non-greedy quantifier in lookbehind

[GoogleCodeExporter]

Hi,
I just noticed some irregularity while using lookbehinds containing greedy and 
non-greedy quantifiers, howewer, I don't know the specification for such cases 
(as re doesn't support variable-length lookbehind at all). cf.:

>>> regex.findall(r"(?<=:\S+ )\w+", ":9 abc :10 def")
['abc', 'def']
>>> regex.findall(r"(?<=:\S+? )\w+", ":9 abc :10 def")
['def']

i.e. \S+? somehow doesn't match a single digit; (\S*? does).

I'd expect that both quantifiers would match in the same way in this case.
And (probably?) the pattern would match consistently with the one where the 
lookbehind part is changed to be part of the matched text:

>>> regex.findall(r":\S+ \w+", ":9 abc :10 def")
[':9 abc', ':10 def']
>>> regex.findall(r":\S+? \w+", ":9 abc :10 def")
[':9 abc', ':10 def']
>>> 

(Using regex-0.1.20120613, python 2.7.3, Win 7.)

best regards,
          vbr

Original issue reported on code.google.com by Vlastimil.Brom@gmail.com on 5 Jul 2012 at 5:00

[GoogleCodeExporter]

Fixed in regex 0.1.20120705.

Original comment by re...@mrabarnett.plus.com on 5 Jul 2012 at 6:30

• Changed state: Fixed

[GoogleCodeExporter]

Wow, that was fast...,
 thanks for the instant fix!
       vbr

Original comment by Vlastimil.Brom@gmail.com on 5 Jul 2012 at 7:36