Evaluate DMN with input that might not match the DMN expected input

BPMN & Modeling Discussion & Questions
1

Out of curiosity, any feedback on my approach to using DMN to evaluate spreadsheet values that might not directly match 1:1?
(e.g. if there is anything from community that can help me with getting input of DMN and any parts of the process or bring it this to next level)

(This feels a bit medieval if I take step back end think about general testability and DMN lifecycle between business users, spreadsheets…)

I have input from excel that I want to use to evaluate DMN.

Challenges:

  • column names might have whitespaces, be in different casing format or have some +/? at the end…

I can’t just call DMN passing those columnName:value key pairs, so need to check what inputs DMN have and fuzzy match them with excel.

Here is my approach:

  1. Create API to retrieve all input fields of DMN. That api should generate template that I need to fill on client side and then pass to the /evaluate api as payload,
{
	"variables": {
		"Country" : {
			"value": "<<Country>>", "type": "String"
		},
	}
}

  1. Rest API will return above template, now on the other side I will do the used to fuzzy match,
    a) find all tokens << >>
    b) fuzzy match tokens with available values
    c) replace tokens in the template

  2. Call /evaluate
    e.g.

{
	"variables": {
		"Country" : {
			"value": "USA", "type": "String"
		},
	}
}