search

netglade / auto_mappr

Code generation for mapping between different objects with ease.
https://pub.dev/packages/auto_mappr
MIT License
71 stars 7 forks source link
dart dart-library flutter mapper

readme

netglade

Developed with 💚 by netglade

ci auto_mappr license: MIT style: netglade analysis Discord

A mapper that maps between different objects with ease. Heavily inspired by C# AutoMapper.

👀 What is this?

AutoMappr is a code-generation package that helps with writing object-to-object mappings, so you don't have to write code by hand.

These mappings work by analyzing source and target objects and creating mapping to selected constructor and setter fields. That is done by code generation, which moves mapping overload from runtime to pre-compile time, so your code is as fast, predictable, and debuggable as if you write it yourself.

The only thing you have to do to make it work is create a mappr class and annotate it with a @AutoMappr annotation. Then for each object mapping, set up a mapping between a source and a target type of those objects like this: MapType<Source, Target>(). This set up the automatic mapping of matching fields. Check the getting started section to learn more about the technical details. While AutoMappr has a lot of customizations, it should work in most cases automatically for you. Despite that, you can still configure default values and custom mappings for both objects and fields, ignoring unwanted fields, setting a rename, forcing a selected constructor etc.

Why should you use it?

Mapping objects to other objects can be for sure done by hand. While it works, it's incredibly boring. Most of the time, object mapping can occur in places like mapping network DTOs from/to domain layer's models, domain layer's models from/to UI models, etc. In other words: if you care about code segregation and single responsibility, you do a lot of mappings. Tools like AutoMappr can help you with reducing boilerplate code and reduce the time you would spend on mapping objects or updating the mappings.

🚀 Getting started

How to use

Create a mapping class with @AutoMappr annotation. You will also need to import the annotation. Then use MapType<Source, Target>() for each mapping.

import 'package:auto_mappr_annotation/auto_mappr_annotation.dart';

import 'my_file.auto_mappr.dart';

@AutoMappr([
  MapType<UserDto, User>(),
])
class Mappr extends $Mappr {}

Depending on your needs, it can also be heavily customized. Below you can see just some of its options. See features for a complete list.

import 'package:auto_mappr_annotation/auto_mappr_annotation.dart';

import 'my_file.auto_mappr.dart';

@AutoMappr([
  MapType<UserDto, User>(
    fields: [
      Field('address', from: 'userAddress'),
      Field('fullName', custom: Mappr.mapFullName),
      Field('age', custom: 42),
      Field('tag', ignore: true)
    ],
  ),
])
class Mappr extends $Mappr {
  static String mapFullName(UserDto dto) => '${dto.firstName} ${dto.lastName}';
}

To actually use the mappr in your code, call the convert method on it's instance. Note that the convert function has two generic parameters — source and target. AutoMappr uses type inference to determine which mapping to use, therefore you should care about a strict type inference. Either assign the result of converting to an explicitly typed variable or explicitly state generics. The function cannot infer the generic parameters just from the source parameter.

It should look like this:

void main() {
  final mappr = Mappr();

  // convert like this
  User user = mappr.convert(UserDto(...));

  // or like this
  final user2 = mappr.convert<UserDto, User>(UserDto(...));
}

To make the Dart analyzer help you with inference failures, you can set up the analyzer in the analysis_options.yaml file. I would also suggest to use some predefined list of lints and analysis settings such as very_good_analysis or our netglade_analysis.

analyzer:
  language:
    strict-casts: true
    strict-inference: true
    strict-raw-types: true

convert vs tryConvert

The convert and tryConvert works the same. The only difference is that if a null value is passed inside as the argument, and whenSourceIsNull is not set for the mapping, convert will throw an exception while tryConvert will return null. Therefore, convert has a return type TARGET and tryConvert has a return type TARGET?. It is an analogy with parse and tryParse methods on int.

try/convert{Iterable, List, Set}

If you have an iterable of source objects and need to transfer them into an iterable of target objects, use either convertIterable, convertList, or convertSet. What to choose depends on what iterable type you need as an output:

All of these function also have generics as <SOURCE, TARGET>, where the source and the target are the types of those inner objects.

