AIX

I have a file with contents:

1|4|oho hosfadu|
1|3|sdfsd fds|
2|2|sdfg|
2|1|sdf a|
3|5|ouhuh hu|

I would like to do three things to it;
1- first, sort it on the first two fields
2- get a unique count on the first field
3- and write the first two unique rows (uniqueness based off the 1st field) to a file:

Step 1:
1|3|sdfsd fds|
1|4|oho hosfadu|
2|1|sdf a|
2|2|sdfg|
3|5|ouhuh hu|

Step 2:
unique count = 3

Step 3:
1|3|sdfsd fds|
1|4|oho hosfadu|
2|1|sdf a|
2|2|sdfg|



Thanks,

- CB

#3 is not clear - the output is not unique, or did you mean "output all rows that match the first two unique numbers in field 1"?

Sorry, let me clarify #3:

write the first two unique ids (first column) to a file.

So that would be:

1|3|sdfsd fds|
1|4|oho hosfadu|
2|1|sdf a|
2|2|sdfg|

This is not fine tuned code

Code:

#!/bin/ksh
echo "Step 1"
sort -n -t'|'  -k1.1,1.4 -k 2.1,2.4 inputfile > outputfile
echo "Step 2: unique count \c"
awk -F'|' '!arr[$1]++' outputfile > tmp.tmp
cat tmp.tmp | wc -l             # not a UUOC done on purpose
set -A arr $( head -2 tmp.tmp | tr -s '\n' ' ')
echo "Step 3:"
grep -e ''^${arr[0]}"  -e "^$arr[1]}"  outputfile

Thanks... can you explain your solution to step# 3. Another thing is that I chose '2' unique ids to be written to a file as an example. That number could change (will be passed through a parameter).

So if this is hard-coded, could you provide an option where I can use a parameter?

Thanks,

- CB