phenobarbital/python-datamodel

DataModel is a reimplementation of python Dataclasses supporting true inheritance (without decorators), true composition and other good features.

DataModel

DataModel is a simple library based on python +3.8 to use Dataclass-syntax for interacting with Data, using the same syntax of Dataclass, users can write Python Objects and work with Data in the same way (like ORM's), is a reimplementation of python Dataclasses supporting true inheritance (without decorators), true composition and other good features.

The key features are:

• Easy to use: No more using decorators, concerns abour re-ordering attributes or common problems with using dataclasses with inheritance.
• Extensibility: Can use other dataclasses, Data objects or primitives as data-types.
• Fast: DataModel is a replacement 100% full compatible with dataclasses, without any overhead.

Requirements

Python 3.8+

Installation

$ pip install python-datamodel
---> 100%
Successfully installed datamodel

Quickstart

from datamodel import Field, BaseModel
from dataclasses import dataclass, fields, is_dataclass


# This pure Dataclass:
@dataclass
class Point:
    x: int = Field(default=0, min=0, max=10)
    y: int = Field(default=0, min=0, max=10)

point = Point(x=10, y=10)
print(point)
print(fields(point))
print('IS a Dataclass?: ', is_dataclass(point))

# Can be represented by BaseModel
class newPoint(BaseModel):
    x: int = Field(default=0, min=0, max=10)
    y: int = Field(default=0, min=0, max=10)

    def get_coordinate(self):
        return (self.x, self.y)

point = newPoint(x=10, y=10)
print(point)
print(fields(point))
print('IS a Dataclass?: ', is_dataclass(point))
print(point.get_coordinate())

Supported types

DataModel support recursive transformation of fields, so you can easily work with nested dataclasses or complex types.

DataModel supports automatic conversion of:

• datetime objects. datetime objects are encoded to str exactly like orjson conversion, any str typed as datetime is decoded to datetime. The same behavior is used to decoding time, date and timedelta objects.

• UUID objects. They are encoded as str (JSON string) and decoded back to uuid.UUID objects.

• Decimal objects. They are also encoded as float and decoded back to Decimal.

Also, "custom" encoders are supported.

import uuid
from typing import (
    List,
    Optional,
    Union
)
from dataclasses import dataclass, field
from datamodel import BaseModel, Field

@dataclass
class Point:
    x: int = Field(default=0, min=0, max=10)
    y: int = Field(default=0, min=0, max=10)

class coordinate(BaseModel, intSum):
    latitude: float
    longitude: float

    def get_location(self) -> tuple:
        return (self.latitude, self.longitude)

def auto_uid():
    return uuid.uuid4()

def default_rect():
    return [0,0,0,0]

def valid_zipcode(field, value):
    return value > 1000

class Address(BaseModel):
    id: uuid.UUID = field(default_factory=auto_uid)
    street: str = Field(required=True)
    zipcode: int = Field(required=False, default=1010, validator=valid_zipcode)
    location: Optional[coordinate]
    box: List[Optional[Point]]
    rect: List[int] = Field(factory=default_rect)


addr = Address(street="Calle Mayor", location=(18.1, 22.1), zipcode=3021, box=[(2, 10), (4, 8)], rect=[1, 2, 3, 4])
print('IS a Dataclass?: ', is_dataclass(addr))

print(addr.location.get_location())

# returns
Address(id=UUID('24b34dd5-8d35-4cfd-8916-7876b28cdae3'), street='Calle Mayor', zipcode=3021, location=coordinate(latitude=18.1, longitude=22.1), box=[Point(x=2, y=10), Point(x=4, y=8)], rect=[1, 2, 3, 4])

• Fast and convenience conversion from-to JSON (using orjson):

import orjson

b = addr.json()
print(b)

{"id":"24b34dd5-8d35-4cfd-8916-7876b28cdae3","street":"Calle Mayor","zipcode":3021,"location":{"latitude":18.1,"longitude":22.1}, "box":[{"x":2,"y":10},{"x":4,"y":8}],"rect":[1,2,3,4]}

# and re-imported from json
new_addr = Address.from_json(b) # load directly from json string
# or using a dictionary decoded by orjson
data = orjson.loads(b)
new_addr = Address(**data)

Inheritance

python-datamodel supports inheritance of classes.

import uuid
from typing import Union, List
from dataclasses import dataclass, field
from datamodel import BaseModel, Column, Field


def auto_uid():
    return uuid.uuid4()

class User(BaseModel):
    id: uuid.UUID = field(default_factory=auto_uid)
    name: str
    first_name: str
    last_name: str


@dataclass
class Address:
    street: str
    city: str
    state: str
    zipcode: str
    country: Optional[str] = 'US'

    def __str__(self) -> str:
        """Provides pretty response of address"""
        lines = [self.street]
        lines.append(f"{self.city}, {self.zipcode} {self.state}")
        lines.append(f"{self.country}")
        return "\n".join(lines)

class Employee(User):
    """
    Base Employee.
    """
    role: str
    address: Address # composition of a dataclass inside of DataModel is possible.

# Supporting multiple inheritance and composition
# Wage Policies
class MonthlySalary(BaseModel):
    salary: Union[float, int]

    def calculate_payroll(self) -> Union[float, int]:
        return self.salary

class HourlySalary(BaseModel):
    salary: Union[float, int] = Field(default=0)
    hours_worked: Union[float, int] = Field(default=0)

    def calculate_payroll(self) -> Union[float, int]:
        return (self.hours_worked * self.salary)

# employee types
class Secretary(Employee, MonthlySalary):
    """Secretary.

    Person with montly salary policy and no commissions.
    """
    role: str = 'Secretary'

class FactoryWorker(Employee, HourlySalary):
    """
    FactoryWorker is an employee with hourly salary policy and no commissions.
    """
    role: str = 'Factory Worker'

class PayrollSystem:
    def calculate_payroll(self, employees: List[dataclass]) -> None:
        print('=== Calculating Payroll === ')
        for employee in employees:
            print(f"Payroll for employee {employee.id} - {employee.name}")
            print(f"- {employee.role} Amount: {employee.calculate_payroll()}")
            if employee.address:
                print('- Sent to:')
                print(employee.address)
            print("")

jane = Secretary(name='Jane Doe', first_name='Jane', last_name='Doe', salary=1500)
bob = FactoryWorker(name='Bob Doyle', first_name='Bob', last_name='Doyle', salary=15, hours_worked=40)
mitch = FactoryWorker(name='Mitch Brian', first_name='Mitch', last_name='Brian', salary=20, hours_worked=35)

payroll = PayrollSystem()
payroll.calculate_payroll([jane, bob, mitch])

A sample of output:

```console
=== Calculating Payroll ===
Payroll for employee 745a2623-d4d2-4da6-bf0a-1fa691bafd33 - Jane Doe
- Secretary Amount: 1500
- Sent to:
Rodeo Drive, Rd
Los Angeles, 31050 CA
US

Contributing

First of all, thank you for being interested in contributing to this library. I really appreciate you taking the time to work on this project.

• If you're just interested in getting into the code, a good place to start are issues tagged as bugs.
• If introducing a new feature, especially one that modifies the public API, consider submitting an issue for discussion before a PR. Please also take a look at existing issues / PRs to see what you're proposing has already been covered before / exists.
• I like to follow the commit conventions documented here

License

This project is licensed under the terms of the BSD v3. license.