Externally-defined enum values

[afeld]

I'm working on the datapackage.json for the data underpinning dataportals.org. There are a few columns that are enums, for which there are long lists of values that come from other sources.

Field	Possible values
country	ISO 3166-1 alpha-2
language*	ISO 639 Set 1
license_id	Open Definition

*This field can have multiple values, though that isn't supported in frictionless-py yet: frictionlessdata/frictionless-py#1741

I could copy-and-paste all those values into the enum array, but:

• that would bloat the JSON
• I'd need to maintain that list
  • Granted, these particular lists are fairly static.

I tried:

1. Adding rdfTypes, exactly like the example
   • I assumed this would add validation, but no such luck.
2. Making a script to populate the enum values, to at least avoid the copying-and-pasting: validate license values okfn/dataportals.org#389

Is this a use case that the spec considers? If not, would it be in scope?

Thanks!

[khusmann]

This sounds like a good application of the Category Tables recipe, which will be available in the v2.1 version of the spec: #1081

We have not extended this functionality to lists of categorical values, but this is an easy logical next step, and would be also useful for multiselect survey items: #1039

[khusmann]

(@afeld if adding Category Tables capabilities to list types matches your use case here, let me know and we can talk about putting together a proposal to add to the spec)

(PS Here's a link to a more easily readable version of the category tables recipe: https://github.com/frictionlessdata/datapackage/blob/next/content/docs/recipes/category-tables.md)

Here's how it might look for list of fruits referencing a list of possible values in another package. (e.g. Apple,Banana, Cherry,Banana, Cherry)

{
  "name": "fruit",
  "type": "list",
  "itemType": "string",
  "categories": {
    "package": "http://example.com/package.json",
    "resource": "fruit-codes",
    "encodedAs": "label"
  }
}

And then in http://example.com/package.json, the fruit-codes resource would have:

{
  "name": "fruit-codes",
  "type": "table",
  "categoryFieldMap": {
    "value": "code",
    "label": "name",
    "ordered": false
  },
  "schema": {
    "fields": [
      { "name": "code", "type": "string" },
      { "name": "name", "type": "string" }
    ]
  },
  "data": [
    { "code": "A", "name": "Apple" },
    { "code": "B", "name": "Banana" },
    { "code": "C", "name": "Cherry" }
  ]
}