Shell Programming and Scripting

Hi There,

First time poster here. I've got a parsing question and have a solution but am sure there is a much better way of doing this with just awk. My knowledge of awk is pretty limited so hope someone out there can give me a better solution. Here's the problem, I'm receiving a file from a vendor that is formatted like so:
INPUT:
###########################################################
################### 11/03/2008 #######################
###########################################################

FILE_NAME TRANSACTION COUNT
------------------------------ ---------------------------- ------------
TESTFILE_1 650 239
651_1 5
------------
244

TESTFILE_2 650 1
651_1 5
651_2 4
------------
10

TESTFILE_3 650 7
------------
7

TESTFILE_4 650 1
------------
1

I want the output to look like this:
TESTFILE_1 244
TESTFILE_2 10
TESTFILE_3 7
TESTFILE_4 1

Let's assume the filename will always have TESTFILE in it. The code I'm currently running now which works but has alot of assumptions is this:
#!/usr/bin/ksh
INPUT_FILE=$1
OUTPUT_FILE=${INPUT_FILE}.converted
rm $OUTPUT_FILE
touch $OUTPUT_FILE

NUM=0
for FILENAME in `cat $INPUT_FILE|grep TESTFILE|awk '{ print $1 }'`
do
FILENAME_ARRAY[$NUM]=$FILENAME
NUM=`expr $NUM + 1`
done
TOTAL_FILENAMES=$NUM

NUM=0
for TOTALS in `cat $INPUT_FILE|grep -v "^$"|grep -v "#"|grep -v "-"|grep -v "FILE_NAME"|grep -v "PEXODEM"|tr -s " "|awk -F" " 'NF == 1'`
do
TOTALS_ARRAY[$NUM]=$TOTALS
NUM=`expr $NUM + 1`
done
TOTAL_TOTALS=$NUM

if [ TOTAL_FILENAMES -ne $TOTAL_TOTALS ]; then
echo "Number of Files and Number of Totals don't match!"
exit 1
fi

NUM=0
while true
do
echo "${FILENAME_ARRAY[$NUM]} ${TOTALS_ARRAY[$NUM]}" >> $OUTPUT_FILE
NUM=`expr $NUM + 1`
if [ $NUM -eq $TOTAL_FILENAMES ]; then
break
fi
done

Someone's gotta have a more efficient way of accomplishing the same thing.

Thanks in advance,

EDIT: The formatting of the input file isn't really showing up correctly in the post but basically you have a filename called TESTFILE* and a record count. Some files have multiple sub files with a record count as well. Then you have a total records for that file and all it's sub files. I just want the filename and the total records for itself and all it's sub files.

Try this:

Code:

awk '
/^TESTFILE_/{s=$1;next}
/----/ && s{getline;print s, $NF}
' file

Regards

Works beautifully. Wow, I really need to learn more about awk.

Thanks alot,