Re: [musl] regex Back reference matching result not same as glibc and tre.

[Szabolcs Nagy <nsz@port70.net>]

* liheng (P) <liheng40@huawei.com> [2020-04-18 11:07:20 +0000]:
> static const char pat[] = "\\(.?\\).?\\1";
> string: "aba";

? is not special in bre

it should be \{0,1\} (i think we support \? as an extension,
but unescaped ? only matches literal ?). try one of

static const char pat[] = "\\(.\\{0,1\\}\\).\\{0,1\\}\\1";
static const char pat[] = "\\(.\\?\\).\\?\\1";

> 
> I tested this pattern by my test case just now.
> 
> musl:
> # ./test
> regexec failed
> test regex failed
> 
> glibc:
> # ./test
> Invalid back reference
> test regex failed
> 
> tre:
> # ./test
> Invalid back reference
> test regex failed
> 
> -----Original Message-----
> From: Florian Weimer [mailto:fw@deneb.enyo.de] 
> Sent: Saturday, April 18, 2020 6:29 PM
> To: liheng (P) <liheng40@huawei.com>
> Cc: Rich Felker <dalias@libc.org>; musl@lists.openwall.com; Xiangrui (Euler) <rui.xiang@huawei.com>; Lizefan <lizefan@huawei.com>
> Subject: Re: [musl] regex Back reference matching result not same as glibc and tre.
> 
> * liheng:
> 
> > static const char pat[] = "(.?).?\\1";
> 
> > This commit reminds me that if i want to use back reference i should 
> > not to tag REG_EXTENDED, but this test case matching still failed.
> 
> 
> Did you change the expression to this for the basic regular expression test?
> 
> static const char pat[] = "\\(.?\\).?\\1";