Alan Ackerman wrote: > Actually, that doesn't work. The CHOP tests only the first occurance > of 'word' in the line.
Ah, right. So instead of using CHOP and testing the two halves of the record, maybe you can use NOT CHOP 1 BEFORE | SPLIT 1 BEFORE and then test the first and n+2th character of each split record. You'd have to INSERT a delimiter at the front of the record to handle the case where the record starts with your phrase. The problem I'm not seeing an easy solution to, though, is the case where two occurrences of your phrase overlap or are separated by less than two characters. (This is about the point where, as an IBM internal user, I resort to using PATTERN.) ¬R
