Skip to content

Latest commit

 

History

History
869 lines (557 loc) · 24.6 KB

json-schema.md

File metadata and controls

869 lines (557 loc) · 24.6 KB

JSON Schema

In a simple way, JSON Schema is an object with validation keywords.

The keywords and their values define what rules the data should satisfy to be valid.

[[toc]]

JSON Schema draft-2019-09

v7 added support for all new keywords in draft-2019-09:

There is also support for $dynamicAnchor/$dynamicRef from the next version of JSON Schema draft that will replace $recursiveAnchor/$recursiveRef.

type

type keyword requires that the data is of certain type (or some of types). Its value can be a string (the allowed type) or an array of strings (multiple allowed types).

Type can be: number, integer, string, boolean, array, object or null.

Examples

  1. schema: {type: "number"}

    valid: 1, 1.5

    invalid: "abc", "1", [], {}, null, true

  1. schema: {type: "integer"}

    valid: 1, 2

    invalid: "abc", "1", 1.5, [], {}, null, true

  1. schema: {type: ["number", "string"]}

    valid: 1, 1.5, "abc", "1"

    invalid: [], {}, null, true

All examples above are JSON Schemas that only require data to be of certain type to be valid.

Most other keywords apply only to a particular type of data. If the data is of different type, the keyword will not apply and the data will be considered valid.

In v7 Ajv introduced Strict types mode that makes these mistakes less likely by requiring that types are constrained with type keyword whenever another keyword that applies to specific type is used.

Keywords for numbers

maximum / minimum and exclusiveMaximum / exclusiveMinimum

The value of keyword maximum (minimum) should be a number. This value is the maximum (minimum) allowed value for the data to be valid.

The value of keyword exclusiveMaximum (exclusiveMinimum) should be a number. This value is the exclusive maximum (minimum) allowed value for the data to be valid (the data equal to this keyword value is invalid).

::: warning Please note Boolean value for keywords exclusiveMaximum (exclusiveMinimum) is no longer supported. :::

Examples

  1. schema: {type: "number", maximum: 5}

    valid: 4, 5

    invalid: 6, 7

  2. schema: {type: "number", minimum: 5}

    valid: 5, 6

    invalid: 4, 4.5

  3. schema: {type: "number", exclusiveMinimum: 5}

    valid: 6, 7

    invalid: 4.5, 5

multipleOf

The value of the keyword should be a number. The data to be valid should be a multiple of the keyword value (i.e. the result of division of the data on the value should be integer).

Examples

  1. schema: {type: "number", multipleOf: 5}

    valid: 5, 10

    invalid: 1, 4

  1. schema: {type: "number", multipleOf: 2.5}

    valid: 2.5, 5, 7.5

    invalid: 1, 4

Keywords for strings

maxLength / minLength

The value of the keywords should be a number. The data to be valid should have length satisfying this rule. Unicode pairs are counted as a single character.

Examples

  1. schema: {type: "string", maxLength: 5}

    valid: "abc", "abcde"

    invalid: "abcdef"

  1. schema: {type: "string", minLength": 2}

    valid: "ab", "😀😀"

    invalid: "a", "😀"

pattern

The value of the keyword should be a string. The data to be valid should match the regular expression defined by the keyword value.

Ajv uses new RegExp(value, "u") to create the regular expression that will be used to test data.

Example

schema: {type: "string", pattern: "[abc]+"}

valid: "a", "abcd", "cde"

invalid: "def", ""

format

The value of the keyword should be a string. The data to be valid should match the format with this name.

Ajv does not include any formats, they can be added with ajv-formats plugin.

Example

schema: {type: "string", format: "ipv4"}

valid: "192.168.0.1"

invalid: "abc"

Keywords for arrays

maxItems / minItems

The value of the keywords should be a number. The data array to be valid should not have more (less) items than the keyword value.

Example

schema: {type: "array", maxItems: 3}

valid: [], [1], ["1", 2, "3"]

invalid: [1, 2, 3, 4]

uniqueItems

The value of the keyword should be a boolean. If the keyword value is true, the data array to be valid should have unique items.

Example

schema: {type: "array", uniqueItems: true}

valid: [], [1], ["1", 2, "3"]

invalid: [1, 2, 1], [{a: 1, b: 2}, {b: 2, a: 1}]

items

The value of the keyword should be an object or an array of objects.

