A more robust parse method #21
Conversation
Cases such as 1.0000001e6 are now handled correctly. scinote and parameters functions are no longer used.
|
I had forgotten to deal with the simplest case possible ("0.0"). Now it should work. |
src/decimal.jl
Outdated
| # Read sign | ||
| s = (str[1] == '-') ? 1 : 0 | ||
| # Unpack scientific notation | ||
| scno = split(str, 'e') |
There was a problem hiding this comment.
IMO longer / more descriptive variable names would be really nice, although I know this repo is bad about this in general (mea culpa...).
There was a problem hiding this comment.
I tried making the variable names more suggestive. I also added checks for excluding double occurrences of '.' or 'e' in the string.
About the exponent separator for scientific notation, originally it was just 'e', but I believe that also 'E' should be allowed since I've seen many tabulated data using it. I just added a call to lowercase at the beginning for this. If you do not agree I can revert it.
* More suggestive variable names * Allow 'E' as a exponent character * Raise an error if '.' or 'e' appear twice in the string to be parsed
The current implementation of the parse method for the Decimal type cannot handle correctly cases such as "1.0000001e6", "30e-2", "0.1234567e-15" or "123456789.1234567899e-11".
I wrote a replacement proposal that handles these cases. It does not treat scientific notation as a separate case, so the function scinote (and the parameters functions) are no longer used.