Add optional simd utf8 validation

ntex-bytes/CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changes
 
+## [0.1.9] (2022-01-10)
+
+* Add optional simd utf8 validation
+
 ## [0.1.8] (2021-12-18)
 
 * Remove futures patch dependency

ntex-bytes/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "ntex-bytes"
-version = "0.1.8"
+version = "0.1.9"
 license = "MIT"
 authors = ["Nikolay Kim <fafhrd91@gmail.com>", "Carl Lerche <me@carllerche.com>"]
 description = "Types and traits for working with bytes (bytes crate fork)"
@@ -11,13 +11,20 @@ keywords = ["buffers", "zero-copy", "io"]
 categories = ["network-programming", "data-structures"]
 edition = "2018"
 
+[features]
+default = []
+
+# simd utf8 check support
+simd = ["simdutf8"]
+
 [dependencies]
 bitflags = "1.3"
 bytes = "1.0.0"
 serde = "1.0.0"
 futures-core = { version = "0.3", default-features = false, features = ["alloc"] }
+simdutf8 = { version = "0.1.3", optional = true }
 
 [dev-dependencies]
 serde_test = "1.0"
 serde_json = "1.0"
-ntex = "0.5.0-b.0"
+ntex = "0.5.0"

ntex-bytes/src/string.rs

@@ -179,7 +179,7 @@ impl ops::Deref for ByteString {
     fn deref(&self) -> &str {
         let bytes = self.0.as_ref();
         // SAFETY:
-        // UTF-8 validity is guaranteed at during construction.
+        // UTF-8 validity is guaranteed during construction.
         unsafe { str::from_utf8_unchecked(bytes) }
     }
 }
@@ -213,42 +213,54 @@ impl<'a> From<borrow::Cow<'a, str>> for ByteString {
 }
 
 impl TryFrom<&[u8]> for ByteString {
-    type Error = str::Utf8Error;
+    type Error = ();
 
     #[inline]
     fn try_from(value: &[u8]) -> Result<Self, Self::Error> {
-        let _ = str::from_utf8(value)?;
+        if utf8::is_valid(&value) {
-        Ok(ByteString(Bytes::copy_from_slice(value)))
+            Ok(ByteString(Bytes::copy_from_slice(value)))
+        } else {
+            Err(())
+        }
     }
 }
 
 impl TryFrom<Vec<u8>> for ByteString {
-    type Error = str::Utf8Error;
+    type Error = ();
 
     #[inline]
     fn try_from(value: Vec<u8>) -> Result<Self, Self::Error> {
-        let buf = String::from_utf8(value).map_err(|err| err.utf8_error())?;
+        if utf8::is_valid(&value) {
-        Ok(ByteString(Bytes::from(buf)))
+            Ok(ByteString(Bytes::from(value)))
+        } else {
+            Err(())
+        }
     }
 }
 
 impl TryFrom<Bytes> for ByteString {
-    type Error = str::Utf8Error;
+    type Error = ();
 
     #[inline]
     fn try_from(value: Bytes) -> Result<Self, Self::Error> {
-        let _ = str::from_utf8(value.as_ref())?;
+        if utf8::is_valid(&value) {
-        Ok(ByteString(value))
+            Ok(ByteString(value))
+        } else {
+            Err(())
+        }
     }
 }
 
 impl TryFrom<BytesMut> for ByteString {
-    type Error = str::Utf8Error;
+    type Error = ();
 
     #[inline]
     fn try_from(value: crate::BytesMut) -> Result<Self, Self::Error> {
-        let _ = str::from_utf8(&value)?;
+        if utf8::is_valid(&value) {
-        Ok(ByteString(value.freeze()))
+            Ok(ByteString(value.freeze()))
+        } else {
+            Err(())
+        }
     }
 }
 
@@ -291,6 +303,20 @@ mod serde {
     }
 }
 
+#[cfg(feature = "simd")]
+mod utf8 {
+    pub(super) fn is_valid(input: &[u8]) -> bool {
+        simdutf8::basic::from_utf8(input).is_ok()
+    }
+}
+
+#[cfg(not(feature = "simd"))]
+mod utf8 {
+    pub(super) fn is_valid(input: &[u8]) -> bool {
+        std::str::from_utf8(input).is_ok()
+    }
+}
+
 #[cfg(test)]
 mod test {
     use std::borrow::ToOwned;

ntex-codec/Cargo.toml

@@ -16,4 +16,4 @@ name = "ntex_codec"
 path = "src/lib.rs"
 
 [dependencies]
-ntex-bytes = "0.1"
+ntex-bytes = "0.1.9"

ntex-io/Cargo.toml

@@ -17,7 +17,7 @@ path = "src/lib.rs"
 
 [dependencies]
 ntex-codec = "0.6.0"
-ntex-bytes = "0.1.8"
+ntex-bytes = "0.1.9"
 ntex-util = "0.1.8"
 ntex-service = "0.3.1"
 

ntex-router/Cargo.toml

@@ -18,7 +18,7 @@ default = ["http"]
 
 [dependencies]
 serde = "1.0"
-ntex-bytes = "0.1"
+ntex-bytes = "0.1.9"
 log = "0.4"
 http = { version = "0.2", optional = true }
 regex = { version = "1.5.4", default-features = false, features = ["std"] }

ntex-tls/Cargo.toml

@@ -25,7 +25,7 @@ openssl = ["tls_openssl"]
 rustls = ["tls_rust"]
 
 [dependencies]
-ntex-bytes = "0.1.8"
+ntex-bytes = "0.1.9"
 ntex-io = "0.1.2"
 ntex-util = "0.1.8"
 ntex-service = "0.3.1"

ntex-tokio/Cargo.toml

@@ -16,9 +16,9 @@ name = "ntex_tokio"
 path = "src/lib.rs"
 
 [dependencies]
-ntex-bytes = "0.1.8"
+ntex-bytes = "0.1.9"
-ntex-io = "0.1.0"
+ntex-io = "0.1.1"
-ntex-util = "0.1.6"
+ntex-util = "0.1.8"
 log = "0.4"
 pin-project-lite = "0.2"
 tokio = { version = "1", default-features = false, features = ["rt", "net", "sync", "signal"] }

ntex/Cargo.toml

@@ -50,7 +50,7 @@ ntex-router = "0.5.1"
 ntex-service = "0.3.1"
 ntex-macros = "0.1.3"
 ntex-util = "0.1.8"
-ntex-bytes = "0.1.8"
+ntex-bytes = "0.1.9"
 ntex-tls = "0.1.1"
 ntex-rt = "0.4.1"
 ntex-io = "0.1.2"