Why is String.equalsIgnoreCase is so slow

Question

I encountered a question in interview to write a method to check for similar words irrespective of character cases.

I answered it by using the difference of ASCII value for each pair of characters. But at home, when I went through the actual implementation of it in String.class, I get disturbed - Why is it implemented that way!

I tried to draw a comparison between inbuilt and my custom method, this way-

public class EqualsIgnoreCase {

    public static void main(String[] args) {
        String str1 = "Srimant @$ Sahu 959s";
        String str2 = "sriMaNt @$ sAhu 959s";

        System.out.println("Avg millisecs with inbuilt () - " + averageOfTenForInbuilt(str1, str2));
        System.out.println("\nAvg millisecs with custom () - " + averageOfTenForCustom(str1, str2));
    }

    public static int averageOfTenForInbuilt(String str1, String str2) {
        int avg = 0;
        for (int itr = 0; itr < 10; itr++) {
            long start1 = System.currentTimeMillis();
            for (int i = 0; i < 100000; i++) {
                str1.equalsIgnoreCase(str2);
            }
            avg += System.currentTimeMillis() - start1;
        }
        return avg / 10;
    }

    public static int averageOfTenForCustom(String str1, String str2) {
        int avg = 0;
        for (int itr = 0; itr < 10; itr++) {
            long start2 = System.currentTimeMillis();
            for (int i = 0; i < 100000; i++) {
                isEqualsIgnoreCase(str1, str2);
            }
            avg += System.currentTimeMillis() - start2;
        }
        return avg / 10;
    }

    public static boolean isEqualsIgnoreCase(String str1, String str2) {
        int length = str1.length();
        if (str2.length() != length) {
            return false;
        }

        for (int i = 0; i < length; i++) {
            char ch1 = str1.charAt(i);
            char ch2 = str2.charAt(i);

            int val = Math.abs(ch1 - ch2);
            if (val != 0) {
                if (isInAlphabetsRange(ch1, ch2)) {
                    if (val != 32) {
                        return false;
                    }
                } else {
                    return false;
                }
            }
        }
        return true;
    }

    public static boolean isInAlphabetsRange(char ch1, char ch2) {
        return (((ch1 <= 122 && ch1 >= 97) || (ch1 <= 90 && ch1 >= 65)) && ((ch2 <= 122 && ch2 >= 97) || (ch2 <= 90 && ch2 >= 65)));
    }

}

Output-

Avg millisecs with inbuilt () - 14

Avg millisecs with custom () - 5

I found that the inbuilt method is hitting efficiency, as because of lots of checks and method calls. Is there any specific reasons behind such an implementation? Or Am I missing something in my logic?

Any suggestions, will be heartily appreciated!

Answer 1

Your routine only handles ASCII characters. The system one handles all unicode characters.

Consider following example:

public class Test {

    public static void main(String[] args) {
        System.out.println((int) 'ě'); // => 283
        System.out.println((int) 'Ě'); // => 282 
    }

}