mjhjmtu

How to replace this word in a file globally which has control characters. how can i do that ? I have tried the below with no success.

Replace "^Aabc_def_a^B" with "^Adef_a^B" globally

Tried the below

sed -i 's/^Aabc_def_a^B/^Adef_a^B/g' file.txt 

sed -i 's/^Aabc_def_a^B/^Adef_a^B/g'file.txt

sed 's/x01abc_def_ax02/x01def_ax02/g'

Seems to work for me:

$ echo -n "^Aword^B" | sed 's/^Aword^B/^Alexical unit^B/' | hexdump -C
00000000 01 6c 65 78 69 63 61 6c 20 75 6e 69 74 02 0a |.lexical unit..|
0000000f

01 represents ^A; 02 represents ^B.

Bear in mind that I used literal control characters, not a caret (^) followed by the A or B. To enter them when crafting your sed command, use Ctrl-V followed by the control character to insert it literally rather than having readline parse it as input.

In other words, to insert a ^A, press Ctrl-V followed by Ctrl-A, and you will see ^A displayed as your input. If you try to arrow past this, you will observe your cursor always treating this as a single unit even though it is comprised of two characters on your screen.

In GNU sed, xHH works:

$ printf '01foo02n' | sed 's/x01foox02/x01barx02/g' | od -c
0000000 001 b a r 002 n

If you don't have GNU sed, but have Bash, you can use the $'' quoting to generate the control characters in the shell:

$ printf '01foo02n' | sed $'s/x01foox02/x01barx02/g' | od -c
0000000 001 b a r 002 n 

If you have neither, there's always Perl which also understands xHH:

$ printf '01foo02n' | perl -pe 's/x01foox02/x01barx02/g' | od -c
0000000 001 b a r 002 n

Of course you can do the same with -i.