Validators

Validators ensure input data (body, query params) is correct before reaching the repository. Validation errors are thrown as ZodError and handled by the global filter.

Structure

Each validator extends RequestValidatorBase and defines a schema getter with Zod:

export class CreatePersonValidator extends RequestValidatorBase<CreatePersonRequest> {
  protected get schema() {
    return z.object({
      name: z.string().min(1),
      address: z.object({
        address: z.string().min(1),
      }),
      contacts: z.array(
        z.object({
          contact: z.string().min(1),
        }),
      ),
    });
  }
}

Usage in the handler

Validators are instantiated inline — they are not NestJS providers:

const person = AutoMapper.map(
  new CreatePersonValidator(req).validate(),
  CreatePersonRequest,
  Person,
);

If validation fails, validate() throws ZodError. ErrorsFilter converts it to an HTTP 400 response via formatZodError() — messages in Portuguese:

{
  "statusCode": 400,
  "message": "O campo name é obrigatório.",
  "errors": [
    { "field": "name", "message": "O campo name é obrigatório." }
  ]
}

With multiple errors, message becomes "Foram encontrados N erros de validação.".

List validation

Query validators combine specific filters with LIST_QUERY_SCHEMA:

// src/application/person/read-many/read-many-person.validator.ts
export class ReadManyPersonValidator extends RequestValidatorBase<ReadManyPersonRequest> {
  protected get schema() {
    return LIST_QUERY_SCHEMA.and(
      z.object({
        name: z.string().optional().nullable(),
      }),
    );
  }
}

Query string array normalization

RequestValidatorBase converts keys with [] into arrays. For example, tags[]=a,b becomes { tags: ['a', 'b'] } before Zod validation.

Best practices

• Validators exist for operations with body or query params: create, update, and read-many. Read and delete receive only the id in the handler.
• Zod schemas mirror the request class structure, not the entity.
• Complex business rules can live in the handler or domain services; validators focus on input format and constraints.