how to deal with an exit code 1

[Baraka69]

I have an AWK script that is being called within a shell script.

The shell script takes a directory as argument and runs the AWK script in that directory (and all subdirectiories) on a specific file (always the same file name).

This file somtimes has lines that are longer than 3,000 bytes and then my AWK script exits with an error message.

I can get rid of the lines that are too long with an extra SED script, but that script is quite slow, because it needs to unzip the original file, make a file containing only the offending lines, another file without the offending lines, backup the original file and finally zip all 3 files again.

I want my script to run normally until the AWK scripts exits with the error and then (and only then) run the SED script and afterwards re-run the AWK script.

Since I am new to shell scripts I have no clue how to get the return/exit code of a script.

Any help is gretly appreciated.



-----SHELL SCRIPT: (relevant parts) HP-UX-11.00
cd $1
find . -name INPUTFILE.gz | while read filename
do
gzcat $filename | awk -f AWKFILE
done
cd -
-----
-----SED SCRIPT:
cd $1
gunzip INPUTFILE.gz
sed -n '/PATTERN/p' INPUTFILE >> OFFENDINGFILE
sed '/PATTERN/d' INPUTFILE >> GOODFILE
mv INPUTFILE BACKUPFILE
mv GOODFILE INPUTFILE
gzip -9 INPUTFILE BACKUPFILE OFFENDINGFILE
cd -
-----

 

[Boria]

Hi,
try add after 'done' (before 'cd -') lines:
awkcode=$?
if [ $awkcode -ne 0 ]; then
# awk ended which error
else
# no error
fi

by the way. You can suppress output of error message. instead done write done 2> /dev/null

regards Boris

 

[Baraka69]

First of all, thanks for the help Boris. Now new questions come up:
How about testing for success within the success loop?

e.g.
-----SHELL SCRIPT
cd $1
find . -name INPUTFILE.gz | while read filename; do
gzcat $filename | awk -f AWKFILE
awkExitCode=$?
if [ awkExitCode -ne 0 ]; then
# step 1 - rename file
mv INPUTFILE.gz INPUTFILE_original.gz
# step 2 - make file with pattern only
gzcat INPUTFILE_original.gz | sed -n '/PATTERN/p' | gzip > INPUTFILE_PATTERN_only.gz
# step 3 - make file without pattern
gzcat INPUTFILE_original.gz | sed '/PATTERN/d' | gzip > INPUTFILE.gz
# step 4 - run the original awk script again
gzcat INPUTFILE.gz | awk -f AWKFILE
done
cd -
-----

Of course, I want to continue to step 2 only, if step 1 was successful. Especially step 4 should not be run (and overwrite the original file) if step 1 failed.

Question 1
How do I get successive success established? Can I use $? again and again?

Question 2
Is "awkexitcode=$?; if [ awkexitcode -ne 0 ]; then" the same as "if [ $? -ne 0 }; then"?

Question 3
Boris (aka Boria) suggested that you can suppress output of error mesages by substituting "done" with "done 2> /dev/null". Does this work for suppressing the output of the error message of my awk script? I assume it only suppresses the output of the shell script, or more precise the output of the do-done loop.

 

[Boria]

Hi,
Q 1. after each shell command $? means error code of this command.
Q 2. Yes.
Q 3. In your case
while [....]; do
.....
done 2> /dev/null
all error message from all command inside loop 'while' will be suppressed.

Regards Boris

 

[Baraka69]

Thanks again Boris, now I'm set and will hopefully be able to finish my script.