1615: Fix bug in getCJKAdjustedLength() that double counts supplement…

…ary code points

getCJKAdjustedLength calculates the column width of text strings.

But it operates on chars which causes it to double count supplementary code
points. Supplimentary code points have a high and a low code unit(char).

This commit reworks this logic to be based on code points. The code assembles
the code points by hand as String.codePoints() was only added in 1.8 and
picocli has a source level of 1.5

src/main/java/picocli/CommandLine.java

@@ -7921,16 +7921,16 @@ public void run() {
              * @since 4.0 */
             public UsageMessageSpec autoWidth(boolean detectTerminalSize) { autoWidth = detectTerminalSize; return this; }
             /**
-             * Given a character, is this character considered to be a CJK character?
+             * Given a codePoint, is this codePoint considered to be a CJK character?
              * Shamelessly stolen from
              * <a href="http://stackoverflow.com/questions/1499804/how-can-i-detect-japanese-text-in-a-java-string">StackOverflow</a>
              * where it was contributed by user Rakesh N. (Upvote! :-) )
-             * @param c Character to test
+             * @param codePoint code point to test
              * @return {@code true} if the character is a CJK character
              */
-            static boolean isCharCJK(char c) {
-                Character.UnicodeBlock unicodeBlock = Character.UnicodeBlock.of(c);
-                return (c == 0x00b1
+            static boolean isCodePointCJK(int codePoint) {
+                Character.UnicodeBlock unicodeBlock = Character.UnicodeBlock.of(codePoint);
+                return (codePoint == 0x00b1
                         || unicodeBlock == Character.UnicodeBlock.HIRAGANA)
                         || (unicodeBlock == Character.UnicodeBlock.KATAKANA)
                         || (unicodeBlock == Character.UnicodeBlock.KATAKANA_PHONETIC_EXTENSIONS)
@@ -7946,7 +7946,7 @@ static boolean isCharCJK(char c) {
                         || (unicodeBlock == Character.UnicodeBlock.CJK_SYMBOLS_AND_PUNCTUATION)
                         || (unicodeBlock == Character.UnicodeBlock.ENCLOSED_CJK_LETTERS_AND_MONTHS)
                         //The magic number here is the separating index between full-width and half-width
-                        || (unicodeBlock == Character.UnicodeBlock.HALFWIDTH_AND_FULLWIDTH_FORMS && c < 0xFF61);
+                        || (unicodeBlock == Character.UnicodeBlock.HALFWIDTH_AND_FULLWIDTH_FORMS && codePoint < 0xFF61);
             }
 
             /** Returns the help section renderers for the predefined section keys. see: {@link #sectionKeys()} */
@@ -18106,10 +18106,27 @@ public int getCJKAdjustedLength() {
                  * @return the number of columns that the specified portion of this Text will occupy on the console, adjusted for wide CJK characters
                  * @since 4.0 */
                 public int getCJKAdjustedLength(int fromPosition, int charCount) {
+                    String lengthOf = plain.substring(from, from + charCount);
+
                     int result = 0;
-                    for (int i = fromPosition; i < fromPosition + charCount; i++) {
-                        result += UsageMessageSpec.isCharCJK(plain.charAt(i)) ? 2 : 1;
+                    int i = 0;
+                    while (i < lengthOf.length()) {
+                        int codePoint;
+                        char c1 = lengthOf.charAt(i++);
+                        if (!Character.isHighSurrogate(c1) || i >= length) {
+                            codePoint = c1;
+                        } else {
+                            char c2 = lengthOf.charAt(i);
+                            if (Character.isLowSurrogate(c2)) {
+                                i++;
+                                codePoint = Character.toCodePoint(c1, c2);
+                            } else {
+                                codePoint = c1;
+                            }
+                        }
+                        result += UsageMessageSpec.isCodePointCJK(codePoint) ? 2 : 1;
                     }
+
                     return result;
                 }
             }

src/test/java/picocli/CJKLengthTest.java

@@ -0,0 +1,39 @@
+
+package picocli;
+
+import static org.junit.Assert.assertEquals;
+import static picocli.CommandLine.Help.Ansi;
+
+import org.junit.Rule;
+import org.junit.Test;
+import org.junit.contrib.java.lang.system.ProvideSystemProperty;
+import org.junit.contrib.java.lang.system.RestoreSystemProperties;
+import org.junit.rules.TestRule;
+
+public class CJKLengthTest {
+
+    // allows tests to set any kind of properties they like, without having to individually roll them back
+    @Rule
+    public final TestRule restoreSystemProperties = new RestoreSystemProperties();
+
+    @Rule
+    public final ProvideSystemProperty ansiOFF = new ProvideSystemProperty("picocli.ansi", "false");
+
+    @Test
+    public void testCJKLengths() {
+      testLength("abc", 3);
+      // some double width Hiragana characters
+      testLength("平仮名", 6);
+
+      // a supplementary code point character (has a high and low code point values)
+      testLength("𝑓", 1);
+    }
+
+    private void testLength(String of, int expectedLength) {
+      Ansi.Text text = Ansi.OFF.text(of);
+      int cjkWidth = text.getCJKAdjustedLength();
+      assertEquals(String.format("Expected '%s' to have width %d but is %d", of, expectedLength, cjkWidth),
+          expectedLength, cjkWidth);
+    }
+
+}