[SOLVED] sed use - extract substring from lines of text file

tastiero · 03-07-2012, 10:17 AM

hello everybody.
i'm looking for a code to use sed command to extract a substring from the lines of a text file, like the example:

.......
ew7yweryu
gh34234hj3.cfg 23l
34gh5435385h45f.cfg gf887
sd634g6347.cfg 775t
3ff83g84tg34tt4t9.cfg 534h
f7s87d6g78g.cfg 4774t
f4fh7384t34334tsdf.cfg 5747
dsfahjr8
.......

i would cut the end of line start from the space only for the lines that contains filename.cfg and cut the oethers like this:

gh34234hj3.cfg
34gh5435385h45f.cfg
sd634g6347.cfg
3ff83g84tg34tt4t9.cfg
f7s87d6g78g.cfg
f4fh7384t34334tsdf.cfg

can anyone help me about the code?
thanks a lot.

grail · 03-07-2012, 11:17 AM

What have you tried? Here is a link if you do not know where to start: http://www.grymoire.com/Unix/Sed.html

tastiero · 03-07-2012, 11:31 AM

i've tried something like this but it's little test, i miss the part to cut the end of line after the space after .cfg
sed -rn '\|cfg|p' output.txt

tastiero · 03-07-2012, 12:32 PM

i've tried also this:

sed -nr '/^[a-z0-9]+[.]+[a-z]+$/ p' output.txt

but i can only see the lines that contains only filename.cfg. the others lines (e.g. filename.cfg 54434) doesn't appear on output.

please help me. i'm new.

crts · 03-07-2012, 12:44 PM

Hi,

this worked for your sample file:

Code:

$ cat file
ew7yweryu
gh34234hj3.cfg 23l
34gh5435385h45f.cfg gf887
sd634g6347.cfg 775t
3ff83g84tg34tt4t9.cfg 534h
f7s87d6g78g.cfg 4774t
f4fh7384t34334tsdf.cfg 5747
dsfahjr8

$ sed '/\.cfg/!d;s/ .*//' file
gh34234hj3.cfg
34gh5435385h45f.cfg
sd634g6347.cfg
3ff83g84tg34tt4t9.cfg
f7s87d6g78g.cfg
f4fh7384t34334tsdf.cfg

Let us know if that is what you are looking for.

PS: Welcome to LQ.

tastiero · 03-07-2012, 03:06 PM

it works very very fine, it's just that i want. thank you very much.

grail · 03-07-2012, 10:05 PM

And another alternative:

Code:

sed -n '/cfg/s/ .*//p' file

cbtshare · 03-07-2012, 11:15 PM

Similar solution with explanation for people who need it

Quote:

sed -e '/.cfg/!d; s/ .*/ /'

/.cfg/ - will match any line with the .cfg and !d says do not delete such a line when it is found.

s/ .* - says substitute .(a character) * (repeat a previous character) and substitute it with a blank line // or / /
(s/ .*/ /)

grail · 03-08-2012, 12:20 AM

@cbtshare - I think you need to look at your definitions again as they are not correct (completely) ... eg. /.cfg/ - will not just match .cfg but also Acfg as the dot (.) here means the same as in your
second explanation.

tastiero · 03-08-2012, 01:29 AM

thanks to everybody for the help.

tastiero · 03-08-2012, 02:07 AM

another little help.
now i have the right file:

asdfa23e23def.cfg
df73rdfdf.cfg
9h9fdhisfgd.cfg
......
......

how can i insert at the start of each line a string of web address:

http://mysite/configuration/asdfa23e23def.cfg
http://mysite/configuration/df73rdfdf.cfg
http://mysite/configuration/9h9fdhisfgd.cfg
......
......

thanks again

grail · 03-08-2012, 09:09 AM

Did you look on the link I provided?

cbtshare · 03-08-2012, 10:29 PM

@grail , thanks for the correction.

tastiero · 03-09-2012, 01:59 AM

i've tried
sed 's/^/http:\/\/mysite/' output.txt

and no problem but when i use metacharacters of web link e.g. / . or web link that starts with numbers i have problems about syntax or about output results.
i've tried to put the web link in a string ant then pass it to the command e.g. $address, but same problems.

i try again a try to learn, i like to learn.
somebody can help me?
thanks.

tastiero · 03-09-2012, 02:43 AM

i've found this good pages.
http://www.ibm.com/developerworks/li...ed1/index.html
http://www.ibm.com/developerworks/li...ed2/index.html

i'm learnig