If the keyword value is an object, then for the data array to be valid each item of the array should be valid according to the schema in this value. In this case the additionalItems keyword is ignored.

If the keyword value is an array, then items with indices less than the number of items in the keyword should be valid according to the schemas with the same indices. Whether additional items are valid will depend on additionalItems keyword.

Examples

  1. schema: {type: "array", items: {type: "integer"}}

    valid: [1,2,3], []

    invalid: [1,"abc"]

  1. schema:

    {
  type: "array",
  items: [{type: "integer"}, {type: "string"}]
}

    valid: [1], [1, "abc"], [1, "abc", 2], []

    invalid: ["abc", 1], ["abc"]

The schema in example 2 will log warning by default (see strictTuples option), because it defines unconstrained tuple. To define a tuple with exactly 2 elements use minItems and additionalItems keywords (see example 1 in additionalItems).

additionalItems

The value of the keyword should be a boolean or an object.

If items keyword is not present or it is an object, additionalItems keyword should be ignored regardless of its value. By default Ajv will throw exception in this case - see Strict mode

If items keyword is an array and data array has not more items than the length of items keyword value, additionalItems keyword is also ignored.

If the length of data array is bigger than the length of "items" keyword value than the result of the validation depends on the value of additionalItems keyword:

  • false: data is invalid
  • true: data is valid
  • an object: data is valid if all additional items (i.e. items with indices greater or equal than "items" keyword value length) are valid according to the schema in "additionalItems" keyword.

The schemas in examples 2-3 log warning by default, use option strictTuples: false to allow)

Examples

  1. schema:

    {
  type: "array",
  items: [{type: "integer"}, {type: "integer"}],
  minItems: 2
  additionalItems: false
}

    valid: [1, 2]

    invalid: [], [1], [1, 2, 3], [1, "abc"] (any wrong number of items or wrong type)

  2. schema:

    {
  type: "array",
  items: [{type: "integer"}, {type: "integer"}],
  additionalItems: true
}

    valid: [], [1, 2], [1, 2, 3], [1, 2, "abc"]

    invalid: ["abc"], [1, "abc", 3]

  3. schema:

    {
  type: "array",
  items: [{type: "integer"}, {type: "integer"}],
  additionalItems: {type: "string"}
}

    valid: [], [1, 2], [1, 2, "abc"]

    invalid: ["abc"], [1, 2, 3]

contains

The value of the keyword is a JSON Schema. The array is valid if it contains at least one item that is valid according to this schema.

Example

schema: {type: "array", contains: {type: "integer"}}

valid: [1], [1, "foo"], any array with at least one integer

invalid: [], ["foo", "bar"], any array without integers

maxContains / minContains

The value of these keywords should be an integer.

Without contains keyword they are ignored (logs error or throws exception in ajv strict mode).

The array is valid if it contains at least minContains items and no more than maxContains items that are valid against the schema in contains keyword.

Example

schema:

{
  type: "array",
  contains: {type: "integer"},
  minContains: 2,
  maxContains: 3
}

valid: [1, 2], [1, 2, 3, "foo"], any array with 2 or 3 integers

invalid: [], [1, "foo"], [1, 2, 3, 4], any array with fewer than 2 or more than 3 integers

unevaluatedItems

The value of this keyword is a JSON Schema (can be a boolean).

This schema will be applied to all array items that were not evaluated by other keywords for items (items, additionalItems and contains) in the current schema and all sub-schemas that were valid for this data instance. It includes:

  • all subschemas schemas in allOf and $ref keywords
  • valid sub-schemas in oneOf and anyOf keywords
  • sub-schema in if keyword
  • sub-schemas in then or else keywords that were applied based on the validation result by if keyword.

The only scenario when this keyword would be applied to some items is when items keyword value is an array of schemas and additionalItems was not present (or did not apply, in case it was present in some invalid subschema).

Some user-defined keywords can also make items "evaluated".

Example

schema:

{
  type: "array",
  items: [
    {type: "number"},
    {type: "number"}
  ],
  unevaluatedItems: false,
  anyOf: [
    {items: [true, true, {type: "number"}]},
    {items: [true, true, {type: "boolean"}]}
  ]
}

valid: [1, 2, 3], [1, 2, true]

invalid:

  • [1, 2] - the third item is not present
  • [1, 2, "3"] - the third item is "unevaluated"

See tests for unevaluatedItems keyword for other examples.

