New in version 0.4

Validations are used to ensure that only valid data is parsed and/or stored from forms and user interactions. For example, it may be important to your application that every user provides a valid email address, or that an input field contains a valid number, or that a date has the right format.

There is one validation mechanism in Emmett, suitable whether you're using independent forms or one from a model. It is built to be easy to use, providing built-in helpers for common needs, and it also allows you to create your own validation methods if you need more customation.

Validations cannot be bypassed by end users, and are triggered every time a form is submitted or when you invoke the Model methods validate and create. But how you define validations for your entities and forms?

The quick way is to use the validation parameter of the Field class, which is used by the Model and Form classes in Emmett. It accepts a dict of instructions:

# ensure the input is a valid email address
Field(validation={'is': 'email'})

And when you're using models, it may be more comfortable to group all the field validations into a single place, with the validation attribute of your Model class:

class Person(Model):
    email = Field()
    website = Field()

    validation = {
        'email': {'is': 'email'},
        'website': {'is': 'url', 'message': 'You must insert a valid url'}

Whether you use Field's validation parameter or Model's validation attribute, the result will be the same.

As you've seen with the website field in the example, you can always customize the validation's error message.

Now, let's see the available built-in validation helpers.

Presence and absence of input

Emmett provides two helpers when you just need to ensure that a specific field is not blank or is blank: 'presence' and 'empty'.

In fact, when you need to ensure a field is not empty, you can use the 'presence' validator:

myfield = Field(validation={'presence': True})

In this case, the presence validator ensures that the contents of the input are valid, so a blank input or some white spaces won't pass the validation.

On the other hand, if you need to ensure your field is blank, you can use the 'emtpy' validator:

myfield = Field(validation={'empty': True})

If you prefer, you may also write:

myfield = Field(validation={'empty': False})

which will be the same as {'presence': True}, but remember that writing {'presence': False} doesn't mean {'empty': True}.

When you're applying the 'presence' validator to a reference field, the behavior of the validator will be quite different: it will also check that the given value exists in the referenced table.

Type of input

In the first examples, we ensured that the input values were emails or URLs. This is done with the 'is' validator, and can be used to ensure several types for your fields:

is values validation details
int ensure int type
float ensure float type
decimal ensure decimal.Decimal type
date ensure type
time ensure datetime.time type
datetime ensure datetime.datetime type
email ensure is a valid email address
url ensure is a valid URL
ip ensure is a valid IP
json ensure is valid JSON content
image (for upload fields) ensure the input is an image file
list:type ensure is a list with elements of given type (available with all 'is' values except for image and json)

Since many options of the 'is' validator ensure a specific Python type on validation, the input values will also be converted to the right type: an input which should be 'int' that comes as a string from the form will be converted to an int object for all the other validators, or for your post-validation code.

Note: the datetime validator returns a pendulum Datetime object, which is a subclass of the standard Python datetime class.

Here are some examples of 'is' validation helper:

price = Field.float(validation={'is': {'float': {'dot': ','}}})
emails = Field.string_list(validation={'is': {'list:email': {'splitter': ';'}}})
urls = Field.string_list(validation={'is': 'list:url'})

Note that, since the 'is' validator ensures the input is valid for the given type, it's like an implicit {'presence': True}, so you don't need to add 'presence' when you use it.

– Dude, what if I want to allow the field to be blank, and when it's not, allow a specific type?
you can use the 'allow' validator, described next

Also remember that Field comes with a default 'is' validator (unless you disabled it with the auto_validation parameter) depending on its type. An int field will have an {'is': 'int'} validator, since Emmett guess you want the input to be valid. We described that in the Field chapter.

Specific values allowance

In the previous section, we saw that the 'is' validator is also a {'presence': True} validation, implicitly. Now, what if we need to allow an int field to be blank, so that when is filled it will be converted to an integer and also allow it to pass the validation if it is blank?

We can use the 'allow' validator:

maybe_number ={'allow': 'blank'})
# or
maybe_number ={'allow': 'empty'})

In this specific case, we are telling the validator to accept blank/empty inputs, but you can also pass specific values to it:

maybe_number ={'allow': None})
maybe_number ={'allow': 'nope'})

Practically speaking, the 'allow' validator allows you to add an exception rule to your validation, and can be applied to any of the validators described in this page.

Length of input

You will need to set some length requirements for your fields. Most commonly, you will do this for a password field, which you may want to be greater than a certain a length and/or not too long:

password = Field.password(validation={'len': {'gt': 5, 'lt': 25}})

As you can see, the 'len' validator is just the easier way to do that. It accepts different arguments, depending on what you need:

parameter value expected example
gt int {'len': {'gt': 5}}
lt int {'len': {'lt': 25}}
gte int {'len': {'gte': 6}}
lte int {'len': {'lte': 24}}
range tuple of int {'len': {'range': (6, 25)}}

Note that the range parameter behaves like the Python builtin range(), including the lower value and excluding the upper one, so the above line is the same of {'len': {'gte': 6, 'lt': 25}}.

Value inclusion

To ensure the value is inside a specific set, you can use the 'in' validator:

myfield = Field(validation={'in': ['a', 'b']})

If you have an int field, you can also use the convenient 'range' option:

number ={'in': {'range': (1, 10)}})

The 'in' validator also accepts some specific options, in particular:

parameter description
labels a list of values to display on form's dropdown menu
multiple allow user to select multiple values

When you want to use these options with a set, you may use the 'set' notation:

number = Field(
    validation={'in': {'set': [0, 1], 'labels': ['zero', 'one']}}

Inclusion in database sets

Added in version 0.8

The 'in' validator accepts also a database set as boundary, thanks to the 'dbset' notation.

