RUST-1798 use simd to optimize utf8 validation (#548)

Co-authored-by: Liyixin95 <[email protected]>

Author: abr-egn

Cargo.lock

@@ -65,6 +65,7 @@ serde_with = { version = "3.1.0", optional = true }
 time = { version = "0.3.9", features = ["formatting", "parsing", "macros", "large-dates"] }
 bitvec = "1.0.1"
 serde_path_to_error = { version = "0.1.16", optional = true }
+simdutf8 = "0.1.5"
 
 [target.'cfg(all(target_arch = "wasm32", target_os = "unknown"))'.dependencies]
 js-sys = "0.3"

Cargo.toml

@@ -1,5 +1,3 @@
-use std::str::Utf8Error;
-
 use crate::spec::ElementType;
 
 /// An error that occurs when attempting to parse raw BSON bytes.
@@ -44,7 +42,7 @@ pub enum ErrorKind {
     MalformedValue { message: String },
 
     /// Improper UTF-8 bytes were found when proper UTF-8 was expected.
-    Utf8EncodingError(Utf8Error),
+    Utf8EncodingError,
 }
 
 impl std::fmt::Display for Error {
@@ -60,7 +58,7 @@ impl std::fmt::Display for Error {
             ErrorKind::MalformedValue { message } => {
                 write!(f, "{}malformed value: {:?}", prefix, message)
             }
-            ErrorKind::Utf8EncodingError(e) => write!(f, "{}utf-8 encoding error: {}", prefix, e),
+            ErrorKind::Utf8EncodingError => write!(f, "{}utf-8 encoding error", prefix),
         }
     }
 }

src/raw/error.rs

@@ -271,7 +271,7 @@ fn read_lenencode(buf: &[u8]) -> Result<&str> {
 }
 
 fn try_to_str(data: &[u8]) -> Result<&str> {
-    std::str::from_utf8(data).map_err(|e| Error::new(ErrorKind::Utf8EncodingError(e)))
+    simdutf8::basic::from_utf8(data).map_err(|_| Error::new(ErrorKind::Utf8EncodingError))
 }
 
 fn usize_try_from_i32(i: i32) -> Result<usize> {