perf(serde_v8): fast path for large strings (#14450)

serde_v8/de.rs

@@ -211,7 +211,7 @@ impl<'de, 'a, 'b, 's, 'x> de::Deserializer<'de>
   {
     if self.input.is_string() {
       let v8_string = v8::Local::<v8::String>::try_from(self.input).unwrap();
-      let string = v8_string.to_rust_string_lossy(self.scope);
+      let string = to_utf8(v8_string, self.scope);
       visitor.visit_string(string)
     } else {
       Err(Error::ExpectedString)
@@ -661,3 +661,66 @@ fn bigint_to_f64(b: v8::Local<v8::BigInt>) -> f64 {
     .sum();
   sign * x
 }
+
+pub fn to_utf8(
+  s: v8::Local<v8::String>,
+  scope: &mut v8::HandleScope,
+) -> String {
+  to_utf8_fast(s, scope).unwrap_or_else(|| to_utf8_slow(s, scope))
+}
+
+fn to_utf8_fast(
+  s: v8::Local<v8::String>,
+  scope: &mut v8::HandleScope,
+) -> Option<String> {
+  // Over-allocate by 20% to avoid checking string twice
+  let len = s.length();
+  let capacity = (len as f64 * 1.2) as usize;
+  let mut buf = Vec::with_capacity(capacity);
+  let mut nchars = 0;
+  let data = buf.as_mut_ptr();
+  let length = s.write_utf8(
+    scope,
+    // SAFETY: we're essentially providing the raw internal slice/buffer owned by the Vec
+    // which fulfills all of from_raw_parts_mut's safety requirements besides "initialization"
+    // and since we're operating on a [u8] not [T] we can safely assume the slice's values
+    // are sufficiently "initialized" for writes
+    unsafe { std::slice::from_raw_parts_mut(data, capacity) },
+    Some(&mut nchars),
+    v8::WriteOptions::NO_NULL_TERMINATION
+      | v8::WriteOptions::REPLACE_INVALID_UTF8,
+  );
+  if nchars < len {
+    return None;
+  }
+  // SAFETY: write_utf8 guarantees `length` bytes are initialized & valid utf8
+  unsafe {
+    buf.set_len(length);
+    Some(String::from_utf8_unchecked(buf))
+  }
+}
+
+fn to_utf8_slow(
+  s: v8::Local<v8::String>,
+  scope: &mut v8::HandleScope,
+) -> String {
+  let capacity = s.utf8_length(scope);
+  let mut buf = Vec::with_capacity(capacity);
+  let data = buf.as_mut_ptr();
+  let length = s.write_utf8(
+    scope,
+    // SAFETY: we're essentially providing the raw internal slice/buffer owned by the Vec
+    // which fulfills all of from_raw_parts_mut's safety requirements besides "initialization"
+    // and since we're operating on a [u8] not [T] we can safely assume the slice's values
+    // are sufficiently "initialized" for writes
+    unsafe { std::slice::from_raw_parts_mut(data, capacity) },
+    None,
+    v8::WriteOptions::NO_NULL_TERMINATION
+      | v8::WriteOptions::REPLACE_INVALID_UTF8,
+  );
+  // SAFETY: write_utf8 guarantees `length` bytes are initialized & valid utf8
+  unsafe {
+    buf.set_len(length);
+    String::from_utf8_unchecked(buf)
+  }
+}