed, 27 Aug 2025 11:45:42 +0800 Subject: Add spin_loop hint for LoongArch --- library/core/src/hint.rs | 56 ++++++++++++++++++++---------------------------- 1 file changed, 23 insertions(+), 33 deletions(-) (limited to 'library/core/src') 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 */ } } } -- cgit 1.4.1-3-g733a5 From dacae07b6d8c42f5eca57de9c18e17349994d654 Mon Sep 17 00:00:00 2001 From: Alice Ryhl Date: Wed, 27 Aug 2025 12:42:49 +0000 Subject: Rename `Location::file_with_nul` to `file_as_c_str` --- library/core/src/panic/location.rs | 4 ++-- tests/ui/rfcs/rfc-2091-track-caller/file-is-nul-terminated.rs | 6 +++--- 2 files changed, 5 insertions(+), 5 deletions(-) (limited to 'library/core/src') 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, 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/tests/ui/rfcs/rfc-2091-track-caller/file-is-nul-terminated.rs b/tests/ui/rfcs/rfc-2091-track-caller/file-is-nul-terminated.rs index 65e61a21f1a..7902f40b09b 100644 --- a/tests/ui/rfcs/rfc-2091-track-caller/file-is-nul-terminated.rs +++ b/tests/ui/rfcs/rfc-2091-track-caller/file-is-nul-terminated.rs @@ -5,12 +5,12 @@ const fn assert_file_has_trailing_zero() { let caller = core::panic::Location::caller(); let file_str = caller.file(); - let file_with_nul = caller.file_with_nul(); - if file_str.len() != file_with_nul.count_bytes() { + let file_cstr = caller.file_as_c_str(); + if file_str.len() != file_cstr.count_bytes() { panic!("mismatched lengths"); } let trailing_byte: core::ffi::c_char = unsafe { - *file_with_nul.as_ptr().offset(file_with_nul.count_bytes() as _) + *file_cstr.as_ptr().offset(file_cstr.count_bytes() as _) }; if trailing_byte != 0 { panic!("trailing byte was nonzero") -- cgit 1.4.1-3-g733a5 From e06cd9f2c82c5bb3999eafda7abe2b1951384f78 Mon Sep 17 00:00:00 2001 From: Pavel Grigorenko Date: Wed, 27 Aug 2025 16:15:08 +0300 Subject: `const`ify (the unstable) `str::as_str` --- library/core/src/str/mod.rs | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'library/core/src') diff --git a/library/core/src/str/mod.rs b/library/core/src/str/mod.rs index 1b6e84175b9..1a055a0a5ad 100644 --- a/library/core/src/str/mod.rs +++ b/library/core/src/str/mod.rs @@ -3072,7 +3072,7 @@ impl str { /// for example references to `Box` or `Arc`. #[inline] #[unstable(feature = "str_as_str", issue = "130366")] - pub fn as_str(&self) -> &str { + pub const fn as_str(&self) -> &str { self } } -- cgit 1.4.1-3-g733a5 From df802ccd2fad88ea617622ae0bf89067a250145e Mon Sep 17 00:00:00 2001 From: neeko-cat <60898184+neeko-cat@users.noreply.github.com> Date: Thu, 28 Aug 2025 22:19:37 +0200 Subject: fix `core::marker::Destruct` doc --- library/core/src/marker.rs | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'library/core/src') 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")] -- cgit 1.4.1-3-g733a5 From 1a33d628df211c92650e357c291097b6ce4f223f Mon Sep 17 00:00:00 2001 From: Jeremy Smart Date: Thu, 28 Aug 2025 18:37:58 -0400 Subject: implement Sum and Product for Saturating(u*) --- library/core/src/iter/traits/accum.rs | 58 +++++++++++++++++++++++++++- library/coretests/tests/iter/traits/accum.rs | 36 +++++++++++++++++ 2 files changed, 93 insertions(+), 1 deletion(-) (limited to 'library/core/src') 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>(iter: I) -> Self { + iter.fold( + $zero, + |a, b| a + b, + ) + } + } + + #[$attr] + #[doc = $doc] + impl Product for $a { + fn product>(iter: I) -> Self { + iter.fold( + $one, + |a, b| a * b, + ) + } + } + + #[$attr] + #[doc = $doc] + impl<'a> Sum<&'a $a> for $a { + fn sum>(iter: I) -> Self { + iter.fold( + $zero, + |a, b| a + b, + ) + } + } + + #[$attr] + #[doc = $doc] + impl<'a> Product<&'a $a> for $a { + fn product>(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/coretests/tests/iter/traits/accum.rs b/library/coretests/tests/iter/traits/accum.rs index f3eeb31fe58..95f299d2680 100644 --- a/library/coretests/tests/iter/traits/accum.rs +++ b/library/coretests/tests/iter/traits/accum.rs @@ -1,4 +1,5 @@ use core::iter::*; +use std::num::Saturating; #[test] fn test_iterator_sum() { @@ -64,3 +65,38 @@ fn test_iterator_product_option() { let v: &[Option] = &[Some(1), None, Some(3), Some(4)]; assert_eq!(v.iter().cloned().product::>(), None); } + +#[test] +fn test_saturating_sum_product() { + let v = (1u32..=10).map(|i| Saturating(i)); + assert_eq!(v.sum::>(), Saturating(55)); + let v = (1u32..=10).map(|i| Saturating(i)); + assert_eq!(v.product::>(), Saturating(3628800)); + let v = [Saturating(usize::MAX), Saturating(2)]; + assert_eq!(v.iter().copied().sum::>(), Saturating(usize::MAX)); + assert_eq!(v.iter().copied().product::>(), Saturating(usize::MAX)); + + let mut cnt = 0; + let v = 250..=u8::MAX; + assert_eq!( + v.map(|i| { + cnt += 1; + Saturating(i) + }) + .sum::>(), + Saturating(u8::MAX) + ); + assert_eq!(cnt, 6); // no short-circuiting + + let mut cnt = 0; + let v = (250..=u8::MAX).chain(0..5); + assert_eq!( + v.map(|i| { + cnt += 1; + Saturating(i) + }) + .product::>(), + Saturating(0) + ); + assert_eq!(cnt, 11); // no short-circuiting +} -- cgit 1.4.1-3-g733a5