Keywords for objects

maxProperties / minProperties

The value of the keywords should be a number. The data object to be valid should have not more (less) properties than the keyword value.

Example

schema: {type: "object", maxProperties: 2 }

valid: {}, {a: 1}, {a: "1", b: 2}

invalid: {a: 1, b: 2, c: 3}

required

The value of the keyword should be an array of unique strings. The data object to be valid should contain all properties with names equal to the elements in the keyword value.

Example

schema: {type: "object", required: ["a", "b"]}

valid: {a: 1, b: 2}, {a: 1, b: 2, c: 3}

invalid: {}, {a: 1}, {c: 3, d: 4}

properties

The value of the keyword should be a map with keys equal to data object properties. Each value in the map should be a JSON Schema. For data object to be valid the corresponding values in data object properties should be valid according to these schemas.

::: warning Please note properties keyword does not require that the properties mentioned in it are present in the object (see examples). :::

Example

schema:

{
  type: "object",
  properties: {
    foo: {type: "string"},
    bar: {
      type: "number",
      minimum: 2
    }
  }
}

valid: {}, {foo: "a"}, {foo: "a", bar: 2}

invalid: {foo: 1}, {foo: "a", bar: 1}

patternProperties

The value of this keyword should be a map where keys should be regular expressions and the values should be JSON Schemas. For data object to be valid the values in data object properties that match regular expression(s) should be valid according to the corresponding schema(s).

When the value in data object property matches multiple regular expressions it should be valid according to all the schemas for all matched regular expressions.

::: warning Please note

  1. patternProperties keyword does not require that properties matching patterns are present in the object (see examples).
  2. By default, Ajv does not allow schemas where patterns in patternProperties match any property name in properties keyword - that leads to unexpected validation results. It can be allowed with option allowMatchingProperties. See Strict mode :::

Example

schema:

{
  type: "object",
  patternProperties: {
    "^fo.*$": {type: "string"},
    "^ba.*$": {type: "number"}
  }
}

valid: {}, {foo: "a"}, {foo: "a", bar: 1}

invalid: {foo: 1}, {foo: "a", bar: "b"}

additionalProperties

The value of the keyword should be either a boolean or a JSON Schema.

If the value is true the keyword is ignored.

If the value is false the data object to be valid should not have "additional properties" (i.e. properties other than those used in "properties" keyword and those that match patterns in "patternProperties" keyword).

If the value is a schema for the data object to be valid the values in all "additional properties" should be valid according to this schema.

Examples

  1. schema:

    {
  type: "object",
  properties: {
    foo: {type: "number"}
  },
  patternProperties: {
    "^.*r$": {type: "number"}
  },
  additionalProperties: false
}

    valid: {}, {foo: 1}, {foo: 1, bar: 2}

    invalid: {a: 3}, {foo: 1, baz: 3}

  2. schema:

    {
  type: "object",
  properties: {
    foo: {type: "number"}
  },
  patternProperties: {
    "^.*r$": {type: "number"}
  },
  additionalProperties: {type: "string"}
}

    valid: {}, {a: "b"}, {foo: 1}, {foo: 1, bar: 2}, {foo: 1, bar: 2, a: "b"}

    invalid: {a: 3}, {foo: 1, baz: 3}

  3. schema:

    {
  type: "object",
  properties: {
    foo: {type: "number"}
  },
  additionalProperties: false,
  anyOf: [
    {
      properties: {
        bar: {type: "number"}
      }
    },
    {
      properties: {
        baz: {type: "number"}
      }
    }
  ]
}

    valid: {}, {foo: 1}

    invalid: {bar: 2}, {baz: 3}, {foo: 1, bar: 2}, etc.

dependencies

This keyword is deprecated. The same functionality is available with keywords dependentRequired and dependentSchemas.

The value of the keyword is a map with keys equal to data object properties. Each value in the map should be either an array of unique property names ("property dependency" - see dependentRequired keyword) or a JSON Schema ("schema dependency" - see dependentSchemas keyword).

