# User defined keywords

# Common attributes of keyword definitions

The usual interface to define all keywords has these properties:

interface _KeywordDef {
  keyword: string | string[]
  type?: JSONType | JSONType[] // data type(s) that keyword applies to,
  // if defined, it is usually "string", "number", "object" or "array"
  schemaType?: JSONType | JSONType[] // the allowed type(s) of value that keyword must have in the schema
  error?: {
    message: string | ((cxt: KeywordCxt) => Code)
    params?: (cxt: KeywordCxt) => Code
  }
}

Keyword definitions may have additional optional properties - see types (opens new window) and KeywordCxt (opens new window).

# Define keyword with code generation function recommended

Starting from v7 Ajv uses CodeGen module (opens new window) for all pre-defined keywords - see codegen.md for details.

This is the best approach for user defined keywords:

  • safe against code injection
  • best performance
  • the precise control over validation process
  • access to the parent data and the path to the currently validated data

While Ajv can be safely used with plain JavaScript, it is strongly recommended to use Typescript for user-defined keywords that generate code - the prevention against code injection via untrusted schemas is partially based on the type system, not only on runtime checks.

The usual keyword definition for keywords generating code extends common interface with "code" function:

interface CodeKeywordDefinition extends _KeywordDef {
  code: (cxt: KeywordCxt, ruleType?: string) => void // code generation function
}

Example even keyword:

import {_, KeywordCxt} from Ajv

ajv.addKeyword({
  keyword: "even",
  type: "number",
  schemaType: "boolean",
  // $data: true // to support [$data reference](./guide/combining-schemas.md#data-reference), ...
  code(cxt: KeywordCxt) {
    const {data, schema} = cxt
    const op = schema ? _`!==` : _`===`
    cxt.fail(_`${data} %2 ${op} 0`) // ... the only code change needed is to use `cxt.fail$data` here
  },
})

const schema = {even: true}
const validate = ajv.compile(schema)
console.log(validate(2)) // true
console.log(validate(3)) // false

Example range keyword:

import {_, nil, KeywordCxt} from Ajv

ajv.addKeyword({
  keyword: "range",
  type: "number",
  code(cxt: KeywordCxt) {
    const {schema, parentSchema, data} = cxt
    const [min, max] = schema
    const eq: Code = parentSchema.exclusiveRange ? _`=` : nil
    cxt.fail(_`${data} <${eq} ${min} || ${data} >${eq} ${max}`)
  },
  metaSchema: {
    type: "array",
    items: [{type: "number"}, {type: "number"}],
    minItems: 2,
    additionalItems: false,
  },
})

You can review pre-defined Ajv keywords in validation (opens new window) folder for more advanced examples - it is much easier to define code generation keywords than it was in the previous version of Ajv.

See KeywordCxt (opens new window) and SchemaCxt (opens new window) type definitions for more information about properties you can use in your keywords.

# Define keyword with "validate" function

Usual keyword definition for validation keywords:

interface FuncKeywordDefinition extends _KeywordDef {
  validate?: SchemaValidateFunction | DataValidateFunction // DataValidateFunction requires `schema: false` option
  schema?: boolean // schema: false makes validate not to expect schema (DataValidateFunction)
  modifying?: boolean
  async?: boolean
  valid?: boolean
  errors?: boolean | "full"
}

interface SchemaValidateFunction {
  (schema: any, data: any, parentSchema?: AnySchemaObject, dataCxt?: DataValidationCxt):
    | boolean
    | Promise<any>
  errors?: Partial<ErrorObject>[]
}

interface DataValidateFunction {
  (this: Ajv | any, data: any, dataCxt?: DataValidationCxt): boolean | Promise<any>
  errors?: Partial<ErrorObject>[]
}

The function should return validation result as boolean. It can return an array of validation errors via .errors property of itself (otherwise a standard error will be used).

validate keywords are suitable for:

  • testing your keywords before converting them to compiled/code keywords
  • defining keywords that do not depend on the schema value (e.g., when the value is always true). In this case you can add option schema: false to the keyword definition and the schemas won't be passed to the validation function, it will only receive the same parameters as compiled validation function.
  • defining keywords where the schema is a value used in some expression.
  • defining keywords that support $data reference - in this case validate or code function is required, either as the only option or in addition to compile or macro.