For example, if you have a reference in your model and you want its values to be in a specific subset of the database rather than just checking their existence in the referred table, you can write:


validation = {
    'article': {
        'in': {
            'dbset': lambda db: db.where( > 5)

As you can see, the value for 'dbset' should be a function, accept the database as a parameter and return a database set.

With the 'dbset' notation, the 'in' validator accepts also some additional options, that will be used in rendering forms for your entity, in particular:

parameter description
orderby define a sorting rule for dropdowns
label_field use a field of the reference model to render dropdowns

For example, if you have a rating field in your Article model, you can order the results by this column, and maybe you also want to use the titles of each article to render the choices:

validation = {
    'doctor': {
        'in': {
            'dbset': lambda db: db( > 5), 
            'orderby': lambda doctor: ~doctor.rating,
            'label_field': 'title'

Remember that the orderby clause should be a function and, accept the referred model as a parameter and return a sorting rule, while the label_field one should be a string identifying the name of the field that you want to use to format results.

Numeric boundaries

When you need to ensure some numeric values on int fields, you may prefer a different approach, using the same notation as 'len':

num_a ={'gt': 0})
num_b ={'lt': 12})
num_c ={'gte': 1, 'lte': 10})

Validate only a specific value

Sometimes you need to validate only a specific value for a field. Emmett provides the 'equals' validator to help you:

accept_terms = Field(validation={'equals': 'yes'})

Basically, the 'equals' validator performs a == check between the input value and the one given.

Match input

When you want to validate against a regex expression, you can use the 'match' validation helper. For example, let say you want to validate a ZIP code:

zip = Field(validation={'match': r'^\d{5}(-\d{4})?$'})

or a phone number:

phone = Field(validation={'match': r'^\+?1?((-)\d{3}-?|\(\d{3}\))\d{3}-?\d{4}$'})

'match' also accepts some parameters:

  • the search parameter (default to False), which will use the regex method search instead of match
  • the strict parameter (default to False), which will only match the beginning of the string.

In this example, due to the strict parameter, the value for the first field will pass validation and the second won't:

normal = Field(validation={'match': 'ab'})
strict = Field(validation={'match': {'expression': 'ab', 'strict': True}})
Model.validate({'normal': 'abc', 'strict': 'abc'})


When you need to do an exclusion operation during the validation process, you can use the 'not' validation helper. Let's say that you want the value to be different from a certain value:

myfield = Field(validation={'not': {'equals': 'somevalue'}})

or you want to exclude a set of values:

color = Field(validation={'not': {'in': ['white', 'black']}})

Basically, the 'not' validator takes another validation as argument and check the opposite result.


Sometimes you need to validate an input value that responds to any of the given validations. Under this circumstances, you can use the 'any' validation helper:

myfield = Field(validation={'any': {'is': 'email', 'in': ['foo', 'bar']}})

Obviously, the 'any' validator takes other validations as arguments and validates if any of the child validations pass.


Emmett provides several validation helpers that let you transform the input value of the field on validation. For example, you may want your string field to always being lowercase, or you need your password field to be encrypted.

Here is the complete list of transformation helpers built into Emmett:


The 'lower' helper turns your string to lowercase:

low = Field(validation={'lower': True})


The 'upper' helper turns your string to uppercase:

up = Field(validation={'upper': True})


The 'clean' helper removes any special characters from your string:

clean = Field(validation={'clean': True})


The 'urlify' helper allows you to create URL valid strings (so that, for example, you can use them for routing purposes):

urldata = Field(validation={'urlify': True})

This helper also accepts the underscore parameter (default set as False) that you can use if you want underscores to be kept in the string:

urldata = Field(validation={'urlify': {'underscore': True}})


The 'crypt' helper becomes handy when you want to encrypt the contents of the field. The easiest way to use it is just to enable it:

password = Field(validation={'crypt': True})

which will encrypt the contents using the sha512 algorithm.

If you want to use a different algorithm, you can choose between md5, sha1, sha224, sha256, sha384, sha512 and just write:

password = Field(validation={'crypt': 'md5'})

But 'crypt' also accepts two more parameters:

  • the key parameter, which allows you to specify your own key to use with the algorithm
  • the salt parameter, which allows you to specify a salt to hash the password with.
password = Field(
    validation={'crypt': {'algorithm': 'md5', 'key': 'MyVerySecretKey'}}

Custom validation

Of course, the builtin validation helpers cannot be enough in many particular cases. When you need to implement your own validation logic on a specific field, you can create your own Validator subclass, and pass an instance of it to the validation parameter:

from emmett.validators import Validator

class MyValidator(Validator):
    message = "Invalid value"

    def __call__(self, value):
        if value == "notallowed":
            return value, self.message
        return value, None

myfield = Field(validation=MyValidator())

and if you need to use multiple Validator classes, you can pass a list of instances to validation:

myfield = Field(validation=[MyValidator1(), MyValidator2()])

When you write down your own Validator, you just have to remember that the validation logic has to be inside the __call__ method. That should return the value and the error message if validation has failed, or the value and None if everything was OK.

The Validator class also has a formatter function, which allows you to format the value to display in the form for particular cases:

class FloatValidator(Validator):
    def formatter(self, value):
        #: shows only 2 values after the separator
        if value is None:
            return None
        val = str(value)
        if '.' not in val:
            val += '.00'
            val += '0' * (2 - len(val.split('.')[1]))
        return val

Combining custom validators with standard ones

New in version 1.0

You can also use custom validators we just saw in combination with the standard ones provided by Emmett. Just use the custom helper:

class OddValidator(Validator):
    message = "value has to be odd"

    def __call__(self, value):
        if value % 2:
            return value, self.message
        return value, None

mynumber ={
    'gte': 0, 'lt': 20, 'custom': OddValidator()})

Note: you can also pass a list of custom validators to the custom helper.