Yet another Java CRC32 implementation related issue

Question

My problem is when using this implementation; the Java CRC32 calculation is different from C#'s CRC32 calculation using a look-up table.

The following is the CRC32 code I am using:

public static int CalculateCRCWithTable(byte[] data){
    int crc = 0;
    for (byte b : data) {
        crc = CRCTable[(crc & 0xff) ^ (b & 0xff)] ^ (crc >>> 8) ;
    }
    //crc = crc ^ 0xffffffff; // flip bit/sign
    return (crc);       
}

Reading from file:

public static byte[] readFromFileToByteArray(String fileName) throws IOException {
    BufferedInputStream fis = new BufferedInputStream(new FileInputStream(fileName));
    byte [] bytes = IOUtils.toByteArray(fis);
    for (byte b : bytes){
        b = (byte) (b & 0xff);          
    }
    return bytes;       
}

When testing the Java CRC32 calculation with a simple text file the CRC calculation is (sometimes) correct as in identical to C#'s, but pointing to an executable/binary file then I start seeing different results.

(Moving sample data from comment into question section): String: "simple CRC32 calculation test" Hexadecimal value:

73 69 6D 70 6C 65 20 43 52 43 33 32 20 63 61 6C 63 75 6C 61 74 69 6F 6E 20 74 65 73 74 

Both implemented with look-up table:

• "Expected/Correct" C#'s CRC32 calc: 3347067236
• Java CRC32 calc : -947900060

In a similar thread someone mentioned implementing unsigned integer 32 for Java should fix the difference.

Any pointers would be greatly appreciated and thank you.

SOLVED: Thank you Mark for your verification! With that in mind the below was just needed to get the right CRC32 calc:

String tmp = Integer.toHexString(crc);      
long lCRC = Long.parseLong(tmp, 16);

Answer 1

The two values you provided in your comment (they should be moved to the question) are exactly equal in their low 32 bits. 3347067236 = 2³² - 947900060. They are both 0xc7803164. You don't have a problem.