platform/packages/orm/README.md

# Protevus ORM

![Pub Version (including pre-releases)](https://img.shields.io/pub/v/protevus_orm?include_prereleases)
[![Null Safety](https://img.shields.io/badge/null-safety-brightgreen)](https://dart.dev/null-safety)
[![Discord](https://img.shields.io/discord/1060322353214660698)](https://discord.gg/3X6bxTUdCM)
[![License](https://img.shields.io/github/license/dart-backend/protevus)](https://github.com/dart-backend/protevus/tree/master/packages/orm/LICENSE)

Source-generated ORM for use with the [Protevus framework](https://github.com/dart-backend/protevus). Now you can combine the power and flexibility of Protevus with a strongly-typed ORM.

Documentation for migrations can be found here:
[ORM Migration](https://protevus-docs.dukefirehawk.com/guides/orm/migrations)

- [Protevus ORM](#protevus-orm)
  - [Usage](#usage)
  - [Models](#models)
  - [Example](#example)
  - [Relations](#relations)
    - [Many to Many Relations](#many-to-many-relations)
  - [Columns](#columns)
    - [Column Types](#column-types)
    - [Indices](#indices)
    - [Default Values](#default-values)

## Usage

You'll need these dependencies in your `pubspec.yaml`:

```yaml
dependencies:
  protevus_orm: ^6.0.0
dev_dependencies:
  protevus_orm_generator: ^6.0.0
  build_runner: ^2.0.0
```

`package:protevus_orm_generator` exports a class that you can include in a `package:build` flow:

- `PostgresOrmGenerator` - Fueled by `package:source_gen`; include this within a `SharedPartBuilder`.

However, it also includes a `build.yaml` that builds ORM files automatically, so you shouldn't
have to do any configuration at all.

## Models

The ORM works best when used with `package:protevus_serialize`:

```dart
library angel_orm.test.models.car;

import 'package:protevus_migration/protevus_migration.dart';
import 'package:protevus_model/protevus_model.dart';
import 'package:protevus_orm/protevus_orm.dart';
import 'package:protevus_serialize/protevus_serialize.dart';
part 'car.g.dart';

@serializable
@orm
abstract class _Car extends Model {
  String get make;

  String get description;

  bool get familyFriendly;

  DateTime get recalledAt;
}

// You can disable migration generation.
@Orm(generateMigrations: false)
abstract class _NoMigrations extends Model {}
```

Models can use the `@SerializableField()` annotation; `package:protevus_orm` obeys it.

After building, you'll have access to a `Query` class with strongly-typed methods that
allow to run asynchronous queries without a headache.

Remember that if you don't need automatic id-and-date fields, you can
simply just not extend `Model`:

```dart
@Serializable
abstract class _ThisIsNotAnAngelModel {
  @primaryKey
  String get username;
}
```

## Example

MVC just got a whole lot easier:

```dart
import 'package:protevus_framework/protevus_framework.dart';
import 'package:protevus_orm/protevus_orm.dart';
import 'car.dart';
import 'car.orm.g.dart';

/// Returns an Protevus plug-in that connects to a database, and sets up a controller connected to it...
AngelConfigurer connectToCarsTable(QueryExecutor executor) {
  return (Protevus app) async {
    // Register the connection with Protevus's dependency injection system.
    // 
    // This means that we can use it as a parameter in routes and controllers.
    app.container.registerSingleton(executor);
    
    // Attach the controller we create below
    await app.mountController<CarController>();
  };
}

@Expose('/cars')
class CarController extends Controller {
  // The `executor` will be injected.
  @Expose('/recalled_since_2008')
  carsRecalledSince2008(QueryExecutor executor) {
    // Instantiate a Car query, which is auto-generated. This class helps us build fluent queries easily.
    var query = CarQuery();
    query.where
      ..familyFriendly.equals(false)
      ..recalledAt.year.greaterThanOrEqualTo(2008);
    
    // Shorter syntax we could use instead...
    query.where.recalledAt.year <= 2008;
    
    // `get()` returns a Future<List<Car>>.
    var cars = await query.get(executor);
    
    return cars;
  }
  
  @Expose('/create', method: 'POST')
  createCar(QueryExecutor executor) async {
    // `package:protevus_orm` generates a strongly-typed `insert` function on the query class.
    // Say goodbye to typos!!!
    var query = CarQuery();
    query.values
      ..familyFriendly = true
      ..make 'Honda';
    var car = query.insert(executor);
    
    // Auto-serialized using code generated by `package:angel_serialize`
    return car;
  }
}
```

