Client-side validation library that aims for the excellent user experience. Do not expect cheap magic here; we will force you to write some amount of code. In return, you get a lot of flexibility and customizability (if you want it); and of course, really great user experience with your validated forms.
You can view the demo examples here.
To run the examples locally:
- Clone the repository:
git clone git@github.com:vacuumlabs/react-custom-validation.git
- Enter the
examples
directory:cd react-custom-validation/examples
npm install
npm run example
- Navigate to
localhost:3000
Code of the examples is located in examples/example/
directory.
Examples 1 and 2 illustrate the usage of this library in very simple cases. It is highly recommended to review and understand the code of these examples, as it can help to understand this documentation and basic features of this library.
Example 3 is more complicated and is used to show that this library can be used even when the requirements are more complex (e.g. when one wants to dynamically add and remove fields). The user is encouraged to review the code of example 3 if they want to create more complex validated forms.
Let us start with a simple registration form that contains three validated fields: email, password and repeated password. The corresponding code without validations can look as follows:
class RegistrationForm extends React.Component {
render() {
let {
fields: {email, password, rePassword}
changeEmail,
...
} = this.props
return (
<form>
<input
type="text"
id="email"
onChange={(e) => changeEmail(e.target.value)}
value={email}
/>
{ /* similar code for password and re-password inputs */ }
</form>
)
}
}
To add some validations for this React component, we need to define a function that calculates validation config from component's props:
function validationConfig(props) {
let {fields: {email, password, rePassword}} = props
return {
fields: ['email', 'password', 'rePassword'],
validations: {
email: [
[isEmail, email], // [function, ...args]
[isUnique, email]
],
password: [[minLength, password, 6]],
rePassword: {
rules: [[areSame, password, rePassword]],
fields: [
// hide validation result if user is typing in one of these fields
['password', 'rePassword'],
// hide validation result until user touched all of these fields
['rePassword']
]
}
},
}
}
@validated(validationConfig)
class Registration extends React.Component {
...
To make the show/hide validity recommendations work properly, one also needs to
notify the validation library about some user actions (changes, blurs, submits).
This can be easily done by using a provided helper function $field
:
@validated(validationConfig)
class RegistrationForm extends React.Component {
render() {
let {
fields: {email, password, rePassword}
$field,
changeEmail,
...
} = this.props
return (
<form>
<input
type="text"
id="email"
{...$field('email', (e) => changeEmail(e.target.value))}
value={email}
/>
{ /* similar code for password and re-password inputs */ }
</form>
)
}
}
In return, the validation library provides data on validity and show/hide
behavior as the prop $validation
. (If needed, one can also specify
onValidation
handler that will be called whenever new validation data is
available. This takes some more coding, but provides full flexibility regarding
validation data handling.)
render() {
let {
...
// validation data provided as props by the validated decorator
$validation: {
email: {
isValid: emailValid,
error: {rule: emailRule, reason: emailReason},
show: emailShow
},
...
}
} = this.props
return (
<div>
...
{emailShow ? `Invalid email (${emailReason})` : 'Valid email!'}
...
</div>
)
}
Note that multiple-field validations (such as passwordsMatch
) are very easy to
specify. Apart from absolutely straighforward definition of rules, one just
needs to provide the list of all fields involved in the validation, so that
correct show/hide recommendations can be given.
With React and proper application state management system (for example, Redux) it is simple to validate things. All the data is available in the application state, hence obtaining the validity is as easy as applying validation functions to the appropriate arguments. Multiple-field validations and asynchronous validations do not complicate the story much.
The real challenge for top-notch validated component is not computing actual validity of individual fields, but computing whether the validation result should be shown to the user. We strongly believe that these two aspects are completely orthogonal and should be treated so. Required field never touched? Invalid, but do not show it. Email field does not look e-mail-ish at all? Invalid, but do not show it if the user is still typing. If the user changes the focus to another field, show it ASAP! You see the picture.
