0.8.0 read escaped surrogate pairs properly, fixes #39

Author: maciejhirsz

src/parser.rs

@@ -254,11 +254,31 @@ impl<'a> Parser<'a> {
     }
 
     fn read_codepoint(&mut self) -> JsonResult<()> {
-        let codepoint = try!(self.read_char_as_hexnumber()) << 12
+        let mut codepoint = try!(self.read_char_as_hexnumber()) << 12
                       | try!(self.read_char_as_hexnumber()) << 8
                       | try!(self.read_char_as_hexnumber()) << 4
                       | try!(self.read_char_as_hexnumber());
 
+        match codepoint {
+            0xD800 ... 0xDFFF => {
+                codepoint -= 0xD800;
+                codepoint <<= 10;
+
+                expect!(self, b'\\');
+                expect!(self, b'u');
+
+                let lower = try!(self.read_char_as_hexnumber()) << 12
+                          | try!(self.read_char_as_hexnumber()) << 8
+                          | try!(self.read_char_as_hexnumber()) << 4
+                          | try!(self.read_char_as_hexnumber());
+
+                codepoint |= lower - 0xDC00;
+
+                codepoint += 0x010000;
+            },
+            _ => {}
+        }
+
         let ch = try!(
             char::from_u32(codepoint).ok_or(JsonError::FailedUtf8Parsing)
         );

tests/lib.rs

@@ -533,6 +533,17 @@ fn parse_escaped_unicode() {
     assert_eq!(data, "❤️");
 }
 
+#[test]
+fn parse_escaped_unicode_surrogate() {
+    let data = parse(r#"
+
+    "\uD834\uDD1E"
+
+    "#).unwrap();
+
+    assert_eq!(data, "𝄞");
+}
+
 #[test]
 fn array_len() {
     let data = array![0, 1, 2, 3];