Validation-sanitizing functionality for web frameworks, browsers, and you-name-it


Validation-sanitizing functionality for web frameworks, browsers, and you-name-it.

Sanidate (SANItize + valiDATE) is a JavaScript library for validating and sanitizing user-supplied data. It uses a developer-defined constraint schema and converts input data into valid and properly formatted data. It can be used both on server- and client-side, and provides middlewares for Express.js, and FragRouter, as well as a jQuery plugin.

Sanidate grew out of author's frustration with existing validation frameworks which tend to produce a lot of cruft, and/or create unreasonably hard-to-read, verbose code. Data validation and sanitizing is a very common task (virtually all user input has to be validated and/or sanitized at some point), repetitive, and outright boring. On top of it, it's not even the most important part of the business logic, but merely an overture.

Sanidate tries to reduce validation and sanitizing to what it's supposed to be, merely a checklist for what the data should look like before it can be consumed by business logic code. It also tries to reduce validation and sanitizing into a single pass (called 'sanidation'), so you don't have to do two mentally very close tasks tasks in two separate passes. Finally, it systematizes data into a convenient package, so you can access it with ease, rather than hunt for it in multiple different places.

The core of Sanidate is its constraints system. Each constraint is a function which does two things:

  • converts data into intended data type
  • checks the validity of the data

These constraints are named, and a sanidation schema can be contructed using the names, instead of references to functions. This makes for concise representation of the sanidation schema (more on that in the following section).

Unlike some of the validation frameworks for JavaScript, Sanidate is fully asynchronous, which means it supports asynchronous operations like database lookups in its constraint functions. In fact, it ships with two Mongoose-specific constraints all good to go.

To sanidate some data, you need two things:

  • The actual data as an object (key-value parirs)
  • The sanidation schema (more on that in next section)

When you have both, simply call the sanidate.check() method:

sanidate.check(data, schema, function(err, data) {
  if (err) { return console.log('Error!'); }
  // Do something useful with the data, like:

Sanidation schema is an object that describes your desired output. Since input is almost always string, Sanidate not only checks the validity of the strings, but also converts them into desired type and/or format. The conversion and validation for a single constraint are both controlled by a single constraint function.

Let's take a look at a simple sanidation schema and then discuss its structure in more detail:

var schema = {
  name: 'required',
  email: 'email',
  age: 'integer',
  username: 'required',
  profilePicture: [
    ['custom', checkStorage]

You will first notice that, for most part, schema is simply a key-value pair, where keys are parameter names, and values are strings or arrays.

Strings values are very simple: they are names of built-in constraints (e.g., 'required', 'email', 'numeric'...).

For specifying multiple constraints, or specifying constraints with parameters, arrays must be used. Each member of the constraint array is a single constraint, and it can be either a string or an array. Strings are the same as string values above: just names of built-in constraints. Constraint represented by an array has the name of the built-in constraint as it's first member, and other members are arguments passed to the constraint function.

In our example, profilePicture parameter has a single constraint called 'custom', with a single argument, which is a reference to a function caleld checkStorage (which, you can imagine, checks the hard drive for profile image file and fails validation if one is not found).

The built-in constraints, such as 'required' or 'email', are all stored in sanidate.funcs object. They are looked up internally when validation takes place, but are exposed for your convenience, if you ever need to customize them, or add new constraints yourself.

One thing to note about the constraints such as 'numeric' is that they will actually convert the input data into an appropriate type (e.g., 'numeric' converts to a float, 'integer' converts to an integer, and 'date' converts to Date object).

Another important note is that all constraints require a value to be present. Currently, there is no support for optional constraints that are applied only when data is present. If you wish to apply constraints only if a value is present, use the 'optional' constraint as the first constraint, followed by other constraints.

If you use more than one constraint on a parameter, you should keep in mind that each constraint transforms data before passing to the next one. For example, if you put a 'match' constraint after the 'date' constraint, this will yield unexpected results, because the value will have been transformed into a Date object before reaching the 'match' constraint.

Think of constraint arrays as layers of filters one on top of the other on which you drip the data. If data is prevented to pass any of the layers, subsequent layers will not even see it. Any layer always sees data as filtered by previous layers. This is also important to remember when handling validation errors, because, for a set of chained constarints, error will only be registered for the first constraint that reports validations failure.

