from the jeffrey friedl report:

(?>...) is not supported

[this is a "stand-alone pattern".  the engine has code 
for this, but the parser doesn't recognize this yet.  
shouldn't be too hard to fix; I just need a couple of 
good test cases before I start /F]

Also referred to as an atomic group: see
http://www.regular-expressions.info/atomic.html for a discussion.

Fredrik, when you say "the engine has code for this", what do you mean?

There's also an alternate syntax for this, called possessive
quantifiers: a*+, a++, a{m,n}+.

Fredrik,

If you're still listening, I am gonna try and tackle this one but I 
would like to know why you or the famous Jeffrey of the Regexp world 
claims that there is already code in the Regexp Engine for Atomic 
Grouping?  Adding a hook for (?>...) should be trivial but I don't wanna 
re-invent the wheel if the proper stack-unwind code already exists.  
Thanks.  Of course, Andrew (a.k.a. A.M. Kuchling) already asked this 
question and you did not answer it, so I guess you're not reading this, 
but if you are, please respond.  Thanks!

I'm digging into the sre_parse.py at the moment and this I have all the 
changes I need for that now.  The rest of the changes I believe are in 
either sre_compile.py or more likely directly in _sre.c, so I will 
examine those files next.  I am attaching a single diff for 
expedience.  This is not an official patch, just a sample to see the 
progress I am making.  I forgot the correct format for patch files but 
I promise to get it right when I have made more progress.

Quick update (see thread 2636 for other updates to the Regex
functionality): I do have a design in my mind for how to accomplish both
Atomic Grouping and Possessive Qualifiers and it would work with either
the existing Engine design or the new design I have been playing with. 
It's just a matter of getting around to implementing it, testing it and
documenting it.

I currently am having problem with Bazaar / Launchpad which are beyond
the scope of this issue but are effecting my work.  Those issues can be
found on the Launchpad site if you look for Answers / Bugs originating
from "timehorse".

I have finished work on the Atomic Grouping / Possessive Qualifiers 
support and am including a patch to achieve this; however, 
http://bugs.python.org/issue2636 should be consulted for the complete list 
of changes in the works for the Regexp engine.

Does the re2 work (#2636) make this outdated?

Implemented in #2636.

Some unrelated to the topic changes from the patch are committed in 30a6c74ad87f.

Reopened because definitely regex will be not adopted in 3.5.

Here is updated to 3.5 Jeffrey's patch.

New changeset 345b390ed69f36681dbc41187bc8f49cd9135b54 by Serhiy Storchaka in branch 'main':
bpo-433030: Add support of atomic grouping in regular expressions (GH-31982)
https://github.com/python/cpython/commit/345b390ed69f36681dbc41187bc8f49cd9135b54