Ygtjki

I have lines like this:

 400 january ####
 304 april ####
 151 may ####
 126 june ####
 115 august ####
 98 december ####

And I want them to look like this:

#### january 400 
#### april 304
#### may 151 
#### june 126 
#### august 115 
#### december 98

I tried using this command, but no avail.

 sed -E 's/(.*)(.*)(.*) /3 2 1/'

Try:

$ sed -E 's/^(.*[0-9])(.*[[:alpha:]])(.*)/3 2 1/' file
 #### january 400
 #### april 304
 #### may 151
 #### june 126
 #### august 115
 #### december 98

How it works

• ^(.*[0-9]) matches from the beginning of the line to the last number of the line.




• (.*[[:alpha:]]) matches from the after the above to the last alphabetic character on the line.




• (.*) matches anything after the last alphabetic character on the line.

Discussion

Consider:

sed -E 's/(.*)(.*)(.*) /3 2 1/'

sed's regexes match leftmost-longest. That means that the first (.*) above matches the entire line. The remaining groups are empty. If we modify the command, this becomes clear:

$ sed -E 's/(.*)(.*)(.*) /3=3 and 2=2 and 1=1/' file
3= and 2= and 1= 400 january ####
3= and 2= and 1= 304 april ####
3= and 2= and 1= 151 may ####
3= and 2= and 1= 126 june ####
3= and 2= and 1= 115 august ####
3= and 2= and 1= 98 december ####

Here is a simpler way to do it:

awk 'print $3," "$2," "$1' file

It just uses awk to print columns 3, 2, and 1 with two spaces between the 3rd and 2nd column and four spaces between the 2nd and 1st column.

Output:

#### january 400
#### april 304
#### may 151
#### june 126
#### august 115
#### december 98

If you want it to edit the file in place, use this command:

awk -i inplace 'print $3," "$2," "$1' file