For property dependency, if the data object contains a property that is a key in the keyword value, then to be valid the data object should also contain all properties from the array of properties.

For schema dependency, if the data object contains a property that is a key in the keyword value, then to be valid the data object itself (NOT the property value) should be valid according to the schema.

Examples

  1. schema (property dependency):

    {
  type: "object",
  dependencies: {
    foo: ["bar", "baz"]
  }
}

    valid: {foo: 1, bar: 2, baz: 3}, {}, {a: 1}

    invalid: {foo: 1}, {foo: 1, bar: 2}, {foo: 1, baz: 3}

  2. schema (schema dependency):

    {
  type: "object",
  dependencies: {
    foo: {
      properties: {
        bar: {type: "number"}
      }
    }
  }
}

    valid: {}, {foo: 1}, {foo: 1, bar: 2}, {a: 1}

    invalid: {foo: 1, bar: "a"}

dependentRequired

The value of this keyword should be a map with keys equal to data object properties. Each value in the map should be an array of unique property names.

If the data object contains a property that is a key in the keyword value, then to be valid the data object should also contain all properties from the corresponding array of properties in this keyword.

Example

schema:

{
  type: "object",
  dependentRequired: {
    foo: ["bar", "baz"]
  }
}

valid: {foo: 1, bar: 2, baz: 3}, {}, {a: 1}

invalid: {foo: 1}, {foo: 1, bar: 2}, {foo: 1, baz: 3}

dependentSchemas

The value of the keyword should be a map with keys equal to data object properties. Each value in the map should be a JSON Schema.

If the data object contains a property that is a key in the keyword value, then to be valid the data object itself (NOT the property value) should be valid according to the corresponding schema in this keyword.

Example

schema:

{
  type: "object",
  dependentSchemas: {
    foo: {
      properties: {
        bar: {type: "number"}
      }
    }
  }
}

valid: {}, {foo: 1}, {foo: 1, bar: 2}, {a: 1}

invalid: {foo: 1, bar: "a"}

propertyNames

The value of this keyword is a JSON Schema.

For data object to be valid each property name in this object should be valid according to this schema.

Example

schema (requires email format from ajv-formats):

{
  type: "object",
  propertyNames: {
    format: "email"
  }
}

valid: {"[email protected]": "any", "[email protected]": "any"}

invalid: {foo: "any value"}

unevaluatedProperties

The value of this keyword is a JSON Schema (can be a boolean).

This schema will be applied to all properties that were not evaluated by other keywords for properties (properties, patternProperties and additionalProperties) in the current schema and all sub-schemas that were valid for this data instance. It includes:

  • all subschemas schemas in allOf and $ref keywords
  • valid sub-schemas in oneOf and anyOf keywords
  • sub-schema in if keyword
  • sub-schemas in then or else keywords that were applied based on the validation result by if keyword.

Some user-defined keywords can also make properties "evaluated".

Example

schema:

{
  type: "object",
  required: ["foo"],
  properties: {foo: {type: "number"}},
  unevaluatedProperties: false,
  anyOf: [
    {
      required: ["bar"],
      properties: {bar: {type: "number"}}
    }
    {
      required: ["baz"],
      properties: {baz: {type: "number"}}
    }
  ]
}

valid: {foo: 1, bar: 2}, {foo: 1, baz: 2}, {foo: 1, bar: 2, baz: 3}

invalid:

  • {foo: 1} - neither bar nor baz are present
  • {foo: 1, bar: 2, boo: 3} - boo is unevaluated
  • {foo: 1, bar: 2, baz: "3"} - not valid against the 2nd subschema, so baz is "unevaluated".

See tests for unevaluatedProperties keyword for other examples.

Keywords for all types

enum

The value of the keyword should be an array of unique items of any types. The data is valid if it is deeply equal to one of items in the array.

Example

schema: {enum: [2, "foo", {foo: "bar" }, [1, 2, 3]]}

valid: 2, "foo", {foo: "bar"}, [1, 2, 3]

invalid: 1, "bar", {foo: "baz"}, [1, 2, 3, 4], any value not in enum

const

The value of this keyword can be anything. The data is valid if it is deeply equal to the value of the keyword.

