Re: [Regexp] a regexp question

[Levent Yilmaz]

Thanks for the thinking Thomas!

Yes I have a unix like environment; cygwin, inparticular. And you 
wouldn't guess how many workarounds I tried for this problem :)
And unfortunately what you recommended here is one of the first things 
that I have tried that did not work.

If you give it a second thought it does not work for more general 
examples, since all lines without 'puppy.*kitty' are not the lines with 
kitty. That is,

grep -v "puppy.*kitty" *

Also matches a line such as

no pets allowed :)

But I want a line that contains kitty. Actually, FYI, what I needed is 
to make grep return the name of the files that contain one or more of 
such lines (no puppy but kitty lines).
Well, I tried many things (with and without regexp) and this is the most 
decent solution I could get:

(in pseudo commands)
find all <kitty>s and feed the pipe with lines in the form
        <filename: the line that has kitty> |
drop the lines with <puppy.*kitty> |
remove everything but the <filename>s |
remove multiple occurances of the <filename>s

(in real unix commands)
grep -i "kitty" * | \                         
grep -iv ":.*puppy.*kitty" | \
sed 's/:.*//' | \
uniq


Well this does the job. But is barely useful as things get complex. So I 
really need an elegant regexp to takes care of this.


FYI, the actual reason why I need such a thing is to find in a list of 
source files(FORTRAN), all occurances of a variable other than where it 
is declared.

Thank you and best of my greetings!
-Levent


Hartwig, Thomas wrote:
> Forget my second email, this was a silly one. Of course you could do 
> something like grep -v "puppy.*kitty" always without sed, but that is 
> not a fine regex of course.
>
> Thomas
>
> Hartwig, Thomas wrote:
>
> > If you have a sort of unix this one could help you:
> >
> > sed 's/puppy.*kitty/§§§/' | grep -v "§§§"
> >
> > §§§ is just a sample.
> >
> > Thomas
> >
> > Levent Yilmaz wrote:
> >
> > > Hello,
> > >
> > > It sounded like a very simple problem but I couldn't come up with a 
> > > solution. I was trying to do this with Basic Regexp in grep:
> > >
> > > Find all lines which contain the word 'kitty'. But only those lines 
> > > that do not have 'puppy' somewhere before 'kitty'. For instance:
> > >
> > > my kitty is very cute
> > > but puppy is cuter than kitty
> > >
> > > What regular expression stands for the first line but not the second? 
> > > Note that the other way is very easy, that is the lines with 'puppy' 
> > > followed by 'kitty': puppy.*kitty
> > >
> > > thank you so very much!
> > > -Levent.
> > >
> > >
> > > PS: What is all this spam on the list by the way?
> > >
> > >
> > >
> > > _______________________________________________
> > > Gnu-regexp-users mailing list
> > > address@hidden
> > > http://mail.gnu.org/mailman/listinfo/gnu-regexp-user
> >
> >
> > s
> >
> > _______________________________________________
> > Gnu-regexp-users mailing list
> > address@hidden
> > http://mail.gnu.org/mailman/listinfo/gnu-regexp-users
>
>
> _______________________________________________
> Gnu-regexp-users mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/gnu-regexp-users