util-validator Guide

Preface

Validating data is a common task that occurs throughout all application layers, from the presentation to the persistence layer. Often the same validation logic is implemented in each layer which is time consuming and error-prone. To avoid duplication of these validations, developers often bundle validation logic directly into the domain model, cluttering domain classes with validation code which is really metadata about the class itself.

The Jakarta Bean Validation 2.0 - defines a metadata model and API for entity and method validation. The default metadata source are Java annotations (with the ability to override and extend the meta-data through the use of XML if desired). The API is not tied to a specific application tier nor programming model and it is specifically not tied to either the web or persistence tier, and is available for both server-side application programming or rich client applications.

The ScalaValidator is a Scala wrapper around the Java Hibernate Validator which itself is the reference implementation of Jakarta Bean Validation specification.

Getting Started

The ScalaValidator depends only on the Hibernate Validator, the Glassfish reference implementation of the Jakarta Expression Language for message interpolation, and json4s-core for Scala reflection.

We highly recommend reading through the Hibernate Validator documentation as our guide is largely based on this information.

In order to use the util-validator add a dependency on the project, e.g. in sbt:

"com.twitter" %% "util-validator" % utilValidatorVersion

or with Maven:

<dependency>
    <groupId>com.twitter</groupId>
    <artifactId>util-validator_2.12</artifactId>
    <version>${util.validator.version}</version>
</dependency>

where utilValidatorVersion and util.validator.version are the expected library version.

Applying constraints

Let’s start with a basic example to see how to apply constraints. Let’s assume you have a case class defined as such:

package com.example

import jakarta.validation.constraints.{Min, NotEmpty, Size}

case class Car(
  @NotEmpty manufacturer: String,
  @NotEmpty @Size(min = 2, max = 14) licensePlate: String,
  @Min(2) seatCount: Int)

The @NotEmpty, @Size and @Min annotations are used to declare the constraints which should be applied to the fields of a Car instance:

  • manufacturer must never be an empty String (and thus also not null)
  • licensePlate must never be an empty String (thus not also not null) and must be between 2 and 14 characters
  • seatCount must be equal to at least 2

Validating constraints

ScalaValidator#validate

To perform a validation of these constraints, you use an instance of the ScalaValidator. Let’s take a look at a test for validating Car instances:

In the test constructor we instantiate an instance of a ScalaValidator. A ScalaValidator instance is thread-safe and may be reused multiple times. It thus can safely be stored in a member variable field and be used in the test cases to validate the different Car instances.

The ScalaValidator#validate method returns a set of ConstraintViolation instances, which you can iterate over in order to see which validation errors occurred. The first three tests show some expected constraint violations:

  • The @NotEmpty constraint on manufacturer is violated in test(“manufacturer is empty”)
  • The @Size constraint on licensePlate is violated in test(“licensePlate is too short”)
  • The @Min constraint on seatCount is violated in test(“seatCount is too small”)

If the object validates successfully, ScalaValidator#validate returns an empty set as you can see in test(“car is valid”).

Note that this method is recursive and will cascade validations as explained in the section on Object graphs.

ScalaValidator#verify

The ScalaValidator also has an API to throw constraint violations as a type of ValidationException rather than returning a Set[ConstraintViolation[_].

Instead of calling ScalaValidator#validate, use ScalaValidator#verify:

Like ScalaValidator#validate, this method is recursive and will cascade validations as explained in the section on Object graphs.

Declaring and validating case class constraints

In this section we show how to declare (see “Declaring case class constraints”) and validate (see “Validating bean constraints”) case class constraints.

You will likely want to review Hibernate’s “Built-in constraints” documentation which provides an overview of all built-in constraints coming with the Hibernate Validator and thus supported here.

If you are interested in applying constraints to method parameters and return values, refer to Declaring and validating method constraints.

Declaring case class constraints

Constraints in Jakarta Bean Validation are expressed via Java annotations. In this section we show how to enhance an object model with these annotations. There are three types of bean constraints:

  • field constraints
  • property constraints
  • container element constraints
  • class constraints

When defining a class field in Scala the compiler creates up to four accessors for it: a getter, a setter, and if the field is annotated with @BeanProperty, a bean getter and a bean setter [reference].

Thus, if you had the following class definition:

class C(@myAnnot var c: Int)

There are six entities which can carry the @myAnnot annotation: the constructor param, the generated field and the four accessors. By default, annotations on constructor parameters end up only on the constructor parameter and not on any other entity. Annotations on fields, by default, only end up on the field.

If you wanted Scala to copy the annotation to the generated field, you would need to apply the scala.annotation.meta.field meta annotation to the annotation.

For example, using class C again from above:

class C(@(myAnnot @field) var c: Int)

Thus to use the Hibernate Validator Java library directly from Scala, you would need to always annotate constraint annotations with a scala.annotation.meta annotation along with translating between Scala and Java collection types.

