Page 1 of 1
Posted: Mon Jul 13, 2015 5:51 am
by ShaneMuir
I assume that this cannot be done at source and/or target?
Posted: Mon Jul 13, 2015 5:56 am
by ShaneMuir
Failing that - doesn't the E represent the number of places to move the decimal?
So using the E as a delimiter split your values into the digit (the first part) and the exponential and just multiply them out?
6.4443433E7 = 6.4443433*10^7
Posted: Mon Jul 13, 2015 6:22 am
by ArndW
The data cannot be correctly read in at the source (thanks to an excel source file with 600K rows and the way that the unstructured data stage works).
The "E" is indeed base 10 exponential; I thought about the mathematical approach and think that it will take longer than the string approach.
Code: Select all
svTrimNoLeading[1,LEN(svTrimNoLeading)-2]*(10**Right(svTrimNoLeading,1)
I should add that the output value is a string column, so the computation above still needs to be converted from a decimal/numeric to a string using some method such as the embedded trims above.
Addendum - I just ran a test and the original solution is faster.
Posted: Mon Jul 13, 2015 6:36 am
by ShaneMuir
So you cannot just convert to dfloat to decimal, then decimal to string and suppress zeros?
Posted: Mon Jul 13, 2015 7:01 am
by ArndW
The decimal-to-string conversion just has the "fix_zero" option to allow a 0 value, but no suppress zeroes.
Posted: Mon Jul 13, 2015 7:11 am
by ShaneMuir
When did this happen. Did they remove the "suppress_zero" option?
Posted: Mon Jul 13, 2015 7:48 am
by ArndW
I can't find it in my documentation for the DecimalToString.
Posted: Mon Jul 13, 2015 7:53 am
by priyadarshikunal
Did you try AsInteger()? I think it works with dfloat too. Truncates the mantissa (decimal part) though.
Posted: Mon Jul 13, 2015 8:19 am
by ArndW
Yes, I did try AsInteger(). It does work with dfloat, but converts to a binary representation and thus isn't feasible in this context.
Posted: Mon Jul 13, 2015 8:25 am
by ShaneMuir
ArndW wrote:I can't find it in my documentation for the DecimalToString.
I am pretty sure its still there. It would be "suppress_zero"
Posted: Mon Jul 13, 2015 9:49 am
by ArndW
Cool - the statement is now
Code: Select all
DecimalToString(DFloatToDecimal(svtest),"suppress_zero")
and that is faster and easier to understand.
Posted: Mon Jul 13, 2015 9:55 am
by ShaneMuir
ArndW wrote:that is faster and easier to understand. ...
Excellent! And good to know that its faster.