optimized converting mutf8 to utf8

benches/compare.rs

@@ -120,10 +120,10 @@ pub fn bench_read_file(filename: &str, c: &mut Criterion) {
 
 fn bench(c: &mut Criterion) {
     // bench_read_file("hello_world.nbt", c);
-    // bench_read_file("bigtest.nbt", c);
-    // bench_read_file("simple_player.dat", c);
+    bench_read_file("bigtest.nbt", c);
+    bench_read_file("simple_player.dat", c);
     bench_read_file("complex_player.dat", c);
-    // bench_read_file("level.dat", c);
+    bench_read_file("level.dat", c);
     // bench_read_file("stringtest.nbt", c);
     // bench_read_file("inttest.nbt", c);
 }

benches/mutf8.rs

@@ -1,6 +1,18 @@
-use criterion::{criterion_group, criterion_main, Criterion};
+use std::hint::black_box;
 
-fn bench(_c: &mut Criterion) {}
+use criterion::{criterion_group, criterion_main, Criterion};
+use simdnbt::Mutf8Str;
+
+fn bench(c: &mut Criterion) {
+    let mut group = c.benchmark_group(format!("mutf8"));
+
+    group.bench_function("to_str", |b| {
+        let input = black_box(Mutf8Str::from_slice(b"hello world"));
+        b.iter(|| {
+            black_box(input.to_str());
+        })
+    });
+}
 
 criterion_group!(benches, bench);
 criterion_main!(benches);

src/lib.rs

@@ -772,6 +772,7 @@ impl<'a> ListTag<'a> {
 mod tests {
     use std::io::Read;
 
+    use byteorder::WriteBytesExt;
     use flate2::read::GzDecoder;
 
     use super::*;
@@ -831,7 +832,6 @@ mod tests {
 
     #[test]
     fn inttest_1024() {
-        use byteorder::WriteBytesExt;
         let mut data = Vec::new();
         data.write_u8(COMPOUND_ID).unwrap();
         data.write_u16::<BE>(0).unwrap();
@@ -854,7 +854,6 @@ mod tests {
 
     #[test]
     fn inttest_1021() {
-        use byteorder::WriteBytesExt;
         let mut data = Vec::new();
         data.write_u8(COMPOUND_ID).unwrap();
         data.write_u16::<BE>(0).unwrap();
@@ -877,7 +876,6 @@ mod tests {
 
     #[test]
     fn longtest_1023() {
-        use byteorder::WriteBytesExt;
         let mut data = Vec::new();
         data.write_u8(COMPOUND_ID).unwrap();
         data.write_u16::<BE>(0).unwrap();
@@ -898,6 +896,15 @@ mod tests {
         assert_eq!(ints.len(), 1023);
     }
 
+    #[test]
+    fn stringtest() {
+        let nbt = Nbt::new(&mut Cursor::new(include_bytes!("../tests/stringtest.nbt")))
+            .unwrap()
+            .unwrap();
+
+        assert_eq!(nbt.list("😃").unwrap().strings().unwrap().len(), 16);
+    }
+
     // #[test]
     // fn generate_inttest() {
     //     use byteorder::WriteBytesExt;
@@ -916,4 +923,18 @@ mod tests {
 
     //     std::fs::write("tests/inttest1023.nbt", out).unwrap();
     // }
+
+    // #[test]
+    // fn generate_stringtest() {
+    //     let mut out = Vec::new();
+    //     out.write_u8(COMPOUND_ID).unwrap();
+    //     out.write_u16::<BE>(0).unwrap();
+    //     out.write_u8(LIST_ID).unwrap();
+    //     out.write_u16::<BE>(0).unwrap();
+    //     out.write_u8(STRING_ID).unwrap();
+    //     out.write_i32::<BE>(16).unwrap();
+    //     out.extend_from_slice(&std::fs::read("tests/stringtest.nbt").unwrap().as_slice()[13..]);
+    //     out.write_u8(END_ID).unwrap();
+    //     std::fs::write("tests/stringtest2.nbt", out).unwrap();
+    // }
 }

src/mutf8.rs

@@ -35,9 +35,22 @@ impl Mutf8Str {
         }
     }
     pub fn to_str(&self) -> Cow<str> {
-        match mutf8::decode(&self.slice).expect("Mutf8Str must alwaus be valid MUTF-8") {
-            Cow::Borrowed(b) => Cow::Borrowed(b),
-            Cow::Owned(o) => Cow::Owned(o),
+        // fast check to skip if none of the bytes have the top bit set or are null
+        let mut is_not_ascii = false;
+        for &byte in self.slice.into_iter() {
+            if byte & 0b1000_0000 != 0 || byte == 0 {
+                is_not_ascii = true;
+            }
+        }
+
+        if is_not_ascii {
+            return match mutf8::decode(&self.slice).expect("Mutf8Str must alwaus be valid MUTF-8") {
+                Cow::Borrowed(b) => Cow::Borrowed(b),
+                Cow::Owned(o) => Cow::Owned(o),
+            };
+        } else {
+            // SAFETY: &[u8] and &str are the same layout.
+            unsafe { Cow::Borrowed(mem::transmute(self.slice.as_ref())) }
         }
     }
 }