The benefit of util-validator is that the library will work with simply annotated case class constructor parameters (i.e., without needing to add a scala.annotation.meta annotation) with an API that uses and expresses Scala collection types.

Constructor param constraints

Constraints can be expressed by annotating the constructor param of a case class. E.g.,

package com.example

import jakarta.validation.constraints.{Min, NotEmpty, Size}

case class Car(
  @NotEmpty manufacturer: String,
  @NotEmpty @Size(min = 2, max = 14) licensePlate: String,
  @Min(2) seatCount: Int)

Important

When using constructor parameter constraints, field access strategy is used to access the value to be validated. This means the validation logic directly accesses the instance variable and will never invoke the generated property accessor method.

Constraints can be applied to fields of any access type (public, private etc.), including inherited members. Neither constraints on companion object fields, nor secondary companion object #apply method constraints are supported.

Secondary constructors

The validation library is not used for case class construction but because of how annotations are carried by default in Scala for classes, the constructor of a case class plays an important role in case class validation. By default, the library only tracks constraint annotations for constructor parameters which are also declared fields within the class. Thus, if you have a primary constructor of Int types with a secondary constructor of constraint-annotated String types (which does some translation into the required Int types), the secondary constructor constraints have no bearing on validation.

Container element constraints

Note that due to a long-standing issue [SCALA-9883] which affects how to access Java annotations from the Scala compiler, that container element constraints are not supported.

For examples of container element constraints, see the Hibernate documentation.

Thus, take note that constraint annotations will apply only to the container and not the type(s) of the container with the one caveat being a field of type Option[_] which is always unwrapped by default and thus the annotation always pertains to the contained type.

With Option[_]

When applying a constraint on the type argument of Option[_], the ScalaValidator will automatically unwrap the Option[_] type and validate the internal value. For example:

Welcome to Scala 2.12.13 (JDK 64-Bit Server VM, Java 1.8.0_242).
Type in expressions for evaluation. Or try :help.

scala> import jakarta.validation.constraints._
import jakarta.validation.constraints._

scala> case class Car(@Min(1000) towingCapacity: Option[Int] = None)
defined class Car

scala> val car = Car(Some(100))
car: Car = Car(Some(100))

scala> import com.twitter.util.validation.ScalaValidator
import com.twitter.util.validation.ScalaValidator

scala> val validator = ScalaValidator()
Mar 30, 2021 11:58:47 AM org.hibernate.validator.internal.util.Version <clinit>
INFO: HV000001: Hibernate Validator 7.0.1.Final
validator: com.twitter.util.validation.ScalaValidator = com.twitter.util.validation.ScalaValidator@574f9e36

scala> val violations = validator.validate(car)
violations: Set[jakarta.validation.ConstraintViolation[Car]] = Set(ConstraintViolationImpl{interpolatedMessage='must be greater than or equal to 1000', propertyPath=towingCapacity, rootBeanClass=class Car, messageTemplate='{jakarta.validation.constraints.Min.message}'})

scala> val violation = violations.head
violation: jakarta.validation.ConstraintViolation[Car] = ConstraintViolationImpl{interpolatedMessage='must be greater than or equal to 1000', propertyPath=towingCapacity, rootBeanClass=class Car, messageTemplate='{jakarta.validation.constraints.Min.message}'}

scala> violation.getMessage
res0: String = must be greater than or equal to 1000

scala> violation.getPropertyPath.toString
res1: String = towingCapacity

scala>

Note

The property path only contains the name of the property, in this way, Options are treated as a “transparent” container.

Class-level constraints

A constraint can also be placed on the class level. In this case not a single property is subject of the validation but the complete object. Class-level constraints are useful if the validation depends on a correlation between several properties of an object.

For instance, if we had a Car case class defined with two field seatCount and passengers and we want to ensure that the list of passengers does not have more entries than available seats. For that purpose the @ValidPassengerCount constraint is added on the class level. The validator of that constraint has access to the complete Car object, allowing to compare the numbers of seats and passengers.

See the corresponding section in Creating custom constraints - Class-level constraints for documentation on how to implement such a custom constraint.

Constraint inheritance

When a class implements a trait or extends an abstract class, all constraint annotations declared on the super-type apply in the same manner as the constraints specified in the class itself. For example, if we had a trait, Car with a RentalCar implementation:

trait Car {
  @NotEmpty def manufacturer: String
}

case class RentalCar(manufacturer: String, @NotEmpty rentalStation: String) extends Car

Here the case class RentalCar implements the Car trait and adds the property rentalStation. If an instance of RentalCar is validated, not only the @NotEmpty constraint on rentalStation is evaluated, but also the constraint on manufacturer from the Car trait.

