refactor(allocator): String type (#8568)

Wrap `bumpalo::collections::String` in a new type instead of exporting it directly. This opens the door to:

1. Replacing it with our own `String` type which wraps our `Vec` type, rather than having 2 different implementations of `Vec` (`String` is just a wrapper around `Vec`, but a *different* `Vec` implementation).
2. Adding additional methods to `String` (`String::from_utf8` added in this PR).

crates/oxc_allocator/src/convert.rs

@@ -49,7 +49,7 @@ impl<'a> FromIn<'a, String> for crate::String<'a> {
 impl<'a> FromIn<'a, String> for &'a str {
     #[inline(always)]
     fn from_in(value: String, allocator: &'a Allocator) -> Self {
-        crate::String::from_str_in(value.as_str(), allocator).into_bump_str()
+        allocator.alloc_str(value.as_str())
     }
 }
 

crates/oxc_allocator/src/lib.rs

@@ -44,7 +44,6 @@ use std::{
     ops::{Deref, DerefMut},
 };
 
-pub use bumpalo::collections::String;
 use bumpalo::Bump;
 
 mod address;
@@ -53,6 +52,7 @@ mod boxed;
 mod clone_in;
 mod convert;
 pub mod hash_map;
+pub mod string;
 mod vec;
 
 pub use address::{Address, GetAddress};
@@ -60,6 +60,7 @@ pub use boxed::Box;
 pub use clone_in::CloneIn;
 pub use convert::{FromIn, IntoIn};
 pub use hash_map::HashMap;
+pub use string::String;
 pub use vec::Vec;
 
 /// A bump-allocated memory arena based on [bumpalo].