It turns out that whether the validation result should be presented to the user depends on many details: what inputs were already touched, when the last keystroke happened, whether the user already attempted to submit the form, etc. Such details are 100% unimportant for anything else than showing validation result, so you do not capture and store these data in any way. Therefore, the validation library creates a higher order component (HOC) that stores this information in its internal component state.
The contract is simple:
- you configure what validation rules exist and what fields affect what validations
- you inform the validation component about all changes, blurs and submits performed by the user
- validation component informs you what is the status of each individual validation: Whether the validation is OK / not OK and whether you should / should not display validation status
- Automatic re-calculation of validity when user changes the input value
- Suggestions on showing/hiding the validation result
- Hide validity while the user is typing
- Hide validity if the field was not touched yet
- Show validity if the user finished typing
- Easy definition and usage of custom validation rules
- Multiple fields validation
- Async validations
- Debouncing / Throttling
- Conditional validations
- Flexibility and extensibility: can be easily combined with other validation approaches
This library is intended to be used with React. It also plays well with Redux, but it can be used without Redux as well. It can be easily integrated with React-Intl or other (custom?) i18n solution.
The validationConfig
function that is used to annotate the validated React
component takes in one argument, the props of that React component, and returns a
config, which is a plain javascript object containing the keys specified below.
Object that maps validation names to validation rules and fields. The keys serve
as identifiers for the validations and will be referred to as validationName
s
in this documentation. For example:
validations: {
email: {
rules: [
['isEmail', isEmail, email],
['isUnique', isUnique, email]
],
fields: [['email'], ['email']]
},
...
}
If the optional fields
part is not provided, one can specify the rules
directly (top-level):
validations: {
email: [
['isEmail', isEmail, email],
['isUnique', isUnique, email]
],
...
}
List of function-call descriptors in the form
[string, fn, arg1, arg2, ...] // rule name, rule function, args
The first item (rule name) is optional. It is used to specify the failed rule in the validation result; if ommitted, the rule function name is used.
[fn, arg1, arg2, ...]
The next item fn
can be any function that complies with the rule function
API. All other items are provided to this function as
arguments in the specified order (i.e. fn(arg1, arg2, ...)
is called
internally).
β Using lambda functions as fn
leads to infinite loop in
validation calculations in some cases. It is best to avoid them completely and
use named functions instead.
π Rule ordering matters, put sync rules first! The rules are evaluated in the order they were specified, and if a rule synchronously returns an invalid result, the later rules are not evaluated at all. This can save a lot of useless server requests.
Fields validated by this validation in the form
[dependsOn, needTouch]
Both dependsOn
and needTouch
are Arrays of field names (strings). These data
are used to calculate show/hide recommendations.
dependsOn
should list all user-input fields that influence the validation
result. Validation library tracks whether user is typing in any of these fields
and if so, it recommends to hide the corresponding validation result.
needTouch
should list all user-input fields that need to be touched (changed,
blurred, submitted) before the validation result can be shown.
In most cases, both fields look the same, and contain one item, the name of the validated field. Therefore, the following defaults are provided:
null
andundefined
resolve to[[validationName], [validationName]]
(e.g. ([['email'], ['email']]
)someString
resolves to[[someString], [someString]]
someArray
resolves to[someArray, someArray]
Array of all field names (strings) that require validation, for example:
fields: ['email', 'password', 'rePassword']
Handler function that is called by the validation library whenever new validation data (i.e. validation results and recommendations for showing/hiding these validation results) is available. It provides the application developer with full control over handling the calculated validation data.
The implementation of this handler is optional and if not provided, it defaults
to empty function. If the validation data is needed only from within the
validated React component, one can use the provided prop $validation
to access
this data. However, if the data is needed elsewhere in the application (e.g. by
some other React components), the onValidation
handler needs to be
implemented.
The onValidation
handler takes in two arguments:
validationName
: name of the corresponding validation as defined invalidations
part of configdata
: newly calculated validation data; javascript object with the following structure
{
// valid or invalid or unknown validity
isValid: true | false | null
error: {
// name of the rule that failed
rule: undefined | String
// data returned by the rule function providing more specific info on error
reason: undefined | any javascipt object
},
// whether the validation result should be displayed to the user
show: true | false
}
Note that rule
and reason
are not undefined only if isValid
is false
.