 Welcome to Scala 2.12.13 (JDK 64-Bit Server VM, Java 1.8.0_242).
 Type in expressions for evaluation. Or try :help.

 scala> import jakarta.validation.constraints._
 import jakarta.validation.constraints._

 scala> trait Car {
      |   @NotEmpty def manufacturer: String
      | }
 defined trait Car

 scala> case class RentalCar(manufacturer: String, @NotEmpty rentalStation: String) extends Car
 defined class RentalCar

 scala> import com.twitter.util.validation.ScalaValidator
 import com.twitter.util.validation.ScalaValidator

 scala> val validator = ScalaValidator()
 Mar 30, 2021 12:21:41 PM org.hibernate.validator.internal.util.Version <clinit>
 INFO: HV000001: Hibernate Validator 7.0.1.Final
 validator: com.twitter.util.validation.ScalaValidator = com.twitter.util.validation.ScalaValidator@64e89bb2

 scala> val rental = RentalCar("", "Hertz")
 rental: RentalCar = RentalCar(,Hertz)

 scala> val violations = validator.validate(rental)
 violations: Set[jakarta.validation.ConstraintViolation[Car]] = Set(ConstraintViolationImpl{interpolatedMessage='must not be empty', propertyPath=manufacturer, rootBeanClass=class RentalCar, messageTemplate='{jakarta.validation.constraints.NotEmpty.message}'})

 scala> violations.size
 res0: Int = 1

 scala> violations.head.getPropertyPath.toString
 res: String = manufacturer

 scala> violations.head.getMessage
 res: String = must not be empty

 scala> val rental = RentalCar("Renault", "")
 rental: RentalCar = RentalCar(Renault,)

 scala> val violations = validator.validate(rental)
 violations: Set[jakarta.validation.ConstraintViolation[Car]] = Set(ConstraintViolationImpl{interpolatedMessage='must not be empty', propertyPath=rentalStation, rootBeanClass=class RentalCar, messageTemplate='{jakarta.validation.constraints.NotEmpty.message}'})

 scala> violations.size
 res: Int = 1

 scala> violations.head.getPropertyPath.toString
 res: String = rentalStation

 scala> violations.head.getMessage
 res: String = must not be empty

 scala> val rental = RentalCar("", "")
 rental: RentalCar = RentalCar(,)

 scala> val violations = validator.validate(rental)
 violations: Set[jakarta.validation.ConstraintViolation[Car]] = Set(ConstraintViolationImpl{interpolatedMessage='must not be empty', propertyPath=manufacturer, rootBeanClass=class RentalCar, messageTemplate='{jakarta.validation.constraints.NotEmpty.message}'}, ConstraintViolationImpl{interpolatedMessage='must not be empty', propertyPath=rentalStation, rootBeanClass=class RentalCar, messageTemplate='{jakarta.validation.constraints.NotEmpty.message}'})

 scala> violations.size
 res: Int = 2

 scala> violations.map(v => s"${v.getPropertyPath}: ${v.getMessage}").mkString("\n")
 res: String =
 manufacturer: must not be empty
 rentalStation: must not be empty

 scala> val rental = RentalCar("Renault", "Hertz")
 rental: RentalCar = RentalCar(Renault,Hertz)

 scala> val violations = validator.validate(rental)
 violations: Set[jakarta.validation.ConstraintViolation[Car]] = Set()

 scala>

The same would be true, if Car was not a trait but an abstract class implemented by RentalCar.

Constraint annotations are aggregated if members are overridden. So if the RentalCar case class also specifies any constraints on the manufacturer field, it would be evaluated in addition to the @NotEmpty constraint from the Car trait:

Object graphs

The Jakarta Bean Validation API does not only allow to validate single case class instances but also complete object graphs via cascaded validation. To do so, just annotate a field or property representing a reference to another case class with @Valid.

case class Person(@NotEmpty name: String)

case class Car(@NotEmpty manufacturer: String, @Valid driver: Person)

If an instance of Car is validated, the referenced Person case class will also be validated, as the driver field is annotated with @Valid. Therefore the validation of a Car will fail if the name field of the referenced Person instance is empty.

The validation of object graphs is recursive, i.e. if a reference marked for cascaded validation points to an object which itself has properties annotated with @Valid, these references will be followed up by the validation logic as well. The validation logic will ensure that no infinite loops occur during cascaded validation, for example if two objects hold references to each other.

Note that null or None values are ignored during cascaded validation.

Iterable[_]

Unlike constraints, object graph validation works for some container elements (those which are an Iterable[_]). That means the field of the container can be annotated with @Valid, which will cause each contained element to be validated when the parent object is validated.

case class Person(@NotEmpty name: String)

case class Car(@NotEmpty manufacturer: String, @Valid drivers: Seq[Person])

If an instance of Car is validated, the referenced sequence of Person case classes will also be validated, as the drivers field is annotated with @Valid. Therefore the validation of a Car will fail if the name field of the referenced Person instance is empty.

