First, some test data (the 'zzz' bit will become clear later)...
load 'zzz' & Rawtext as Rawtext inline [
Rawtext
CatDogRat
HorsePig
HorseHare
ZebraAntelope];
... then we need to look at each string and place a comma before the capital letters. I decided the easiest way was to loop over each capital letter, replacing them as needed with [comma+capital letter], so as a preceding load to the test data ...
Replace(peek('String'), chr(iterno()+64), ',' & chr(iterno()+64))) as String while iterNo() >=1 and iterNo() <= 26;
... this will expand the data, creating a row for each iterno() 1 to 26 (one for each letter of the alphabet). If it's the first iteration, then use the string, otherwise 'peek' the last value of the string. We need to pass the ASCII code for each letter which starts at 65 for uppercase A (or iterno() + 64) to 90 for uppercase Z. After this we only need the last pass, so we can inner join on a set value of pass (26) to get rid of pass values 1 to 25 ...
inner join
Load 26 as pass autogenerate 1;
... then we just need to eliminate the rogue comma introduced at the front of the string. I placed an identifier 'zzz' to make it simple using another Replace in a preceding load (you will see the text to remove in the string is 'zzz,' and not just 'zzz') ...
Load Replace(Rawtext, 'zzz','') as Rawtext, pass, Replace(String, 'zzz,','') as String;
... finally, remove the 'pass' field as not needed.