crates/oxc_allocator/src/string.rs

@@ -0,0 +1,249 @@
+//! Arena String.
+//!
+//! See [`String`] for more details.
+
+// All methods which just delegate to `bumpalo::collections::String` methods marked `#[inline(always)]`
+#![expect(clippy::inline_always)]
+
+use std::{
+    fmt::{self, Debug, Display},
+    hash::{Hash, Hasher},
+    mem::ManuallyDrop,
+    ops::{Deref, DerefMut},
+};
+
+use bumpalo::collections::String as BumpaloString;
+use simdutf8::basic::from_utf8;
+pub use simdutf8::basic::Utf8Error;
+
+use crate::{Allocator, Vec};
+
+/// Arena String.
+///
+/// UTF-8 encoded, growable string. Identical to [`std::string::String`] except that it stores
+/// string contents in arena allocator.
+#[derive(PartialOrd, Eq, Ord)]
+pub struct String<'alloc>(BumpaloString<'alloc>);
+
+impl<'alloc> String<'alloc> {
+    /// Creates a new empty [`String`].
+    ///
+    /// Given that the `String` is empty, this will not allocate any initial
+    /// buffer. While that means that this initial operation is very
+    /// inexpensive, it may cause excessive allocation later when you add
+    /// data. If you have an idea of how much data the `String` will hold,
+    /// consider the [`with_capacity_in`] method to prevent excessive
+    /// re-allocation.
+    ///
+    /// [`with_capacity_in`]: String::with_capacity_in
+    #[inline(always)]
+    pub fn new_in(allocator: &'alloc Allocator) -> String<'alloc> {
+        Self(BumpaloString::new_in(allocator))
+    }
+
+    /// Creates a new empty [`String`] with specified capacity.
+    ///
+    /// `String`s have an internal buffer to hold their data. The capacity is
+    /// the length of that buffer, and can be queried with the `capacity`
+    /// method. This method creates an empty `String`, but one with an initial
+    /// buffer that can hold `capacity` bytes. This is useful when you may be
+    /// appending a bunch of data to the `String`, reducing the number of
+    /// reallocations it needs to do.
+    ///
+    /// If the given capacity is `0`, no allocation will occur, and this method
+    /// is identical to the [`new_in`] method.
+    ///
+    /// [`capacity`]: String::capacity
+    /// [`new_in`]: String::new_in
+    #[inline(always)]
+    pub fn with_capacity_in(capacity: usize, allocator: &'alloc Allocator) -> String<'alloc> {
+        Self(BumpaloString::with_capacity_in(capacity, allocator))
+    }
+
+    /// Construct a new [`String`] from a string slice.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use oxc_allocator::{Allocator, String};
+    ///
+    /// let allocator = Allocator::default();
+    ///
+    /// let s = String::from_str_in("hello", &allocator);
+    /// assert_eq!(s, "hello");
+    /// ```
+    #[inline(always)]
+    pub fn from_str_in(s: &str, allocator: &'alloc Allocator) -> String<'alloc> {
+        Self(BumpaloString::from_str_in(s, allocator))
+    }
+
+    /// Convert `Vec<u8>` into [`String`].
+    ///
+    /// # Errors
+    /// Returns [`Err`] if the `Vec` does not comprise a valid UTF-8 string.
+    pub fn from_utf8(bytes: Vec<'alloc, u8>) -> Result<String<'alloc>, Utf8Error> {
+        // Check vec comprises a valid UTF-8 string.
+        from_utf8(&bytes)?;
+        // SAFETY: We just checked it's a valid UTF-8 string
+        let s = unsafe { Self::from_utf8_unchecked(bytes) };
+        Ok(s)
+    }
+
+    /// Convert `Vec<u8>` into [`String`], without checking bytes comprise a valid UTF-8 string.
+    ///
+    /// Does not copy the contents of the `Vec`, converts in place. This is a zero-cost operation.
+    ///
+    /// # SAFETY
+    /// Caller must ensure this `Vec<u8>` comprises a valid UTF-8 string.
+    //
+    // `#[inline(always)]` because this is a no-op at runtime
+    #[expect(clippy::missing_safety_doc, clippy::unnecessary_safety_comment)]
+    #[inline(always)]
+    pub unsafe fn from_utf8_unchecked(bytes: Vec<'alloc, u8>) -> String<'alloc> {
+        // Cannot use `bumpalo::String::from_utf8_unchecked` because it takes a `bumpalo::collections::Vec`,
+        // and our inner `Vec` type is `allocator_api2::vec::Vec`.
+        // SAFETY: Conversion is safe because both types store data in arena in same way.
+        // Lifetime of returned `String` is same as lifetime of original `Vec<u8>`.
