Comments on: What’s an alternative to %SCAN?

By: tomliotta

tomliotta — Sat, 22 May 2010 00:54:40 +0000

…thanks for the reply.

You’re welcome. Happy to help, and thank you for returning to post a followup.

Tom

By: bogeybetsy

bogeybetsy — Fri, 21 May 2010 17:37:27 +0000

Hi Tom, thanks for the reply. You were right about the %Xlate. We found out that the problem was not the %SCAN. There was some other new code inserted that caused the program to run for about 22 hours. The %SCAN did just fine. But I did replace the %SCAN code with the %XLATE. There was a whole new subroutine using %SCAN. With %XLATE, there were only a few lines (3 statements). Thanks very much for your help!

By: tomliotta

tomliotta — Wed, 12 May 2010 21:23:01 +0000

First, does the field have to be 5000 bytes or could it be variable length? Is the value often less than 5000 long? (I.e., is it right-padded with blanks? Do you know how long the value is?)

Next, have you tried a simple %xlate() instead? And since this is a file field, why not just read it through a LF that applies a basic translation table (the DDS TRNTBL() keyword) to the field that translates the hex values to blanks?

By moving a pointer through the field, you could probably make good use of the strpbrk() C run-time function in a proc that starts with the pointer at position 1 and resets it in a loop every time it gets a hit. You might also use the Test and Replace (TESTRPL) MI builtin function, but I’d want to know if strpbrk() or TESTRPL is faster.

Or any of the Translate (XLATE), Translate Bytes (XLATEB), Translate Bytes One Byte at a Time (XLATEB1), Translate Multiple Bytes (XLATEMB) or Translate with Table (XLATEWT) MI builtins might be useful.

I’d start by seeing how %xlate worked. Then I’d see about using a LF to do the translation for me.

Tom