Java .charAt(i) comparison issue

Question

Why when comparing a char against another it must be taken also from a string? For example;

This does not work

   while(i < t.length() && zeroCount < 5) {
        if(t.charAt(i) == 0){
            zeroCount++;
        }
        i++;
    }

Nor does this

char zero = 0;

      while(i < t.length() && zeroCount < 5) {
            if(t.charAt(i) == zero){
                zeroCount++;
            }
            i++;
        }

The only way I managed to get it working is like this...

String zeros = "0000000000";

      while(i < t.length() && zeroCount < 5) {
            if(t.charAt(i) == zeros.charAt(i)){
                zeroCount++;
            }
            i++;
        }

Can anyone explain if am doing something wrong, or if it is just not acceptable to do it like the top 2 examples. If so, why?

Solution 1

You're confusing

char zero = 0;

with

char zero = '0';

The former is the null-character (ASCII value of zero), whereas the latter is the character representing the digit zero.

This confusion is a rather unfortunate hang-over from C, with char variables being treated as numbers as well as characters.

Solution 2

You're comparing against Unicode value 0 (aka U+0000, the "null" character) - which is not the same as the Unicode character representing the digit 0.

Use '0' instead of 0:

while(i < t.length() && zeroCount < 5) {
    if(t.charAt(i) == '0'){
        zeroCount++;
    }
    i++;
}

Solution 3

You are looking for the character '0'? Then compare to '0', not 0.

Solution 4

The simple answer is that the value 0 is not the same as the character '0' which has an ASCII code of 48 (IIRC).

You should compare it with the char value charAt(i) == '0' or subtract the char before comparison charAt(i) - '0' == 0

Solution 5

Use '0' instead of 0.

View more solutions

Comments

[tchrist]

I’m not comfortable with telling people to use charAt: it’s broken by design.

[tchrist]

I don’t think you should tell people to use the old broken charAt method. Too many problems.

[Ishtar]

Your answer is confusing for someone who doesn't know about charsets. For finding 0 (or any BMP character) there is nothing wrong with charAt, or what problems do you mean?

[tchrist]

@Ishtar: The problems I have are that character counts will be off. Something like "\uD83D\uDCA9\uD83D\uDCA9\u0000\uD83D\uDCA9\uD83D\uDCA9" has only 5 code points in it, with the NUL at number 3. That’s why the Java Pattern ^.....$ matches it: because the Pattern class always deals in logical units, as is proper.