pass shell parameters to awk does not work - bash

Question

Why does this work
for myfile in `find . -name "R*VER" -mtime +1`
do
SHELLVAR=`grep ^err $myfile || echo "No error"`
ECHO $SHELLVAR
done
and outputs
No error
err ->BIST Login Fail 3922 err
No error
err ->IR Remote Key 1 3310 err
But this does not
for myfile in `find . -name "R*VER" -mtime +1`
do
SHELLVAR=`grep ^err $myfile || echo "No error"`
awk -v awkvar=${SHELLVAR} '{print awkvar}'
done
and outputs
awk: cmd. line:1: fatal: cannot open file `{print awkvar}' for reading (No such file or directory)
What am I missing?

Answer 1

Does $SHELLVAR contain a space? If so, your awk script is getting misparsed, and the {print awkvar} is being assumed to be a file name and not the actual AWK program.
You also have a problem where both your for loop and the awk program are both slurping STDIN. In fact, your for loop would only be executed once since the AWK will read in all the STDIN until it finishes. In the end, you'll get the same line over and over, and then your program will stop running as the awk awaits for more STDIN.
I think you want to do something like this...
find . -name "R*VER" -mtime +1 | while read myfile
do
echo "$SHELLVAR" | awk '{print $0}'
done
This way, your echo command feeds into your awk which will prevent awk from reading from the find statement.
As an aside, you're better off doing this:
find . -name "R*VER" -mtime +1 | while read myfile
do
...
done
Rather than this:
for myfile in `find . -name "R*VER" -mtime +1`
do
...
done
This is for several reasons:
The command line buffer could overflow, and you'll lose file names, but you'll never see an error.
The find command in the second example must first complete before the for loop can start to execute. In the first example, the find feeds into the while loop.
ADDENDUM
Now that I saw just-my-correct-opinion's answer, I realize what you've really done wrong: You forgot the file name in the awk command:
find . -name "R*VER" -mtime +1 | while read myfile
do
SHELLVAR=`grep ^err $myfile || echo "No error"`
awk -v awkvar=${SHELLVAR} '{print awkvar}' $myfile
done
Now, the question is what exactly are you doing with the awk. You're not printing anything except the value of $SHELVAR for each and every line in the file. That's probably not what you want to do. In fact, why not simply do this:
find . -name "R*VER" -mtime +1 | while read myfile
do
SHELLVAR=$(grep -q "^err" $myfile")
if [ "x$SHELLVAR" != "x" ]
then
echo "$SHELLVAR"
fi
done
That way, you print out $SHELLVAR, but only if $SHELLVAR is empty.
Or, you can use awk to print out only those lines that match your regex:
find . -name "R*VER" -mtime +1 | while read myfile
do
awk '/^err/ {print $0}' $myfile
done

Answer 2

What you're trying to do is possible but ... a bit quirky. Here's an alternative for you:
for f in $(find . -name "R*VER" -mtime +1)
do
echo "$f"
awk 'BEGIN{ec=0} /^err/{ec+=1;print} END{if(ec==0){print "No error"}}' "$f"
done
This way you don't have to worry about grep, don't have to worry about shell variables and can keep your logic all in one place in one language.
Note that this code is only partially tested since I don't have your data files, but the testing I did worked fine.
If you'd like you can even go a step farther and write the whole thing in awk. It is, after all, a full, general-purpose programming language (with, IMO, a far cleaner syntax than bash). This way you avoid the need for find, grep and bash entirely. I won't be writing that script, however. You'll have to pick up the awk man page and read up on file I/O yourself.

Answer 3

You need to quote $SHELLVAR, to prevent the shell from splitting it.
awk -v awkvar="${SHELLVAR}" '{print awkvar}'

Answer 4

You have already got alternative solutions. Regardless of that, I just want to answer your question, ie the thing that you are missing:
awk -v awkvar=${SHELLVAR} '{print awkvar}'
Here awk is seeking to read the input from STDIN. And that's the problem. Awk seeks to read input from STDIN, unless you specify input file(s). The given commands to awk are executed for each RECORD in the input (and by default a record is a line). See man awk to read more on awk.
But here is a hack, if you want it to proceed without any input:
awk -v awkvar=${SHELLVAR} 'BEGIN{print awkvar}'
BEGIN block is executed as soon as the awk is called. And if awk doesn't find any other block except BEGIN, it executes those BEGIN blocks and exits.
I hope you got the problem behind the error, and a quick solution for that as well.