I have the following command in a Bash shell script:

cat file | grep -w $1 | some other commands

If grep does not find any result, an error message is displayed. How do I test if grep fails and still be able to pipe it's output to the rest of the commands?


I have the following solution:

a=`who | grep -w $1`
if [ $? -eq 0 ]
then
display error message
fi
printf "$a" | some other commands


Is this the best solution, or is there another method?
Just in case I have some other set of commands, and the variable is assigned some other output, I read somewhere that its not a good idea to assign variables to a large amount of data, or binary data.

I don't want to run the first portion twice, because there are further error conditions in the subsequent set of commands. The idea is, I have a long sequence of commands, connected with pipes. If the execution fails at grep, then output an error message. After grep, I retrieve further information from the text file, and if some other condition is not met, output another error message, and so on and so forth. There are multiple error conditions in the commands, and if it fails at any point, I would like to know where it fails, and output an appropriate error message.

Hi.

Don't know if this is what you're after, but have you considered rediricting std err to std you, like this:

cat file | grep -w $1 2>&1 | some other commands

Please put code in "Code:" blocks -- it's much easier to read.

What do you want to do if your grep fails?
Do you still want to pipe its output to the rest of the commands, or only if it succeeds?

Nope, piping the output to the rest of the commands is not necessary if grep fails.

Essentially, what I want to do is something like the following:

command1 | command2 | command3 | command4 | .....

If command2 fails for whatever reason, execute some piece of code and exit;
If command4 fails for whatever reason, execute some piece of code and exit, etc etc.


My question is how to place a conditional within the series of pipes to test for the failure of a command, execute some piece of code if it fails, and still be able to pipe the output to the rest of the commands if it succeeds.

This is a quick, untested suggestion. I don't know if the '{', '}' form will even work; or, if it will, if it's preferable to '(', ')'. If there is a preference, it is based on spawning/not spawning subshells.

I must say that I believe that making a shell-script like you suggested in your first post seems like a more neat way of doing this..