+        let inner = ManuallyDrop::into_inner(bytes.0);
+        let (ptr, len, capacity, bump) = inner.into_raw_parts_with_alloc();
+        Self(BumpaloString::from_raw_parts_in(ptr, len, capacity, bump))
+    }
+
+    /// Creates a new [`String`] from a length, capacity, and pointer.
+    ///
+    /// # SAFETY
+    ///
+    /// This is highly unsafe, due to the number of invariants that aren't checked:
+    ///
+    /// * The memory at `ptr` needs to have been previously allocated by the same [`Allocator`].
+    /// * `length` needs to be less than or equal to `capacity`.
+    /// * `capacity` needs to be the correct value.
+    ///
+    /// Violating these may cause problems like corrupting the allocator's internal data structures.
+    ///
+    /// The ownership of `ptr` is effectively transferred to the `String` which may then deallocate,
+    /// reallocate or change the contents of memory pointed to by the pointer at will. Ensure that
+    /// nothing else uses the pointer after calling this function.
+    ///
+    /// # Examples
+    ///
+    /// Basic usage:
+    ///
+    /// ```
+    /// use std::mem;
+    /// use oxc_allocator::{Allocator, String};
+    ///
+    /// let allocator = Allocator::default();
+    ///
+    /// unsafe {
+    ///     let mut s = String::from_str_in("hello", &allocator);
+    ///     let ptr = s.as_mut_ptr();
+    ///     let len = s.len();
+    ///     let capacity = s.capacity();
+    ///
+    ///     mem::forget(s);
+    ///
+    ///     let s = String::from_raw_parts_in(ptr, len, capacity, &allocator);
+    ///
+    ///     assert_eq!(s, "hello");
+    /// }
+    /// ```
+    #[expect(clippy::missing_safety_doc, clippy::unnecessary_safety_comment)]
+    #[inline(always)]
+    pub unsafe fn from_raw_parts_in(
+        buf: *mut u8,
+        length: usize,
+        capacity: usize,
+        allocator: &'alloc Allocator,
+    ) -> String<'alloc> {
+        // SAFETY: Safety conditions of this method are the same as `BumpaloString`'s method
+        Self(BumpaloString::from_raw_parts_in(buf, length, capacity, allocator))
+    }
+
+    /// Convert this `String<'alloc>` into an `&'alloc str`. This is analogous to
+    /// [`std::string::String::into_boxed_str`].
+    ///
+    /// # Example
+    ///
+    /// ```
+    /// use oxc_allocator::{Allocator, String};
+    ///
+    /// let allocator = Allocator::default();
+    ///
+    /// let s = String::from_str_in("foo", &allocator);
+    /// assert_eq!(s.into_bump_str(), "foo");
+    /// ```
+    #[inline(always)]
+    pub fn into_bump_str(self) -> &'alloc str {
+        self.0.into_bump_str()
+    }
+}
+
+// Provide access to all `bumpalo::String`'s methods via deref
+impl<'alloc> Deref for String<'alloc> {
+    type Target = BumpaloString<'alloc>;
+
+    #[inline]
+    fn deref(&self) -> &Self::Target {
+        &self.0
+    }
+}
+
+impl<'alloc> DerefMut for String<'alloc> {
+    #[inline]
+    fn deref_mut(&mut self) -> &mut BumpaloString<'alloc> {
+        &mut self.0
+    }
+}
+
+impl PartialEq for String<'_> {
+    #[inline]
+    fn eq(&self, other: &String) -> bool {
+        PartialEq::eq(&self[..], &other[..])
+    }
+}
+
+// `impl_eq!` macro copied from `bumpalo`
+macro_rules! impl_eq {
+    ($lhs:ty, $rhs: ty) => {
+        impl<'a, 'alloc> PartialEq<$rhs> for $lhs {
+            #[inline]
+            fn eq(&self, other: &$rhs) -> bool {
+                PartialEq::eq(&self[..], &other[..])
+            }
+        }
+
+        impl<'a, 'alloc> PartialEq<$lhs> for $rhs {
+            #[inline]
+            fn eq(&self, other: &$lhs) -> bool {
+                PartialEq::eq(&self[..], &other[..])
+            }
+        }
+    };
+}
+
+impl_eq! { String<'alloc>, str }
+impl_eq! { String<'alloc>, &'a str }
+impl_eq! { std::borrow::Cow<'a, str>, String<'alloc> }
+impl_eq! { std::string::String, String<'alloc> }
+
+impl Display for String<'_> {
+    #[inline]
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        Display::fmt(self.as_str(), f)
+    }
+}
+
+impl Debug for String<'_> {
+    #[inline]
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        Debug::fmt(self.as_str(), f)
+    }
+}
+
+impl Hash for String<'_> {
+    #[inline]
+    fn hash<H: Hasher>(&self, hasher: &mut H) {
+        self.as_str().hash(hasher);
+    }
+}

