Question

I would like to combine entries from the second field from two files using awk, sed or similar.

File0:

string:data:moredata

File1:

string:random:moredata

If the first field, string in file0 has a matching entry in file1 then print

$random:$data

Selecting the fields seems trivial:

$ awk -F':' '{print $2}' filename

Need to match rows and print matching column $2

Answer 1

How about this one –

awk -F":" 'NR==FNR {x[$1] = $0; y[$1] = $2; next} ($1 in x) {print $2":"y[$1]}' file1 file2

Execution:

[jaypal~/Temp]$ cat file1
string:data:moredata

[jaypal~/Temp]$ cat file2
string:random:moredata

[jaypal~/Temp]$ awk -F":" 'NR==FNR {x[$1] = $0; y[$1] = $2; next} ($1 in x) {print $2":"y[$1]}' file1 file2
random:data

In this solution, we are loading the whole record of file1 in to the array indexed at column 1. We do a quick check in the next file to see if the column 1 is present. If it is then print statement is executed.

Negative Test:

[jaypal~/Temp]$ cat file1
string:data:moredata
man:woman:child

[jaypal~/Temp]$ cat file2
man:random:moredata
string:woman:child

[jaypal~/Temp]$ awk -F":" 'NR==FNR {x[$1] = $0; y[$1] = $2; next} ($1 in x) {print $2":"y[$1]}' file1 file2
random:woman
woman:data

Just to add to the explanation, NR and FNR are awk’s in-built variables. NR gives line number and does not get reset to 0 when looped over two files. FNR on the contrary is also a line number that gets reset to 0 when second file starts. Thus this allows us to store the file 1 into the array because that action is only done when NR==FNR. As soon as this condition becomes false, it means the second file has started and next pattern action statement begins to execute.