Validation

Ryx validates data at two levels: field-level (automatic) and model-level (custom).

Field-Level Validation

Built into field definitions:

class Product(Model):
    name = CharField(
        max_length=200,
        min_length=3,
        blank=False,
    )
    price = DecimalField(
        max_digits=10,
        decimal_places=2,
        min_value=0,
        max_value=999999,
    )
    status = CharField(
        max_length=20,
        choices=["draft", "published", "archived"],
    )
    email = CharField(max_length=254, validators=[EmailValidator()])
    website = CharField(max_length=200, null=True, validators=[URLValidator()])

Built-in Validators

Validator	Description
MaxLengthValidator(n)	String/list length ≤ n
MinLengthValidator(n)	String/list length ≥ n
MaxValueValidator(n)	Value ≤ n
MinValueValidator(n)	Value ≥ n
RangeValidator(min, max)	Value in range
NotBlankValidator()	Not empty string
RegexValidator(pattern)	Matches regex
EmailValidator()	Valid email format
URLValidator()	Valid URL format
ChoicesValidator(values)	Value in allowed list
NotNullValidator()	Not None
UniqueValueValidator()	Unique in table

Model-Level Validation

Override clean() for cross-field validation:

class Event(Model):
    start_date = DateTimeField()
    end_date   = DateTimeField()
    status     = CharField(max_length=20, choices=["draft", "published"])

    async def clean(self):
        errors = {}
        if self.end_date and self.end_date < self.start_date:
            errors["end_date"] = ["End date must be after start date"]
        if self.status == "published" and not self.start_date:
            errors["start_date"] = ["Published events need a start date"]
        if errors:
            raise ValidationError(errors)

Running Validation

product = Product(name="x", price=-1)

# Manual
try:
    await product.full_clean()
except ValidationError as e:
    print(e.errors)
    # → {
    #     "name": ["Ensure this value has at least 3 characters."],
    #     "price": ["Ensure this value is greater than or equal to 0."],
    #   }

# Automatic — save() calls full_clean() by default
await product.save()  # Raises ValidationError

# Skip validation (for bulk ops)
await product.save(validate=False)

Collecting All Errors

full_clean() collects ALL errors from ALL fields before raising — you get the complete picture at once:

try:
    await product.full_clean()
except ValidationError as e:
    for field, errors in e.errors.items():
        for error in errors:
            print(f"{field}: {error}")

Next Steps

→ Signals — Lifecycle event observers