perf(web): Optimize `TextDecoder` by adding a new `U16String` type (#13923)

5 files changed, 128 insertions, 0 deletions

diff --git a/serde_v8/src/lib.rs b/serde_v8/src/lib.rs
index 9d0715195..203e1d004 100644
--- a/serde_v8/src/lib.rs
+++ b/serde_v8/src/lib.rs
@@ -14,6 +14,7 @@ pub use keys::KeyCache;
 pub use magic::buffer::MagicBuffer as Buffer;
 pub use magic::bytestring::ByteString;
 pub use magic::string_or_buffer::StringOrBuffer;
+pub use magic::u16string::U16String;
 pub use magic::Value;
 pub use ser::{to_v8, Serializer};
 pub use serializable::{Serializable, SerializablePkg};
diff --git a/serde_v8/src/magic/mod.rs b/serde_v8/src/magic/mod.rs
index 941a69d41..e90b5ab60 100644
--- a/serde_v8/src/magic/mod.rs
+++ b/serde_v8/src/magic/mod.rs
@@ -3,6 +3,7 @@ pub mod buffer;
 pub mod bytestring;
 mod field;
 pub mod string_or_buffer;
+pub mod u16string;
 mod value;
 pub mod zero_copy_buf;
 
diff --git a/serde_v8/src/magic/u16string.rs b/serde_v8/src/magic/u16string.rs
new file mode 100644
index 000000000..a75af016b
--- /dev/null
+++ b/serde_v8/src/magic/u16string.rs
@@ -0,0 +1,62 @@
+use std::ops::{Deref, DerefMut};
+
+use serde::Serialize;
+
+pub const NAME: &str = "$__v8_magic_u16string";
+pub const FIELD_PTR: &str = "$__v8_magic_u16string_ptr";
+pub const FIELD_LEN: &str = "$__v8_magic_u16string_len";
+
+#[derive(Default, PartialEq, Eq)]
+pub struct U16String(pub Vec<u16>);
+
+impl U16String {
+  pub fn with_zeroes(length: usize) -> U16String {
+    U16String(vec![0u16; length])
+  }
+
+  pub fn truncate(&mut self, new_length: usize) {
+    self.0.truncate(new_length);
+    self.0.shrink_to_fit()
+  }
+}
+
+impl Deref for U16String {
+  type Target = [u16];
+  fn deref(&self) -> &[u16] {
+    self.0.deref()
+  }
+}
+
+impl DerefMut for U16String {
+  fn deref_mut(&mut self) -> &mut [u16] {
+    self.0.deref_mut()
+  }
+}
+
+impl AsRef<[u16]> for U16String {
+  fn as_ref(&self) -> &[u16] {
+    self.0.as_ref()
+  }
+}
+
+impl AsMut<[u16]> for U16String {
+  fn as_mut(&mut self) -> &mut [u16] {
+    self.0.as_mut()
+  }
+}
+
+impl Serialize for U16String {
+  fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
+  where
+    S: serde::Serializer,
+  {
+    use serde::ser::SerializeStruct;
+
+    let mut s = serializer.serialize_struct(NAME, 3)?;
+    s.serialize_field(FIELD_PTR, &(self.0.as_ptr() as usize))?;
+    s.serialize_field(FIELD_LEN, &self.0.len())?;
+    s.end()
+  }
+}
+
+// TODO: Deserialize
diff --git a/serde_v8/src/ser.rs b/serde_v8/src/ser.rs
index 8829b9fc3..ce7e8c707 100644
--- a/serde_v8/src/ser.rs
+++ b/serde_v8/src/ser.rs
@@ -339,11 +339,64 @@ impl<'a, 'b, 'c> ser::SerializeStruct
   }
 }
 
