Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > comp.lang.python > #20413
| Date | 2012-02-14 18:05 +0000 |
|---|---|
| From | MRAB <python@mrabarnett.plus.com> |
| Subject | Re: re module: Nothing to repeat, but no sre_constants.error: nothing to repeat ? |
| References | <CABicbJLzLM9As_Ji3EV3kZoV_5_OS129jnFdzZtbgBQTiNBDQA@mail.gmail.com> <CAHzaPEMbxr9K9i_qUouAmrk3VVm0_W23tL_z+9RxO1SQiLOxJA@mail.gmail.com> <CABicbJLpyVhDJfq7omb1n1UnwZoGt3NCJ+fdG8nvHh+a2m5BbA@mail.gmail.com> |
| Newsgroups | comp.lang.python |
| Message-ID | <mailman.5808.1329242751.27778.python-list@python.org> (permalink) |
On 14/02/2012 15:53, Devin Jeanpierre wrote: > On Tue, Feb 14, 2012 at 10:05 AM, Vlastimil Brom > <vlastimil.brom@gmail.com> wrote: >> However, is there any realistic usecase for repeated zero-width anchors? > > Maybe. There is a repeated zero-width anchor is used in the Python re > test suite, which is what made me notice this. I assume that came from > some actual use-case. (see: > http://hg.python.org/cpython/file/096e856a01aa/Lib/test/test_re.py#l599 > ) > > And yeah, even something as crazy as ()* works, but as soon as it > becomes (a*)* it doesn't work. Weird. > I think it's a combination of warning the user about something that's pointless, as in the case of "$*", and producing a pattern which could cause the internal regex engine to get stuck in an infinite loop. It is inconsistent in that it warns about "$*" but not "(?=$)*" even though they are basically equivalent.
Back to comp.lang.python | Previous | Next | Find similar | Unroll thread
Re: re module: Nothing to repeat, but no sre_constants.error: nothing to repeat ? MRAB <python@mrabarnett.plus.com> - 2012-02-14 18:05 +0000
csiph-web