Example: constant keyword (a synonym for draft-06 keyword const, it is equivalent to enum keyword with one item):

ajv.addKeyword({
  keyword: "constant",
  validate: (schema, data) =>
    typeof schema == "object" && schema !== null ? deepEqual(schema, data) : schema === data,
  errors: false,
})

const schema = {
  constant: 2,
}
const validate = ajv.compile(schema)
console.log(validate(2)) // true
console.log(validate(3)) // false

const schema = {
  constant: {foo: "bar"},
}
const validate = ajv.compile(schema)
console.log(validate({foo: "bar"})) // true
console.log(validate({foo: "baz"})) // false

const keyword is already available in Ajv.

Keywords that do not define errors

If the keyword does not define errors (see Reporting errors) pass errors: false in its definition; it will make generated code more efficient.

To add asynchronous keyword pass async: true in its definition.

# Define keyword with "compile" function

The keyword is similar to "validate", with the difference that "compile" property has function that will be called during schema compilation and should return validation function:

interface FuncKeywordDefinition extends _KeywordDef {
  compile?: (schema: any, parentSchema: AnySchemaObject, it: SchemaObjCxt) => DataValidateFunction
  schema?: boolean // schema: false makes validate not to expect schema (DataValidateFunction)
  modifying?: boolean
  async?: boolean
  valid?: boolean
  errors?: boolean | "full"
}

In some cases it is the best approach to define keywords, but it has the performance cost of an extra function call during validation. If keyword logic can be expressed via some other JSON Schema then macro keyword definition is more efficient (see below).

Example. range and exclusiveRange keywords using compiled schema:

ajv.addKeyword({
  keyword: "range",
  type: "number",
  compile([min, max], parentSchema) {
    return parentSchema.exclusiveRange === true
      ? (data) => data > min && data < max
      : (data) => data >= min && data <= max
  },
  errors: false,
  metaSchema: {
    // schema to validate keyword value
    type: "array",
    items: [{type: "number"}, {type: "number"}],
    minItems: 2,
    additionalItems: false,
  },
})

const schema = {
  range: [2, 4],
  exclusiveRange: true,
}
const validate = ajv.compile(schema)
console.log(validate(2.01)) // true
console.log(validate(3.99)) // true
console.log(validate(2)) // false
console.log(validate(4)) // false

See note on errors and asynchronous keywords in the previous section.

# Define keyword with "macro" function

Keyword definition:

interface MacroKeywordDefinition extends FuncKeywordDefinition {
  macro: (schema: any, parentSchema: AnySchemaObject, it: SchemaCxt) => AnySchema
}

"Macro" function is called during schema compilation. It is passed schema, parent schema and schema compilation context and it should return another schema that will be applied to the data in addition to the original schema.

It is an efficient approach (in cases when the keyword logic can be expressed with another JSON Schema), because it is usually easy to implement and there is no extra function call during validation.

In addition to the errors from the expanded schema macro keyword will add its own error in case validation fails.

Example. range and exclusiveRange keywords from the previous example defined with macro:

ajv.addKeyword({
  keyword: "range",
  type: "number",
  macro: ([minimum, maximum]) => ({minimum, maximum}), // schema with keywords minimum and maximum
  // metaSchema: the same as in the example above
})

Macro keywords can be recursive - i.e. return schemas containing the same keyword. See the example of defining a recursive macro keyword deepProperties in the test (opens new window).

# Schema compilation context

Schema compilation context SchemaCxt (opens new window) is available in property it of KeywordCxt (opens new window) (and it is also the 3rd parameter of compile and macro keyword functions). See types in the source code on the properties you can use in this object.

# Validation time variables

All function scoped variables available during validation are defined in names (opens new window).

# Reporting errors

All keywords can define error messages with KeywordErrorDefinition object passed as error property of keyword definition:

interface KeywordErrorDefinition {
  message: string | ((cxt: KeywordErrorCxt) => Code)
  params?: (cxt: KeywordErrorCxt) => Code
}

code keywords can pass parameters to these functions via cxt.setParams (see implementations of pre-defined keywords), other keywords can only set a string message this way.

Another approach for reporting errors can be used for validate and compile keyword - they can define errors by assigning them to .errors property of the validation function. Asynchronous keywords can return promise that rejects with new Ajv.ValidationError(errors), where errors is an array of validation errors (if you don't want to create errors in asynchronous keyword, its validation function can return the promise that resolves with false).

Each error object in errors array should at least have properties keyword, message and params, other properties will be added.

If keyword doesn't define or return errors, the default error will be created in case the keyword fails validation.