Important
Renewing maintanance of this library!
After a long pause, I see the library is still in use and receives more stars. Thank you all who likes and uses it. So, I renew the py-automapper maintanance. Expect fixes and new version soon.
Table of Contents:
Check CHANGELOG.md
Python auto mapper is useful for multilayer architecture which requires constant mapping between objects from separate layers (data layer, presentation layer, etc).
Inspired by: object-mapper
The major advantage of py-automapper is its extensibility, that allows it to map practically any type, discover custom class fields and customize mapping rules. Read more in documentation.
Read CONTRIBUTING.md guide.
Install package:
pip install py-automapper
Let's say we have domain model UserInfo
and its API representation PublicUserInfo
without exposing user age
:
class UserInfo:
def __init__(self, name: str, profession: str, age: int):
self.name = name
self.profession = profession
self.age = age
class PublicUserInfo:
def __init__(self, name: str, profession: str):
self.name = name
self.profession = profession
user_info = UserInfo("John Malkovich", "engineer", 35)
To create PublicUserInfo
object:
from automapper import mapper
public_user_info = mapper.to(PublicUserInfo).map(user_info)
print(vars(public_user_info))
# {'name': 'John Malkovich', 'profession': 'engineer'}
You can register which class should map to which first:
# Register
mapper.add(UserInfo, PublicUserInfo)
public_user_info = mapper.map(user_info)
print(vars(public_user_info))
# {'name': 'John Malkovich', 'profession': 'engineer'}
If source object is dictionary:
source = {
"name": "John Carter",
"profession": "hero"
}
public_info = mapper.to(PublicUserInfo).map(source)
print(vars(public_info))
# {'name': 'John Carter', 'profession': 'hero'}
If your target class field name is different from source class.
class PublicUserInfo:
def __init__(self, full_name: str, profession: str):
self.full_name = full_name # UserInfo has `name` instead
self.profession = profession
Simple map:
public_user_info = mapper.to(PublicUserInfo).map(user_info, fields_mapping={
"full_name": user_info.name
})
Preregister and map. Source field should start with class name followed by period sign and field name:
mapper.add(UserInfo, PublicUserInfo, fields_mapping={"full_name": "UserInfo.name"})
public_user_info = mapper.map(user_info)
print(vars(public_user_info))
# {'full_name': 'John Malkovich', 'profession': 'engineer'}
Very easy if you want to field just have different value, you provide a new value:
public_user_info = mapper.to(PublicUserInfo).map(user_info, fields_mapping={
"full_name": "John Cusack"
})
print(vars(public_user_info))
# {'full_name': 'John Cusack', 'profession': 'engineer'}
By default, py-automapper performs a recursive copy.deepcopy()
call on all attributes when copying from source object into target class instance.
This makes sure that changes in the attributes of the source do not affect the target and vice versa.
If you need your target and source class share same instances of child objects, set use_deepcopy=False
in map
function.
from dataclasses import dataclass
from automapper import mapper
@dataclass
class Address:
street: str
number: int
zip_code: int
city: str
class PersonInfo:
def __init__(self, name: str, age: int, address: Address):
self.name = name
self.age = age
self.address = address
class PublicPersonInfo:
def __init__(self, name: str, address: Address):
self.name = name
self.address = address
address = Address(street="Main Street", number=1, zip_code=100001, city='Test City')
info = PersonInfo('John Doe', age=35, address=address)
# default deepcopy behavior
public_info = mapper.to(PublicPersonInfo).map(info)
print("Target public_info.address is same as source address: ", address is public_info.address)
# Target public_info.address is same as source address: False
# disable deepcopy
public_info = mapper.to(PublicPersonInfo).map(info, use_deepcopy=False)
print("Target public_info.address is same as source address: ", address is public_info.address)
# Target public_info.address is same as source address: True
py-automapper
has few predefined extensions for mapping support to classes for frameworks:
Out of the box Pydantic models support:
from pydantic import BaseModel
from typing import List
from automapper import mapper
class UserInfo(BaseModel):
id: int
full_name: str
public_name: str
hobbies: List[str]
class PublicUserInfo(BaseModel):
id: int
public_name: str
hobbies: List[str]
obj = UserInfo(
id=2,
full_name="Danny DeVito",
public_name="dannyd",
hobbies=["acting", "comedy", "swimming"]
)
result = mapper.to(PublicUserInfo).map(obj)
# same behaviour with preregistered mapping
print(vars(result))
# {'id': 2, 'public_name': 'dannyd', 'hobbies': ['acting', 'comedy', 'swimming']}
Out of the box TortoiseORM models support:
from tortoise import Model, fields
from automapper import mapper
class UserInfo(Model):
id = fields.IntField(pk=True)
full_name = fields.TextField()
public_name = fields.TextField()
hobbies = fields.JSONField()
class PublicUserInfo(Model):
id = fields.IntField(pk=True)
public_name = fields.TextField()
hobbies = fields.JSONField()
obj = UserInfo(
id=2,
full_name="Danny DeVito",
public_name="dannyd",
hobbies=["acting", "comedy", "swimming"],
using_db=True
)
result = mapper.to(PublicUserInfo).map(obj)
