---

yaml
---
r: 160471
b: refs/heads/master
c: f6607a2
h: refs/heads/master
i:
  160469: d2396fe
  160467: d97dcbc
  160463: 7ba3e70
v: v3

Author: brson

[refs]

@@ -1,5 +1,5 @@
 ---
-refs/heads/master: 0150fa4b1b3e30b1f763905bd1af2d2ccd73c47e
+refs/heads/master: f6607a20c4abbd03a806c1320d059e0911dd0cdb
 refs/heads/snap-stage1: e33de59e47c5076a89eadeb38f4934f58a3618a6
 refs/heads/snap-stage3: c9f6d696420107f82304b992cf623b806995fe18
 refs/heads/try: 225de0d60f8ca8dcc62ab2fd8818ebbda4b58cfe

trunk/src/libcore/char.rs

@@ -323,6 +323,10 @@ pub trait Char {
     /// UTF-8.
     fn len_utf8(&self) -> uint;
 
+    /// Returns the amount of bytes this character would need if encoded in
+    /// UTF-16.
+    fn len_utf16(&self) -> uint;
+
     /// Encodes this character as UTF-8 into the provided byte buffer,
     /// and then returns the number of bytes written.
     ///
@@ -363,6 +367,12 @@ impl Char for char {
     #[inline]
     fn len_utf8(&self) -> uint { len_utf8_bytes(*self) }
 
+    #[inline]
+    fn len_utf16(&self) -> uint {
+        let ch = *self as u32;
+        if (ch & 0xFFFF_u32) == ch { 1 } else { 2 }
+    }
+
     #[inline]
     fn encode_utf8<'a>(&self, dst: &'a mut [u8]) -> Option<uint> {
         // Marked #[inline] to allow llvm optimizing it away

trunk/src/libcoretest/char.rs

@@ -197,6 +197,14 @@ fn test_encode_utf16() {
     check('\U0001f4a9', &[0xd83d, 0xdca9]);
 }
 
+#[test]
+fn test_len_utf16() {
+    assert!('x'.len_utf16() == 1);
+    assert!('\u00e9'.len_utf16() == 1);
+    assert!('\ua66e'.len_utf16() == 1);
+    assert!('\U0001f4a9'.len_utf16() == 2);
+}
+
 #[test]
 fn test_width() {
     assert_eq!('\x00'.width(false),Some(0));