I am trying to automate the swapping of characters 1 and 2, 3 and 4, etc, in a an arbitrary-length string of hex characters.
Example:
A627E39B
becomes:
6A723EB9
It seems like this should be possible using sed or one of the other *nix text utilities, but so far I have not been able to find any examples.
You can use back-references:
echo A627E39B | sed 's/\(.\)\(.\)/\2\1/g'
This finds all occurrences of two characters and swaps them.
As glenn jackman pointed out, enabling extended regular expressions (-E w/ GNU, BSD, or toybox seds) avoids having to escape the parentheses:
echo A627E39B | sed -E 's/(.)(.)/\2\1/g'
There’s a tool that specifically, natively supports that function:
dd with the swab (swap bytes) conversion:
% echo ABCD | dd conv=swab
BADC
0+1 records in
0+1 records out
5 bytes (5 B) copied, 0.000109277 s, 45.8 kB/s
You’ll want to add 2> /dev/null to discard the log output:
% echo ABCD | dd conv=swab 2> /dev/null
BADC
Warning: This will do pretty much what you want if you send it
an even number of (printable) characters. But, if you give it
an odd number of printable characters — let’s say 2n+1 —
then dd will take the newline at the end
as the (2n+2)th character,
and swap it with the (2n+1)th. So,
% echo ABCDE | dd conv=swab 2> /dev/null
sends
A B C D E \n
to dd, which gives you
B A D C \n E
back; this will look like
BADC
E
You can work around this by specifying the -n option to echo
to suppress the newline at the end
(or just use printf without adding a newline).
If you’re displaying the output to the screen, then, for completeness,
you should add a newline at the very end:
% echo -n ABCDE | dd conv=swab 2> /dev/null; echo
BADCE
% printf "%s" ABCDE | dd conv=swab 2> /dev/null; printf "\n"
BADCE
You can also use perl:
$ echo A627E39B | perl -CLS -nle 'print map {~~reverse} unpack "(a2)*"'
6A723EB9
