[SOLVED] remove certain lines from file based on start of line except beginning and ending

nwalsh88 · 02-19-2013, 11:13 AM

Hi, I have multiple files which all have multiple entries like below..

Quote:

00..................
06..................
06..................
06..................
06..................
70..................
80..................
00..................
06..................
06..................
06..................
06..................
70..................
80..................
00..................
06..................
06..................
06..................
06..................
70..................
80..................

I want to have a script that will delete all the 00 records except for the first one and delete all the 80 records except for the last one.

Any help will be greatly appreciated.

Thanks

colucix · 02-19-2013, 11:26 AM

Maybe using something like:

Code:

sed '1!{/^00/d};$!{/^80/d}' file

It means don't delete a line starting with 00 if it is the first line in the file (but delete from the rest of the file) and don't delete a line starting with 80 if it is the last line in the file (but delete from the rest).

nwalsh88 · 02-20-2013, 03:02 AM

Thanks for your reply.

When i run the above command this is the result i get:

Quote:

sed: 0602-404 Function 1!{/^00/d};$!{/^80/d} cannot be parsed.

chrism01 · 02-20-2013, 04:14 AM

Works for me, maybe its a version thing

Code:

 sed --version
GNU sed version 4.2.1

colucix · 02-20-2013, 10:00 AM

Indeed, googlin' around it seems an error from UNIX sed. I don't know how to modify it to let it work in not-GNU sed, but until someone more experienced steps in, here is an awk solution:

Code:

awk '/^00/ && !_[00]++; /^80/{rec = $0}; !/^00/ && !/^80/; END { print rec }' file

The difference between the awk and the sed solution is that the former works even if the first 00 line is not the first line of the file and the last 80 line is not the last one, whereas the latter makes these assumptions.

nwalsh88 · 02-20-2013, 10:20 AM

Thanks for your help colucix

I eventually got it working with the help of a colleague.

the script we used was:

Quote:

awk '
BEGIN { zz = ""; ee = "" }
$0 ~ /^00/ && zz == "" { zz = $0 ; print ; next }
$0 ~ /^80/ { ee = $0 ; next }
$0 !~ /^00/ { print }
END { print ee }
' file

exit 0