 Welcome to Scala 2.12.13 (JDK 64-Bit Server VM, Java 1.8.0_242).
 Type in expressions for evaluation. Or try :help.

 scala> import jakarta.validation.constraints._
 import jakarta.validation.constraints._

 scala> case class Person(@NotEmpty name: String)
 defined class Person

 scala> import jakarta.validation.Valid
 import jakarta.validation.Valid

 scala> case class Car(@NotEmpty manufacturer: String, @Valid drivers: Seq[Person])
 defined class Car

 scala> val car = Car("Renault", Seq(Person("")))
 car: Car = Car(Renault,List(Person()))

 scala> import com.twitter.util.validation.ScalaValidator
 import com.twitter.util.validation.ScalaValidator

 scala> val validator = ScalaValidator()
 Mar 30, 2021 3:59:53 PM org.hibernate.validator.internal.util.Version <clinit>
 INFO: HV000001: Hibernate Validator 7.0.1.Final
 validator: com.twitter.util.validation.ScalaValidator = com.twitter.util.validation.ScalaValidator@6dcc7696

 scala> val violations = validator.validate(car)
 violations: Set[jakarta.validation.ConstraintViolation[Car]] = Set(ConstraintViolationImpl{interpolatedMessage='must not be empty', propertyPath=drivers[0].name, rootBeanClass=class Car, messageTemplate='{jakarta.validation.constraints.NotEmpty.message}'})

 scala> violations.head.getPropertyPath.toString
 res: String = drivers[0].name

 scala> violations.head.getMessage
 res: String = must not be empty

 scala> val car = Car("Renault", Seq(Person("Lupin"), Person("")))
 car: Car = Car(Renault,List(Person(Lupin), Person()))

 scala> val violations = validator.validate(car)
 violations: Set[jakarta.validation.ConstraintViolation[Car]] = Set(ConstraintViolationImpl{interpolatedMessage='must not be empty', propertyPath=drivers[1].name, rootBeanClass=class Car, messageTemplate='{jakarta.validation.constraints.NotEmpty.message}'})

 scala> violations.head.getPropertyPath.toString
 res: String = drivers[1].name

 scala> violations.head.getMessage
 res: String = must not be empty

 scala>

Validating case class constraints

The ScalaValidator does not directly implement the Jakarta Bean Validation interface but provides the same methods (and a few more) with Scala collection types. Note, however, since the ScalaValidator is a wrapper that the underlying Jakarta Bean Validation is available for direct access for Java users or to use functionality not exposed in the ScalaValidator. Also note that this is a Java library and bypassing the ScalaValidator methods will thus not support all Scala types.

This section shows how to obtain a configured ScalaValidator instance and then will walk through the different methods of the ScalaValidator.

Obtaining a Validator instance

The ScalaValidator has an apply() function which will return an instance of a ScalaValidator configured with defaults.

val validator: ScalaValidator = ScalaValidator()

To apply custom configuration to create a ScalaValidator, there is a builder for constructing a customized validator.

E.g., to configure a custom size for the descriptor cache:

Or to add custom validations:

or to specify multiple constraint mappings:

You are not required to add a constraint mapping if the constraint annotation specifies a validatedBy() class.

Important

Attempting to add multiple mappings for the same annotation will result in a ValidationException.

ScalaValidator API

The ScalaValidator contains methods that can be used to either validate entire entities or just single properties of the entity. All methods but ScalaValidator#verify return a Set[ConstraintViolation[_]]. The set is empty if the validation succeeds. Otherwise, a ConstraintViolation instance is added for each violated constraint. In the case of ScalaValidator#verify, when the returned Set[ConstraintViolation[_]] is non-empty a ConstraintViolationException is thrown which wraps the Set[ConstraintViolation[_]].

All the validation methods have a version which takes a groups: Seq[Class[_]] parameter that is used to specify validation groups to be considered when performing the validation. If the parameter is empty, the default validation group (jakarta.validation.groups.Default) is used.

The topic of validation groups is discussed in detail in Hibernate’s Chapter 5, Grouping constraints.

ScalaValidator#validate

Use validate to perform validation of all constraints of a given case class.

Using the given definition of a Car, the above example shows an instance which fails to satisfy the @NotEmpty constraint on the manufacturer property. The validation call therefore returns one ConstraintViolation[Car] object.

ScalaValidator#verify

The verify method is the same as validate but instead of returning Set[ConstraintViolation[T]] the method throws a ConstraintViolationException.

ScalaValidator#validateValue

By using the validateValue method you can check whether a single field of a given case class can be validated successfully if the field had the given value:

There is also a version of the method which instead of accepting a Class[T], takes a CaseClassDescriptor[T] of a given class.

