AWK move lines of a file down

[judkil]

Hello,

I own
input

AB-0004 XXX XXX
A8-0005 XXX
AB-0007 XXX XXX
AB-0008 XXX

and I would like to get

Output

AB-0004 XXX XXX
AB-0007 XXX XXX
A8-0005 XXX
AB-0008 XXX

I would like to be able to move lines of a file down according to the number of columns whether with an awk or a sed but without relying on the line number

If anyone had an idea! Thank you very much !

 

[mikrom]

My only idea is, that you can read all lines into an array and in the END{} print the array elements in order you need

 

[mikrom]

You can use awk with sort and cut:

For example we have this file:

$ cat judkil_sortfile.txt 
AB-0004 XXX XXX
A8-0005 XXX
A8-0006 XXX XXX XXX
AB-0007 XXX XXX
AB-0008 XXX

and we want to sort the lines according to the number of columns.

First, we can use awk to print number of columns (NF = Number of Fields) in every line as the the 1st column

$ awk '{print NF, $0}' judkil_sortfile.txt
3 AB-0004 XXX XXX
2 A8-0005 XXX
4 A8-0006 XXX XXX XXX
3 AB-0007 XXX XXX
2 AB-0008 XXX

Then, we can use sort, to sort awk's output according to the 1st column in reverse order

$ awk '{print NF, $0}' judkil_sortfile.txt | sort -nr
4 A8-0006 XXX XXX XXX
3 AB-0007 XXX XXX
3 AB-0004 XXX XXX
2 AB-0008 XXX
2 A8-0005 XXX

Finally, we can use cut, to select from sort's output only the second through last columns (fields)

$ awk '{print NF, $0}' judkil_sortfile.txt | sort -nr | cut -f2- -d' '
A8-0006 XXX XXX XXX
AB-0007 XXX XXX
AB-0004 XXX XXX
AB-0008 XXX
A8-0005 XXX

.. and that seems to be what you need

 

[judkil]

It's ok !! Thank you very much !