topic Re: Replace two lines with one using AWK in Operating System - Linux

Replace two lines with one using AWK

JUP — Mon, 10 Jan 2005 01:46:12 GMT

Hi

I need to replace two lines with one. I tried SED but looks like SED does not accept Ctrl chars as input (eg. linefeed) as required for replacing the two lines.

I need to scan text files and replace the following:

#ifdef ABC
#123456

with

# Test

There is a carriage return between the #ifdef and the #123456.

Have tried with awk without any luck - can someone suggest how to do this.

Thanks in advance
JUP

Re: Replace two lines with one using AWK

NiCK_76 — Mon, 10 Jan 2005 02:30:45 GMT

Hi,

sed -e 's/#ifdef ABC//g' -e 's/#123456/# Test/g' yourfilename

NiCK

Re: Replace two lines with one using AWK

JUP — Mon, 10 Jan 2005 17:07:25 GMT

Thanks Nick

however your SED command leaves a blank line in the file. In total I wish to replace 4 lines with 1 so I will have 3 blank lines.

The only way to do this is with AWK - I have come up with this command but its not working. It is:

awk '{gsub("#ifdef ABC"\r\n#123456, "# Test"); print $0}' file.h > file.out

Does anyone know whats going wrong here ?

Thanks in advance

Re: Replace two lines with one using AWK

Stuart Browne — Mon, 10 Jan 2005 20:56:20 GMT

Awk is literally 'line-by-line'. It can't pattern match over multiple lines.

Perl can (although in my quick tests, I couldn't get it to work (phooey)), so this leaves doing it the hard way.

awk '
END {
print THIS
}
{
LAST=THIS
THIS=$0
if (SKIP == 1) {
SKIP=0
next
}
if ((LAST == "#ifdef ABC") && (THIS == "#123456")) {
print "#Test"
SKIP=1
} else {
print LAST
}
}' < yourfile

I'm sure there's a neater, shorter way out there though.