## Relations

`protevus_orm` supports the following relationships:

- `@HasOne()` (one-to-one)
- `@HasMany()` (one-to-many)
- `@BelongsTo()` (one-to-one)
- `@ManyToMany()` (many-to-many, using a "pivot" table)

The annotations can be abbreviated with the default options (ex. `@hasOne`), or supplied
with custom parameters (ex. `@HasOne(foreignKey: 'foreign_id')`).

```dart
@serializable
@orm
abstract class _Author extends Model {
  @HasMany // Use the defaults, and auto-compute `foreignKey`
  List<_Book> books;
  
  // Also supports parameters...
  @HasMany(localKey: 'id', foreignKey: 'author_id', cascadeOnDelete: true)
  List<_Book> books;
  
  @SerializableField(alias: 'writing_utensil')
  @hasOne
  _Pen pen;
}
```

The relationships will "just work" out-of-the-box, following any operation. For example, after fetching an `Author` from the database in the above example, the `books` field would be populated with a set of deserialized `Book` objects, also fetched from the database.

Relationships use joins when possible, but in the case of `@HasMany()`, two queries are used:

- One to fetch the object itself
- One to fetch a list of related objects

### Many to Many Relations

A many-to-many relationship can now be modeled like so. `RoleUser` in this case is a pivot table joining `User` and `Role`.

Note that in this case, the models must reference the private classes (`_User`, etc.), because the canonical versions (`User`, etc.) are not-yet-generated:

```dart
@serializable
@orm
abstract class _User extends Model {
  String get username;
  String get password;
  String get email;

  @ManyToMany(_RoleUser)
  List<_Role> get roles;
}

@serializable
@orm
abstract class _RoleUser {
  @belongsTo
  _Role get role;

  @belongsTo
  _User get user;
}

@serializable
@orm
abstract class _Role extends Model {
  String name;

  @ManyToMany(_RoleUser)
  List<_User> get users;
}
```

TLDR:

1. Make a pivot table, C, between two tables, table A and B
2. C should `@belongsTo` both A and B. C *should not* extend `Model`.
3. A should have a field: `@ManyToMany(_C) List<_B> get b;`
4. B should have a field: `@ManyToMany(_C) List<_A> get a;`

Test: <https://raw.githubusercontent.com/protevus-dart/orm/master/angel_orm_generator/test/many_to_many_test.dart>

## Columns

Use a `@Column()` annotation to change how a given field is handled within the ORM.

### Column Types

Using the `@Column()` annotation, it is possible to explicitly declare the data type of any given field:

```dart
@serializable
@orm
abstract class _Foo extends Model {
  @Column(type: ColumnType.bigInt)
  int bar;
}
```

### Indices

Columns can also have an `index`:

```dart
@serializable
@orm
abstract class _Foo extends Model {
  @Column(index: IndexType.primaryKey)
  String bar;
}
```

### Default Values

It is also possible to specify the default value of a field.
**Note that this only works with primitive objects.**

If a default value is supplied, the `SqlMigrationBuilder` will include it in the generated schema. The `PostgresOrmGenerator` ignores default values; it does not need them to function properly.

```dart
@serializable
@orm
abstract class _Foo extends Model {
  @Column(defaultValue: 'baz')
  String bar;
}
```