The built-in constraints are constructed in such a way that you usually only need one of them at a time, so chaining should generally not be required. For example, 'max' constraint may also serve as either 'numeric' or 'integer' constraint becuase it will fail if value is not a valid number.

Sanidate ships with a handful of (arguably) very useful constraints. These are (with arguments in square brackets):

  • required: Simply fails if value is not supplied
  • optional: [def] If no value is supplied, return the default value (def), and interrupt execution of further constraints; must be the first constraint if used with other constraints
  • optionalIfPresent: [params, def] Makes parameter optional if parameters in params array are present, otherwise requires it
  • match: [pattern] Fails if value does not match the regexp pattern
  • enum: [allowed] Fails if value does not appear in allowed array of strings
  • numeric: Forces conversion to float, and fails when conversion fails
  • integer: Forces conversion to integer and fails when conversion fails
  • max: [x, integer, equality] Forces conversion to float or integer (if integer flag is set to true), and, depending on equality boolean falg, fails if greater than or greater and equal to x
  • min: [x, integer, equality] Same as max, but fails if less than / less than or equal tox
  • date: [resetTime] Forces conversion to Date, and fails if conversion fails (does not work with any particular date format, so if date formatting is required, insert the match constraint before date constraint)
  • email: Fails if value is not an email
  • zip: Fails if value is not a 5-digit number (such as US zip code)
  • phone: [digitsOnly] Fails if value does not contain 10 digits (disregaring any other non-numeric characters), and returns either only the digits (if digitsOnly flag is set to true), or formatted phone number (currently only supports US phone format)
  • isTrue: Returns true if value is 'true', 'yes', 'on', or '1', otherwise returns false, and never fails
  • isNotTrue: Returns false if value is 'false', 'no', 'off', or '0', otherwise returns true, and never fails
  • strictBoolean: [optional, def] Requires parameter to be strictly boolean, and fails if it's not, or always passes if optional flag is set.
  • isDocument: [Model, key] Looks up Mongoose model using either supplied optional key, or parameter name as key name, and fails if no documents are found; value is converted to returned document
  • isNotDocument: [Model, key] Same as 'isDocument', but fails if document is found, and returns original value on success.
  • custom: [func] Uses the func function as constraint
  • derive: [paramName, func] Uses parameter paramName from original user-supplied data, and applies func validation function to its value

Note that you can use multiple 'custom' constraints for any user-supplied data.

Sanidation errors are returned as first argument to the callback you pass to sanidate.check(). If there are no errors, you will receive null instead.

If there are any errors, the error object will have two keys.

The count represents an integer count of errors. This number can never be higher than the number of parameters that were sanidated.

The errors key will contain a object mapping between parameter names and constraint names. Constraint names in the mapping represent the names of constraints for which the parameter values failed validation.

It is possible to write custom constraint functions and use them with the 'custom' constraint.

Let's take a look at a simple constraint function that checks the age of a date field, and makes sure it is not older than 20 days:

function minAge(v, next) {
  var now =;
  v = new Date(v);
  var maxDifference = 20 * 24 * 60 * 60 * 1000; // 20 days in ms
  var difference = now - v.getTime();
  next(null, difference < maxDifference ? v : null, 'minAge');

The function takes two arguments. First argument, v, is the current value of the data, as sanidated by all previous constraints. The second argument is a callback function which allows Sanidate to continue applying other constraints.

What goes on inside the function is pretty self-evident, so we won't go into details. The important bit is the data we pass on to the callback. Callback takes three arguments.

The first argument is an error object. It is currently not important, as it makes little difference whether you pass an error object. Presence of an error will be treated as failed validation. This behavior will change in future, so it's best to always pass the error object if there has been a non-validation-related error (such as filesystem or database failure). In later versions, Sanidate will be able to somehow tell you that there has been an unexpected condition.

The second argument is sanidated value. It's important to note that only null value is considered a failure. Falsy values like 0, empty string, and even undefined, are not considered failure conditions. So, if you want to let Sanidate know that validation has failed, you must pass a null value as sanidated value.

