Skip to content
This repository

JSON Schema validator for Node.js

branch: master

issue #41

latest commit c9e42bd5ba
Martin Zagora authored

z-schema validator

NPM version Dependency Status

JSON Schema validator for Node.js (draft4 version)

Coded according to:

json-schema documentation, json-schema-core, json-schema-validation, json-schema-hypermedia

Passing all tests here (even optional, except zeroTerminatedFloats and some URI tests, see more info in #18):


Will try to maintain this as much as possible, all bug reports welcome.

Grunt automatization

If you need to automatize validation of your schemas, there's a Grunt plugin grunt-z-schema by Petr Bela

How does it compare to others?

Basic Usage

var ZSchema = require("z-schema");
ZSchema.validate(json, schema)
        // successful validation
        // there might be warnings: console.log(report.warnings)

There is also support for sync mode like this:

var validator = new ZSchema({ sync: true });
var valid = validator.validate(json, schema);
if (!valid) {
    var error = validator.getLastError();

Using traditional callback:

ZSchema.validate(json, schema, function(err, report){
    // successful validation
    // there might be warnings: console.log(report.warnings)

If you need just to validate your schema, you can do it like this:

var validator = new ZSchema();

Or with Node.js style callback:

var validator = new ZSchema();
validator.validateSchema(schema, function (err, report) {
    if (err) ...

Remote references in schemas

Your schemas can include remote references that should be real URIs (more on that here) so validator can make a request and download the schema needed. Validator automatically caches these remote requests so they are not repeated with every validation.

In case you don't have a real server or you'd like to load files from different location, you can preload remote locations into the validator like this:

var fileContent = fs.readFileSync(__dirname + '/../json_schema_test_suite/remotes/integer.json', 'utf8');
ZSchema.setRemoteReference('http://localhost:1234/integer.json', fileContent);

http://localhost:1234/integer.json doesn't have to be online now, all schemas referencing it will validate against string that was passed to the function.

Advanced Usage

You can pre-compile schemas (for example on your server startup) so your application is not bothered by schema compilation and validation when validating ingoing / outgoing objects.


var validator = new ZSchema();

Or callback:

var validator = new ZSchema();
validator.compileSchema(schema, function (err, compiledSchema) {

Then you can re-use compiled schemas easily just the same way as non-compiled.

var validator = new ZSchema();
validator.validate(json, compiledSchema)
        // ...

Custom format validators

You can add validation for your own custom string formats like this: (these are added to all validator instances, because it would never make sense to have multiple functions to validate format with the same name)

var validator = new ZSchema();

ZSchema.registerFormat('xstring', function (str) {
    return str === 'xxx'; // return true/false as a result of validation

validator.validate('xxx', {
    'type': 'string',
    'format': 'xstring'

Custom validators can also be async:

Using promises:

ZSchema.registerFormat('xstring', function (str) {
    return Q.delay(1000).thenResolve(return str === 'xxx'); // return a promise for validation result

Using classic callback:

ZSchema.registerFormat('xstring', function (str, callback) {
        callback(null, str === 'xxx');
        // or return custom error: callback(new Error('Bad, bad value!'))
    }, 2000)

Any exception thrown (or returned via classic callback) in custom validation function is written into validation error:

ZSchema.registerFormat('xstring', function (str) {
    throw new Error('Bad, bad value!');

And then expect errors to contain something like this:

[{ code: 'FORMAT',
    message: 'xstring format validation failed: Error: Bad, bad value!',
    path: '#/test',
    params: { format: 'xstring', error: [Error: Bad, bad value!] } } ]

Strict validation

When creating new instance of validator, you can specify some options that will alter the validator behaviour like this:

var validator = new ZSchema({
    option: true
  • noExtraKeywords: true/false

when true, do not allow unknown keywords in schema

  • noZeroLengthStrings: true/false

when true, always adds minLength: 1 to schemas where type is string

  • noTypeless: true/false

when true, every schema must specify a type

  • forceAdditional: true/false

when true, forces not to leave out some keys on schemas (additionalProperties, additionalItems)

  • forceProperties: true/false

when true, forces not to leave out properties or patternProperties on type-object schemas

  • forceItems: true/false

when true, forces not to leave out items on array-type schemas

  • forceMaxLength: true/false

when true, forces not to leave out maxLength on string-type schemas, when format or enum is not specified

Alternatively, you can turn on all of the above options with:

var validator = new ZSchema({
    strict: true

More options

  • noSchemaCache: true/false

when true, disables caching of compiled schemas - cache is used to resolve references to other schemas by their ID

  • strictUris: true/false

when true, uri's need to be in full rfc3986 format, by default checks for uri-fragment, more info in #18

Pull requests

Avoid JSHint errors - settings for the JSHint are specified in .jshintrc. You can check for errors using grunt command which runs both jshint and mocha tests. Please check for errors before opening any pull requests.


Something went wrong with that request. Please try again.