# same behaviour with preregistered mapping
# filtering out protected fields that start with underscore "_..."
print({key: value for key, value in vars(result) if not key.startswith("_")})
# {'id': 2, 'public_name': 'dannyd', 'hobbies': ['acting', 'comedy', 'swimming']}
Out of the box SQLAlchemy models support:
from sqlalchemy.orm import declarative_base
from sqlalchemy import Column, Integer, String
from automapper import mapper
Base = declarative_base()
class UserInfo(Base):
__tablename__ = "users"
id = Column(Integer, primary_key=True)
full_name = Column(String)
public_name = Column(String)
hobbies = Column(String)
def __repr__(self):
return "<User(full_name='%s', public_name='%s', hobbies='%s')>" % (
self.full_name,
self.public_name,
self.hobbies,
)
class PublicUserInfo(Base):
__tablename__ = 'public_users'
id = Column(Integer, primary_key=True)
public_name = Column(String)
hobbies = Column(String)
obj = UserInfo(
id=2,
full_name="Danny DeVito",
public_name="dannyd",
hobbies="acting, comedy, swimming",
)
result = mapper.to(PublicUserInfo).map(obj)
# same behaviour with preregistered mapping
# filtering out protected fields that start with underscore "_..."
print({key: value for key, value in vars(result) if not key.startswith("_")})
# {'id': 2, 'public_name': 'dannyd', 'hobbies': "acting, comedy, swimming"}
When you first time import mapper
from automapper
it checks default extensions and if modules are found for these extensions, then they will be automatically loaded for default mapper
object.
What does extension do? To know what fields in Target class are available for mapping, py-automapper
needs to know how to extract the list of fields. There is no generic way to do that for all Python objects. For this purpose py-automapper
uses extensions.
List of default extensions can be found in /automapper/extensions folder. You can take a look how it's done for a class with __init__
method or for Pydantic or TortoiseORM models.
You can create your own extension and register in mapper
:
from automapper import mapper
class TargetClass:
def __init__(self, **kwargs):
self.name = kwargs["name"]
self.age = kwargs["age"]
@staticmethod
def get_fields(cls):
return ["name", "age"]
source_obj = {"name": "Andrii", "age": 30}
try:
# Map object
target_obj = mapper.to(TargetClass).map(source_obj)
except Exception as e:
print(f"Exception: {repr(e)}")
# Output:
# Exception: KeyError('name')
# mapper could not find list of fields from BaseClass
# let's register extension for class BaseClass and all inherited ones
mapper.add_spec(TargetClass, TargetClass.get_fields)
target_obj = mapper.to(TargetClass).map(source_obj)
print(f"Name: {target_obj.name}; Age: {target_obj.age}")
You can also create your own clean Mapper without any extensions and define extension for very specific classes, e.g. if class accepts kwargs
parameter in __init__
method and you want to copy only specific fields. Next example is a bit complex but probably rarely will be needed:
from typing import Type, TypeVar
from automapper import Mapper
# Create your own Mapper object without any predefined extensions
mapper = Mapper()
class TargetClass:
def __init__(self, **kwargs):
self.data = kwargs.copy()
@classmethod
def fields(cls):
return ["name", "age", "profession"]
source_obj = {"name": "Andrii", "age": 30, "profession": None}
try:
target_obj = mapper.to(TargetClass).map(source_obj)
except Exception as e:
print(f"Exception: {repr(e)}")
# Output:
# Exception: MappingError("No spec function is added for base class of <class 'type'>")
# Instead of using base class, we define spec for all classes that have `fields` property
T = TypeVar("T")
def class_has_fields_property(target_cls: Type[T]) -> bool:
return callable(getattr(target_cls, "fields", None))
mapper.add_spec(class_has_fields_property, lambda t: getattr(t, "fields")())
target_obj = mapper.to(TargetClass).map(source_obj)
print(f"Name: {target_obj.data['name']}; Age: {target_obj.data['age']}; Profession: {target_obj.data['profession']}")
# Output:
# Name: Andrii; Age: 30; Profession: None
# Skip `None` value
target_obj = mapper.to(TargetClass).map(source_obj, skip_none_values=True)
print(f"Name: {target_obj.data['name']}; Age: {target_obj.data['age']}; Has profession: {hasattr(target_obj, 'profession')}")
# Output:
# Name: Andrii; Age: 30; Has profession: False