Using sed in ksh script with condition

I'm using a sed command for replacing some words in text files. This command is embedded in a ksh script. I would like to reach that not everytime will every replacement rule of sed fire, only if some conditions are filled. E.g. depending on a value of a shell variable.

In other words I would like to write this script without the if statement, I would rather include the conditional expression 开发者_Python百科inside the sed. Is it possible?

REPLACE_A=TRUE

if [ "$REPLACE_A" = "TRUE" ]
then
    cat myfile \
    | sed 's/A/B/g;
          's/C/D/g;
          's/E/F/g;'
else
    cat myfile \
    | sed 's/C/D/g;
          's/E/F/g;'    
fi

The only solution that pops into my mind is to store the sed command in a variable, like

# in your script
SEDCMD=""
# ...
SEDCMD="s/A/B/g;"
# later 
SEDCMD="${SEDCMD}s/C/D/g;"
# ...
# finally
sed "$SEDCMD" FILE

But it still not the solution you want.

You can't do it with sed, but with awk. If you like to try, I suggest you have a look at this:

http://www.grymoire.com/Unix/Awk.html