Converting JSON Schema to CUE | CUE

Paul Jolly

Introduction

CUE is a concise and powerful language for describing and constraining data, but not all systems speak CUE. Some systems can only produce schemas formatted in JSON Schema. This isn’t a problem for the cue command because it speaks JSON Schema - letting you use JSON Schema alongside schemas written in CUE.

In this tutorial you’ll use cue to convert a JSON Schema to CUE, and then use the result to validate some data.

Prerequisites

The cue binary – follow the installation instructions if you don’t already use cue
A tool to edit text files – any text editor you have will be fine, such as VSCode, Notepad, or Vim
A command terminal – cue works on all platforms, so you can use any Linux or macOS terminal, or a Windows terminal such as PowerShell, cmd, or WSL to run commands.
Some awareness of CUE schemata – the language tour’s pages on Constraints and Definitions are a good refresher

This tutorial is written using the following version of cue:

TERMINAL

$ cue version
cue version v0.9.0-alpha.4
...

Steps

This tutorial takes you through converting some JSON Schema to a CUE definition using the cue import command, and then using the result to validate some YAML.

Create a JSON schema file:

schema.json

{
    "$schema": "http://json-schema.org/draft-07/schema#",
    "type": "object",
    "additionalProperties": false,
    "required": [
        "name",
        "cuisine",
        "tables"
    ],
    "properties": {
        "name": {
            "type": "string"
        },
        "cuisine": {
            "type": "string"
        },
        "tables": {
            "type": "array",
            "items": {
                "$ref": "#/$defs/table"
            }
        }
    },
    "definitions": {
        "table": {
            "type": "object",
            "additionalProperties": false,
            "required": [
                "seats"
            ],
            "properties": {
                "seats": {
                    "type": "number",
                    "minimum": 2,
                    "maximum": 10
                },
                "view": {
                    "type": "boolean"
                }
            }
        }
    }
}

This schema validates data that defines a restaurant.

Convert the JSON Schema to a CUE definition called #restaurant:

TERMINAL

$ cue import -l '#restaurant:' -p cuisine schema.json

It’s good practise to tell cue to place imported CUE inside a package. Here, we choose the cuisine package.

cue import creates the file schema.cue containing this CUE:

schema.cue

package cuisine

#restaurant: {
	@jsonschema(schema="http://json-schema.org/draft-07/schema#")
	name:    string
	cuisine: string
	tables: [...#table]

	#table: {
		seats: >=2 & <=10
		view?: bool
	}
}

CUE natively understands JSON Schema, and is able to convert all of the JSON Schema constraints into CUE constraints. Notice the CUE schema is also considerably more concise and readable.

Create some data files that contain restaurant details:

split_pea.yml

name: The Split Pea
cuisine: Contemporary
tables:
  - seats: 4
    view: true
  - seats: 6

pomodoro.yml

name: Il Pomodoro Marcio
cuisine: Italian
tables:
  - seats: 100
  - seats: 8
    view: true

One of these files contains a deliberate problem that cue will catch for us, shortly. If you spotted the problem while entering the data, make sure you didn’t correct it!

Validate the data using the schema and constraints:

TERMINAL

$ cue vet -d '#restaurant' schema.cue *.yml
tables.0.seats: invalid value 100 (out of bound <=10):
    ./schema.cue:10:16
    ./pomodoro.yml:4:12

cue vet outputs nothing when validation succeeds. But as you can see, there are some validation errors.

Fix the data validation error by updating your pomodoro.yml file:

pomodoro.yml

name: Il Pomodoro Marcio
cuisine: Italian
tables:
  - seats: 10
  - seats: 8
    view: true

Re-validate the data using the schema and constraints:

TERMINAL

$ cue vet -d '#restaurant' schema.cue *.yml

The “silent” lack of output from cue vet confirms that the fixed data validates successfully.

Conclusion

You’ve completed this tutorial - well done!

In this tutorial you converted JSON Schema to CUE using cue import, and then used the resulting CUE to catch and fix an error in some data.