File matches with structured information

Question

The following grep command that returns the files that contain the string ptrn. The sed command skips files using a step stored in stp.

grep -rl "${isufx[@]}" -e "$ptrn" -- "${fdir[@]}" |
  sed "${sta}~${stp}!d" >> $logfl

This would be the result

cat /home/flora/logs/27043-T13:09:44.893003954.log
file1.sh
file2.sh
file3.sh
file4.sh
file5.sh
file6.sh

I would like to end up with the following

cat /home/flora/logs/27043-T13:09:44.893003954.log
File: file1.sh
+ file2.sh
+ file3.sh
+ file4.sh
+ file5.sh
+ file6.sh

I do not think one can do that with grep though. Some will likely recommend using awk on the file logfl.

Answer 1

Using GNU rectools (which you mention in comments to another of your questions):

: >output

recins -f File -v "$(cat logfile)" output

This creates or empties the file called output, and then inserts a single field called File with a value that is the contents of the file logfile.

Note that the + counts as a newline character in the GNU recfile format, so what the above does is to insert a single value, containing embedded newline characters.

If you want to insert each filename as a separate element, you could use

: >output

xargs -I {} recins -f File -v {} output <logfile

The above calls recins once per line in logfile and inserts a separate key-value pair for each.

This could be done more efficiently if we let the input be a CSV file:

{ echo File; cat logfile; } | csv2rec >output

The echo command is used here to prepend the CSV header to the input data for csv2rec.

Answer 2

I solved the problem this way

 grep -rl "${isufx[@]}" -e "$ptrn" -- "${fdir[@]}" |
    sed '1s/^/File: /;1!s/^/+ /' >> $logfl