Can't understand how xargs is treating unicode literals

Question

I'm trying to convert unicode literals into readable format. I'm using the -t flag to see what command xargs is going to run before it runs it. I see the following weirdness:

$ echo -n "\\\t\\\u0042\\\u0065\\\u006e" | xargs -t -I '{}' echo -e '"'{}'"'
echo -e "\t\u0042\u0065\u006e"
"       \u0042\u0065\u006e"

Yet when I run it manually:

$ echo -e "\t\u0042\u0065\u006e"
        Ben
$

The \t is being correctly processed in both examples. but the \u00XX values are being treated as a plain string in the first example but is being encoded correctly in the second example. I see the same weirdness when I use printf as well.

$echo -n "\\\t\\\u0042\\\u0065\\\u006e" | xargs -t -I '{}' printf \"{}\"
printf "\t\u0042\u0065\u006e"
"       printf: invalid universal character name \u0042
$ printf "\t\u0042\u0065\u006e"
        Ben

Any ideas on what's happening and how to fix it?

Answer 1

Dave is correct, the shell isn't involved, so the echo is /bin/echo, try this:

echo  "\\\t\\\u0042\\\u0065\\\u006e" | od -c

echo "\\\t\\\u0042\\\u0065\\\u006e" | xargs -t -I '{}' echo -e '"'{}'"'
 
# echo -e '"'\t\u0042\u0065\u006e'"'
/bin/echo -e '"'\\t\\u0042\\u0065\\u006e'"'

Result:

0000000   \   \   t   \   \   u   0   0   4   2   \   \   u   0   0   6
0000020   5   \   \   u   0   0   6   e  \n
0000031
"   \u0042\u0065\u006e"
"   \u0042\u0065\u006e"
echo -e "\t\u0042\u0065\u006e" 

Tried with online bash shell