LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Programming (http://www.linuxquestions.org/questions/programming-9/)
-   -   Filtering the output of a process (http://www.linuxquestions.org/questions/programming-9/filtering-the-output-of-a-process-4175415397/)

littleplane 07-07-2012 08:22 AM

Filtering the output of a process
 
Hi,

I'd like to write a bash script what launches a program with a lot of output (this program doesn't stop unless you kill it, but it runs like a daemon). So I need to filter its output in runtime and kill it if this output contains a specific string. Unfortunatelly

./dameon_program | grep -Ei "string"

doesn't a good solution because it cannot capture this string.
Any idea how can I solve this?

littleplane 07-07-2012 08:59 AM

Okay, the fist part is solved:
./daemon_program 2>&1 | grep -Eio "LOADFAIL"

This command captures the proper string but how can I kill the program if I detect this?
I tried the following:

if [ `./daemon_profram 2>&1 | grep -Eio "LOADFAIL"` != "" ]; then
kill `pidof deamon_program`
fi

But it doesn't work. Ideas?

pan64 07-07-2012 10:02 AM

grep will wait until your daemon_program ends, so you need to tell grep to stop at the first match.
you do not need -E and -o to grep, but you would need -L. Try:
./daemon_program 2>&1 | grep -Li "LOADFAIL" && pkill daemon_program

bigearsbilly 07-10-2012 03:50 AM

You probably need to rethink the problem.


something like this could do:

Code:

SEARCH=profile

find $HOME | while read x;do

    echo $x
    [ "${x#*$SEARCH*}" ] || break # do kill here

done

# or do kill here

obviously replace the find with your daemon

theNbomr 07-10-2012 09:39 AM

bigearsbilly has posted a superior solution, which allows the pipeline to accumulate very little data by breaking the output of the child process into individual lines. I think the other solutions will accumulate the child process output in a buffer until an EOF is received (the child terminates), and then the grep will receive the full complement of buffered data.

--- rod.

pan64 07-10-2012 12:20 PM

Quote:

Originally Posted by theNbomr (Post 4723979)
bigearsbilly I think the other solutions will accumulate the child process output in a buffer until an EOF is received (the child terminates), and then the grep will receive the full complement of buffered data.
--- rod.

I made a "slow" cat:
Code:

while read a
do
    echo $a
    # this line for debug only
    echo $a > /dev/tty
    sleep 1
done < searchfile | grep -L 'pattern'

without -L it will run until the end of searchfile - regardless of the pattern, but with -L it will be stopped after the first match, grep does not wait "any longer". After the match even the while loop will be finished because the pipe has been broken. (Actually the while loop will print out a few additional debug line by then)

grail 07-10-2012 10:20 PM

pan64 - is -L the correct solution? From the man page:
Quote:

-L, --files-without-match
Suppress normal output; instead print the name of each input file from which no output would normally have been printed. The scanning will stop on the first match.
I would suggest either -l (that is lower case 'el') or -m1 for first match

pan64 07-11-2012 02:08 AM

I think they will differ only how grep will exit (what will be the exit code). All of them will stop on the first match.

bigearsbilly 07-11-2012 03:00 AM

Quote:

Originally Posted by theNbomr (Post 4723979)
bigearsbilly has posted a superior solution,

I couldn't agree more ;)


All times are GMT -5. The time now is 08:29 PM.