ScalaValidator#validateProperty

With help of the validateProperty you can validate a single named field of a given case class instance.

Note

The difference between validateValue and validateProperty is that the former takes a class type, a field in the class, and a presumed value in order to perform validation. Thus, no actual instance of the class is required for performing validation.

Whereas the latter performs validation of a given field on an instance of a case class.

Important

The @Valid annotation for cascading validation is not honored by validateValue or validateProperty.

ConstraintViolation

As mentioned, all methods but ScalaValidator#verify return a Set[ConstraintViolation[_]]. The ConstraintViolation contains information about the cause and location of the validation failure.

See the Hibernate documentation for details.

Built-in constraints

In addition to the constraints defined by the Jakarta Bean Validation API and the Hibernate Built-in constraints we provide several useful custom constraints which are listed below. These constraints all apply to the field/property level.

  • @CountryCode

    Checks that the annotated field represents a valid ISO 3166 country code or a collection of ISO 3166 country codes.

    • Supported data types

      String, Array[String], Iterable[String]

  • @OneOf(value=)

    Checks that the annotated field represents a value or a set of values from the given array of values. Note that the #toString representations are checked for equality.

    • Supported data types

      Any, Array[Any], Iterable[Any]

  • @UUID

    Checks if the annotated field is a valid String representation of a java.util.UUID.

    • Supported data types

      String

Declaring and validating method constraints

Constraints can not only be applied to case classes and their fields, but also to the parameters and return values of methods within the case class. Following the Jakarta Bean Validation specification, constraints can be used to specify

  • the preconditions that must be satisfied by the caller before a method or constructor may be invoked (by applying constraints to the parameters of an executable)
  • the postconditions that are guaranteed to the caller after a method or constructor invocation returns (by applying constraints to the return value of an executable)

Important

In all cases except when using @MethodValidation, declaring method or constructor constraints itself does not automatically cause their validation upon validation of a case class instance. Instead, the ScalaExecutableValidator API must be used to perform the validation.

Declaring method constraints

@MethodValidation constraint

The @MethodValidation constraint annotation is a special method validation constraint provided by the library.

Unlike other executable specific constraints, @MethodValidation-annotated methods will be validated along with all field-level and class-level validations when calling ScalaValidate#validate or ScalaValidator#verify. Note, that the method validation is strictly performed after all field level validations (and before class-level validations). Annotated methods MUST have no specified parameters and MUST return a MethodValidationResult.

To manually validate a given method or methods, you can use ScalaExecutableValidator#validateMethod or ScalaExecutableValidator#validateMethods.

You can find details in “Creating custom constraints - @MethodValidation constraints” which shows how to implement a @MethodValidation constraint.

Method Parameter constraints

You can specify the preconditions of a method by adding constraint annotations to its parameters:

import jakarta.validation.constraints.{Min, NotEmpty, NotNull}
import java.time.LocalDate

case class RentalStation(@NotEmpty name: String) {

  def rentCar(
    @NotNull customer: Customer,
    @NotNull @Future startDate: LocalDate,
    @Min(1) durationInDays: Int
  ): Unit = ???
}

The following preconditions are specified here:

  • The name passed to the RentalCar constructor must not be empty.
  • When invoking the rentCar() method, the given customer must not be null, the rental’s start date must not be null as well as be in the future and finally the rental duration must be at least one day.

Note

Constraints may only be applied to instance methods, i.e. declaring constraints on static methods is not supported.

Cross-parameter constraints

Sometimes validation does not only depend on a single parameter but on several or even all parameters of a method. This kind of requirement can be fulfilled with help of a cross-parameter constraint.

Cross-parameter constraints can be considered as the method validation equivalent to class-level constraints. Both can be used to implement validation requirements which are based on several elements. While class-level constraints can apply to several fields of a case class, cross-parameter constraints can apply to several parameters of a method.

In contrast to single-parameter constraints, cross-parameter constraints are declared on the method. In the example below, the cross-parameter constraint @LuggageCountMatchesPassengerCount declared on the load() method is used to ensure that no passenger has more than two pieces of luggage.

case class Car(@NotEmpty manufacturer: String, @AssertTrue isRegistered: Boolean) {

  @LuggageCountMatchesPassengerCount(piecesOfLuggagePerPassenger = 2)
  def load(passengers: Seq[Person], luggage: Seq[PieceOfLuggage]): Unit = ???
}

We’ll see in the next section that return value constraints are also declared on the method level. Thus, in order to distinguish cross-parameter constraints from return value constraints, the constraint target is configured in the ConstraintValidator implementation using the @SupportedValidationTarget annotation.

More details in “Creating custom constraints - Cross-parameter constraints” which shows how to implement a cross-parameter constraint.

