I would approach this by first normalizing the character that is separating the values and then using the subfield() function to extract the parts. This does mean that you have to anticipate the characters that are used to separate the fields. Something like this:
NoConcatenate
Comments:
Load Load Key, replace(replace(replace(replace(Comments,'#',','),' ',','),'/',','),'-',',') as Comments
Resident [table with comments];
CommentParts:
Load subfield(Comments,',') as CommentPart
Resident Comments;
NoConcatenate
ServiceNumber:
Load CommentPart as ServiceNumber
Resident CommentParts
where IsNum(CommentPart) and len(CommentPart) >= 7 and len(CommentPart) <=9;
drop table CommentParts;
If you need to have the Service number in the same table as the original table, you can in the key field so that you can join it back to the original table.
