Conditional transposition in awk based on column values - awk

Question

I'm trying to make the below transformation using awk.
Input:
status,parent,child,date
first,foo,bar,2019-01-01
NULL,foo,bar,2019-01-02
NULL,foo,bar,2019-01-03
last,foo,bar,2019-01-04
NULL,foo,bar,2019-01-05
blah,foo,bar,2019-01-06
NULL,foo,bar,2019-01-07
first,bif,baz,2019-01-02
NULL,bif,baz,2019-01-03
last,bif,baz,2019-01-04
Expected output:
parent,child,first,last
foo,bar,2019-01-01,2019-01-04
bif,baz,2019-01-02,2019-01-04
I'm pretty stumped by this problem, and haven't got anything to show yet - any pointers would be very helpful.

Answer 1

Could you please try following.
awk '
BEGIN{
FS=OFS=SUBSEP=","
print "parent,child,first,last"
}
$1=="first" || $1=="last"{
a[$1,$2,$3]=$NF
b[$2,$3]
}
END{
for(i in b){
print i,a["first",i],a["last",i]
}
}
' Input_file
Output will be as follows.
parent,child,first,last
bif,baz,2019-01-02,2019-01-04
foo,bar,2019-01-01,2019-01-04
Explanation: Adding detailed explanation for above code.
awk ' ##Starting awk program from here.
BEGIN{ ##Starting BEGIN section from here.
FS=OFS=SUBSEP="," ##Setting Fs, OFS and SUBSEP as comma here.
print "parent,child,first,last" ##Printing header values as per OP request here.
} ##Closing BEGIN BLOCK for this progam here.
$1=="first" || $1=="last"{ ##Checking condition if $1 is either string first or last then do following.
a[$1,$2,$3]=$NF ##Creating an array named a whose index is $1,$2,$3 and its value is $NF(last column of current line).
b[$2,$3] ##Creating an array named b whose index is $2,$3 from current line.
} ##Closing main BLOCK for main program here.
END{ ##Starting END BLOCK for this awk program.
for(i in b){ ##Starting a for loop to traverse through array here.
print i,a["first",i],a["last",i] ##Printing variable it, array a with index of "first",i and value of array b with index of "last",i.
} ##Closing BLOCK for, for loop here.
} ##Closing BLOCK for END block for this awk program here.
' Input_file ##Mentioning Input_file name here.

Answer 2

$ cat tst.awk
BEGIN { FS=OFS="," }
{ key = $2 OFS $3 }
FNR==1 { print key, "first", "last" }
$1=="first" { first[key] = $4 }
$1=="last" { print key, first[key], $4 }
$ awk -f tst.awk file
parent,child,first,last
foo,bar,2019-01-01,2019-01-04
bif,baz,2019-01-02,2019-01-04
If you can have a first without a last or vice-versa or they can occur out of order then include those cases in the example in your question.

Answer 3

Not awk, you already have that, but here's an option in bash alone, just for kicks.
#!/usr/bin/env bash
declare -A first=()
printf 'parent,child,first,last\n'
while IFS=, read pos a b date; do
case "$pos" in
first) first[$a,$b]=$date ;;
last) printf "%s,%s,%s,%s\n" "$a" "$b" "${first[$a,$b]}" "$date" ;;
esac
done < input.csv
Requires bash 4+ for the associative array.