RUST-1798 use simd to optimize utf8 validation

Cargo.toml

@@ -28,7 +28,7 @@ exclude = [
     "rustfmt.toml",
     ".travis.yml",
     ".evergreen/**",
-    ".gitignore"
+    ".gitignore",
 ]
 
 [features]
@@ -54,20 +54,31 @@ name = "bson"
 
 [dependencies]
 ahash = "0.8.0"
-chrono = { version = "0.4.15", features = ["std"], default-features = false, optional = true }
+chrono = { version = "0.4.15", features = [
+    "std",
+], default-features = false, optional = true }
 rand = "0.8"
 serde = { version = "1.0", features = ["derive"] }
 serde_json = { version = "1.0", features = ["preserve_order"] }
+simdutf8 = "0.1.4"
 indexmap = "1.6.2"
 hex = "0.4.2"
 base64 = "0.13.0"
 once_cell = "1.5.1"
-uuid-0_8 = { package = "uuid", version = "0.8.1", features = ["serde", "v4"], optional = true }
+uuid-0_8 = { package = "uuid", version = "0.8.1", features = [
+    "serde",
+    "v4",
+], optional = true }
 uuid = { version = "1.1.2", features = ["serde", "v4"] }
 serde_bytes = "0.11.5"
 serde_with = { version = "1.3.1", optional = true }
 serde_with-3 = { package = "serde_with", version = "3.1.0", optional = true }
-time = { version = "0.3.9", features = ["formatting", "parsing", "macros", "large-dates"] }
+time = { version = "0.3.9", features = [
+    "formatting",
+    "parsing",
+    "macros",
+    "large-dates",
+] }
 bitvec = "1.0.1"
 
 [target.'cfg(target_arch = "wasm32")'.dependencies]
@@ -78,7 +89,11 @@ criterion = "0.3.0"
 pretty_assertions = "0.6.1"
 proptest = "1.0.0"
 serde_bytes = "0.11"
-chrono = { version = "0.4", features = ["serde", "clock", "std"], default-features = false }
+chrono = { version = "0.4", features = [
+    "serde",
+    "clock",
+    "std",
+], default-features = false }
 
 [package.metadata.docs.rs]
 all-features = true

src/de/error.rs

@@ -1,6 +1,7 @@
-use std::{error, fmt, fmt::Display, io, string, sync::Arc};
+use std::{error, fmt, fmt::Display, io, sync::Arc};
 
 use serde::de::{self, Unexpected};
+use simdutf8::basic::Utf8Error;
 
 use crate::Bson;
 
@@ -13,7 +14,7 @@ pub enum Error {
 
     /// A [`std::string::FromUtf8Error`](https://doc.rust-lang.org/std/string/struct.FromUtf8Error.html) encountered
     /// while decoding a UTF-8 String from the input data.
-    InvalidUtf8String(string::FromUtf8Error),
+    InvalidUtf8String(Utf8Error),
 
     /// While decoding a [`Document`](crate::Document) from bytes, an unexpected or unsupported
     /// element type was encountered.
@@ -44,8 +45,8 @@ impl From<io::Error> for Error {
     }
 }
 
-impl From<string::FromUtf8Error> for Error {
-    fn from(err: string::FromUtf8Error) -> Error {
+impl From<Utf8Error> for Error {
+    fn from(err: Utf8Error) -> Error {
         Error::InvalidUtf8String(err)
     }
 }

src/de/mod.rs

@@ -42,7 +42,7 @@ use crate::{
     Decimal128,
 };
 
-use ::serde::{
+use serde::{
     de::{DeserializeOwned, Error as _, Unexpected},
     Deserialize,
 };
@@ -108,14 +108,12 @@ pub(crate) fn read_string<R: Read + ?Sized>(reader: &mut R, utf8_lossy: bool) ->
         ));
     }
 
+    let mut buf = Vec::with_capacity(len as usize - 1);
+    reader.take(len as u64 - 1).read_to_end(&mut buf)?;
     let s = if utf8_lossy {
-        let mut buf = Vec::with_capacity(len as usize - 1);
-        reader.take(len as u64 - 1).read_to_end(&mut buf)?;
         String::from_utf8_lossy(&buf).to_string()
     } else {
-        let mut s = String::with_capacity(len as usize - 1);
-        reader.take(len as u64 - 1).read_to_string(&mut s)?;
-        s
+        to_string(buf)?
     };
 
     // read the null terminator
@@ -152,7 +150,13 @@ fn read_cstring<R: Read + ?Sized>(reader: &mut R) -> Result<String> {
         v.push(c);
     }
 
-    Ok(String::from_utf8(v)?)
+    to_string(v)
+}
+
+fn to_string(v: Vec<u8>) -> Result<String> {
+    let _ = simdutf8::basic::from_utf8(&v)?;
+    // Safety: `v` is a valid UTF-8 string.
+    unsafe { Ok(String::from_utf8_unchecked(v)) }
 }
 
 #[inline]

src/de/raw.rs

@@ -1759,7 +1759,7 @@ impl<'a> BsonBuf<'a> {
         let s = if utf8_lossy_override.unwrap_or(self.utf8_lossy) {
             String::from_utf8_lossy(bytes)
         } else {
-            Cow::Borrowed(std::str::from_utf8(bytes).map_err(Error::custom)?)
+            Cow::Borrowed(simdutf8::basic::from_utf8(bytes).map_err(Error::custom)?)
         };
 
         // consume the null byte

src/raw/error.rs

@@ -1,4 +1,4 @@
-use std::str::Utf8Error;
+use simdutf8::basic::Utf8Error;
 
 use crate::spec::ElementType;
 

src/raw/mod.rs

@@ -255,7 +255,8 @@ fn read_lenencoded(buf: &[u8]) -> Result<&str> {
 }
 
 fn try_to_str(data: &[u8]) -> Result<&str> {
-    std::str::from_utf8(data).map_err(|e| Error::new_without_key(ErrorKind::Utf8EncodingError(e)))
+    simdutf8::basic::from_utf8(data)
+        .map_err(|e| Error::new_without_key(ErrorKind::Utf8EncodingError(e)))
 }
 
 fn usize_try_from_i32(i: i32) -> Result<usize> {