If you need output of nullable objects in an iterable, use tryConvert{Iterable, List, Set} versions.

Install

To use AutoMappr, install these three packages:

For a Flutter project:

flutter pub add auto_mappr_annotation
flutter pub add --dev build_runner
flutter pub add --dev auto_mappr

For a Dart project:

dart pub add auto_mappr_annotation
dart pub add --dev build_runner
dart pub add --dev auto_mappr

Run the generator

For Dart or Flutter projects:

dart run build_runner build

✨ Features

Primitive objects mapping

Primitive objects like num, int, double, String, bool, Symbol, Type, and null are copied to target object with no additional processing.

Complex objects mapping

Complex (or nested) objects are mapped according to their MapType<SOURCE, TARGET>() mapping setup. The generator generates mapping methods for each MapType. These mapping methods are used in nested objects.

Field mapping

When target and source fields' name do not match, you can change source field by using the from argument in a Field() mapping. Alternatively, you can use the Field.from() constructor which hides other then-invalid parameters.

@AutoMappr([
  MapType<UserDto, User>(
    fields: [
      Field('name', from: 'myName'),
      Field.from('age', from: 'myAge'),
    ],
  ),
])
class Mappr extends $Mappr {}

Custom mapping

When you need to assign a custom function or a const value as a value for given target field, you can use the custom argument in a Field mapping. Alternatively, you can use the Field.custom() constructor which hides other then-invalid parameters. Provide const Target value or custom mapping function.

The custom function has to follow one of these formals:

@AutoMappr([
  MapType<UserDto, User>(
    fields: [
      Field('name', custom: Mappr.mapName), // Static Mappr method.
      Field('age', custom: mapAge), // Global method.
      Field.custom('note', custom: 'constant value'), // Constant value.
    ],
  ),
])
class Mappr extends $Mappr {
  static String mapName(UserDto dto) => dto.name.toUpperCase();
}

/// Return always 42
int mapAge() => 42;

Ignore mapping

To completely ignore some target field, so it is not mapped into a constructor or into a setter, you can use the ignore argument in a Field mapping. Alternatively, you can use the Field.ignore() constructor which hides other then-invalid parameters.

@AutoMappr([
  MapType<UserDto, User>(
    fields: [
      Field('name', ignore: true),
      Field.ignore('age', ignore: true),
    ],
  ),
])
class Mappr extends $Mappr {}

Iterable objects mapping

Values in iterables like List, Set, or Iterable are mapped using the .map() method when the values are complex types. When needed, mostly after mapping, .toList() or .toSet() methods are called to cast an Iterable into a List/Set.

Specialized variants of List<int>

AutoMappr will automatically convert between List<int> and its specialized variants Uint8List, Uint16List, Uint32List and Uint64List.

Convesion between these specialized variants are not handled and its developer responsibility to configure mapping.

Map objects mapping

Maps are a specific case of Iterables, that has to be handled a bit differently. For example, we must make sure that both keys and values of MapEntry are mapped correctly based on whether they are primitive or complex types.

Default field value

To make sure that a default value is assigned to a target field when a source field is null you can set up a whenNull property on Field which takes a constant value of target field type.

You can set up default field value by using Target Function() function or const Target value on Field, Field.from, and Field.custom, constructors but not on Field.ignore constructor.

@AutoMappr([
  MapType<UserDto, User>(
    fields: [
      Field('name', whenNull: 'John Smith'),
    ],
  ),
])
class Mappr extends $Mappr {}

Default object value

When the whole source object is null, you can set up a default value for it using the whenSourceIsNull property on MapType. It can also take a constant value of target object type.

You can set up Target Function() function or const Target value.

@AutoMappr([
  MapType<UserDto, User>(
    whenSourceIsNull: User(name: 'Neo', age: 28),
  ),
])
class Mappr extends $Mappr {}

Constructor selection

When you want to specifically select a certain constructor, set the constructor property on MapType. Otherwise the mapping automatically selects a constructor with the most parameters. It prioritizes non factory constructors over factory ones and never selects fromJson factory constructor.

