Problem with SRE's regular expressions

[Christophe Delord]

Thanks for your explanation!

In my real program, I had a more complex regexp. I need to match text 
enclosed in << ... >>
I first have used such an expression : <<([^>]|>[^>])*>>
but I prefered <<.*?>> because it is more readable.

I'll listen to the advice and use my old regexp to see how faster it is 
this way ;-)

Thanks,
Christophe.

Fredrik Lundh wrote:

> Christophe Delord wrote:
> 
>>I'm using regular expressions in python in a parser generator. I have
>>some troubles with the module named 're'. In fact it uses the 'sre'
>>module. When I explicitly use the old 'pre' module it works fine.
>>So my problem is when big strings are scanned using non greedy operator
>>(for example ".*?") 'sre' seems to be recursivle and python stack limit
>>is exceeded.
>>
> 
> the SRE engine is trying to tell you that you're using the wrong
> tool for the task, and probably should think of a better way to
> do the right thing...
> 
> 
>>big_string = "<" + "that's a very very big string!"*1000 + ">"
>>
>>if re.match('<.*?>', big_string):
>>
> 
> here's the same thing, without any backtracking:
> 
>     if re.match('<[^>]*>', big_string):
> 
> under PRE, it's about an order of a magnitude faster than your
> version.  and SRE is over twice as fast as PRE on this one...
> 
> </F>
> 
> 


-- 
Christophe Delord
http://christophe.delord.free.fr/