Removing CRLF from message using RegSub not working

[Weston Olmstead]

I’m having a bit of an issue using a pre-processing script on an HL7 to remove carriage return line feed (0D0D) and replace it with just carriage return (0D).

I have a message in the following format:

Segment [0D0A]

Segment [0D0A]

Segment [0D0A]

Segment [0D0A]

that I’m trying to run a pre-processor against using regsub to get the following output:

Segment [0D]

Segment [0D]

Segment [0D]

Segment [0D]

I’m using the following script, but for some reason when it translates the message it replaces the 0D0A with a space. This makes me think its identifying it correctly, but not performing the replace. I even tried changing the string that was supposed to be the text inserted to “TESTTEXT” just to see if I had indicated the special character incorrectly but this didn’t seem to help.

I’m sure I’m overlooking something simple.

proc tpsConvertCRLFToLF { args } {

   set argList [keylget args ARGS]    

   keylget args MODE mode               ;# Fetch mode

   switch -exact — $mode {

       run {

           set msgID [keylget args MSGID]    

           set msgText [msgget $msgID]        

           regsub -all  {rn} $msgText {r} msgText

           msgset $msgID [join $newmsg]

           return “{CONTINUE $msgID}”

            }  

       default {

       error “Unknown mode ‘$mode’ in tpsMessageFilter.”

       }

   }

}

Topic

[James Cobane]

Weston,

I think the issue may be with your ‘msgset’ command; I don’t see where you are valuing the ‘newmsg’ variable and also the [join $newmsg] has no delimiter specified.

[bill bearden]

Along with what James said, something else to look at…

Try replacing the {r} with “r”.

I struggle sometimes with the curly braces in Tcl. It seems like sometimes it means “take this string literally with no variable or command or escape sequence substitution” and sometimes it doesn’t.

I tested with the following. It appears to me that, when using {r}, regsub is replacing the crlf with literally slash-r. The last value of b looks like it is just def but it is actually abcdef. In Windows, plain old just puts you back at the beginning of the current line so the def is overwriting the abc.

Code:

tcl>set a “abcrndefrn”
abc
def

tcl>regsub — {rn} $a {r} b
1
tcl>echo $b
abcrdef

tcl>regsub -all {rn} $a {r} b
2
tcl>echo $b
abcrdefr
tcl>regsub -all {rn} $a “r” b
2
tcl>echo $b
def
tcl>echo [string length $b]
8
tcl>

[Charlie Bursell]

Bill is correct.  The output side of a regsub is interpreted by Tcl.  The difference in quoting with curly braces and quotes is Tcl will not do substitution of the characters inside the braces.  For example {$hello} is simply $hello, not a variable  while $hello reference the varible.  Try this at the Tcl prompt

tcl>set str {hellorThere}

hellorThere

tcl>set str “hellorThere”

There

With that said why use a regsub at all when string map is much more ergonomic and less arcane

set msgText [string map “n {}” $msgText]

or better

set msgText [string map “n {}” [msgget $msgID] ]

[Weston Olmstead]

Thank you all for the assistance. Its amazing how another set of eyes can catch a simple mistake!

I also have a difficult time determining to use quotes or curley brackets, but that example was very helpful. Thank you also for the advice on string mapping, I’ll look into that!

---

Edit: Forgot to include the solution that worked for me in case others find this thread. The regsub is currently working, but I’m looking into character maps as that seems the proper way to do this now.

proc tpsConvertCRLFToLF { args } {

  set argList [keylget args ARGS]    

  keylget args MODE mode               ;# Fetch mode

  switch -exact — $mode {

      run {

          set msgID [keylget args MSGID]    

          set msgText [msgget $msgID]        

regsub -all  {rn} $msgText “r” msgText

          msgset $msgID $msgText

          return “{CONTINUE $msgID}”

           }  

      default {

      error “Unknown mode ‘$mode’ in tpsMessageFilter.”

      }

  }

}

[Author]

Replies