crates/oxc_allocator/src/vec.rs

@@ -19,7 +19,7 @@ use allocator_api2::vec::Vec as InnerVec;
 use bumpalo::Bump;
 #[cfg(any(feature = "serialize", test))]
 use serde::{ser::SerializeSeq, Serialize, Serializer};
-use simdutf8::basic::{from_utf8, Utf8Error};
+use simdutf8::basic::Utf8Error;
 
 use crate::{Allocator, Box, String};
 
@@ -32,7 +32,7 @@ use crate::{Allocator, Box, String};
 /// Note: This is not a soundness issue, as Rust does not support relying on `drop`
 /// being called to guarantee soundness.
 #[derive(PartialEq, Eq)]
-pub struct Vec<'alloc, T>(ManuallyDrop<InnerVec<T, &'alloc Bump>>);
+pub struct Vec<'alloc, T>(pub(crate) ManuallyDrop<InnerVec<T, &'alloc Bump>>);
 
 /// SAFETY: Not actually safe, but for enabling `Send` for downstream crates.
 unsafe impl<T> Send for Vec<'_, T> {}
@@ -190,16 +190,12 @@ impl<'alloc, T> Vec<'alloc, T> {
 }
 
 impl<'alloc> Vec<'alloc, u8> {
-    /// Convert `Vec<u8>` into `String`.
+    /// Convert `Vec<u8>` into [`String`].
     ///
     /// # Errors
     /// Returns [`Err`] if the `Vec` does not comprise a valid UTF-8 string.
     pub fn into_string(self) -> Result<String<'alloc>, Utf8Error> {
-        // Check vec comprises a valid UTF-8 string.
-        from_utf8(&self.0)?;
-        // SAFETY: We just checked it's a valid UTF-8 string
-        let s = unsafe { self.into_string_unchecked() };
-        Ok(s)
+        String::from_utf8(self)
     }
 
     /// Convert `Vec<u8>` into [`String`], without checking bytes comprise a valid UTF-8 string.
@@ -211,13 +207,8 @@ impl<'alloc> Vec<'alloc, u8> {
     #[expect(clippy::missing_safety_doc, clippy::unnecessary_safety_comment)]
     #[inline(always)] // `#[inline(always)]` because this is a no-op at runtime
     pub unsafe fn into_string_unchecked(self) -> String<'alloc> {
-        // Cannot use `bumpalo::String::from_utf8_unchecked` because it takes a `bumpalo::collections::Vec`,
-        // and our inner `Vec` type is `allocator_api2::vec::Vec`.
-        // SAFETY: Conversion is safe because both types store data in arena in same way.
-        // Lifetime of returned `String` is same as lifetime of original `Vec<u8>`.
-        let inner = ManuallyDrop::into_inner(self.0);
-        let (ptr, len, cap, bump) = inner.into_raw_parts_with_alloc();
-        String::from_raw_parts_in(ptr, len, cap, bump)
+        // SAFETY: Caller guarantees vec comprises a valid UTF-8 string.
+        String::from_utf8_unchecked(self)
     }
 }
 

crates/oxc_ast/src/ast_builder_impl.rs

@@ -8,7 +8,7 @@
 
 use std::{borrow::Cow, mem};
 
-use oxc_allocator::{Allocator, Box, FromIn, String, Vec};
+use oxc_allocator::{Allocator, Box, FromIn, Vec};
 use oxc_span::{Atom, Span, SPAN};
 use oxc_syntax::{number::NumberBase, operator::UnaryOperator, scope::ScopeId};
 
@@ -78,7 +78,7 @@ impl<'a> AstBuilder<'a> {
     /// in the heap.
     #[inline]
     pub fn str(self, value: &str) -> &'a str {
-        String::from_str_in(value, self.allocator).into_bump_str()
+        self.allocator.alloc_str(value)
     }
 
     /// Allocate an [`Atom`] from a string slice.

crates/oxc_prettier/src/macros.rs

@@ -47,7 +47,7 @@ macro_rules! text {
 #[macro_export]
 macro_rules! dynamic_text {
     ($p:ident, $str:expr) => {{
-        let s = oxc_allocator::String::from_str_in($str, $p.allocator).into_bump_str();
+        let s = $p.allocator.alloc_str($str);
         $crate::ir::Doc::Str(s)
     }};
 }

crates/oxc_span/src/atom.rs

@@ -79,7 +79,7 @@ impl<'alloc> FromIn<'alloc, &Atom<'alloc>> for Atom<'alloc> {
 
 impl<'alloc> FromIn<'alloc, &str> for Atom<'alloc> {
     fn from_in(s: &str, allocator: &'alloc Allocator) -> Self {
-        Self::from(oxc_allocator::String::from_str_in(s, allocator))
+        Self::from(&*allocator.alloc_str(s))
     }
 }