Last argument is the constraint name. Because there can be multiple 'custom' constraints, Sanidate allows constraint functions to return a different constraint name in the callback. As with all other arguments, null is, again, the special value. If you pass null as constraint name, any further constraint processing will be abandoned, and the value passed to the callback will be used without questions. That is, for example, what 'optional' does. It passes on any value that user has provided, and passes null as constraint name to foce Sanidate to use any provided value (including null).

Custom constraint functions have access to details about the parameter that is being processed (like the parameter name), the original parameter value, and the original values of all other parameters that are being sanidated. This data is stored in this as properties:

  • this.paramName: parameter name
  • this.originalValue: original (raw) value of the parameter
  • this.originalData: all data being sanidated as name-value pairs

Although the 'derive' constraint can be used effectively for the same purpose, we will write our own 'custom' constraint that checks email confirmation, just to demonstrate usage of originalData property.

function emailConfirmation(v, next) {
  next(null, v === this.originalData['email'] ? v : null,

We assume that the email was first entered in a field named email, and then confirmed in a field called emailconfirm:

var schema = {
  email: 'email',
  emailconfirm: [['custom', emailConfrimation]]

To demonstrate the difference between 'custom' constraint and 'derive' constraint in this scenario, let's rewrite the above to use 'derive':

function emailConfirmation(v, o) { return v === o ? v : null; }

var schema = { email: 'email', emailconfirm: [['derive', 'email', emailConfrimation]] };

Note that 'emailconfirm' paramter does not have an 'email' constraint. We don't need to add that, since we know that if 'email' constraint on 'email' field failed, it must also fail for 'emailconfirm' field if it is identical to 'email' field.

Apart from writing function for use with 'custom' constraint, you can also add your constraint as a named constraint to make it act as one of the built-in constraints. To do that, siply add your constraint to sanidate.funcs object:

sanidate.funcs.minAge = function(maxDays) {
  return function minAge(v, next) {
    var now =;
    v = new Date(v);
    var maxDifference = maxDays * 24 * 60 * 60 * 1000; // 20 days in ms
    var difference = now - v.getTime();
    next(null, difference < maxDifference ? v : null, 'minAge');

As you can see, a built-in constraint is a function that returns functions. The outer function is called a setup function, which prepares the validation function. The outer function has access to a paramObject (via this), which contains more information

var data = {
  email: '',
  number: '11'

var schema = {
  name: 'required',
  email: 'email',
  number: 'integer'

sanidate.check(data, schema, function(err, data) {
  if (err) {
    console.log('There were ' + err.count + ' errors:', err.errors);
  // data is:
  //   {email: '', number: 11}
var schema = {
  name: 'required',
  email: 'email',
  number: 'integer'

// Set up middleware using `` call'/user',, function(req, res) {
  if (req.dataError) { return req.send('Error', 400); }

  // Sanidated data is now available as

  req.send('Success!', 200);

function myHandler() {
  var schema = {
    name: 'required',
    email: 'email',
    number: 'integer'

  // get `data` from somewhere

  this.sanidate(data, schema)

Although Sanidate is an AMD module, it doesn't list jQuery as its dependency, since it doesn't actually need it.

If you want to use Sanidate's jQuery plugin, you need to make sure Sanidate is loaded after jQuery, or configure your AMD loader to load jQuery as Sanidate's dependency.

Alternatively, you can run this line after jQuery has been loaded:


After you've successfully activated the plugin, you can use it like this:

var schema = {
  name: 'required',
  email: 'email',
  number: 'integer'

$('#myForm').sanidate(schema, function(err, data) {
  // `err` and `data` are the usual objects

This object contains all of the built-in constraints that can be used by name in the sanidation schema. Each property is a function which may take one or more arguments, and which returns a sanidation function.

Sanidation functions take two arguments, of which first is the value as sanidized by any previous constraints, and second is a callback which chains to next sanidizer or finishes the chain. The sanidation function works more or less the same way as custom functions (see 'Writing custom constraints' section for more details).

You can customize any of the built-in constraints by overloading the function assigned to its key in this object. You can also add new built-in constraints by adding new keys.

Makes value required, and won't pass if there is no value.

Fails if value does not appear in allowed array.

