Use code point count for truncate as well

2 files changed, 7 insertions, 33 deletions

diff --git a/vespajlib/src/main/java/com/yahoo/text/Text.java b/vespajlib/src/main/java/com/yahoo/text/Text.java
index e133407a967..fe931ef34a3 100644
--- a/vespajlib/src/main/java/com/yahoo/text/Text.java
+++ b/vespajlib/src/main/java/com/yahoo/text/Text.java
@@ -170,15 +170,15 @@ public final class Text {
     }
 
     /**
-     * Returns a string which is never larger than the given number of characters.
+     * Returns a string which is never larger than the given number of code points.
      * If the string is longer than the given length it will be truncated.
      * If length is 4 or less the string will be truncated to length.
      * If length is longer than 4, it will be truncated at length-4 with " ..." added at the end.
      */
     public static String truncate(String s, int length) {
-        if (s.length() <= length) return s;
-        if (length <= 4) return safeSubstring(s, length);
-        return safeSubstring(s, length - 4) + " ...";
+        if (s.codePointCount(0, s.length()) <= length) return s;
+        if (length <= 4) return substringByCodepoints(s, 0, length);
+        return substringByCodepoints(s, 0, length - 4) + " ...";
     }
 
     public static String substringByCodepoints(String s, int fromCP, int toCP) {
@@ -209,13 +209,4 @@ public final class Text {
 	return String.format(Locale.US, format, args);
     }
 
-    /** Like {@link String#substring(int)}, but if this would split a surrogate pair at the end, the leading high surrogate is also cut. */
-    public static String safeSubstring(String s, int length) {
-        boolean pairCut =    0 < length
-                          && length < s.length()
-                          && Character.isHighSurrogate(s.charAt(length - 1))
-                          && Character.isLowSurrogate(s.charAt(length));
-        return s.substring(0, length - (pairCut ? 1 : 0));
-    }
-
 }
diff --git a/vespajlib/src/test/java/com/yahoo/text/TextTestCase.java b/vespajlib/src/test/java/com/yahoo/text/TextTestCase.java
index b4324797086..9bb4668b7cb 100644
--- a/vespajlib/src/test/java/com/yahoo/text/TextTestCase.java
+++ b/vespajlib/src/test/java/com/yahoo/text/TextTestCase.java
@@ -83,24 +83,6 @@ public class TextTestCase {
     }
 
     @Test
-    public void testSafeSubstring() {
-        String withSurrogates = "abc\uD83D\uDE48\uD83D\uDE49\uD83D\uDE4Adef";
-        assertEquals("", Text.safeSubstring(withSurrogates, 0));
-        assertEquals("a", Text.safeSubstring(withSurrogates, 1));
-        assertEquals("ab", Text.safeSubstring(withSurrogates, 2));
-        assertEquals("abc", Text.safeSubstring(withSurrogates, 3));
-        assertEquals("abc", Text.safeSubstring(withSurrogates, 4));
-        assertEquals("abc\uD83D\uDE48", Text.safeSubstring(withSurrogates, 5));
-        assertEquals("abc\uD83D\uDE48", Text.safeSubstring(withSurrogates, 6));
-        assertEquals("abc\uD83D\uDE48\uD83D\uDE49", Text.safeSubstring(withSurrogates, 7));
-        assertEquals("abc\uD83D\uDE48\uD83D\uDE49", Text.safeSubstring(withSurrogates, 8));
-        assertEquals("abc\uD83D\uDE48\uD83D\uDE49\uD83D\uDE4A", Text.safeSubstring(withSurrogates, 9));
-        assertEquals("abc\uD83D\uDE48\uD83D\uDE49\uD83D\uDE4Ad", Text.safeSubstring(withSurrogates, 10));
-        assertEquals("abc\uD83D\uDE48\uD83D\uDE49\uD83D\uDE4Ade", Text.safeSubstring(withSurrogates, 11));
-        assertEquals("abc\uD83D\uDE48\uD83D\uDE49\uD83D\uDE4Adef", Text.safeSubstring(withSurrogates, 12));
-    }
-
-    @Test
     public void testIsDisplayable() {
         assertTrue(Text.isDisplayable('A'));
         assertTrue(Text.isDisplayable('a'));
@@ -122,8 +104,9 @@ public class TextTestCase {
         assertEquals("",   Text.truncate("ab", 0));
         assertEquals("ab c",  Text.truncate("ab cde", 4));
         assertEquals("a ...", Text.truncate("ab cde", 5));
-        assertEquals("abc\uD83D\uDE48 ...", Text.truncate("abc\uD83D\uDE48\uD83D\uDE49\uD83D\uDE4Adef", 9));
-        assertEquals("abc\uD83D\uDE48 ...", Text.truncate("abc\uD83D\uDE48\uD83D\uDE49\uD83D\uDE4Adef", 10));
+        assertEquals("abc ...", Text.truncate("abc\uD83D\uDE48\uD83D\uDE49\uD83D\uDE4Adef", 7));
+        assertEquals("abc\uD83D\uDE48 ...", Text.truncate("abc\uD83D\uDE48\uD83D\uDE49\uD83D\uDE4Adef", 8));
+        assertEquals("abc\uD83D\uDE48\uD83D\uDE49\uD83D\uDE4Adef", Text.truncate("abc\uD83D\uDE48\uD83D\uDE49\uD83D\uDE4Adef", 9));
     }
 
     @Test