Question

I’m trying to understand some String class functions in Java. So, here’s is a simple code:

/* different experiments with String class */

public class TestStrings {
    public static void main(String[] args) {
        String greeting = "Hello\uD835\uDD6b";

        System.out.println("Number of code units in greeting is " + greeting.length());
        System.out.println("Number of code points " + greeting.codePointCount(0,greeting.length()));

        int index = greeting.offsetByCodePoints(0,6);
        System.out.println("index = " + index);
        int cp = greeting.codePointAt(index);
        System.out.println("Code point at index is " + (char) cp);
    }
}

\uD835\uDD6b is an ℤ symbol, so it’s ok surrogate pair.

So, the string has 6(six) code points and 7(seven) code units (2-byte chars). As it’s in documentation:

offsetByCodePoints

public int offsetByCodePoints(int index,
                              int codePointOffset)

Returns the index within this String that is offset from the given index by codePointOffset code points.
Unpaired surrogates within the text range given by index and codePointOffset count as one code point each.

Parameters:

index – the index to be offset

codePointOffset – the offset in code points

So we do give an argument in code points. But, with given arguments (0,6) it still works fine, without exceptions. But fails for codePointAt(), because it returns 7 which is out of bounds. So, maybe the function gets its args in code units? Or I’ve missed something.

Answer 1

codePointAt takes a char index.

The index refers to char values (Unicode code units) and ranges from 0 to length() - 1.

There are six code-points in that string. The offsetByCodePoints call returns the index after 6 code-points which is char-index 7. You then try to get the codePointAt(7) which is at the end of the string.

To see why, consider what

"".offsetByCodePoints(0, 0) == 0

because to count past all 0 code-points, you have to count past all 0 chars.

Extrapolating that to your string, to count past all 6 code-points, you have to count past all 7 chars.

Maybe seeing codePointAt in use will make this clear. This is the idiomatic way to iterate over all code-points in a string (or CharSequence):

for (var charIndex = 0, nChars = s.length(), codepoint;
     charIndex < nChars;
     charIndex += Character.charCount(codepoint)) {
  codepoint = s.codePointAt(charIndex);
  // Do something with codepoint.
}