Trying to add multiple grep commands within an exec

Question

I am trying to search all files in a directory (and all subdirectories) for a string. If the file has that string, I then want to search for the number of times another string occurs, in this case "snake" and output it. I would also like to output the lines on which "snake" occurs within the file.

echo "Files with Reptile:"
find . -type f -name "*.txt" -exec grep -l "Reptile" {} \;

My challenge is how I add additional grep commands after the curly brackets. I have tried a couple of things and keep getting a "missing argument to -exec" error.

Am I trying to stuff too much into this find?

Thanks

Answer 1

You could try the following:

$ find . -type f -name "*.txt" -exec sh -c "grep -l "Reptile" {} | xargs -I% grep -Hn snake %" \;
./rep1.txt:2:snake
./rep2.txt:5:another snake

Output contains colon-delimited lists in which the first argument is the file name (from the -H argument to grep), the second argument is the line number on which the desired term appears (from the -n argument to grep), and the third argument is the line itself.

The xargs can be moved outside the find, giving you:

$ find . -type f -name "*.txt" -exec grep -l "Reptile" {} \; | xargs -i grep -Hn snake {}

Note that the -i argument to xargs (which is equivalent to -I{} is deprecated but I use it often for convenience.

Input files:

$ tail -n+1 rep*.txt
==> rep1.txt <==
Reptile
snake
iguana
crocodile
==> rep2.txt <==
Reptile
alligator
turtle
another snake
komodo dragon
==> rep3.txt <==
Reptile
lizard
gecko

If you need to deal with malformed file names, you can think about incorporating print0 and the -0 option to xargs.

Answer 2

Multiple -exec predicates. Since -exec is only true if the command succeeds, you don't need to do anything special with the first grep.

find ... -exec grep -q "Reptile" {} \; -exec grep -Hn snake {} \;

• user001: +1 for pointing out that -exec predicates are conditionally dependent on preceding ones

Answer 3

find . -type f -name "*.txt" -exec sh -c 't=$(mktemp)
 for fi
 do
 if grep -q Reptile "$fi"
 then
 echo "$fi contains Reptile"
 grep snake "$fi" | tee "$t"
 echo "snake appeared on $(wc -l < "$t") line(s)"
 fi
 done
 rm -f "$t"
 ' sh {} +

This reports the names of files that contain the word/string "Reptile", even if they don’t contain the word "snake", and independently reports the number of lines that contain "snake".

Oops; I misread the question, and thought that was a requirement. Taking it out is easy:

find . -type f -exec sh -c '
 for fi
 do
 if grep -q Reptile "$fi"
 then
 echo "$fi contains Reptile"
 grep snake "$fi"
 fi
 done
 ' sh {} +

You can add a -n option to the grep snake if you want.

This should handle just about any valid filename. Check the other answer(s) with filenames beginning with space and/or containing newline.