parenthesis

[Michele Simionato]

bokr at oz.net (Bengt Richter) wrote in message 
> For speed, I think it might be faster to count position but not use it to get the
> characters, e.g.,
> 
>  >>> def get_1st(exp):
>  ...     parens=i=0
>  ...     for c in exp:
>  ...         i += 1
>  ...         if c=='(': parens+=1
>  ...         elif c==')':
>  ...             parens -=1
>  ...             if not parens: return exp[:i]
>  ...
>  >>> exp = '(a*(b+c*(2-x))+d)+f(s1)'
>  >>> get_1st(exp)
>  '(a*(b+c*(2-x))+d)'
> 
> This is a little different from Lee Harr's post (which BTW looks to me
> like it depends on the first char being '('). I'd expect the above to run a
> little faster.
> 

You are right:

parenthesized_group:  130-140 microseconds
Addelim:           620-640 microseconds
Harr:               90-95 microseconds
get_1st:            60-65 microseconds

get_1st wins as the faster approach, more than twice my original
version
using lists (which however was not intended for speed).

> There are dangers in drawing too general conclusions from particular
> examples too ;-)
> 

I agree, still it has been instructive to me to compare the various 
approaches: I learned something more that just the fact that this is
not a problem for regular expressions (thing that I suspected).

Thanks to everybody,

                                                Michele