On 2010-11-16, at 09:29, André Bargull wrote:
>> static var PercentPattern = new
>> RegExp("^\\s*(1?\\d?\\d?\\.?\\d*)\\s*%\\s*$");
>> static var NumberPattern = new RegExp("^\\s*(\\d{0,3}\\.?\\d*)\\s*$");
>
> These patterns actually accept any number, because of the \\d*. And the
> percent pattern also accept this string ".%" or simply "%".
>
> Percent pattern: ^\\s*(100(?:\\.0*)?|\\d{1,2}(?:\\.\\d*)?|\\.\\d+)\\s*%\\s*$
> 1) "100", possibly followed by "." and any number of "0".
> 2) Any number in range [0,99], possibly followed by "." and any number of
> "0". (This part allows leading "0", is that ok? For example "01%")
> 3) Or numbers without leading digits as in ".5%"
>
> A similar for the number pattern: ^\\s*(\\d{1,3}(?:\\.\\d*)?|\\.\\d+)\\s*$
Thanks. I rushed this out so Fred could proceed. Clearly it needs more work.
I don't know how rigorous we really have to be on the pattern. We could just
allow any number of digits before/after an optional `.` and then test the
output of parseFloat not being NaN.
If we were using parseInt, a leading 0 could be a problem due to some runtimes
parsing that as octal.
