Mailinglist Archive: opensuse-programming (68 mails)

[Mark Gray <markgray-temp-1072781312@xxxxxxxxxxxx>]

Trey Gruel <drathos-suse@xxxxxxxxxxxxxx> writes:

>> Secondly, it is almost always a mistake to try to do all your work
>> using a single regex -- use multiple regex's and program logic based
>> on the results.

> One of the reasons I'm trying to move things to regex is so the simpler 
> checks can be edited in config files that can be reloaded on the fly.

Have you considered embedding a perl interpreter in your program?  Not
such overkill if you have a lot of configuration that you would like
to be that flexible.  (For one regex, it would be overkill though:-)

> If this check ends up being a problem, I may end up making an option for a
> "negative check" where it'll match invalid values.  Prolly not a bad idea 
> anyway..

This is what the perl cookbook 2nd edition (page 61) gives for a VALID
C float:

        ^([+-]?)(?=\d|\.\d)\d*(\.\d*)?([Ee]([+-]?\d+))?$

and pcretest seems to like it:

markgray@soyo:/ARK/BOOKS/perlcookbook2/ch02> pcretest 
PCRE version 4.4 21-August-2003

  re> /^([+-]?)(?=\d|\.\d)\d*(\.\d*)?([Ee]([+-]?\d+))?$/
data> 123.45
 0: 123.45
 1: 
 2: .45
data> -123.45
 0: -123.45
 1: -
 2: .45
data> unkn
No match
data> NaN
No match
data> abcdef
No match
data> +123.456e+02
 0: +123.456e+02
 1: +
 2: .456
 3: e+02
 4: +02
data>