This loop statement:
for (char digit: charArr) {
iterates the values in the array. The values have type char and can be anything from 0 to 65535. However, this statement
if (!Character.isDefined(charArr[digit])) {
uses digit as an index for the array. For that to "work" (i.e. not throw an exception), the value needs to be in the range 0 to charArr.length - 1. Clearly, for the input string you are using, some of those values are not acceptable as indexes (e.g. value >= charArr.length) and an exception ensues.
But you don't want to fix that by testing value is in the range required. The values of value are not (from a semantic perspective) array indexes anyway. (If you use them as if they are indexes, you will end up missing some positions in the array.)
If you want to index the values in the array, do this:
for (int i = 0; i < charArr.length; i++) {
and then use i as the index.
Even when you have fixed that, there is still a problem with your code ... for some usecases.
If your input is encoded using UTF-8 (for example) it could include Unicode codepoints (characters) that are greater than 65535, and are encoded in the Java string as two consective char values. (A so-called surrogate pair.) If your string contains surrogate pairs, then isDefined(char) is not a valid test. Instead you should be using isDefined(int) and (more importantly) iterating the Unicode codepoints in the string, not the char values.