Removing non-standard text when cleaning data - Substring, PRXCHANGE?

[vbylsma]

I am standardizing unit names, however many of them contain a non-standard indicator at the end of the unit name (it denotes geographically separated but part of the same unit). I need to remove this final string so these units can later be combined into one unit. Typically I would substring this, but the indicator 1) does not occur on every unit name, 2) is not in a standard position in the string, and 3) the final sting itself isn't consistent in the use of digit vs character starting in position 3 (see below). 

Examples of the unit names:

SURGICAL OPS OLA900

SURGICAL OPS OLHC00

EDU AND TRNG OL0IZH

EDU AND TRNG OL0PZR

LEGAL OLCS00

LEGAL OLA200

The final sting always starts with " OL". Searches have brought me to PRXCHANGE, but as I've read in forums and found out myself, the syntax is particularly confusing. What is your best practice recommendation in this situation? 

[Reeza]

Your examples all seem to have it at the end and same pattern. Can you provide a more representative example and also include what you're looking for as your final output.

[Astounding]

Given that the string always begins the final word, here's one way:

delete_here = index(unit, ' OL');
if delete_here then
unit = substr(unit, 1, delete_here);

[vbylsma]

This also worked, thank you so much!

[Ksharp · Accepted solution]

data have;
input have $40.;
want=prxchange('s/\bOL\w+$//i',1,strip(have));
cards;
SURGICAL OPS OLA900
SURGICAL OPS OLHC00
EDU AND TRNG OL0IZH
EDU AND TRNG OL0PZR
LEGAL OLCS00
LEGAL OLA200
;

[vbylsma]

This worked AND I learned a lot about PRXCHANGE. Thank you!