How Java gets the number of decimal digits of a double

Objective

Look at the title is not think this is a very simple problem, I think so at the beginning, but the actual situation, in various cases we have tested, found that many of the actual situation can not be unsatisfactory.

Method analysis

What you can think of now is easier to have the following several
1, directly using double processing
2. Convert double to string for processing

Method One: The double is processed directly, the calculation is calculated by the result of the modulo operation to obtain the scale, as follows:

     Public Static intGetnumberdecimaldigits (DoubleNumber ) {        if(Number = = (Long) {return0; }        inti = 0;  while(true) {i++; if(Number * MATH.POW (ten, i)% 1 = = 0) {                returni; }        }    }

Method Two: After the double is converted to a string, then use the "." The segmentation is judged as follows: (here a double converts a string there are several ways we choose one of these)

 Public Static intGetnumberdecimaldigits (DoubleNumber ) {String Moneystr=string.valueof (number); string[] Num= Moneystr.split ("\ \.")); if(Num.length = = 2) {         for (;;) {            if(Num[1].endswith ("0") ) {num[1] = num[1].substring (0, Num[1].length ()-1); }Else {                 Break; }        }        returnNum[1].length (); }Else {        return0; }}

Test method:

 Public Static voidMain (string[] args) {System.out.print (Getnumberdecimaldigits (0)); System.out.print ("" + getnumberdecimaldigits (0.0)); System.out.print ("" + getnumberdecimaldigits (1)); System.out.print ("" + getnumberdecimaldigits (-11.1)); System.out.print ("" + getnumberdecimaldigits (1.01)); System.out.print ("" + getnumberdecimaldigits (0.0100)); System.out.print ("" + getnumberdecimaldigits (0.123456)); System.out.print ("" + getnumberdecimaldigits (11111111.1234)); System.out.print ("" + getnumberdecimaldigits (11.0123456789));}

Results analysis

Test results:
Method One: 0 0 0 1 2 2 6 5 11
Method Two: 0 0 0 1 2 2 6 13 10

Why is the final result different from what we imagined???
There are three reasons:
1, double calculation loss accuracy:
2.01 * 10 = 20.099999999999998

2, double to string loss of precision:
String.valueof (11111111.1234) = 1.11111111234E7
11111111.1234 + "" = 1.11111111234E7
Double.valueof (11111111.123) = 1.1111111123E7
...

Summarize

So in the current situation, there are a variety of problems when judging with double, because of the problem of double precision.
So the only solution is, do not use a double, directly using the string, whether it is the value passed from the client or the production value is directly using the string for processing and judgment, the method is similar to the method two.
If you do not want to make calculations, use BigDecimal for calculations. New BigDecimal. Note Be sure to use a string as a parameter for the new

If you have a better way, can pass the above test, then hurriedly comment let me worship worship ~

Problems raised and resolved abroad:
Https://stackoverflow.com/questions/6264576/number-of-decimal-digits-in-a-double?utm_medium=organic&utm_ Source=google_rich_qa&utm_campaign=google_rich_qa

How Java gets the number of decimal digits of a double