Keep WordUtils.abbreviate from splitting a surrogate pair

src/main/java/org/apache/commons/text/WordUtils.java

@@ -93,13 +93,21 @@ public static String abbreviate(final String str, int lower, int upper, final St
         final StringBuilder result = new StringBuilder();
         final int index = Strings.CS.indexOf(str, " ", lower);
         if (index == -1) {
-            result.append(str, 0, upper);
+            int end = upper;
+            if (splitsSurrogatePair(str, end)) {
+                end--;
+            }
+            result.append(str, 0, end);
             // only if abbreviation has occurred do we append the appendToEnd value
             if (upper != str.length()) {
                 result.append(StringUtils.defaultString(appendToEnd));
             }
         } else {
-            result.append(str, 0, Math.min(index, upper));
+            int end = Math.min(index, upper);
+            if (splitsSurrogatePair(str, end)) {
+                end--;
+            }
+            result.append(str, 0, end);
             result.append(StringUtils.defaultString(appendToEnd));
         }
         return result.toString();
@@ -426,6 +434,17 @@ public static boolean isDelimiter(final int codePoint, final char[] delimiters)
         return false;
     }
 
+    /**
+     * Tests whether the given index splits a surrogate pair, that is, whether it falls between a high surrogate and its trailing low surrogate.
+     *
+     * @param str   The String to check.
+     * @param index The index to test.
+     * @return Whether cutting {@code str} at {@code index} would split a surrogate pair.
+     */
+    private static boolean splitsSurrogatePair(final String str, final int index) {
+        return index > 0 && index < str.length() && Character.isHighSurrogate(str.charAt(index - 1)) && Character.isLowSurrogate(str.charAt(index));
+    }
+
     /**
      * Swaps the case of a String using a word based algorithm.
      * <ul>

src/test/java/org/apache/commons/text/WordUtilsTest.java

@@ -86,6 +86,17 @@ void testAbbreviateForUpperLimitAndAppendedString() {
         assertEquals("0123456789", WordUtils.abbreviate("0123456789", 0, -1, ""));
     }
 
+    @Test
+    void testAbbreviateSurrogatePairs() {
+        // an upper limit landing inside a surrogate pair must not leave a lone surrogate behind
+        assertEquals("\uD83D\uDE00", WordUtils.abbreviate("\uD83D\uDE00\uD83D\uDE00\uD83D\uDE00", 0, 3, ""));
+        assertEquals("\uD83D\uDE00-", WordUtils.abbreviate("\uD83D\uDE00\uD83D\uDE00\uD83D\uDE00", 0, 3, "-"));
+        // the same applies when the abbreviation stops short of a space
+        assertEquals("\uD83D\uDE00-", WordUtils.abbreviate("\uD83D\uDE00\uD83D\uDE00 0", 0, 3, "-"));
+        // an upper limit falling between two pairs is already whole and must be left alone
+        assertEquals("\uD83D\uDE00\uD83D\uDE00", WordUtils.abbreviate("\uD83D\uDE00\uD83D\uDE00\uD83D\uDE00", 0, 4, ""));
+    }
+
     @Test
     void testAbbreviateUpperLessThanLowerValues() {
         assertThrows(IllegalArgumentException.class, () -> WordUtils.abbreviate("0123456789", 5, 2, ""));