The recommended implementation of the onValidation
handler should simply save
the provided data in the application state so that they can be accessed there
when needed. Example:
onValidation: (validationName, data) => dispatch(
// Here we dispatch a function that defines how app state should be modified
// using update from immutability-helper library
(state) => update(state, {validations: {[validationName]: {$set: data}}})
)
Validity of the whole form. This information is used to provide the $submit
prop.
The value can be
true
: all fields are validfalse
: at least one field is invalidnull
: validity is unknown (some validations are pending)
This field is optional; if it is not specified, isFormValid($validation)
is
used, where $validation
is validation data that is also provided as the prop
$validation
to the validated form component and isFormValid
is a helper
function for calculating overall form validity.
Object that further configures the validation library.
asyncThrottle
sets throttling for async validity computations; number (milliseconds)typingDebounce
specifies the time to wait after last user's type before validation result is shown. Allowed values are:- number (millisecons)
[number1, number2]
When user enters the field for the second time (typically to correct the value), you may want to lower the typing debounce so that the feedback is more immediate. Therefore, you can specify two typing debounces:number1
will be used on the first visit of the field andnumber2
on the second and all subsequent visits of the field
If some of the options is not specified, the following default values are used:
{
asyncThrottle: 500,
typingDebounce: [2500, 1000]
}
π Note that setting typingDebounce
to infinite (very long) time will
result in the validation results being shown only on blur or submit.
The validated React component gets four new props (apart from all other props
that were passed to it) from the validation library. Prop names starting with
$
were chosen to prevent name collisions. To further ensure that these prop
names do not collide with other prop names passed to the validated component,
they are checked during render and error is thrown if some collision is
detected.
It is recommended to call this function on form submit. It internally waits
until the form validity calculation is finished and calls one of the provided
handlers: onValid
in case the form validity is true
or onInvalid
in case
the form validity is false
. If the validity of the form is known already when
user clicks on the submit button, the onValid
(or onInvalid
) handler is
called right away. Both, onValid
and onInvalid
arguments are optional,
defaulting to empty functions.
Since it is the desired behavior in most cases, calling $submit
also calls
$fieldEvent('submit')
. This default behavior can be turned off by passing
false
as the third (fieldEvent
) argument.
Example of usage:
class RegistrationForm extends React.Component {
render() {
...
<form onSubmit={
(e) => {
e.preventDefault()
this.props.$submit(
() => alert('Registration successful!'),
() => alert('There are errors in the form!')
)
}
}>
...
}
}
Since invalid data submission can be easily avoided by using this $submit
function, it is recommended to keep the submit button enabled while the user is
filling out the form.
If the user submits the form while validity calculation is in progress and the
user continues typing (and thus changing the form field values), the
onValid
/onInvalid
handler call will be canceled. It will also be canceled if
anything in validations
part of the validation config changes. This is to
prevent possible mismatch between validated and actually submitted fields.
Note that the validation library does not prevent multiple submits while
onValid
handler is running; the submit button should be therefore disabled
while the onValid
handler is running.
For information on form validity calculation see here.
π If no async validations are used, the form validity is always known
(it never equals null
). In such a case, one can simply use
isFormValid($validity)
instead of $submit
to find out form validity at
submit (see code in example 2).
Handler used to notify the validation library about user actions which are used in show/hide validation calculations. Takes in the following arguments:
event
:'blur'
or'change'
or'submit'
or'reset'
field
(optional): string, one of field names specified infields
part of the validation configdebounce
(optional):true
orfalse
or number ornull
; overrides typing debounce set in validation config; only available for'change'
events
The event 'reset'
causes the validation library to "forget" all past
field events. For example, suppose that the field 'email'
was changed, blurred
and/or submitted and the user is not typing right now. One will thus get
{show: true}
for this email validation. After $fieldEvent('reset', 'email')
is called, the email field will be considered untouched (all changes, blurs,
submits being forgotten) and one will get {show: false}
for the email
validation. The most usual case is perhaps calling $fieldEvent('reset')
which
is useful for resetting all recorded field data (for example, after successful
submit).
