RE: Replace 1 Char with 3Char's

["Dennis Lovelady" <iseries@xxxxxxxxxxxx>]

Robert Poinsatte wrote:
> I use sed once in a while. It seems to always takes me forever to get
> the command to work.  I had a ASCII file that I needed to take out the
> tabs (hex 09).  The output file that sed created was EBCDIC.  Finally
> figured out I had to remove hex 05 not hex 09 to get it work.  My 2
> cents.

This is not an issue with sed... this will always happen character sets are
mixed.

Scott Klement wrote:
> ...   really makes it ugly.  But it should work:
>
> sed "s:\\\\\.br\\\\:$(print '\\\r\\\n'):g" in.hl7 > out.hl7
>
> (I also replaced the % with : -- IMHO, % is ugly)
>
> I was hoping you'd come up with a solution, since mine (above) is so ugly.
:)

Best I can do is:
   CRLF=$(print '\\r\\n'); sed "s:[\\][.]br[\\]:${CRLF}:g" < in.hl7 >
out.hl7

A little less ugly, methinks.  But GNU has it beat hands down!

Joe Wells wrote:
> However, when I use this against a live file I have a
> problem.  The live file contains stand alone cr (x'0d')
> as line delimiters.

A minor modification should resolve.  Below is not tested.
   CRLF=$(print '\\r\\n');
   CR=$(print '\\r');
   sed -e "s:${CR}:${CRLF}:g" -e "s:[\\][.]br[\\]:${CRLF}:g" < in.hl7 >
out.hl7

Dennis Lovelady
http://www.linkedin.com/in/dennislovelady
--
Patience - a minor form of despair disguised as a virtue.
        -- Ambrose Bierce