awk: 0602-562 Field $() is not correct. in a ksh script on AIX

Question

I have a text file that has information for several locations. Each new section in the file has a header. I am trying to take all the information under a single header, strip out the header, and move it to a separate file. All sections move to a separate file based on header information. I am programming this in a shell script using ksh running on aix.

awk '/FILE-HDRPQ/{x=substr ($0,11,6)}NR>1{print $0 > $TRANSDIR"/"x"_prchgrpt.txt";}' $TRANSDIR/$prcfile

FILE-HDRPQ is the first part of the header. $0 contains all columns from that header until the line before the next header. x=an identifying number in the header which is used to create the filename that $0 (minus the top header line) will dump into.

Variables $TRANSDIR and $prcfile help us name the file and place it in the proper directory. These variables are called out earlier in the script.

I am now getting

awk: 0602-562 Field $() is not correct.

when I run the script. Can't for the life of me figure out what is wrong. I have tried playing with single and double quotes.

Answer 1

awk is always trying to interpret $<something> as a field value (e.g. $0 - the whole record, $3 - the third field, $NF - the last field value)

Pass directory name via variable:

awk -v dir="$TRANSDIR" '/FILE-HDRPQ/{x=substr ($0,11,6)}
       NR>1{print $0 > dir"/"x"_prchgrpt.txt";}' "$TRANSDIR/$prcfile"

Answer 2

You should pass the TRANSDIR variable to awk explicitly, like that:

awk -v TRANSDIR="$TRANSDIR"

and then use it like this:

print $0 > TRANSDIR

inside the awk script.