CODEWITHSUNDEEP
grep
Razer
Razer

Reputation: 8211

How to grep and execute a command (for every match)

How to grep in one file and execute for every match a command?

File:

foo
bar
42
foo
bar

I want to execute to execute for example date for every match on foo.

Following try doesn't work:

grep file foo | date %s.%N

How to do that?

Upvotes: 59

Views: 82112

Answers (6)

Enrique Palacio
Enrique Palacio

Reputation: 1771

There is an interesting command in linux for that: xargs, It allows You to use the output from previous command(grep, ls, find, etc.) as the input for a custom execution but with several options that allows You to even execute the custom command in parallel. Below some examples:

Based in your question, here is how to print the date with format "%s.%N" for each "foo" match in file.txt: 

grep "foo" file.txt | xargs -I {} date +%s.%N

A more interesting use is creating a file for each match, but in this case if matches are identical the file will be override:

grep "foo" file.txt | xargs -I {} touch {}

If You want to concatenate a custom date to the file created

grep "foo" file.txt | xargs -I {} touch "{}`date +%s.%N`"

Imagine the matches are file names and You want to make a backup of them:

grep "foo" file.txt | xargs -I {} cp {} "{}.backup"

And finally for xargs using the custom date in the backupName

grep "foo" file.txt | xargs -I {} cp {} "{}`date +%s.%N`"

For more info about options like parallel execution of xargs visit: https://en.wikipedia.org/wiki/Xargs and for date formats: https://www.thegeekstuff.com/2013/05/date-command-examples/

Extra I have found also a normal for command useful in this scenarios It is simpler but less versatile below are the equivalent for above examples:

for i in `grep "foo" test.txt`; do date +%s.%N; done
for i in `grep "foo" test.txt`; do touch ${i}; done
for i in `grep "foo" test.txt`; do touch "${i}`date +%s.%N`"; done
for i in `grep "foo" test.txt`; do cp ${i} "${i}.backup2"; done
for i in `grep "foo" test.txt`; do cp ${i} "${i}.backup2`date +%s.%N`"; done

Have Fun!!!

Upvotes: 5

Raveen Kumar
Raveen Kumar

Reputation: 71

    grep search_string files_to_search | sh

Upvotes: 0

Emmanuel Blot
Emmanuel Blot

Reputation: 58

grep may need --line-buffered option to emit each matching line when it matches it, otherwise it buffers up to 4K byte before printing match lines, which defeats the goal here, e.g.

tail -f source | grep --line-buffered "expression | xargs ...

Upvotes: 1

James
James

Reputation: 79

grep command_string file | sh -

Upvotes: 7

Marko Kukovec
Marko Kukovec

Reputation: 711

What you really need is a xargs command. http://en.wikipedia.org/wiki/Xargs

grep file foo | xargs date %s.%N

example of matching some files and converting matches to the full windows path in Cygwin environment

$ find $(pwd) -type f -exec ls -1 {} \; | grep '\(_en\|_es\|_zh\)\.\(path\)$' | xargs cygpath -w

Upvotes: 25

Phil H
Phil H

Reputation: 20151

grep file foo | while read line ; do echo "$line" | date %s.%N ; done

More readably in a script:

grep file foo | while read line
do
    echo "$line" | date %s.%N
done

For each line of input, read will put the value into the variable $line, and the while statement will execute the loop body between do and done. Since the value is now in a variable and not stdin, I've used echo to push it back into stdin, but you could just do date %s.%N "$line", assuming date works that way.

Avoid using for line in `grep file foo` which is similar, because for always breaks on spaces and this becomes a nightmare for reading lists of files:

 find . -iname "*blah*.dat" | while read filename; do ....

would fail with for.

Upvotes: 53

Related Questions