can anyone help me on this one (sed)

nukey · 08-17-2005, 09:20 AM

I have 50 lines of text, all containing this pattern:

random1-random2-random3-random4

All the "randoms" contain letters and numbers
None of the "randoms" have the same number of characters.

I want to take away random 2 3 and 4 and the dash before random2

random1 sometimes also contains a dash, so i started with sed and working from the end of the line by using a $

sed s/-.$//g for example gets rid of the part of the line where a dash is followed by only one character. But I want any number of character after the dash removed, same goes for random2 and random3.

After a google search I noticed that you can use the + symbol to define "any number of characters, so i tried:

sed s/-[a-z][0-9]+$//g

But that didn't work

sed s/-[a-z]+$//g doesn't even remove the characters after the dash

I'm kinda lost here and missing something I guess, any thoughts ?

PenguinPwrdBox · 08-17-2005, 10:40 AM

Your quantifier was in the wrong place.
It should read:

Code:

sed s/-[a-z]+$//g

or, the cleaner way to do it is:

Code:

sed s/-\w*$//g

nukey · 08-17-2005, 10:53 AM

Thanks for replying but it doesn't seem to work

$ echo "random1-random2-random3-random" | sed s/-+[a-z]$//g
$ echo "random1-random2-random3-random" | sed s/-[a-z]+$//g
$ echo "random1-random2-random3-random" | sed s/-\w*$//g
all return:

random1-random2-random3-random

And i removed the number from random4 since we are only removing letters here, but every character should be removed, maybe that can be done easier.

nukey · 08-17-2005, 05:55 PM

More sed-experts on this forum ?

nukey · 08-18-2005, 07:33 AM

Got it, not nice, but it works

sed 's/$.*$-.*/\1/' | sed 's/$.*$-.*/\1/' | sed 's/$.*$-.*/\1/'

keefaz · 08-18-2005, 07:51 AM

Why not use cut ?

Code:

echo "random1-random2-random3-random" | cut -d'-' -f1

Maybe I don't well get it

nukey · 08-18-2005, 08:02 AM

Quote:

Originally posted by keefaz
Why not use cut ?

Code:

echo "random1-random2-random3-random" | cut -d'-' -f1

Maybe I don't well get it

cut could work, but as i said, in some lines random1 also contains a dash, so it doesn't work for lines where random1 contains a dash.

keefaz · 08-18-2005, 08:04 AM

Is the dash at the end or at the beginning of random1 or at random position ?
[edit]
Sorry, if you want to keep the dash in random1, your solution seems better than cut

nukey · 08-18-2005, 08:05 AM

Quote:

Originally posted by keefaz
Is the dash at the end or at the beginning of random1 or at random position ?

ra-ndo-m1-random2-random3-random4
r-a-n-dom1-random2-random3-random4

I think you get the picture

keefaz · 08-18-2005, 08:08 AM

Maybe :

Code:

echo "r-a-n-dom1-random2-random3-random4" | sed 's/\(.*\)-\(.*\)-\(.*\)-.*/\1/'

nukey · 08-18-2005, 08:10 AM

Quote:

Originally posted by keefaz
Maybe :

Code:

echo "r-a-n-dom1-random2-random3-random4" | sed 's/$.*$-$.*$-$.*$-.*/\1/'

Yup, thanks, way shorter

theYinYeti · 08-18-2005, 09:34 AM

From your spec, here's what I'd write:

Code:

sed 's/-[^-]*-[^-]*-[^-]*$//'

Yves.