Imagine that you have a User(String name, int age, String note) and User.fromDto(String name, int age) constructors. Default algorithm selects the default constructor because it has the most parameters. To change the selected constructor, do:

@AutoMappr([
  MapType<UserDto, User>(
    constructor: 'fromDto',
  ),
])
class Mappr extends $Mappr {}

Enum mapping

AutoMappr also supports mapping of enums. You register them as usual with MapType<SourceEnum, TargetEnum> and AutoMappr will convert enum options based on name.

The target enum can either be a superset of the source or has to define whenSourceIsNull which will be used for unknown enum values. If the target is not a superset of the source enum the generator will throw.

E.g. in the example below, RemotePerson.alien will be mapped to LocalPerson.unknown.

enum RemotePerson { student, employee, alien }
enum LocalPerson { student, employee, unknown }

@AutoMappr([
  MapType<RemotePerson, LocalPerson>(
    whenSourceIsNull: LocalPerson.unknown,
  ),
])
class Mappr extends $Mappr {}

Mapping works for both simple and enhanced enums.

Positional and named constructor parameters

The mapping automatically assigns source getters to constructor parameters no matter if they are positional or named.

Mapping to target

Mapping into a target object can be done in two places. First, the mapping tries to map all the fields to selected constructor. And for the target fields that have not been mapped, it tries to set them using public setters (both explicit ones or implicit ones created by fields), if they have any.

Mapping from source

Mapping from a source object can be done from either public getters or static getters. Getters can be both explicit ones or implicit ones created by fields.

Nullability handling

For each MapType<SOURCE, TARGET>() mapping the generator generates at most two mapping methods. First method is a method with non-nullable return type TARGET. Second method is a method with nullable return type TARGET? that is being generated only when other methods use it. If the object mapping has whenSourceIsNull parameter set, the nullable method is not generated.

Note that convert cannot return null. The value null can only be returned for nested object mappings.

Forced non-nullable field for nullable source

It is possible to configure automappr use bang operator (!) in cases where SOURCE's field is nullable but TARGET's field not. In that case mappr will generate 

sourceField: targetField!

By default mappr would generate invalid code 

sourceField: targetField

This is by design and it is up to developer decide if bang operator is appropiate. This behavior can be configured

Precedense is global configuration -> MapType -> Field -> defaults to false.

Safe mapping

Safe mapping is a auto_mapr feature that automatically handles exceptions thrown during the mapping. Typically these exceptions can be caused by mapping from source with a nullable variable to a target with a corresponding non-nullable variable. If you want to avoid propagating the exception through the app, you can enable safe mapping and auto_mappr can catch the exception and return null as a result of mapping instead. In case you want to know that something went wrong during the mapping there is a onMappingError callback that is invoked each time error occurs. Enabling safe mapping can be done either on the global level in build.yaml with safeMapping option

   $default:
    builders:
      auto_mappr:
        options:
          safeMapping: true

or individually for each MapType with safeMapping option:

@AutoMappr([
  MapType<UserDto, User>(
    safeMapping: true,
  ),
])
class Mappr extends $Mappr {}

Precedense is global configuration -> MapType -> defaults to false.

Safe mapping can be done only through tryConvert, tryConvertIterable, tryConvertList or tryConvertSet methods. All of them have an optional parameter void Function(Object error, StackTrace stackTrace, SOURCE? source)? onMappingError, that is a callback function triggered whenever an exception occurs during the mapping process. This can be used fo example for logging the error.

Generics

AutoMappr can handle generics, so you can map objects with type arguments with ease.

