[SOLVED] String substitution with sed

String substitution with sed

I'd like to do the following transformation within a file (named inputfile) with sed

Searched line :

$CmdVar $ENV\util.exe firstparm secondparm

Searched pattern is util.exe ; firstparm and secondparm are always the last two words. The job consists in commenting the matching whole line with ###, a tag (mytag), ###, the searched pattern, ###, and the two parameters followed by ### again.

So expected output would be :

###mytag###util.exe###firstparm secondparm### $CmdVar $ENV\util.exe firstparm secondparm

i'm thinking of combining some grep/awk and a loop in order to collect relevant information to use sed by specifying line number.

info=`grep -n "util.exe" inputfile | awk '{print $1,$(NF-1),$NF","}'`
if [ ! -z $info ]; then
  tr , '\n' <<<"$info" | while read;
  do
    words=${REPLY}
  # use sed here by splitting $words  
  done
fi

Because of the high number of files i have to handle, this might take a lot of time.

Is there any a more straightforward way to use sed or awk to achieve this? Thanks in advance

Solution

If query is a regex:

awk '
    $0 ~ q {
        printf "###%s###%s###%s %s### ",
            tag, q, $(NF-1), $(NF)
    }
    { print }
' tag='mytag' q='util.exe' inputfile >outputfile

If query is a plain string:

awk '
    index(q) {
        printf "###%s###%s###%s %s### ",
            tag, q, $(NF-1), $(NF)
    }
    { print }
' tag='mytag' q='util.exe' inputfile >outputfile