From: Tom Lane <tgl@sss.pgh.pa.us>
To: Laurenz Albe <laurenz.albe@cybertec.at>
cc: Dominique Devienne <ddevienne@gmail.com>, pgsql-general@postgresql.org
Subject: Re: Latest patches break one of our unit-test, related to RLS
In-reply-to: <2c58dcdba889f5b28df5ad9d21b5ea2d0ac63a9a.camel@cybertec.at>
References: <CAFCRh-8NwJd0jq6P=R3qhHyqU7hw0BTor3W0SvUcii24et+zAw@mail.gmail.com> <CAFCRh-93hsRvoJqQCwXq06=AJJJ_s783gVsGHKc-QjgOO6jGVg@mail.gmail.com> <77a63548783dd4007ee479a4c5ed300629aaa776.camel@cybertec.at> <CAFCRh--DPWEHpRmdr-Jfw4guYqjc9TM_QyVH7SR4B=Le5gGe3Q@mail.gmail.com> <CAFCRh--M1y2YpJJMb6undzvyvhd4geKtRWFkDyr1shHvvoOkpQ@mail.gmail.com> <CAFCRh-_OBKtabe8MO0vUocNSq_O3_b852AXknPiNhL0tNUBGQg@mail.gmail.com> <CAFCRh-_U+8fmRoc-sYM5ahp=9dNjs=Dri1LFLcHCgEJX8Zq_oA@mail.gmail.com> <2109533.1757686026@sss.pgh.pa.us> <2c58dcdba889f5b28df5ad9d21b5ea2d0ac63a9a.camel@cybertec.at>
Comments: In-reply-to Laurenz Albe <laurenz.albe@cybertec.at>
	message dated "Sat, 13 Sep 2025 01:34:08 +0200"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <2673229.1757722323.1@sss.pgh.pa.us>
Date: Fri, 12 Sep 2025 20:12:03 -0400
Message-ID: <2673230.1757722323@sss.pgh.pa.us>
Archived-At: <https://www.postgresql.org/message-id/2673230.1757722323%40sss.pgh.pa.us>
Precedence: bulk

Laurenz Albe <laurenz.albe@cybertec.at> writes:
> On Fri, 2025-09-12 at 10:07 -0400, Tom Lane wrote:
>> The _ and % are not getting converted to their POSIX equivalents
>> ("." and ".*").

> Indeed, and I have to take the blame for introducing a bug in a minor
> release :^(
> The attached patch should fix the problem.

I had not particularly studied the new charclass-parsing logic.
Looking at it now, this bit further down (lines 871ff) looks
fishy:

            if (pchar == ']' && charclass_start > 2)
                charclass_depth--;
            else if (pchar == '[')
                charclass_depth++;

            /*
             * If there is a caret right after the opening bracket, it negates
             * the character class, but a following closing bracket should
             * still be treated as a normal character.  That holds only for
             * the first caret, so only the values 1 and 2 mean that closing
             * brackets should be taken literally.
             */
            if (pchar == '^')
                charclass_start++;
            else
                charclass_start = 3;    /* definitely past the start */

Should not we be setting charclass_start to 1 after incrementing
charclass_depth?  That is, I'd be more comfortable if this logic
looked like

            if (pchar == ']' && charclass_start > 2)
                charclass_depth--;
            else if (pchar == '[')
            {
                /* start of a nested character class */
                charclass_depth++;
                charclass_start = 1;
            }
            else if (pchar == '^')
                charclass_start++;
            else
                charclass_start = 3;    /* definitely past the start */

I haven't experimented, but it looks like this might misprocess
^ or ] at the start of a nested character class.

			regards, tom lane