perf(encoding): UTF-8 encoder walks code units inline instead of for char in src

crypto/pkg.generated.mbti

@@ -74,10 +74,10 @@ pub impl ByteSource for Bytes
 pub impl ByteSource for BytesView
 
 pub(open) trait CryptoHasher {
-  size(Self) -> Int
-  block_size(Self) -> Int
-  reset(Self) -> Unit
-  update(Self, BytesView) -> Unit
-  finalize_into(Self, FixedArray[Byte], offset~ : Int) -> Unit
+  fn size(Self) -> Int
+  fn block_size(Self) -> Int
+  fn reset(Self) -> Unit
+  fn update(Self, BytesView) -> Unit
+  fn finalize_into(Self, FixedArray[Byte], offset~ : Int) -> Unit
 }
 

encoding/encoding.mbt

@@ -54,9 +54,28 @@ pub fn encode(encoding : Encoding, src : String) -> Bytes {
     UTF16BE => write_utf16be_char
     _ => abort("unreachable")
   }
-  for char in src {
-    // SAFETY: Assume String are always valid UTF16LE
-    write(new_buf, char)
+  // Walk by UTF-16 code unit and assemble surrogate pairs inline. This
+  // avoids `for char in src`, which goes through String::iter +
+  // Iter::next per char (~30% of self time on an ASCII-heavy 64 KiB
+  // bench even though the inner body is small).
+  let len = src.length()
+  let mut i = 0
+  while i < len {
+    let cu = src.unsafe_get(i).to_int()
+    i += 1
+    let cp = if cu >= 0xD800 && cu <= 0xDBFF && i < len {
+      let cu2 = src.unsafe_get(i).to_int()
+      if cu2 >= 0xDC00 && cu2 <= 0xDFFF {
+        i += 1
+        (cu - 0xD800) * 0x400 + (cu2 - 0xDC00) + 0x10000
+      } else {
+        cu
+      }
+    } else {
+      cu
+    }
+    // SAFETY: Assume String contains valid UTF-16
+    write(new_buf, cp.unsafe_to_char())
   }
   new_buf.to_bytes()
 }

num/pkg.generated.mbti

@@ -12,13 +12,13 @@ package "moonbitlang/x/num"
 // Traits
 #deprecated
 pub trait Num {
-  from_int(Int) -> Self
-  op_add(Self, Self) -> Self
-  op_sub(Self, Self) -> Self
-  op_mul(Self, Self) -> Self
-  op_div(Self, Self) -> Self
-  op_neg(Self) -> Self
-  abs(Self) -> Self
-  signum(Self) -> Self
+  fn from_int(Int) -> Self
+  fn op_add(Self, Self) -> Self
+  fn op_sub(Self, Self) -> Self
+  fn op_mul(Self, Self) -> Self
+  fn op_div(Self, Self) -> Self
+  fn op_neg(Self) -> Self
+  fn abs(Self) -> Self
+  fn signum(Self) -> Self
 }