i want a command for my script!!!

say file consists of character 123 125 127.

i need a query to replace the number 2 with 0

so the output should be 103 105 107.

i use unix-aix

tr -s '2'  '0' < oldfile > newfile

I am not sure about what you mean: ist this the text as it is in your file or are these decimal (octal?) values for three characters in your file?

If this is text you can use the following:

sed 's/12\([0-9]\)/10\1/g' /path/to/file > /path/to/other/file

No.

I will tell you clearly.

i want to delete a invalid character(which i told you as example '2'.)

The query should check for characters which is not present in the following
A to Z,
a to z ,
0 to 9
and all the symbols present in the keyboard.

If you have any doubts reply me.

You have already solved the problem almost - all on your own. Just construct a simple regular expression from your goal:

[^A-Za-z0-9.,#&@]

sed '/[^A-Za-z0-9.,#&@]//g' /path/to/sourcefile > /path/to/newfile

Myscript

#!/usr/bin/ksh
sed '/[^A-Za-z0-9.,#&@]//g' /home/1.txt > /home/2.txt


it throws error like this
"sed: 0602-403 /[^A-Za-z0-9.,#&@]//g is not a recognized function."


why is it so.
did i do any mistake?

My bad, i had a typo in the script:

sed '/[^A-Za-z0-9.,#&@]//g' /path/to/sourcefile > /path/to/newfile

sed 's/[^A-Za-z0-9.,#&@]//g' /path/to/sourcefile > /path/to/newfile