Hi John,

Interesting use case and requirements. I'll bet you are not alone.

As of Release 23, Orchestrator provides the "Map Inputs from a CSV File" feature, which is a special case of the Custom step that allows you to easily map CSV file input into arrays without writing a custom script to do it. Is that how your current solution works? If so, you'll notice that that step allows you to specify a delimiter; however, it does not enable you to set the delimiter as a variable. (Sounds like a valid enhancement request!)

Therefore, you would have to "clone" that custom step for each delimiter you might use, and put those steps behind a Rule to direct to the appropriate one based on the delimiter.

You could establish the delimiter as an orchestration input, which is then passed into the Rule for evaluation. Or, you could get fancy (as you suggest) and create a script to sniff the contents of the CSV file to automatically determine its delimiter, and then pass that into the Rule for evaluation. I tried asking ChatGPT to generate such a JRuby script, and it seemed to give me a valid result (I didn't actually try to run it).

As for a script to read a CSV file and produce an array, yes, prior to Release 23 and the enhancement above, we used scripts to read CSVs and map to arrays. There is even an Oracle-by-Example tutorial with a sample script. Check out Using Files As Orchestration Input and Output (Exercise #3, step 29).

Hope that helps. Keep us posted on your progress.

AJ