If field
is not specified, all fields specified in the fields
part of the
validation config are assumed. This is useful especially for the 'submit
'
event. Note that $fieldEvent('reset')
is an exception; it resets all the
recorded field events (even for fields that are not specified in the validation
config).
Examples of usage:
<input
type="text"
id="email"
onChange={(e) => {
this.handleEmailChange(e.target.value)
this.props.$fieldEvent('change', 'email')
}}
onBlur={() => this.props.$fieldEvent('blur', 'email')}
value={this.props.fields.email}
/>
<form onSubmit={
(e) => {
e.preventDefault()
this.props.$fieldEvent('submit')
...
Syntactic sugar that saves manual calling of the $fieldEvent
function. Takes in
the following arguments:
- field name
onChange
handleronBlur
handlerdebounce
(optional):true
orfalse
or number ornull
; overrides typing debounce set in validation config for the'change'
fieldEvent; does not affect'blur
' field event. Default value is'true'
Provides modified onChange
and onBlur
that take care of calling the
$fieldEvent
function. Both handlers can be null, empty functions are then used
as a default.
Examples of usage:
<input
type="text"
id="email"
{...this.props.$field('email', (e) => this.handleEmailChange(e.target.value))}
value={this.props.fields.email}
/>
<input
type="checkbox"
id="isOther"
{...this.props.$field(
'isOther',
(e) => this.handleIsOtherChange(e.target.value), // onChange
null, // onBlur
false // we do not want typing debounce for a checkbox
)}
checked={this.props.fields.isOther}
/>
Object that maps each validationName
specified in the validations
part of the
config to the corresponding validation result and show/hide recommendation.
The data structure is the same as what is provided by the onValidation
handler. For example:
let {
$validation: {
email: {isValid, error: {reason, rule}},
password: {...}
}
} = this.props
Returns validity of multiple validation results. The result is false if any
single validation contains valid = false, null if any validation contains valid
= null (and none is false) and true otherwise. The argument $validation
should
be a dict of validation results as provided by the validation library.
Returns initial status of validation data as provided by this library. Can be used to initialize the app state. It is recommended (but not necessary) to keep the validation data in the app state structured in the same way.
It is very easy to create new rule functions. The rule function is any function that:
- returns a valid value or a
Promise
that resolves to a valid value if the arguments satify the rule - returns an error description (not equal to any valid value) or a
Promise
of such error description if the arguments do not satisfy the rule
The valid values are null
, undefined
, true
, []
and {}
. This
convention enables one to use functions provided by some popular javascript
validation libraries without any additional boilerplate code.
Examples:
function areSame({value1, value2}) {
return (value1 === value2) ? null : 'Values are different'
}
function isUnique({value}) {
let isValid = value.indexOf('used') === -1
let response = isValid ? null : 'The value is not unique.'
return Promise.delay(10).then(() => response)
}
In larger and more serious projects one might need to translate the validation messages to other languages, or one might want to provide very specific validation messages. The recommended way of achieving this is outlined below.
Write rule functions that return object:
function hasLength({value, min, max}) {
if (min != null && value.length < min) {
return {code: 'too short', args: {min}, msg: `Length should be at least ${min}.`}
}
if (max != null && value.length > max) {
return {code: 'too long', args: {max}, msg: `Length should be at most ${max}.`}
}
return null
}
Write a displayMessage
function that can handle all types of errors that you
are using, for example:
function displayMessage(error, dictionary) {
if (typeof error === 'string') {
return error
}
{code, args, msg} = error
if (dictionary[code] != null) {
return dictionary[code](args)
} else {
return msg // falling back to provided message
}
}
The dictionary in this case might be something like:
{
'too short': ({min}) => `The provided value is very short. Please enter at
least ${min} characters`,
...
}