Handle supplementary code points in splitByCharacterType (#1734)

Author: alhudz

src/main/java/org/apache/commons/lang3/StringUtils.java

@@ -7259,14 +7259,17 @@ private static String[] splitByCharacterType(final String str, final boolean cam
         final char[] c = str.toCharArray();
         final List<String> list = new ArrayList<>();
         int tokenStart = 0;
-        int currentType = Character.getType(c[tokenStart]);
-        for (int pos = tokenStart + 1; pos < c.length; pos++) {
-            final int type = Character.getType(c[pos]);
+        int currentType = Character.getType(Character.codePointAt(c, tokenStart));
+        for (int pos = tokenStart + Character.charCount(Character.codePointAt(c, tokenStart)); pos < c.length;) {
+            final int codePoint = Character.codePointAt(c, pos);
+            final int type = Character.getType(codePoint);
+            final int count = Character.charCount(codePoint);
             if (type == currentType) {
+                pos += count;
                 continue;
             }
             if (camelCase && type == Character.LOWERCASE_LETTER && currentType == Character.UPPERCASE_LETTER) {
-                final int newTokenStart = pos - 1;
+                final int newTokenStart = pos - Character.charCount(Character.codePointBefore(c, pos));
                 if (newTokenStart != tokenStart) {
                     list.add(new String(c, tokenStart, newTokenStart - tokenStart));
                     tokenStart = newTokenStart;
@@ -7276,6 +7279,7 @@ private static String[] splitByCharacterType(final String str, final boolean cam
                 tokenStart = pos;
             }
             currentType = type;
+            pos += count;
         }
         list.add(new String(c, tokenStart, c.length - tokenStart));
         return list.toArray(ArrayUtils.EMPTY_STRING_ARRAY);

src/test/java/org/apache/commons/lang3/StringUtilsTest.java

@@ -2355,6 +2355,18 @@ void testSplitByCharacterType() {
 
         assertTrue(Objects.deepEquals(new String[]{"ASFR", "ules"},
                 StringUtils.splitByCharacterType("ASFRules")));
+
+        // Supplementary code points are classified by their own type, not split apart as surrogates.
+        // U+1D400 MATHEMATICAL BOLD CAPITAL A is an upper-case letter, like ASCII 'A'.
+        final String boldA = new String(Character.toChars(0x1D400));
+        // U+1D7D3 MATHEMATICAL BOLD DIGIT FIVE is a decimal digit, like ASCII '5'.
+        final String boldFive = new String(Character.toChars(0x1D7D3));
+        assertTrue(Objects.deepEquals(new String[]{"A" + boldA},
+                StringUtils.splitByCharacterType("A" + boldA)));
+        assertTrue(Objects.deepEquals(new String[]{"5" + boldFive},
+                StringUtils.splitByCharacterType("5" + boldFive)));
+        assertTrue(Objects.deepEquals(new String[]{boldA, "5" + boldFive, "z"},
+                StringUtils.splitByCharacterType(boldA + "5" + boldFive + "z")));
     }
 
     @Test
@@ -2382,6 +2394,16 @@ void testSplitByCharacterTypeCamelCase() {
 
         assertTrue(Objects.deepEquals(new String[]{"ASF", "Rules"},
                 StringUtils.splitByCharacterTypeCamelCase("ASFRules")));
+
+        // A supplementary upper-case letter immediately before a lower-case run joins the following token,
+        // exactly as a BMP upper-case letter does. U+1D400 MATHEMATICAL BOLD CAPITAL A is an upper-case letter.
+        final String boldA = new String(Character.toChars(0x1D400));
+        assertTrue(Objects.deepEquals(new String[]{boldA + "bc"},
+                StringUtils.splitByCharacterTypeCamelCase(boldA + "bc")));
+        assertTrue(Objects.deepEquals(new String[]{"AB", boldA + "cd"},
+                StringUtils.splitByCharacterTypeCamelCase("AB" + boldA + "cd")));
+        assertTrue(Objects.deepEquals(new String[]{"foo", boldA + "bar"},
+                StringUtils.splitByCharacterTypeCamelCase("foo" + boldA + "bar")));
     }
 
     @Test