Don't decompose Hangul in the svar iterator.

sunfishcode · sunfishcode · commit 41dc717459ad · 2020-12-07T11:56:16.000-08:00
diff --git a/src/normalize.rs b/src/normalize.rs
@@ -47,11 +47,27 @@ pub fn decompose_compatible<F: FnMut(char)>(c: char, emit_char: F) {
 /// [Unicode 6.3 Release Summary](https://www.unicode.org/versions/Unicode6.3.0/#Summary)
 /// for more information.
 #[inline]
-pub fn decompose_svar<F>(c: char, emit_char: F)
+pub fn decompose_svar<F>(c: char, mut emit_char: F)
 where
     F: FnMut(char),
 {
-    decompose(c, svar_fully_decomposed, emit_char)
+    // 7-bit ASCII never decomposes
+    if c <= '\x7f' {
+        emit_char(c);
+        return;
+    }
+
+    // Don't perform decomposition for Hangul
+
+    if let Some(decomposed) = svar_fully_decomposed(c) {
+        for &d in decomposed {
+            emit_char(d);
+        }
+        return;
+    }
+
+    // Finally bottom out.
+    emit_char(c);
 }
 
 #[inline]
diff --git a/tests/svar.rs b/tests/svar.rs
@@ -66,3 +66,12 @@ fn test_standardized_variations_for_cjk_singleton_decompositions() {
     assert_eq!(svar_nfkd_iter.next(), Some('\u{fe00}'));
     assert_eq!(svar_nfkd_iter.next(), None);
 }
+
+/// `svar` shouldn't decompose Hangul.
+#[test]
+fn test_svar_hangul() {
+    assert_eq!(
+        "중국어 (홍콩)".chars().svar().collect::<String>(),
+        "중국어 (홍콩)"
+    );
+}