Makes value match a pattern.

Makes value numeric.

Makes value integer.

Makes value be greater than or equal to x. If integer is true, it will only do integer comparison.


Makes value be less than or equal to x. If integer is true, it will only do integer comparison.


Makes value a date.

Makes value match a canonical email format.

Makes vlaue match 5-digit number. It returns the original value on match, not converted to numbers.

Makes value match any of the following:

  • true
  • yes
  • on
  • 1

If the value matches, true is returned, otherwise false is returned.

This validator never fails.

Makes value not match any of the following:

  • false
  • no
  • off
  • 0

If the value matches any of the above, false is returned. Otherwise, true is returned.

This validator never fails.

Makes value match any of the strict boolean values. Fails if optional flag is not true, otherwise stops further sanidation and returns the original value or def is supplied, and original value is absent.

If def is a function, it will be executed with no arguments, and its return value will be used as the default value. Function is always executed after it is determined that default value is needed.


Tests if 10 digits are contained in the value, and returns only the digits as string (digitsOnly === true), or a formatted US phone number (digitsOnly === false).


Tests if a document that has the key-value pair matching the value of the parameter exists in a Mongoose/MongoDB database.

The validator only succeeds if there is a match, and returns the document as new value.

Key name will default to parameter name.

Tests if a document that has the key-value part matching the value of the parameter does not exist in a Mongoose/MongoDB database.

The validator fails if there is such a document, otherwise it returns the value of the parameter intact.

Key name will default to parameter name.

Run a custom validation function. The custom function must have the same signature as normal validation functions, and must call the next callback.

Interrupts sanidation immediately if no value is found, but does not fail validation.

If def is a function, it will be executed with no arguments, and its return value will be used as the default value. Function is always executed after it is determined that default value is needed.

value is found.

Interrupts sanidation immediately if no value is found, and values are found for parameters specified in params array. All parameters listed in params array must be present to make this field optional.

If params is a single string, it will be treated as an array with single member.

Default value of def can be used.

If def is a function, it will be executed with no arguments, and its return value will be used as the default value. Function is always executed after it is determined that default value is needed.


Uses data from another parameter to either derive its own value, or validate its value.

The func function accepts two arguments. First argument is the sanidated value of the current parameter. Second paramter is the original (unsanidated) value of the other parameter. The func must return a value, which is then used for validation. To fail validation, you must return null. All other values (including ) will be considered valid.

Note that 'derive' only operates on the original value of the other parameter, and not the sanidated value.


Runs constraints on a parameter with paramName name, value value, and executes a cb callback when finished. The callback should expect three arguments:

  • error object (null if there were no errors)
  • final value after running sanidators over the original value
  • constraint name if there was a failure of no value was returned

The constraint name is the name of the sanidator that was run last in case there is no value returned from it, or it threw an error.

This method, although exposed, is a private method, and you should not rely on the stability of its API.

Object containing recipes used in sanidate.prepareSchema. Exposed to allow customization. Object maps constraint names with preparation functions which take two arguments:

  • constraint (array whose first member is the constraint name, followed by parameters)
  • module which is used in preparation (see sanidate.prepareSchema documentation)

Function simply performs replacement, and does not need to explicitly return a value. See built-in recipes for examples.

Takes a schema as string or object, and replaces any string function references by matching methods from module object.

This method is meant to be used in scenarios where schema is stored as JSON (and passed either parsed or unparsed), and therefore function and object references could not be made part of the schema.

In such cases, any custom functions and objects have to be made available as a separate module or object whose properties (methods) represent the missing functions or model objects. Instead of using function references, we can use strings representing property names, and this method replaces them with actual function references or model objects.

This prepares the schema for actual usage with schema.check method.

Sanidates the data from data object using schema validation schema, and calls the cb callback.


Express.js middleware for automatic sanidation of data prior to request handling.

The sanidated data will be stored in, and can be accessed normally, as you would with req.query or

Any validation errors that occur will cause the dataErrors property to appear in req object, so you can check for presence of this property when testing for possible errors.

FragRouter middleware. Currently, passing schema does absolutely nothing.

Using this middleware, you can call sanidate.check as this.sanidate(). The method signature is the same as for sanidate.check.