Avoid surrogates when generating char using Standard distribution

Author: Pazzaz

benches/misc.rs

@@ -150,3 +150,14 @@ fn gen_1k_fill(b: &mut Bencher) {
     });
     b.bytes = 1024;
 }
+
+#[bench]
+fn misc_gen_chars(b: &mut Bencher) {
+    use std::iter;
+    let mut rng = StdRng::from_rng(&mut thread_rng()).unwrap();
+    b.iter(|| {
+        let v: Vec<char> = iter::repeat(()).map(|()| rng.gen()).take(128).collect();
+        v
+    });
+    b.bytes = 512;
+}

src/distributions/other.rs

@@ -44,15 +44,18 @@ pub struct Alphanumeric;
 impl Distribution<char> for Standard {
     #[inline]
     fn sample<R: Rng + ?Sized>(&self, rng: &mut R) -> char {
-        let range = Uniform::new(0u32, 0x11_0000);
-        loop {
-            match char::from_u32(range.sample(rng)) {
-                Some(c) => return c,
-                // About 0.2% of numbers in the range 0..0x110000 are invalid
-                // codepoints (surrogates).
-                None => {}
-            }
+        // A valid `char` is either in the interval `[0, 0xD800)` or
+        // `(0xDFFF, 0x11_0000)`. All `char`s must therefore be in
+        // `[0, 0x11_0000)` but not in the "gap" `[0xD800, 0xDFFF]` which is
+        // reserved for surrogates. This is the size of that gap.
+        const GAP_SIZE: u32 = 0xDFFF - 0xD800 + 1;
+
+        let range = Uniform::new(GAP_SIZE, 0x11_0000);
+        let mut n = range.sample(rng);
+        if n <= 0xDFFF {
+            n -= GAP_SIZE;
         }
+        unsafe { char::from_u32_unchecked(n) }
     }
 }