LightSchema

Lightweight and pluggable schema validator. As minimum opinionated as possible.

Getting started

$ npm install --save lightschema

Then in your code import and instantiate it:

import LightSchema from 'lightschema'
 
const createSchema = (definition) => new LightSchema(definition)
const schema = createSchema({
  title: String,  // shorthand definition
  age: {          // detailed definition
    type: Number,
    optional: true
  }
})
 
schema.validate({ title: 'the title' }) // returns [], means no errors found
schema.validate({}) // returns [ { key: 'title', value: undefined, reason: 'keyRequired' }], means one error found

Internal normalization

When creating a schema, the definition model is internally normalized, if shorthands were used. Consider the following example:

{
  title: String,
  complexObject: {
    type: Object,
    entries: {
      name: String,
      value: Number
    }
  },
  optional: {
    type: String,
    optional: true
  },
  arrayObject: {
    type: Array,
    entries: {
      name: String,
      id: String
    }
  }
}

which will be transformed to

{
  title: {
    type: String,
    optional: false
  },
  complexObject: {
    type: Object,
    optional: false
    entries: {
      name: {
        type: String,
        optional: false
      },
      value: {
        type: Number,
        optional: false
      }
    }
  },
  optional: {
    type: String,
    optional: true
  },
  arrayObject: {
    type: Array,
    optional: false
    entries: {
      name: {
        type: String,
        optional: false
      },
      id: {
        type: Number,
        optional: false
      }
    }
  }
}

Access the normalzed and original models

This normalized model can be accessed via schemaInstance.model whereas the original definition is available via schemaInstance.original.

Circular references

Any definition or validation input that contains a circular reference will be denied by a thrown error. If you want to reuse schema pattern you may write a small factory and use Object.assign.

Use plugins

The library comes with a default checking for the type property with the builtin types undefined, null, Number, String, Boolean, Array, Object, Date and the optional property to define keys that can be omitted. To extend the range of types and/or properties or to add custom validation you can statically register plugins on the LightSchema class.

Register Props

The following example registers a plugin for allowing labels:

LightSchema.registerPlugin({
  name: 'labels',
  props: {
    label: {
      type: String
    }
  }
})

now you can define label props on any key:

import LightSchema from 'lightschema'
 
const createSchema = (definition) => new LightSchema(definition)
const schema = createSchema({
  title: {
    type: String,
    label: 'The title'
  },
})
 
schema.validate({ title: 'the title' }) // returns [], means no errors found

this does not alter the validation behavior at all, but your model has now more static information. Now let's register a prop that affects validation.

Register props and validators

Let's say we want to have min/max values to be checked on all Strings, Numbers, Arrays, we can define this plugin like the following:

LightSchema.registerPlugin({
  name: 'minMax',
  props: {
    min: {
      type: Number,
      default: -100 // defaults are possible but be careful as they apply as fallback on all keys and could break validation
    },
    max: {
      type: Number
    }
 },
  validate (key, value, { min, max }) {
    let target
    switch (Object.prototype.toString.call(value)) {
      case '[object String]':
      case '[object Array]':
        target = value.length
        break
      case '[object Number]':
        target = value
        break
      default:
        throw new Error('unexpected undefined type')
    }
    if (typeof min === 'number' && target < min) return { key, value, min, reason: 'valueTooSmall' }
    if (typeof max === 'number' && target > max) return { key, value, max, reason: 'valueTooLarge' }
    // return nothing / void / null if your validator passed
  }
})

Now let's try this plugin with some basic schema:

const minMaxSchema = create({
title: {
  type: String,
  min: 1, // your actual min value for this key
  max: 10
}
})
minMaxSchema.validate({ title: 'The title' }) // []
minMaxSchema.validate({ title: '' }) // [{ key: 'title', value: '', min: 1, reason: 'valueTooSmall' }]
minMaxSchema.validate({ title: 'The title too long' }) // [{ key: 'title', value: '', min: 1, reason: 'valueTooLarge' }]

Register types

You can also register the range of types, that are allowed to be defined by the schema. Let's add an ArrayBuffer type:

LightSchema.registerPlugin({
  name: 'bufferTypes',
  type: ArrayBuffer
})
 
const bufferSchema = create({
  file: ArrayBuffer
})
 
bufferSchema.validate({ file: new ArrayBuffer(8) }) // [] 

Remove plugins

Plugins can be removed. The are returned in case you already lost the plugin context:

LightSchema.registerPlugin({
  name: 'bufferTypes',
  type: ArrayBuffer
})
 
const removed = LightSchema.removePlugin({ name: 'bufferTypes' })
console.log(removed) // { name: 'bufferTypes', type: [object ArrayBuffer] }

Contribution, testing, coverage, rollup

Code of conduct = be nice, don't add too much new stuff, keep it short and simple.

Testing is done using mocha via

$ npm run test

Coverage is done using

$ npm run test-coverage

Rollup is used to bundle the files into one file in the dist folder.

License

MIT, see license file