Method Return value constraints

The postconditions of a method or constructor are declared by adding constraint annotations to the executable:

case class RentalStation @ValidRentalStation(@NotEmpty name: String) {

  @NotEmpty
  @Size(min = 1)
  def getCustomers: Seq[Customer] = ???
}

In the example above, the following constraints apply to executables (constructor and a method in this case) of RentalStation: * A new RentalStation instance must satisfy the @ValidRentalStation constraint * The Seq[Customer] returned by getCustomers must not be empty and must contain at least on element

Cascaded Validation

Like the cascaded validation of case class fields (see “Object graphs”), the @Valid annotation can be used to mark executable parameters and return values for cascaded validation. When validating a parameter or return value annotated with @Valid, the constraints declared on the parameter or return value object are also validated.

case class Car(@NotEmpty manufacturer: String, @NotEmpty @Size(min = 2, max = 14) licensePlate: String)

case class Garage @Valid (@NotEmpty name: String) {

  def checkCar(@NotNull @Valid car: Car): Boolean = ???
}

Above, the car parameter of the method Garage#checkCar as well as the return value of the Garage constructor are marked for cascaded validation.

When validating the arguments of the checkCar() method, the constraints on the properties of the passed Car object are evaluated as well. Similarly, the @NotEmpty constraint on the name field of Garage is checked when validating the return value of the Garage constructor.

Generally, the cascaded validation works for executables in the same way as it does for case class fields.

In particular, null or None values are ignored during cascaded validation (which is impossible for constructor return values) and cascaded validation is performed recursively, i.e. if a parameter or return value object which is marked for cascaded validation itself has properties marked with @Valid, the constraints declared on the referenced elements will be validated as well.

Also, the same issues with annotating container elements apply because of the Scala compiler limitations but also note that more generally, the majority of executable methods proxy to the underlying Hibernate Java library.

Method constraints in inheritance hierarchies

Note there are some rules for method constraints and inheritance as outlined in the Hibernate documentation.

Validating method constraints

The validation of method constraints is done using the ScalaExecutableValidator interface.

Obtaining a ScalaExecutableValidator instance

The ScalaValidator provides a ScalaExecutableValidator which implements a set of “executable” methods that mirrors the bean validation ExecutableValidator API.

All of the corollary ExecutableValidator methods proxy directly to the underlying Hibernate Validator for support since it properly handles annotations in these cases and no special logic is needed.

The ScalaExecutableValidator can be obtained from the ScalaValidator by calling ScalaValidator#forExecutables

import com.twitter.util.validation.ScalaValidator
import com.twitter.util.validation.executable.ScalaExecutableValidator

val validator: ScalaValidator = ScalaValidator()
val executableValidator: ScalaExecutableValidator = validator.forExecutables

ScalaExecutableValidator methods

The ScalaExecutableValidator interface offers altogether six methods:

  • validateMethods(), validateMethod(), validateParameters(), and validateReturnValue() for method validation
  • validateConstructorParameters() and validateConstructorReturnValue() for constructor validation
ScalaExecutableValidator#validateMethods

Validates all @MethodValidation-annotated methods of a given object.

ScalaExecutableValidator#validateMethod

Validates the given @MethodValidation-annotated method of an object.

ScalaExecutableValidator#validateParameters

Validates all constraints placed on the parameters of the given method. This method directly proxies to the underlying Hibernate validator.

ScalaExecutableValidator#validateReturnValue

Validates return value constraints of the given method. This method directly proxies to the underlying Hibernate validator.

ScalaExecutableValidator#validateConstructorParameters

Validates all constraints placed on the parameters of the given constructor. This method directly proxies to the underlying Hibernate validator.

ScalaExecutableValidator#validateConstructorReturnValue

Validates a constructor’s return value. This method directly proxies to the underlying Hibernate validator.

Built-in method constraints

In addition to the built-in case class and field-level constraints discussed in “Built-in constraints”, this library provides one method-level constraint, @MethodValidation. This is a generic constraint which allows for implementation of validation of the (possibly even internal) state of a case class instance.

This is different from a class-level constraint since being implemented as a method within the case class means that the method implementation potentially has access to private or internal state for performing validation whereas a class-level constraint only has access to publicly exposed fields and methods.

  • @MethodValidation

    Used to validate the (possibly internal) state of an entire object. It is important to note that any annotated method will be invoked when performing validation and should thus be side-effect free.

    • Supported data types

      No-arg case class method that returns a MethodValidationResult.

Grouping constraints

See the Hibernate documentation which explains constraint grouping in detail.

Creating custom constraints

In cases where none of the built-in constraints suffice, you can create custom constraints to implement your specific validation requirements.

