jayschema

A comprehensive JSON Schema validator for Node.js

JaySchema

A JSON Schema validator for Node.js.

  • Complete validation coverage of JSON Schema Draft v4.
  • Optional dynamic loader for referenced schemas (load schemas from a database or the web)
  • Useful error messages.
  • NEW: Supports custom validators for the format keword.
npm install jayschema
var JaySchema = require('jayschema');
var js = new JaySchema();
var instance = 64;
var schema = { "type": "integer", "multipleOf": 8 };
 
// synchronous… 
console.log('synchronous result:', js.validate(instance, schema));
 
// …or async 
js.validate(instance, schema, function(errs) {
    if (errs) { console.error(errs); }
    else { console.log('async validation OK!'); }
});

Here the Geographic Coordinate schema is loaded over HTTP. You can also supply your own loader—for example, to load schemas from a database.

var JaySchema = require('jayschema');
var js = new JaySchema(JaySchema.loaders.http);     // we provide the HTTP loader here 
                                                    // you could load from a DB instead 
 
var instance = { "location": { "latitude": 48.8583, "longitude": 2.2945 } };
var schema = {
    "type": "object",
    "properties": {
        "location": { "$ref": "http://json-schema.org/geo" }
    }
};
 
js.validate(instance, schema, function(errs) {
  if (errs) { console.error(errs); }
  else { console.log('validation OK!'); }
});

Create a custom validator for the JSON Schema format keyword:

var JaySchema = require('jayschema');
var js = new JaySchema();
 
js.addFormat('phone-us', function(value) {
  var PHONE_US_REGEXP = /^\(?([0-9]{3})\)?[-. ]?([0-9]{3})[-. ]?([0-9]{4})$/;
  if (PHONE_US_REGEXP.test(value)) { return null; }
  return 'must be a US phone number';
});
 
var instance = '212-555-';
var schema = { "type": "string", "format": "phone-us" };
 
console.log(js.validate(instance, schema));
// fails with error description: must be a US phone number 
  • Validate JSON server-side:
    • For your JSON-based API
    • For data that you want to store in a NoSQL database
  • No ORM required. Change databases or store data in multiple databases using the same schemas. For example, session data in Redis, permanent data in MongoDB.
  • JSON Schema has a really nice declarative syntax. See the official examples.

(Constructor) The optional loader will be called each time an external $ref is encountered. It should load the referenced schema and return it.

If you don’t reference any external schemas, or if you pre-register all the schemas you’re using, you don’t need to provide a loader.

If you provide a loader, you should call the validate() function asynchronously. That’s because loading involves disk or network I/O, and I/O operations in Node are asynchronous.

Sample loader skeleton:

function loader(refcallback) {
    // ref is the schema to load 
    // [ load your schema! ] 
    if (errorOccurred) {
        callback(err);
    } else {
        callback(null, schema);
    }
}

Validate a JSON object, instance, against the given schema. If you provide a callback, validation will be done asynchronously.

schema can be an actual JSON Schema (a JSON object), or it can be the id of a previously-registered schema (a string).

  • async: Uses the standard Node callback signature. The first argument will be an array of errors, if any errors occurred, or undefined on success.
  • synchronous: If you don’t provide a callback, an array of errors will be returned. Success is indicated by an empty array.

Manually register schema. Useful if you have several related schemas you are working with. The optional id can be used to register a schema that doesn’t have an id property, or which is referenced using a unique id.

Returns: an array of missing schemas. A missing schema is one that was $referenced by the registered schema, but hasn’t been regisetered yet. If no missing schemas were referenced, an empty array is returned.

See Schema loading.

Returns an array of missing schemas. A missing schema is one that was $referenced by a register()ed schema, but the referenced schema has not yet been loaded.

See Schema loading.

Return boolean indicating whether the specified schema id has previously been registered.

See Schema loading.

Add a custom handler for the format keyword. Whenever a schema uses the format keyword, with the given formatName, your handler will be called.

The handler receives the value to be validated, and returns null if the value is valid, or a string (description of the error) if the value is not valid.

A loader can be passed to the constructor, or you can set the loader property at any time. You can define your own loader. JaySchema also includes one built-in loader for your convenience:

Loads external $refs using HTTP. :warning: Caveat: HTTP is inherently unreliable. For example, the network or site may be down, or the referenced schema may not be available any more. You really shouldn’t use this in production, but it’s great for testing.

The maximum depth to recurse when retrieving external $ref schemas using a loader. The default is 5.

The schema loader to use, if any. (The same schema loader that was passed to the JaySchema constructor.) You can change or override this at any time.

JaySchema provides several ways to register externally-referenced schemas.

You use the $ref keyword to pull in an external schema. For example, you might reference a schema that’s available in a local database.

Validation will fail if JaySchema encounters a validation rule that references an external schema, if that schema is not registered.

There are several ways to ensure that all referenced schemas are registered:

Pass a loader callback to the JaySchema constructor. When an external schema is needed, JaySchema will call your loader. See the constructor documentation, above. Using a loader requires you to validate asynchronously.

This works with synchronous or async code.

  1. First, register() the main schemas you plan to use.
  2. Next, call getMissingSchemas, which returns an array of externally-referenced schemas.
  3. Retrieve and register() each missing schema.
  4. Repeat from step 2 until there are no more missing schemas.

This works with synchronous or async code.

Each time you call register(schema), the return value will be an array of missing external schemas that were referenced. You can use this to register the missing schemas.

In other words: calling register(schemaA); will (1) register schemaA and (2) return a list of missing schemas that were referenced by schemaA.

If, instead, you want the list of all missing schemas referenced by all registrations that have been done so far, use the getMissingSchemas() method, above.

For the format keyword, JaySchema allows you to add custom validation functions (see addFormat). In addition, the following built-in formats are supported:

  • date-time: Must match the date-time specification given in RFC 3339, Section 5.6. This expects both a date and a time. For date-only validation or time-only validation, JaySchema supports the older draft v3 date and time formats.
  • hostname: Must match the “Preferred name syntax” given in RFC 1034, Section 3.5, with the exception that hostnames are permitted to begin with a digit, as per RFC 1123 Section 2.1.
  • email: Must match RFC 5322, Section 3.4.1, with the following limitations: quoted-strings, domain-literals, comments, and folding whitespace are not supported; the domain portion must be a hostname as in the hostname keyword.
  • ipv4: Must be a dotted-quad IPv4 address.
  • ipv6: Must be a valid IPv6 address as per RFC 2373 section 2.2.
  • uri: As in RFC 3986 Appendix A, including relative URIs (no scheme part, fragment-only), with the exception that well-formedness of internal elements, including percent encoding and authority strings, is not verified.