perf: use simdutf8 to validate UTF-8 when reading files (#5196)

closes #5191

---------

Co-authored-by: overlookmotel <j@dummett.org>

Cargo.lock

@@ -1020,7 +1020,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "4979f22fdb869068da03c9f7528f8297c6fd2606bc3a4affe42e6a823fdb8da4"
 dependencies = [
  "cfg-if",
- "windows-targets 0.48.5",
+ "windows-targets 0.52.6",
 ]
 
 [[package]]
@@ -1624,6 +1624,7 @@ dependencies = [
  "schemars",
  "serde",
  "serde_json",
+ "simdutf8",
  "url",
 ]
 
@@ -2680,6 +2681,12 @@ dependencies = [
  "libc",
 ]
 
+[[package]]
+name = "simdutf8"
+version = "0.1.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f27f6278552951f1f2b8cf9da965d10969b2efdea95a6ec47987ab46edfe263a"
+
 [[package]]
 name = "similar"
 version = "2.6.0"

Cargo.toml

@@ -170,6 +170,7 @@ seq-macro           = "0.3.5"
 serde               = "1.0.206"
 serde_json          = "1.0.124"
 serde-wasm-bindgen  = "0.6.5"
+simdutf8            = { version = "0.1.4", features = ["aarch64_neon"] }
 similar             = "2.6.0"
 syn                 = { version = "2.0.74", default-features = false }
 tempfile            = "3.12.0"

crates/oxc_linter/Cargo.toml

@@ -52,6 +52,7 @@ once_cell           = { workspace = true }
 memchr              = { workspace = true }
 json-strip-comments = { workspace = true }
 schemars            = { workspace = true, features = ["indexmap2"] }
+simdutf8            = { workspace = true }
 
 [dev-dependencies]
 insta        = { workspace = true }

crates/oxc_linter/src/config/mod.rs

@@ -17,7 +17,9 @@ pub use self::{
     settings::{jsdoc::JSDocPluginSettings, OxlintSettings},
 };
 use crate::{
-    rules::RuleEnum, utils::is_jest_rule_adapted_to_vitest, AllowWarnDeny, RuleWithSeverity,
+    rules::RuleEnum,
+    utils::{is_jest_rule_adapted_to_vitest, read_to_string},
+    AllowWarnDeny, RuleWithSeverity,
 };
 
 /// Oxlint Configuration File
@@ -68,7 +70,7 @@ impl OxlintConfig {
     ///
     /// * Parse Failure
     pub fn from_file(path: &Path) -> Result<Self, OxcDiagnostic> {
-        let mut string = std::fs::read_to_string(path).map_err(|e| {
+        let mut string = read_to_string(path).map_err(|e| {
             OxcDiagnostic::error(format!("Failed to parse config {path:?} with error {e:?}"))
         })?;
 

crates/oxc_linter/src/service.rs

@@ -19,6 +19,7 @@ use rustc_hash::FxHashSet;
 
 use crate::{
     partial_loader::{JavaScriptSource, PartialLoader, LINT_PARTIAL_LOADER_EXT},
+    utils::read_to_string,
     Fixer, Linter, Message,
 };
 
@@ -176,7 +177,7 @@ impl Runtime {
             return None;
         }
         let source_type = source_type.unwrap_or_default();
-        let file_result = fs::read_to_string(path).map_err(|e| {
+        let file_result = read_to_string(path).map_err(|e| {
             Error::new(OxcDiagnostic::error(format!(
                 "Failed to open file {path:?} with error \"{e}\""
             )))

crates/oxc_linter/src/utils/mod.rs

@@ -9,6 +9,8 @@ mod tree_shaking;
 mod unicorn;
 mod vitest;
 
+use std::{io, path::Path};
+
 pub use self::{
     config::*, jest::*, jsdoc::*, nextjs::*, promise::*, react::*, react_perf::*, tree_shaking::*,
     unicorn::*, vitest::*,
@@ -37,3 +39,17 @@ pub fn is_jest_rule_adapted_to_vitest(rule_name: &str) -> bool {
 
     jest_rules.contains(&rule_name)
 }
+
+pub fn read_to_string(path: &Path) -> io::Result<String> {
+    // `simdutf8` is faster than `std::str::from_utf8` which `fs::read_to_string` uses internally
+    let bytes = std::fs::read(path)?;
+    if simdutf8::basic::from_utf8(&bytes).is_err() {
+        // Same error as `fs::read_to_string` produces (`io::Error::INVALID_UTF8`)
+        return Err(io::Error::new(
+            io::ErrorKind::InvalidData,
+            "stream did not contain valid UTF-8",
+        ));
+    }
+    // SAFETY: `simdutf8` has ensured it's a valid UTF-8 string
+    Ok(unsafe { String::from_utf8_unchecked(bytes) })
+}