regex searching hangs

[Fortepianissimo]

Could someone explains why the following code hangs (Python 2.3.3)?

--- CODE STARTS ---
import re

p=re.compile(r' \S+(?:\.\S+)+\. com')

t='............ ............... ...........'
p.search(t)
--- CODE ENDS ---

From the syntax and semantics of regex, the entire t should be
consumed by the pattern '\S+(?:\.\S+)+' and the search() call should
return None. Is search() doing a depth-first search and can't pulling
itself out? It doesn't seem right to me, or I'm missing something?

BTW, I don't think any regex matching should ever hang in any
circumstances - correct me if I'm wrong: is halting problem a problem
here?

Thanks!

[Jan Erik Breimo]

Re: regex searching hangs


"Fortepianissim o" <fortepianissim o@yahoo.com.tw> wrote in message
news:ef08387c.0 402080729.62163 2ca@posting.goo gle.com...[color=blue]
> Could someone explains why the following code hangs (Python 2.3.3)?
>
> --- CODE STARTS ---
> import re
>
> p=re.compile(r' \S+(?:\.\S+)+\. com')
>
> t='............ ............... ...........'
> p.search(t)
> --- CODE ENDS ---
>[/color]

The search doesn't hang, it just takes a long time to compute - the time it
takes almost doubles for each dot you add to t.
If you replace \S in the expression with [^ \t\n\r\f\v.] (ie. no whitespace
and no dots) you'll get the computation time you expect.

[Rob Renaud]

Re: regex searching hangs

fortepianissimo @yahoo.com.tw (Fortepianissim o) wrote in message news:<ef08387c. 0402080729.6216 32ca@posting.go ogle.com>...[color=blue]
> Could someone explains why the following code hangs (Python 2.3.3)?
>
> --- CODE STARTS ---
> import re
>
> p=re.compile(r' \S+(?:\.\S+)+\. com')
>
> t='............ ............... ...........'
> p.search(t)
> --- CODE ENDS ---
>
> From the syntax and semantics of regex, the entire t should be
> consumed by the pattern '\S+(?:\.\S+)+' and the search() call should
> return None. Is search() doing a depth-first search and can't pulling
> itself out? It doesn't seem right to me, or I'm missing something?
>
> BTW, I don't think any regex matching should ever hang in any
> circumstances - correct me if I'm wrong: is halting problem a problem
> here?
>
> Thanks![/color]


It doesn't halt, it just seems you found an exponential time search,
searching at length i seems to take about 1.6 times the time it takes
to search at length i-1.

import re
import time
p=re.compile(r' \S+(?:\.\S+)+\. com')
t = '.'
for i in range(100):
start = time.time()
p.search(t*i)
print i, time.time() - start

How should you improve it? I don't know, I can't even read the regexp
:)