Merge branch 'main' of https://github.com/Ygg01/simd-json

Author: Ygg01

.github/workflows/pr-perf.yml

@@ -5,17 +5,21 @@ jobs:
     name: run benchmark
     runs-on: ubuntu-latest
     permissions:
-        contents: write
-        pull-requests: write
-        repository-projects: write    
+      contents: write
+      pull-requests: write
+      repository-projects: write
     steps:
       - uses: actions/checkout@v3
+
+      - name: Check Rust Version
+        run: rustc --version
+
       - uses: boa-dev/criterion-compare-action@v3 # https://github.com/marketplace/actions/criterion-compare-prs
         env:
-            RUSTFLAGS: "-C target-cpu=native"
+          RUSTFLAGS: "-C target-cpu=native"
         with:
           # Optional. Compare only this benchmark target
           benchName: "to_tape"
           features: "bench-all"
           # Needed. The name of the branch to compare with. This default uses the branch which is being pulled against
-          branchName: ${{ github.base_ref }}
+          branchName: ${{ github.base_ref }}

.github/workflows/quality.yaml

@@ -18,7 +18,7 @@ jobs:
     steps:
       - uses: actions/checkout@v3
 
-      - uses: dtolnay/rust-toolchain@1.85 # do clippy chekcs with the minimum supported version
+      - uses: dtolnay/rust-toolchain@1.88 # do clippy chekcs with the minimum supported version
         with:
           components: rustfmt, clippy
 

.github/workflows/tests.yml

@@ -24,7 +24,7 @@ jobs:
     steps:
       - uses: actions/checkout@v3
 
-      - uses: dtolnay/rust-toolchain@1.85
+      - uses: dtolnay/rust-toolchain@1.88
         with:
           components: llvm-tools-preview
 
@@ -120,7 +120,7 @@ jobs:
     steps:
       - uses: actions/checkout@v3
 
-      - uses: dtolnay/rust-toolchain@1.85
+      - uses: dtolnay/rust-toolchain@1.88
         with:
           targets: wasm32-wasip1
 

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "simd-json"
-version = "0.15.1"
+version = "0.16.0"
 authors = ["Heinz N. Gies <heinz@licenser.net>", "Sunny Gleason"]
 edition = "2024"
 exclude = ["data/*", "fuzz/*"]
@@ -9,10 +9,7 @@ description = "High performance JSON parser based on a port of simdjson"
 repository = "https://github.com/simd-lite/simd-json"
 readme = "README.md"
 documentation = "https://docs.rs/simd-json"
-rust-version = "1.85"
-
-[target.'cfg(target_family = "wasm")'.dependencies]
-getrandom = { version = "0.3", features = ["wasm_js"] }
+rust-version = "1.88"
 
 [dependencies]
 simdutf8 = { version = "0.1.4", features = ["public_imp", "aarch64_neon"] }
@@ -33,6 +30,8 @@ alloc_counter = { version = "0.0.4", optional = true }
 colored = { version = "3.0", optional = true }
 getopts = { version = "0.2", optional = true }
 jemallocator = { version = "0.5", optional = true }
+
+[target.'cfg(target_arch = "x86_64")'.dependencies]
 perfcnt = { version = "0.8", optional = true }
 
 ref-cast = "1.0"
@@ -44,7 +43,7 @@ core_affinity = { version = "0.8" }
 
 
 [target.'cfg(not(target_family = "wasm"))'.dev-dependencies]
-criterion = { version = "0.5" }
+criterion = { version = "0.7" }
 proptest = "1.0"
 
 [lib]

examples/perf.rs

@@ -1,6 +1,6 @@
 use std::env;
 
-#[cfg(feature = "perf")]
+#[cfg(all(feature = "perf", target_arch = "x86_64", target_os = "linux"))]
 mod int {
     const ROUNDS: usize = 2000;
     const WARMUP: usize = 200;
@@ -216,10 +216,14 @@ mod int {
     }
 }
 
-#[cfg(not(feature = "perf"))]
+#[cfg(not(all(feature = "perf", target_arch = "x86_64", target_os = "linux")))]
 mod int {
     pub fn bench(_name: &str, _baseline: bool) {
-        println!("Perf requires linux to run and the perf feature to be enabled")
+        if std::env::consts::ARCH != "x86_64" {
+            println!("This Perf example only supports x86_64 for now.");
+        } else if std::env::consts::OS != "linux" {
+            println!("Perf requires linux to run and the perf feature to be enabled");
+        }
     }
 }
 

src/error.rs

@@ -158,15 +158,15 @@ pub struct Error {
     /// Current character
     character: Option<char>,
     /// Type of error
-    error: ErrorType,
+    err_type: ErrorType,
 }
 
 impl Error {
-    pub(crate) fn new(index: usize, character: Option<char>, error: ErrorType) -> Self {
+    pub(crate) fn new(index: usize, character: Option<char>, err_type: ErrorType) -> Self {
         Self {
             index,
             character,
-            error,
+            err_type,
         }
     }
     pub(crate) fn new_c(index: usize, character: char, error: ErrorType) -> Self {
@@ -179,7 +179,7 @@ impl Error {
         Self {
             index: 0,
             character: None,
-            error: t,
+            err_type: t,
         }
     }
 