@AutoMappr([
  MapType<SomeObjectDto<num>, SomeObject<num>>(),
  MapType<SomeObjectDto<int>, SomeObject<int>>(),
  MapType<SomeObjectDto<String>, SomeObject<String>>(),
])
class Mappr extends $Mappr {}

Type parameters are not limited. You can use them as a direct type, in a nested object types, in collections, ... AutoMappr will automatically handles them for you.

class Something<A, B, C> {
  final A first;
  final Nested<A, B> second;
  final List<C> third;
  final List<Nested<A, C>> fourth;
// ...
}

class Nested<X, Y> {
  final X alpha;
  final Y beta;
// ...
}

Library import aliases

In cases when you have two libraries with classes with the same name, mapping works as expected by using import aliases.

import 'my_api.dart' as api;
import 'my_domain.dart' as entity;

@AutoMappr([
  MapType<api.User, entity.User>(),
])
class Mappr extends $Mappr {}

Modules

Each AutoMappr can be considered as a module. The only rule is that the mappr must be constant, and that most of the time means you have to add an const constructor to be able to use it. Other modules (AutoMappr classes) can then use it in two ways.

Including

Including a module means that you want to "absorb" it's mappings and use them later anywhere inside any mapped object. Basically imagine copy-pasting definitions from included module to yours.

That can be handy when you have a common/shared mappr with mappings between objects shared across the app. Since you want to use these common/shared mappings, you include them in your mappr.

// file: shared_mappr.dart
@AutoMappr(
  [
    MapType<AddressDto, Address>(),
  ],
)
class SharedMappr extends $SharedMappr {
  const SharedMappr(); // must be const!
}

// file: user_mappr.dart
@AutoMappr(
  [
    MapType<UserDto, User>(), // UserDto uses AddressDto inside
  ],
  includes: [SharedMappr()], // include shared mappr
)
class UserMappr extends $UserMappr {}

// file: settings_mappr.dart
@AutoMappr(
  [
    MapType<ProfileDto, Profile>(), // ProfileDto uses AddressDto inside
  ],
  includes: [SharedMappr()], // include shared mappr
)
class SettingsMappr extends $SettingsMappr {}

Delegating

Delegating to a module is a bit different from including them. A mappr delegates to a standalone module. When your mappr does not know how to convert the top level object (the object you put inside mappr.convert() method), it asks delegates to do it. So you can think of them as disjunctive units that are grouped together.

This is usefull when you have an app with a mappr for each feature and you want to create one main mappr using other feature mapprs. The main mappr may not have any mapping at all and it delegates everything to feature mapprs.

// file: main_mappr.dart
@AutoMappr(
  [],
  delegates: [
    UserFeatureMappr(),
    SettingsFeatureMappr(),
    // other features
  ], 
)
class MainMappr extends $MainMappr {}

// file: user_feature_mappr.dart
@AutoMappr(
  [
    MapType<UserDto, User>(),
    MapType<AddressDto, Address>(),
  ],
)
class UserFeatureMappr extends $UserFeatureMappr {}

// file: settings_feature_mappr.dart
@AutoMappr(
  [
    MapType<UserDto, User>(),
    MapType<AddressDto, Address>(),
  ],
)
class SettingsFeatureMappr extends $SettingsFeatureMappr {}

Then you can use this main mappr to map between objects specified from every included mappr.

final mappr = MainMappr();

final Settings settings = mappr.convert(SettingsDto(...)); // delegates to settings feature mappr
final User user = mappr.convert(UserDto(...)); // delegates to user feature mappr

That can be handy for example with dependency injection, so you can only provide one main mappr that can handle everything by delegating to other mapprs.

Type converters

MapTypes are usefull for mappings that you can use from the outside. AutoMappr mapps one object to another using constructors and fields and you use the .convert() method on it.

But when you need to customize an inner converting of types, there are TypeConverters that help you with that. Boxing, String to DateTime, and stuff like that, TypeConverters are your friend. Note since type converters are only used internally, you cannot in any way use them using the .convert() method.

Global type converters are also "absorbed" from included modules. To make the priority crystal clear:

  1. (local) type converters from MapType, in order
  2. (global) type converters from AutoMappr, in order
  3. (global included) type converters from included modules, in order

Use MapType for most of the things. Use TypeConverter only for cases that cannot be solve otherwise.

Imagine you have an UserDto with a date in a String and a User model with a date in DateTime. MapType will handle mapping of all the constructor parameters and other fields, while TypeConverter will convert String to DateTime.