Creating a simple constraint

To create a custom constraint, the following three steps are required:

  • Create a constraint annotation
  • Implement a ConstraintValidator
  • Define a default error message

The constraint annotation

This is detailed in the Hibernate documentation and all of the steps here are the same for use with this library. The linked documentation creates a CheckCase annotation with two modes for validating a String is either uppercase or lowercase. We’ll assume the CaseMode enum and the CheckCase constraint annotation exists.

The constraint validator

Having defined the annotation, we now need to create an implementation of a ConstraintValidator, which is able to validate fields with a @CheckCase annotation. To do so, we implement the Jakarta Bean Validation interface ConstraintValidator:

package com.example.constraint

import jakarta.validation.{ConstraintValidator, ConstraintValidatorContext}

class CheckCaseConstraintValidator extends ConstraintValidator[CheckCase, String] {
  @volatile private[this] var caseMode: CaseMode = _

  override def initialize(constraintAnnotation: CheckCase): Unit = {
    this.caseMode = constraintAnnotation.value()
  }

  def isValid(
    obj: String,
    constraintContext: ConstraintValidatorContext): Boolean = {
    if (obj == null) true
    else {
      caseMode match {
        case UPPER =>
          obj == obj.toUpperCase
        case LOWER =>
          obj == obj.toLowerCase
      }
    }
  }
}

The ConstraintValidator interface defines two type parameters which are set in the implementation. The first one specifies the annotation type to be validated (in this case CheckCase), the second one the type of elements the validator can handle (in this case, String).

In case a constraint supports several data types, you can create a ConstraintValidator implementation over the Any or AnyRef types and pattern-match within the implementation. If the implementation receives a type that is unsupported you can throw an UnexpectedTypeException.

The implementation of the validator is straightforward: the initialize method gives you access to the attribute values of the validated constraint and allows you to store them in a field of the validator as shown in the example.

The isValid method contains the validator logic. For @CheckCase this is the check as to whether a given string is either completely lower case or upper case, depending on the caseMode set from the constraint annotation value in initialize.

Important

Note that the Jakarta Bean Validation specification recommends consideration of null values as being valid. If null is not a valid value for an element, it should be explicitly annotated with @NotNull.

This library also extends the same consideration for None values during validation as detailed in With Option[_].

Thread-safety

As noted in the javadocs: access to isValid can happen concurrently and thus thread-safety must be ensured by the implementation. Also note that isValid implementations should not alter the state of the value to validate.

The error message

The last piece is an error message which should be used when a @CheckCase constraint is violated. You have a few options. You can specify the default error message directly within the message() attribute of the constraint annotation, e.g.,

String message() default "Case mode must be {value}";

However, if you want to be able to internationalize an error message, it is generally recommended to place the value in a Resource Bundle, specifying the resource key as the message() attribute in the constraint annotation.

String message() default "{com.example.constraint.CheckCase.message}";

Create a properties file, ValidationMessages.properties with the following contents:

com.example.constraint.CheckCase.message=Case mode must be {value}

Make sure the ValidationMessages.properties is packaged on your classpath to be picked up by the library.

For details, see the Hibernate documentation: “Default message interpolation”.

The TwitterConstraintValidatorContext

First, see the documentation about the ConstraintValidatorContext for background on its usage within a ConstraintValidator implementation.

This library provides a utility for customizing any returned ConstraintViolation from a ConstraintValidator by using the c.t.util.validation.constraintvalidation.TwitterConstraintValidatorContext.

The TwitterConstraintValidatorContext uses the HibernateConstraintValidatorContext described in the Custom contexts Hibernate documentation. See the referenced documentation for details.

The TwitterConstraintValidatorContext has a simple DSL to create custom constraint violations. The DSL allows you to set a dynamic Payload, set message parameters or expression variables for an error message, or specify a fully custom error message template (that is a message template different than the default specified by the constraint annotation).

Note

As mentioned in the Hibernate documentation:

[T]he main difference between message parameters and expression variables is that message parameters are simply interpolated whereas expression variables are interpreted using the Expression Language engine.

In practice, use message parameters if you do not need the advanced features of an Expression Language.

By default, Expression Language support is not enabled for custom violations created via the ConstraintValidatorContext. However, for some advanced requirements, using Expression Language may be necessary. If you add an expression variable via the TwitterConstraintValidatorContext, Expression Language support will be enabled when adding a constraint violation.

Using the constraint

To use the custom @CheckCase constraint, annotate a case class field:

During validation, a licensePlate field with an invalid case will now cause validation to fail.

// invalid license plate
val car: Car = Car("Morris", "dd-ab-123", 4)

val violations: Set[ConstraintViolation[Car]] = validator.validate(car)
assert(violations.size == 1)
assert(violations.head.getMessage == "Case mode must be UPPER")

// valid license plate
val car: Car = Car("Morris", "DD-AB-123", 4)
assert(validator.validate(car).size == 0)

Class-level constraints

As mentioned previously, constraints can also be applied on the class-level to validate the state of the entire case class instance. These constraints are defined in the same way as field constraints.