@@ -198,7 +198,7 @@ impl Error {
     /// Returns the type of error that occurred.
     #[must_use]
     pub fn error(&self) -> &ErrorType {
-        &self.error
+        &self.err_type
     }
 
     // These make it a bit easier to fit into a serde_json context
@@ -209,7 +209,7 @@ impl Error {
     #[must_use]
     pub fn is_io(&self) -> bool {
         // We have to include InternalError _somewhere_
-        match &self.error {
+        match &self.err_type {
             ErrorType::Io(_) | ErrorType::InputTooLarge => true,
             ErrorType::InternalError(e) if !matches!(e, crate::InternalError::TapeError) => true,
             _ => false,
@@ -219,7 +219,7 @@ impl Error {
     /// Indicates if the error that occurred was an early EOF
     #[must_use]
     pub fn is_eof(&self) -> bool {
-        matches!(self.error, ErrorType::Eof)
+        matches!(self.err_type, ErrorType::Eof)
     }
 
     /// Indicates if the error that occurred was due to a data shape error
@@ -234,7 +234,7 @@ impl Error {
     pub fn is_syntax(&self) -> bool {
         // Lazy? maybe but if it aint something else...
         matches!(
-            self.error,
+            self.err_type,
             ErrorType::InternalError(crate::InternalError::TapeError) | //This seems to get thrown on some syntax errors
             ErrorType::BadKeyType |
             ErrorType::ExpectedArrayComma |
@@ -268,9 +268,9 @@ impl std::error::Error for Error {}
 impl fmt::Display for Error {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         if let Some(c) = self.character {
-            write!(f, "{:?} at character {} ('{c}')", self.error, self.index)
+            write!(f, "{:?} at character {} ('{c}')", self.err_type, self.index)
         } else {
-            write!(f, "{:?} at character {}", self.error, self.index)
+            write!(f, "{:?} at character {}", self.err_type, self.index)
         }
     }
 }

src/impls/avx2/deser.rs

@@ -157,7 +157,7 @@ pub(crate) unsafe fn parse_str<'invoke, 'de>(
 
                     if o == 0 {
                         return Err(Deserializer::error_c(src_i, 'u', InvalidUnicodeCodepoint));
-                    };
+                    }
                     // We moved o steps forward at the destination and 6 on the source
                     src_i += s;
                     dst_i += o;

src/impls/native/stage1.rs

@@ -445,7 +445,7 @@ impl Stage1Parse for SimdInput {
         base.reserve(64);
         let final_len = l + cnt;
 
-        let is_unaligned = l % 4 != 0;
+        let is_unaligned = !l.is_multiple_of(4);
         let write_fn = if is_unaligned {
             std::ptr::write_unaligned
         } else {

src/impls/sse42/deser.rs

@@ -148,7 +148,7 @@ pub(crate) unsafe fn parse_str<'invoke, 'de>(
                     };
                     if o == 0 {
                         return Err(Deserializer::error_c(src_i, 'u', InvalidUnicodeCodepoint));
-                    };
+                    }
                     // We moved o steps forward at the destination and 6 on the source
                     src_i += s;
                     dst_i += o;

src/lib.rs

@@ -11,7 +11,7 @@
 )]
 #![allow(
     clippy::module_name_repetitions,
-    unused_unsafe // for nightly
+    unused_unsafe, // for nightly
 )]
 #![doc = include_str!(concat!(env!("CARGO_MANIFEST_DIR"), "/README.md"))]
 
@@ -123,7 +123,7 @@ impl Buffers {
 ///
 /// Will return `Err` if `s` is invalid JSON.
 #[cfg_attr(not(feature = "no-inline"), inline)]
-pub fn to_tape(s: &mut [u8]) -> Result<Tape> {
+pub fn to_tape(s: &mut [u8]) -> Result<Tape<'_>> {
     Deserializer::from_slice(s).map(Deserializer::into_tape)
 }
 
@@ -916,7 +916,7 @@ impl<'de> Deserializer<'de> {
         // expensive carryless multiply in the previous step with this work
         let mut structurals: u64 = 0;
 
-        let lenminus64: usize = if len < 64 { 0 } else { len - 64 };
+        let lenminus64: usize = len.saturating_sub(64);
         let mut idx: usize = 0;
         let mut error_mask: u64 = 0; // for unescaped characters within strings (ASCII code points < 0x20)