Schemaless Models

On occasion, MultiValue records will be used in a manner that does not lend itself to defining a schema. The flexibility of the MultiValue data structure allows records to take amorphous shapes. For instance, a MultiValue record might be a dynamically sized list of strings with each attribute containing one item from the list. It is not possible to define a Schema for these types of records, but consumers of MVOM might still wish to work with records of that nature. Schemaless models allow for that functionality. To create a Model that does not have a schema, you should provide null for the schema parameter in the call to connection.model.

A Model instance that was created with a schema will have properties that match the schema. That is, the shape of the Model object will align with that of the schema. However, schemaless models cannot work that way since they do not have schemas. Instead, a Model that is created without a Schema will instead have a property _raw on it.

The _raw Property

The _raw property is fairly simple. After reading a record using a schemaless model, the _raw property will be populated with an array that matches the MultiValue data structure. Attributes will be converted to values of the array. If attributes contain values or the values contain subvalues, then the array will be multidimensional.

Consumers can work with the _raw property of the Model instance as they would any other array. Upon saving, the _raw property will be turned back into a MultiValue record by converting the array into attributes, values, and subvalues.

Example

Suppose you had a database record structured as follows:

Attribute	Value
ID	0001
1	foo
2	bar
3	baz
4	qux{vm}quux

You could create and read using a schemaless Model as follows:

const Schemaless = connection.model(null, 'SOME_FILE');

const schemaless = await Schemaless.findById('0001');
console.log(schemaless._raw); // outputs ['foo', 'bar', 'baz', ['qux', 'quux']]

schemaless._raw.push('last in line');
schemaless.save();