Bug 15851

Summary:	Sed ranges are broken in non-C locales
Product:	ALT Linux Desktop	Reporter:	Julia Jomantaite <juliette>
Component:	bugs	Assignee:	Anton V. Boyarshinov <boyarsh>
Status:	CLOSED WONTFIX	QA Contact:	Andrey Cherepanov <cas>
Severity:	normal
Priority:	P2
Version:	4.0.2
Hardware:	all
OS:	Linux

Description

Julia Jomantaite 2008-05-30 18:47:52 MSD

When i use sed ranges (like [a-z]) in locales other than "C" uppercase letters
are matched by lowercase ranges.
In "C" locale this problem doesn't appear.

In other linux distributions (tested on Debian, Gentoo, Fedora) lowercase ranges
match only lowercase letters as one would expect. It seems this is because they
configure sed with --without-included-regex, while Alt does not.

Other utilities (like grep and tr) are not affected by this problem.
Steps to Reproduce:
1.Make sure that locale is not "C" (for example "ru_RU.UTF8")
2.echo abcdABCD | sed -e "s/[a-c]/0/g"
Actual Results:  
000d00CD

Expected Results:  
000dABCD

Comment 1

Andrey Rahmatullin 2008-05-30 19:42:29 MSD

https://bugs.gentoo.org/show_bug.cgi?id=149526#c4

*** This bug has been marked as a duplicate of 13870 ***

Comment 2

Julia Jomantaite 2008-05-30 20:37:09 MSD

(In reply to comment #1)
> https://bugs.gentoo.org/show_bug.cgi?id=149526#c4
> 
This problem has been solved in Gentoo (key --without-included-regex is used now).

Anyway sorting order should be the same for all utilities.

The following results are from the same locale:

[altlinux@localhost ~]$ echo abcdABCD | sed -e "s/[a-c]/0/g"
000d00CD
[altlinux@localhost ~]$ echo abcdABCD | grep -o "[a-c]"
a
b
c
[altlinux@localhost ~]$ echo abcdABCD | tr "[a-c]" 0
000dABCD

It can be obtained in both ru_RU.UTF-8 and ru_RU.KOI8-r locales so it's not the
UTF8 issue.

Comment 3

Mikhail Gusarov 2008-06-13 13:00:03 MSD

Comment 4

Andrey Cherepanov 2020-07-09 12:03:20 MSK

Поддержка дистрибутива завершена. Используйте новые версии.