Multi line replace in SED

Question

I know there are quite a few questions asked on this topic. But I need help in a case basis. This is my scenario. I am having a file with the following string

TRACE(HELLO_WORLD, GEN, INFO, "Hello world")
TRACE(
    HELLO_WORLD, GEN, INFO, "Hello World")

I want to match all the lines containing HELLO_WORLD and in those lines I want to change TRACE to TRACE_CLEANUP in sed.

My sed command sed -i '/HELLO_WORLD/s/TRACE/TRACE_CLEANUP/g' abc.txt matches the first case but not second.

Can you please help get the proper command.

Thanks

Chid

Answer 1

$ sed 'H;1h;$!d;x; s/TRACE\(([[:space:]]*HELLO_WORLD\)/TRACE_CLEANUP\1/g' file
TRACE_CLEANUP(HELLO_WORLD, GEN, INFO, "Hello world")
TRACE_CLEANUP(
    HELLO_WORLD, GEN, INFO, "Hello World")

How it works

• H;1h;$!d;x;

  This reads the whole file in at once.

  If your file were huge (too big for memory), we would want a different approach. If it is not huge, this approach is quite simple.

• s/TRACE\\(([[:space:]]*HELLO_WORLD\\)/TRACE_CLEANUP\\1/g

  This looks for TRACE( followed by any whitespace followed by HELLO_WORLD and substitutes in TRACE_CLEANUP . The final g means that this substitution is done "globally," meaning as many times as needed.

Answer 2

Perl solution:

perl -e 'undef $/;
         $s = <>;
         $s =~ s/TRACE(\(\s*HELLO_WORLD)/TRACE_CLEANUP$1/g;
         print $s;
        ' input > output

• undef $/ makes Perl read the whole input at once.
• <> reads the input.
• s/// works similarly as in sed.