+pub struct MagicU16StringSerializer<'a, 'b, 'c> {
+  scope: ScopePtr<'a, 'b, 'c>,
+  ptr: Option<std::ptr::NonNull<u16>>,
+  len: Option<usize>,
+}
+
+impl<'a, 'b, 'c> MagicU16StringSerializer<'a, 'b, 'c> {
+  pub fn new(scope: ScopePtr<'a, 'b, 'c>) -> Self {
+    Self {
+      scope,
+      ptr: None,
+      len: None,
+    }
+  }
+}
+
+impl<'a, 'b, 'c> ser::SerializeStruct for MagicU16StringSerializer<'a, 'b, 'c> {
+  type Ok = JsValue<'a>;
+  type Error = Error;
+
+  fn serialize_field<T: ?Sized + Serialize>(
+    &mut self,
+    key: &'static str,
+    value: &T,
+  ) -> Result<()> {
+    // Get u64 chunk
+    let transmuted = value.serialize(magic::FieldSerializer {})?;
+    match key {
+      magic::u16string::FIELD_PTR => {
+        self.ptr = std::ptr::NonNull::new(transmuted as *mut u16)
+      }
+      magic::u16string::FIELD_LEN => self.len = Some(transmuted as usize),
+      _ => unreachable!(),
+    }
+    Ok(())
+  }
+
+  fn end(self) -> JsResult<'a> {
+    // SAFETY: This function is only called from U16String::serialize(), which
+    // guarantees the Vec is still alive.
+    let slice = unsafe {
+      std::slice::from_raw_parts(self.ptr.unwrap().as_ptr(), self.len.unwrap())
+    };
+    let scope = &mut *self.scope.borrow_mut();
+
+    let v8_value =
+      v8::String::new_from_two_byte(scope, slice, v8::NewStringType::Normal)
+        .unwrap();
+    Ok(v8_value.into())
+  }
+}
+
 // Dispatches between magic and regular struct serializers
 pub enum StructSerializers<'a, 'b, 'c> {
   Magic(MagicSerializer<'a>),
   MagicBuffer(MagicBufferSerializer<'a, 'b, 'c>),
   MagicByteString(MagicByteStringSerializer<'a, 'b, 'c>),
+  MagicU16String(MagicU16StringSerializer<'a, 'b, 'c>),
   Regular(ObjectSerializer<'a, 'b, 'c>),
 }
 
@@ -360,6 +413,7 @@ impl<'a, 'b, 'c> ser::SerializeStruct for StructSerializers<'a, 'b, 'c> {
       StructSerializers::Magic(s) => s.serialize_field(key, value),
       StructSerializers::MagicBuffer(s) => s.serialize_field(key, value),
       StructSerializers::MagicByteString(s) => s.serialize_field(key, value),
+      StructSerializers::MagicU16String(s) => s.serialize_field(key, value),
       StructSerializers::Regular(s) => s.serialize_field(key, value),
     }
   }
@@ -369,6 +423,7 @@ impl<'a, 'b, 'c> ser::SerializeStruct for StructSerializers<'a, 'b, 'c> {
       StructSerializers::Magic(s) => s.end(),
       StructSerializers::MagicBuffer(s) => s.end(),
       StructSerializers::MagicByteString(s) => s.end(),
+      StructSerializers::MagicU16String(s) => s.end(),
       StructSerializers::Regular(s) => s.end(),
     }
   }
@@ -607,6 +662,10 @@ impl<'a, 'b, 'c> ser::Serializer for Serializer<'a, 'b, 'c> {
         let m = MagicByteStringSerializer::new(self.scope);
         Ok(StructSerializers::MagicByteString(m))
       }
+      magic::u16string::NAME => {
+        let m = MagicU16StringSerializer::new(self.scope);
+        Ok(StructSerializers::MagicU16String(m))
+      }
       _ => {
         let o = ObjectSerializer::new(self.scope, len);
         Ok(StructSerializers::Regular(o))
diff --git a/serde_v8/src/serializable.rs b/serde_v8/src/serializable.rs
index df6912db0..533b3f83f 100644
--- a/serde_v8/src/serializable.rs
+++ b/serde_v8/src/serializable.rs
@@ -4,6 +4,7 @@ use std::mem::transmute_copy;
 
 use crate::Buffer;
 use crate::ByteString;
+use crate::U16String;
 
 /// Serializable exists to allow boxing values as "objects" to be serialized later,
 /// this is particularly useful for async op-responses. This trait is a more efficient
@@ -63,6 +64,7 @@ pub enum Primitive {
   String(String),
   Buffer(Buffer),
   ByteString(ByteString),
+  U16String(U16String),
 }
 
 impl serde::Serialize for Primitive {
@@ -86,6 +88,7 @@ impl serde::Serialize for Primitive {
       Self::String(x) => x.serialize(s),
       Self::Buffer(x) => x.serialize(s),
       Self::ByteString(x) => x.serialize(s),
+      Self::U16String(x) => x.serialize(s),
     }
   }
 }
@@ -130,6 +133,8 @@ impl<T: serde::Serialize + 'static> From<T> for SerializablePkg {
       Self::Primitive(Primitive::Buffer(tc(x)))
     } else if tid == TypeId::of::<ByteString>() {
       Self::Primitive(Primitive::ByteString(tc(x)))
+    } else if tid == TypeId::of::<U16String>() {
+      Self::Primitive(Primitive::U16String(tc(x)))
     } else {
       Self::Serializable(Box::new(x))
     }