Detecting Control Characters

I ended up doing it in SQL Server through the ODBC linked server.

The criteria can be set up as an I-descriptor (or EVAL) to avoid the line feed being acted on but AFAIK an I-descriptor cannot be used in a LIKE expression.

I would still be curious if it can be done in UniVerse on a terminal.

------------------------------
Greg Clitheroe
Rocket Forum Shared Account
------------------------------

UniVerse or UniData?

I'm assuming you are having a problem with a hashed file and not a directory although the solution will work for any file type.  The quickest solution would be to create an I-Type or Virtual dictionary item that scans a record for a given character.  That will allow you to build selectlists to be fed into other cleanup routines.  

In the dictionary of the suspect data file, create a dictionary item that searches for the offending character.  Below is an example of a UniVerse I-Type that looks for a tab character.

>ED DICT DATA.FILE TAB.CHAR
This is a Type "I" Descriptor last compiled on 01/02/25 at 19:43.
20 lines long.

----: P
0001: I
0002: IF INDEX(@RECORD,CHAR(9),1) THEN 1 ELSE ''
0003:
0004: Tab
0005: 3R
0006: S
0007:
0008:

You can then execute a SELECT statement to find your bad records.

Example: SELECT DATA.FILE WITH TAB.CHAR

The result will be an active selectlist of records that contain a tab character.

You can create separate dict items for each of the characters or combine your character search in the I-Type by nesting your IF statement.

If you don't know exactly what non-printing character your looking for, the solution is a bit different.  Let me know if that's the case and I'll offer a different solution.

If the problem exists in multiple files, copy your bad character dictionary item to the VOC file and it will be usable on all of your data files.

In the past I have used the following select to find unprintable characters in a file,

SELECT "filename" WITH EVAL "CONVERT(@AM:@VM:@SM,'',@RECORD)" NE EVAL "OCONV(CONVERT(@AM:@VM:@SM,'',@RECORD),'MCP')"

I was just about to post a similar answer!  The trick is that the "MCP" conversion converts all non-printable characters to periods. It makes for a quick way to find strings with any non-printable characters.