Here we’ll see how to implement the @ValidPassengerCount annotation seen in an earlier example:

First the annotation definition could be written as:

package com.example.constraint;

@Target({ TYPE, ANNOTATION_TYPE })
@Retention(RUNTIME)
@Constraint(validatedBy = { ValidPassengerCountValidator.class })
@Documented
public @interface ValidPassengerCount {

    String message() default "{com.example.constraint.ValidPassengerCount.message}";

    Class<?>[] groups() default { };

    Class<? extends Payload>[] payload() default { };
}

We define an error message in a Validation.properties file:

com.example.constraint.ValidPassengerCount.message=invalid number of passengers

Then the ConstraintValidator implementation which is specified to a Car type:

package com.example

import com.example.constraint.ValidPassengerCount
import jakarta.validation.{ConstraintValidator

class ValidPassengerCountValidator extends ConstraintValidator[ValidPassengerCount, Car] {

  def isValid(
    car: Car,
    constraintContext: ConstraintValidatorContext
  ): Boolean =
    if (car == null) true
    else car.passengers.size <= car.seatCount
}

Note as the example shows, you need to use the element type TYPE in the @Target annotation on the constraint annotation. This allows the constraint to be put on type definitions. The ValidPassengerCountValidator receives an instance of a Car in the isValid() method and can access the complete object state to decide whether the given instance is valid or not.

During validation, an instance which does not satisfy the constraint will fail the validation.

package com.example

import com.twitter.util.validation.ScalaValidator
import jakarta.validation.ConstraintViolation

val car = Car(
  seatCount = 2,
  passengers = Seq(
    Person("abcd1234", "J Doe", DefaultAddress),
    Person("abcd1235", "K Doe", DefaultAddress),
    Person("abcd1236", "L Doe", DefaultAddress),
    Person("abcd1237", "M Doe", DefaultAddress),
    Person("abcd1238", "N Doe", DefaultAddress)
  )
)

val validator: ScalaValidator = ScalaValidator()

val violations: Set[ConstraintViolation[Car]] =
  validator.validate(car)
assert(violations.size == 1)
assert(violations.head.getMessage == "invalid number of passengers")
assert(violations.head.getPropertyPath.toString == "Car")

Custom property paths

By default for a class-level constraint, any constraint violation is reported on the level of the annotated type, e.g. in this case the Car type. If the class is the top-level, the property path will be empty, otherwise it will be the field name referencing the class type.

See the Hibernate documentation for details on how to set custom property paths when implementing a class-level constraint.

Cross-parameter constraints

The Bean validation specification distinguishes two types of constraints: generic constraints which apply to the annotated element (e.g., a type, field, container element, method parameter or return value, etc.), and cross-parameter constraints. Cross-parameter constraints by contrast, apply to an array of parameters of a method or constructor and can be used to express validation logic which depends on several parameter values.

See the Hibernate documentation for details on implementing cross-parameter constraints.

Constraint composition

To keep the annotation of case class fields from becoming unwieldy or confusing, you can use a composing constraint. See the Hibernate documentation for details on creating a composing constraint.

@MethodValidation constraints

Important

The @MethodValidation constraint annotation MUST be applied to a no-arg method that returns a com.twitter.util.validation.engine.MethodValidationResult.

These annotated methods will be invoked during validation and should be side-effect free.

In a case class, define a no-arg method that returns a MethodValidationResult. Annotate the method with @MethodValidation, optionally specifying validated fields of the case class for error-reporting.

Value Extraction

Value extraction is the process of extracting values from a container so that they can be validated.

It is used when dealing with container element constraints and cascaded validation inside containers.

Built-in value extractors

This library provides built-in value extraction for Scala Iterable[_] types. See the Hibernate documentation for more details on value extraction including how to define and register a new value extractor.

Best Practices

  • Case classes used for validation should be side-effect free under property access. That is, accessing a case class field should be able to be done eagerly and without side-effects.

  • Case class methods for @MethodValidation validation should be side-effect free under method access. That is, accessing a @MethodValidation-annotated method should be able to be done eagerly and without side-effects.

  • Case classes with generic type params should be considered not supported for validation. E.g.,

    case class GenericCaseClass[T](@NotEmpty @Valid data: T)
    

    This may appear to work for some category of data but in practice the way the library caches reflection data does not discriminate on the type binding and thus validation of genericized case classes is not fully guaranteed to be successful for differing type bindings of a given genericized class.

  • While Iterable collections are supported for cascaded validation when annotated with @Valid, this does not include Map types. Annotated Map types will be ignored during validation.

  • More generally, types with multiple type params, e.g. Either[T, U], are not supported for validation of contents when annotated with @Valid. Annotated unsupported types will be ignored during validation.