Stuck on a very basic regex. [Shell]

Prev: Posix fortran and the gnu toolchain
Next: Need help with awk regular expression to process chess games

From: Georg Klein on 11 Apr 2010 14:06

Hi all,

Harry <simonsharry(a)gmail.com> wrote:

> Hi,
>
> My intent here is to be able to look for fully upper-cased words only,
> and filter out the rest.
> I was, thus, expecting the following egrep to fail.
>
> $ echo "Abby" | egrep '^[A-Z]+$'
> Abby
>
> But for some reason, egrep is able to match the extended regex for the
> mixed-case input.
>
> Does anyone know what I'm missing here?
>
> Many thanks in advance,
> /HS
>
> PS:
> I'm using Gnu bash 4.0.23 and Gnu grep 2.5.3 on Fedora 11.

Having read the thread, i think you did run into a egrep bug, that is fixed
at least with 2.5.4. See

http://savannah.gnu.org/bugs/?16179

for a similar problem. I am interested in this, because we have quite a lot
of code relying on this, and I am glad to know that we do not have to change
everything now.

On systems with gnu grep version < 2.5.4 and >= 2.5 i have the following
behavior:

georg(a)ux4ever:~$ locale
LANG=en_US.UTF-8
LC_CTYPE="en_US.UTF-8"
[...]
LC_ALL=
georg(a)ux4ever:~$ egrep --version
egrep (GNU grep) 2.5

georg(a)ux4ever:~$ echo "Abby" | egrep '^[A-Z]+$'
Abby
georg(a)ux4ever:~$ echo "ABBY" | egrep '^[A-Z]+$'
ABBY
georg(a)ux4ever:~$ echo 'The The' | egrep -i '(the) \1'

.... and the last should return a result.

On systems with egrep == 2.5.4 I get the following:

georg(a)mediabox:~$ locale
LANG=en_US.UTF-8
LC_CTYPE="en_US.UTF-8"
[...]
LC_ALL=
georg(a)mediabox:~$ egrep --version
GNU grep 2.5.4

georg(a)mediabox:~$ echo "Abby" | egrep '^[A-Z]+$'
georg(a)mediabox:~$ echo "ABBY" | egrep '^[A-Z]+$'
ABBY
georg(a)mediabox:~$ echo 'The The' | egrep -i '(the) \1'
The The

.... which again shows the right (and expected) behavior. So it is worth to
upgrade to a newer version of grep or to Fedora 12.

hth

--
Georg
Take nothing too serious ...

From: Seebs on 11 Apr 2010 14:26

On 2010-04-11, Harry <simonsharry(a)gmail.com> wrote:
> On Apr 11, 1:17�pm, Sidney Lambe <sidneyla...(a)nospam.invalid> wrote:
>> I'll bet I've killfiled a score of your aliases already.

>> I have most common first names killfiled. Missed "harry".
>> Thanks. This way I eliminate every stinking troll using
>> the name.

> ???

Sidney is not particularly sane. He freaks out at people, killfiles
them for being right when he was laughably wrong, and invents elaborate
conspiracies to explain the fact that he keeps meeting new people who
don't make him feel warm and fuzzy.

Occasionally he says something interesting and topical about shell
programming, but he's of marginal utility at best, because his answers
are nearly always closely tied to his militant belief that anything
that doesn't work precisely the way he first tried it is clearly an
inferior piece of software designed by and for idiots.

In short, keep one of those fifty-pound blocks of salt on hand if you're
gonna read his posts[*].

-s
[*] Because taking it with a grain of salt is unlikely to be strong
enough.
--
Copyright 2010, all wrongs reversed. Peter Seebach / usenet-nospam(a)seebs.net
http://www.seebs.net/log/ <-- lawsuits, religion, and funny pictures
http://en.wikipedia.org/wiki/Fair_Game_(Scientology) <-- get educated!

From: Harry on 12 Apr 2010 01:28

On Apr 11, 11:06 pm, Georg Klein <n...(a)ux.ennuie.org> wrote:
> Having read the thread, i think you did run into a egrep bug, that is fixed
> at least with 2.5.4. See http://savannah.gnu.org/bugs/?16179
> hth
> Georg
> Take nothing too serious ...

You're most definitely an angel, Georg! Can't tell you how terrible I
was feeling about myself, having had to ask such a basic grep/regex
question after having used it for years now. I was thinking that may
be I'm getting too old and senile for programming :-( However, looks
like, I could may be continue for few more years :-)

I'm not being critical here, but I think that grep (and other such
fundamentally important tools... that are verily the life and breath
of *nix users/programmers) ought to be covered by automated regression
testing... may be TDD style. grep is the perfect example of a program
to which TDD could be applied successfully and tremendously
fruitfully. (Note that I'm quite a sceptic when it comes to TDD's
practicality in regular, real-world development/maintenance.)

Thanks much.

From: Harry on 12 Apr 2010 01:28

First | Prev |
Pages: 1 2 3
Prev: Posix fortran and the gnu toolchain
Next: Need help with awk regular expression to process chess games