Refactor low-level UTF-16 decoding.

* Rename `utf16_items` to `decode_utf16`. "Items" is meaningless.
* Move it to `rustc_unicode::char`, exposed in `std::char`.
* Generalize it to any `u16` iterable, not just `&[u16]`.
* Make it yield `Result` instead of a custom `Utf16Item` enum that was isomorphic to `Result`. This enable using the `FromIterator for Result` impl.
* Add a `REPLACEMENT_CHARACTER` constant.
* Document how `result.unwrap_or(REPLACEMENT_CHARACTER)` replaces `Utf16Item::to_char_lossy`.

1 files changed, 6 insertions, 6 deletions

diff --git a/src/libserialize/json.rs b/src/libserialize/json.rs
index e474f47a1b5..09f98978653 100644
--- a/src/libserialize/json.rs
+++ b/src/libserialize/json.rs
@@ -209,8 +209,6 @@ use std::str::FromStr;
 use std::string;
 use std::{char, f64, fmt, str};
 use std;
-use rustc_unicode::str as unicode_str;
-use rustc_unicode::str::Utf16Item;
 
 use Encodable;
 
@@ -1712,11 +1710,13 @@ impl<T: Iterator<Item=char>> Parser<T> {
                                 _ => return self.error(UnexpectedEndOfHexEscape),
                             }
 
-                            let buf = [n1, try!(self.decode_hex_escape())];
-                            match unicode_str::utf16_items(&buf).next() {
-                                Some(Utf16Item::ScalarValue(c)) => res.push(c),
-                                _ => return self.error(LoneLeadingSurrogateInHexEscape),
+                            let n2 = try!(self.decode_hex_escape());
+                            if n2 < 0xDC00 || n2 > 0xDFFF {
+                                return self.error(LoneLeadingSurrogateInHexEscape)
                             }
+                            let c = (((n1 - 0xD800) as u32) << 10 |
+                                     (n2 - 0xDC00) as u32) + 0x1_0000;
+                            res.push(char::from_u32(c).unwrap());
                         }
 
                         n => match char::from_u32(n as u32) {