From ffc28f9157acba746b5b2eac7f8b8f535f6a44e6 Mon Sep 17 00:00:00 2001
From: overlookmotel <theoverlookmotel@gmail.com>
Date: Wed, 18 Dec 2024 13:48:38 +0000
Subject: [PATCH] feat(allocator): `Vec<u8>::into_string`

---
 Cargo.lock                      |  1 +
 crates/oxc_allocator/Cargo.toml |  1 +
 crates/oxc_allocator/src/vec.rs | 35 ++++++++++++++++++++++++++++++++-
 3 files changed, 36 insertions(+), 1 deletion(-)

diff --git a/Cargo.lock b/Cargo.lock
index 622b708445953..f9707090841d0 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1452,6 +1452,7 @@ dependencies = [
  "bumpalo",
  "serde",
  "serde_json",
+ "simdutf8",
 ]
 
 [[package]]
diff --git a/crates/oxc_allocator/Cargo.toml b/crates/oxc_allocator/Cargo.toml
index ff524d718d3fb..734ba63d3ba49 100644
--- a/crates/oxc_allocator/Cargo.toml
+++ b/crates/oxc_allocator/Cargo.toml
@@ -21,6 +21,7 @@ doctest = false
 [dependencies]
 allocator-api2 = { workspace = true }
 bumpalo = { workspace = true, features = ["allocator-api2", "collections"] }
+simdutf8 = { workspace = true }
 
 serde = { workspace = true, optional = true }
 
diff --git a/crates/oxc_allocator/src/vec.rs b/crates/oxc_allocator/src/vec.rs
index 2feea6d088a2c..983b38e281b99 100644
--- a/crates/oxc_allocator/src/vec.rs
+++ b/crates/oxc_allocator/src/vec.rs
@@ -15,8 +15,9 @@ use allocator_api2::vec;
 use bumpalo::Bump;
 #[cfg(any(feature = "serialize", test))]
 use serde::{ser::SerializeSeq, Serialize, Serializer};
+use simdutf8::basic::{from_utf8, Utf8Error};
 
-use crate::{Allocator, Box};
+use crate::{Allocator, Box, String};
 
 /// A `Vec` without [`Drop`], which stores its data in the arena allocator.
 ///
@@ -180,6 +181,38 @@ impl<'alloc, T> Vec<'alloc, T> {
     }
 }
 
+impl<'alloc> Vec<'alloc, u8> {
+    /// 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)
+    }
+
+    /// 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.
+    #[expect(clippy::missing_safety_doc, clippy::unnecessary_safety_comment)]
+    #[inline] // `#[inline]` 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)
+    }
+}
+
 impl<'alloc, T> ops::Deref for Vec<'alloc, T> {
     type Target = vec::Vec<T, &'alloc Bump>;