Example

schema: {const: "foo"}

valid: "foo"

invalid: any other value

The same can be achieved with enum keyword using the array with one item. But const keyword is more than just a syntax sugar for enum. In combination with the $data reference it allows to define equality relations between different parts of the data. This cannot be achieved with enum keyword even with $data reference because $data cannot be used in place of one item - it can only be used in place of the whole array in enum keyword.

Example

schema:

{
  type: "object",
  properties: {
    foo: {type: "number"},
    bar: {const: {$data: "1/foo"}}
  }
}

valid: {foo: 1, bar: 1}, {}

invalid: {foo: 1}, {bar: 1}, {foo: 1, bar: 2}

Compound keywords

not

The value of the keyword should be a JSON Schema. The data is valid if it is invalid according to this schema.

Example

schema: {type: "number", not: {minimum: 3}}

valid: 1, 2

invalid: 3, 4

oneOf

The value of the keyword should be an array of JSON Schemas. The data is valid if it matches exactly one JSON Schema from this array. Validators have to validate data against all schemas to establish validity according to this keyword.

Example

schema:

{
  type: "number",
  oneOf: [{maximum: 3}, {type: "integer"}]
}

valid: 1.5, 2.5, 4, 5

invalid: 2, 3, 4.5, 5.5

anyOf

The value of the keyword should be an array of JSON Schemas. The data is valid if it is valid according to one or more JSON Schemas in this array. Validators only need to validate data against schemas in order until the first schema matches (or until all schemas have been tried). For this reason validating against this keyword is faster than against "oneOf" keyword in most cases.

Example

schema:

{
  type: "number",
  anyOf: [{maximum: 3}, {type: "integer"}]
}

valid: 1.5, 2, 2.5, 3, 4, 5

invalid: 4.5, 5.5

allOf

The value of the keyword should be an array of JSON Schemas. The data is valid if it is valid according to all JSON Schemas in this array.

Example

schema:

{
  type: "number",
  allOf: [{maximum: 3}, {type: "integer"}]
}

valid: 2, 3

invalid: 1.5, 2.5, 4, 4.5, 5, 5.5

if/then/else

These keywords allow to implement conditional validation. Their values should be valid JSON Schemas (object or boolean).

If if keyword is absent, the validation succeeds.

If the data is valid against the sub-schema in if keyword, then the validation result is equal to the result of data validation against the sub-schema in then keyword (if then is absent, the validation succeeds).

If the data is invalid against the sub-schema in if keyword, then the validation result is equal to the result of data validation against the sub-schema in else keyword (if else is absent, the validation succeeds).

Examples

  1. schema:

    {
  type: "object",
  if: {properties: {foo: {minimum: 10}}},
  then: {required: ["bar"]},
  else: {required: ["baz"]}
}

    valid:

    • {foo: 10, bar: true }
    • {}
    • {foo: 1, baz: true }

    invalid:

    • {foo: 10} (bar is required)
    • {foo: 10, baz: true } (bar is required)
    • {foo: 1} (baz is required)

  1. schema:

    {
  type: "integer",
  minimum: 1,
  maximum: 1000,
  if: {minimum: 100},
  then: {multipleOf: 100},
  else: {
    if: {minimum: 10},
    then": {multipleOf: 10}
  }
}

    valid: 1, 5, 10, 20, 50, 100, 200, 500, 1000

    invalid:

    • -1, 0 (<1)
    • 2000 (>1000)
    • 11, 57, 123 (any integer with more than one non-zero digit)
    • non-integers

Metadata keywords

JSON Schema specification defines several metadata keywords that describe the schema itself but do not perform any validation.

  • title and description: information about the data represented by that schema
  • $comment: information for developers. With option $comment Ajv logs or passes the comment string to the user-supplied function. See Options.
  • default: a default value of the data instance, see Assigning defaults.
  • examples: an array of data instances. Ajv does not check the validity of these instances against the schema.
  • readOnly and writeOnly: marks data-instance as read-only or write-only in relation to the source of the data (database, api, etc.).
  • contentEncoding: RFC 2045, e.g., "base64".
  • contentMediaType: RFC 2046, e.g., "image/png".

::: warning Please note Ajv does not implement validation of the keywords examples, contentEncoding and contentMediaType but it reserves them. If you want to create a plugin that implements any of them, it should remove these keywords from the instance. :::