Module: Hecks::Bluebook::Behaviour::ValueObject

Included in:
ValueObject
Defined in:
lib/hecks/bluebook/behaviour/value_object.rb

Overview

WHAT A VALUE OBJECT DOES. EXTENDED, not included — a value object is a CLASS (Class.new(self), one per declared shape), so its behaviour is singleton behaviour, and the holding half's absorb is what a generated constructor would be.

Instance Method Summary collapse

Instance Method Details

#attribute(named) ⇒ Object 



16
 
# File 'lib/hecks/bluebook/behaviour/value_object.rb', line 16

def attribute(named) = attributes.find { |held| held.name == named.to_sym }

#closed_set? ⇒ Boolean

A one_of DECLARED but left empty used to be indistinguishable from no one_of at all — both are members: [] — so the rule about it could only live in the builder. Recording the declaration lets the language judge it, the same way an empty attribute NAME survives into the IR and is judged there.

Returns:

  • (Boolean) 


14
 
# File 'lib/hecks/bluebook/behaviour/value_object.rb', line 14

def closed_set? = @closed_set

#sole_attribute ⇒ Object

A single-attribute value object (EmailAddressaddress, CustomerNumbervalue) is a NAME for a scalar, not a genuine group — [[feedback_name_the_scalar_field]]. adapters/driven/ sql_query_builder.rb and fuzzing/invalid_value_generator.rb now read through this rather than inlining the check.

forms/field_shape.rb's own two attributes.first.name sites (closed_set_options/closed_set_field) look identical but are NOT the same question — they pick a closed set's DISCRIMINANT column, and a closed set can be genuinely multi-attribute (Runtime::Value::Admission#member_matches?'s own comment names a real one: StatementFrequency's cadence/ retention_months/paper_fee_cents). sole_attribute would return nil for that shape and break the discriminant lookup — left as .first on purpose, not a missed migration.



33
34
35
 
# File 'lib/hecks/bluebook/behaviour/value_object.rb', line 33

def sole_attribute
  attributes.first if attributes.size == 1
end