Hex String Replacement Using sed

Question

I'm having some trouble getting sed to do a find/replace of some hex characters. I want to replace all instances within a file of the following hexadecimal string:

0x0D4D5348

with the following hexadecimal string:

0x0D0A4D5348

How can I do that?

EDIT: I'm trying to do a hex find/replace. The input file does not have the literal value of "0x0D4D5348" in it, but it does have the ASCII representation of that in it.

Answer 1

GNU sed v3.02.80, GNU sed v1.03, and HHsed v1.5 by Howard Helman all support the notation \xNN, where "NN" are two valid hex numbers, 00-FF.

Here is how to replace a HEX sequence in your binary file:

$ sed 's/\x0D\x4D\x53\x48/\x0D\x0A\x4D\x53\x48/g' file > temp; rm file; mv temp file

As @sputnik pointed out, you can use sed's in place functionality. One caveat though, if you use it on OS/X, you'd have to add an empty set of quotes:

$ sed '' 's/\x0D\x4D\x53\x48/\x0D\x0A\x4D\x53\x48/g' file

As sed in place on OS/X takes a parameter to indicate what extension to add to the file name when making a backup, since it does create a temp file first. But then.. OS/X's sed doesn't support \x.

Answer 2

This worked for me on Linux and OSX.

Replacing in-place:

sed -i '.bk' 's'/`printf "\x03"`'/foo/g' index.html

(See @Ernest's comment in the answer by @tolitius)

Answer 3

In OS/X system's Bash, You can use command like this:

# this command will crate a variable named a which contains '\r\n' in it
a=`echo -e "hello\r\nworld\r\nthe third line\r\n"`

echo "$a" | sed $'s/\r//g' | od -c

and now you can see the output characters :

0000000    h   e   l   l   o  \n   w   o   r   l   d  \n   t   h   e
0000020    t   h   i   r   d       l   i   n   e  \n
0000033

You should notice the difference between 's/\r//g' and $'s/\r//g'. Based on the above practices, you can use command like this to replace hex String

echo "$a" | sed $'s/\x0d//g'  | od -c