Shell Programming and Scripting

Code:

$ cat key.txt
a=AE
e=AA
i=AU
o=AI
u=AO

$ echo "Testing" | awk -F= 'NR==FNR{R[$1]=$2;next}
{for(i=1;i<=length;i++) {
   if(substr($0,i,1) in R) printf R[substr($0,i,1)]
   else printf substr($0,i,1)
  }
  printf "\n"}' key.txt - 
TAAstAUng

This is indeed a problem. It is possible, to use a two-step approach by "protecting" the characters you have already changed:

Suppose, you want to change "x" to "yz" and "y" to "ab" - you will have to distinguish between the "y"s from the original text and the "y"s from your first translation. Simply append or prepend a character you don't use normally to signify already translated characters. Basically this is called "escaping" and traditionally the backslash "\" is used for that.:

"x" -> "\y\z", "y" -> "\a\b"

You can now refine your search string to search and translate only for characters not prepended by a "\":

"\x" -> "\x", "x" -> "\y\z", "y" -> "\a\b", "\y" -> "\y"

Of course you have to finally - as the very last step - to remove all the backslashes. Now that in sed code, which uses "\" for escaping too - even for escaping "\" themselves:

Code:

sed 's/^x/\\y\\z/
     s/[^\]x/\\y\\z/g
     s/^y/\\y\\z/
     s/[^\]y/\\a\\b/g
     ... <more translations here>...
     s/\\//g' /path/to/inputfile

As you see you need two rules for every character you want to translate: one for the character at the beginning of a line and one for everywhere else.

I hope this helps.

bakunin