@AutoMappr(
  [
    MapType<UserDto, User>(
      // MapType specific
      converters: [
        TypeConverter<String, DateTime>()
      ],
    ),
  ],
  converters: [
    // or globally here
  ],
)
class Mappr extends $Mappr {
  static DateTime stringToDateTime(String source) {
    // ...
  }
}

You can also create methods that can convert "any" (to "any") using Object. But if you work with type parameters (generics) note that you have to either return the correct type with correct type parameters or initialize it inside correctly. It cannot be cast successfully otherwise. Therefore if you need a method that converts "any" to Value("any"), and to make it work for int and String, it must look like one of these:

static Value<T> objectToValueObject<T extends Object>(T source) {
  return Value<T>(source);
}

static Value<Object> objectToValueObject2(Object source) {
  if (source is int) {
    return Value<int>(source);
  }

  if (source is String) {
    return Value<String>(source);
  }

  return Value(source);
}

To make it more clear, here is a list of type converters and for which source field type -> target field type combinations they can be used. In these examples, we've used int and String as a reference, but the TypeConverters can be adapted to various data types.

Reverse mapping

When you want to create a bidirectional mapping (e.g. normal: source to target and reversed: target to source), you can use reverse option on MapType.

Note that it's your responsibility to make sure those objects support normal and reverse mapping and to keep them in sync.

When Field's mapping with from parameter is used, in reverse mapping those properties are automatically switched too.

For example

class A {
  final String a;
}

class B {
  final String b;
}

// Mapping
MapType<A,B>(fields: [
  Field('b', from: 'a')
], reverse: true)

// generated mapping
A.a mapts to B.b
B.b maps to A.a

Also note that reverse mapping might not work properly when additional configuration such as whenSourceIsNull or constructor is used.

For more complicated scenarios two separate mappings are recommended instead.

Records

Converting records is supported for both positional and named record's fields.

Target positional fields must have their source field equivalent. Target named fields must have their source field equivalent determined by name. Both positional and named target fields without source equivalent must be nullable in order for mapping to be created successfully and then thier value will be null.

Note that we do not have a function similar to convertList for records due to Dart 3 "limitation" as we cannot iterate throught it's positional or named fields on the fly.

Works with equatable

Mapping works with the Equatable package. Some mapping tools tries to map the props getter, but since AutoMappr maps only to public explicit or implicit setters, Equatable and other packages with similar conditions implicitly works.

Works with json_serializable

AutoMappr uses a LibraryBuilder with .auto_mappr.dart file output. That means it does not interfere with shared .g.dart file with packages like JSON Serializable to generate other code to the generated super class.

Works with generated source and target classes

Mapping can be set up for source or target classes which are generated by another package, like Drift. For that, you have to customize a builder to set custom dependencies on the other package's outputs.

You can also use this package as an output for another package's builder. Disable the default auto_mappr builder and enable the auto_mappr:not_shared builder. Check the customizing the build chapter to learn more.

⚙ Customizing the build

By default, AutoMappr uses the auto_mappr:auto_mappr builder that works with LibraryBuilder, which generates .auto_mappr.dart file.

Modify your build.yaml file:

targets:
  $default:
    # You can disable all default builders.
    auto_apply_builders: false
    builders:
      # Or disable specific ones.
      auto_mappr:
        enabled: true

Builder options

Default dependencies

By default the auto_mappr builder has defined required inputs for freezed and drift classes.

 required_inputs: [".freezed.dart", ".drift.dart"]

This allows to depend on generated classes from these packages without need to modify project's build.yaml.

Drift integration

If you are using packages like Drift which generates classes you need to use as a source or a target in your mappings, use their not-shared builder, if they have any. With that, the builder can generate files like .drift.dart which you can add a input dependency to. Specify the required_inputs dependency on your local AutoMappr builder and disable the builders provided by AutoMappr.

targets:
  $default:
    auto_apply_builders: true
    builders:
      # Enable their generators according to their documentation.
      drift_dev:not_shared:
        enabled: true
      drift_dev:preparing_builder:
        enabled: true
      # Disable Drift's shared builder
      drift_dev:drift_dev:
        enabled: false

      auto_mappr:
        enabled: true

👏 Contributing

Your contributions are always welcome! Feel free to open pull request.