[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH] regex: fix backreference matching
From: |
Dmitry V. Levin |
Subject: |
Re: [PATCH] regex: fix backreference matching |
Date: |
Mon, 5 Jul 2021 15:12:02 +0300 |
On Tue, Jun 29, 2021 at 11:51:13AM +0300, Egor Ignatov wrote:
> Well, then I have a few questions about matching and capturing
> groups.
>
> 1. "ab" -> "^(a*)*(.)"
> So, from your test case I can assume that:
> regs[0] = (0, 2]
> regs[1] = (0, 1]
> regs[2] = (1, 2]
>
> But if we add backref at the end:
> 2. "ab" -> "^(a*)*(.)\1"
> check_matching matches the whole string "ab",
> this means that the first group accepted 'a' but in fact is empty,
> otherwise it could not match backref later on.
> What is the correct match here? Is check_matching wrong and
> should match only "a" in the 2nd group (as it would be with
> "^(a*)(.)\1")? or should set_regs check for this and shrink the
> match?
My test-regex.c entry for a similar but a bit simplified case was:
/* Test for ** match with backreferences. */
{ "^(a*)*\\1", "a", REG_EXTENDED, 2, { { 0, 0 }, { 0, 0 } } }
I suppose the corresponding entry for your example would be
{ "^(a*)*(.)\1", "ab", REG_EXTENDED, 3, { { 0, 1 }, { 0, 0 }, { 0, 1 } } }
> Next,
> 3. "aaba" -> "^(a*)*(.)\1"
> Again check_matching matches "aaba", then the first group
> is "a", and were the 2nd 'a' goes?
I suppose the corresponding test-regex.c entry for this case would be
{ "^(a*)*(.)\1", "aaba", REG_EXTENDED, 3, { { 0, 4 }, { 0, 1 }, { 2, 3 } } }
--
ldv
- Re: [PATCH] regex: fix backreference matching,
Dmitry V. Levin <=