'db' within pydantic - A single model for shaping, creating, accessing, storing data within a Database
Key Features
- Automatic Migration on Schema Changes
- Flexible Data Types
- One Model for type validation & database access
- Dynamic Model Relationships
- Integrated Redis Caching Support
Documentation
https://pydbantic.readthedocs.io/en/latest/
Setup
$ pip install pydbantic
$ pip install pydbantic[sqlite]
$ pip install pydbantic[mysql]
$ pip install pydbantic[postgres]Basic Usage - Model
from typing import List, Optional, Union
from pydbantic import DataBaseModel, PrimaryKey
class Department(DataBaseModel):
department_id: str = PrimaryKey()
name: str
company: str
is_sensitive: bool = False
positions: List[Optional['Positions']] = [] # One to Many
class Positions(DataBaseModel):
position_id: str = PrimaryKey()
name: str
department: Department = None # One to One mapping
employees: List[Optional['Employee']] = [] # One to Many
class EmployeeInfo(DataBaseModel):
ssn: str = PrimaryKey()
first_name: str
last_name: str
address: str
address2: Optional[str]
city: Optional[str]
zip: Optional[int]
new: Optional[str]
employee: Optional[Union['Employee', dict]] = None # One to One
class Employee(DataBaseModel):
employee_id: str = PrimaryKey()
employee_info: Optional[EmployeeInfo] = None # One to One
position: List[Optional[Positions]] = [] # One to Many
salary: float
is_employed: bool
date_employed: Optional[str]Basic Usage - Connecting a Database to Models
import asyncio
from pydbantic import Database
from models import Employee, EmployeeInfo, Positions, Department
async def main():
db = await Database.create(
'sqlite:///test.db',
tables=[
Employee,
EmployeeInfo,
Positions,
Department
]
)
if __name__ == '__main__':
asyncio.run(main())Model Usage
Import and use the models where you need them. As long as DB as already been created, the Models can accesss the & Use the connected DB
from models import (
Employee,
EmployeeInfo,
Position,
Department
)Model - Creation
# create department
hr_department = await Department.create(
id='d1234',
name='hr'
company='abc-company',
is_sensitive=True,
)Via instance using insert or save
hr_department = Department.create(
id='d1234',
name='hr'
company='abc-company',
is_sensitive=True,
)
await hr_department.insert()
await hr_department.save()Insert with related models
# create a Position in Hr Department
hr_manager = Position.create(
id='p1234',
name='manager',
department=hr_department
)
# create instance on an hr employee
hr_emp_info = EmployeeInfo.create(
ssn='123-456-789',
first_name='john',
last_name='doe',
address='123 lane',
city='snake city',
zip=12345
)
# create an hr employee
hr_employee = await Employee.create(
id='e1234',
employee_info=hr_emp_info,
position=hr_manager,
is_employed=True,
date_employed='1970-01-01'
)Filtering
# get all hr managers currently employed
managers = await Employee.filter(
Employee.position==hr_manager, # conditional
is_employed=True # key-word argument
)
first_100_employees = await Employee.all(
limit=100
)See also filtering operators
Deleting
# remove all managers not employed anymore
for manager in await Employee.filter(
position=hr_manager,
is_employed=False
):
await manager.delete()Updating
# raise salary of all managers
for manager in await Employee.filter(
position=hr_manager,
is_employed=False
):
manager.salary = manager.salary + 1000.0
await manager.update() # or manager.save().save() results in a new row created in Employee table as well as the related EmployeeInfo, Position, Department tables if not yet created. s
What is pydbantic
pydbantic was built to solve some of the most common pain developers may face working with databases.
- migrations
- model creation / managment
- dynamic relationships
- caching
pydbantic believes that related data should be stored together, in the shape the developer plans to use
pydbantic knows data is rarely flat or follows a set schema
pydbantic understand migrations are not fun, and does them for you
pydbantic speaks many types
Pillars
- pydantic - Models, Type validation
- databases - Database Connection Abstractions
- sqlalchemy - Core Database Query and Database Model
Models
pydbantic most basic object is a DataBaseModel. This object may be comprised of almost any pickle-able python object, though you are encouraged to stay within the type-validation land by using pydantic's BaseModels and validators.
Primary Keys
DataBaseModel 's also have a priamry key, which is the first item defined in a model or marked with = PrimaryKey()
class NotesBm(DataBaseModel):
id: str = PrimaryKey()
text: Optional[str] # optional
data: DataModel # required
coridinates: tuple # required
items: list # required
nested: dict = {'nested': True} # Optional - w/ DefaultModel Types & Typing
DataBaseModel items are capable of being multiple layers deep following pydantic model validation
- Primary Key - First Item, must be unique
- Required - items without default values are assumed required
- Optional - marked explicitly with
typing.Optionalor with a default value - Union - Accepts Either specified input type Union[str|int]
- List[item] - Lists of specified items
Input datatypes without a natural / built in serialization path are serialized using pickle and stored as bytes. More on this later.
Migrations
pydbantic handles migrations automatically in response to detected model changes: New Field, Removed Field, Modified Field, Renamed Field, Primary Key Changes
Renaming an exiting column
Speical consideration is needed when renaming a field in a DataBaseModel, extra metadata __renamed__ is needed to ensure existing data is migrated:
# field `first_name` is renamed to `first_names`
class EmployeeInfo(DataBaseModel):
__renamed__= [{'old_name': 'first_name', 'new_name': 'first_names'}]
ssn: str = PrimaryKey()
first_names: str
last_name: str
address: str
address2: Optional[str]
city: Optional[str]
zip: Optional[int]Cache
Adding cache with Redis is easy with pydbantic, and is complete with built in cache invalidation.
db = await Database.create(
'sqlite:///test.db',
tables=[Employee],
cache_enabled=True,
redis_url="redis://localhost"
)Models with arrays of Foreign Objects
DataBaseModel models can support arrays of both BaseModels and other DataBaseModel. Just like single DataBaseModel references, data is stored in separate tables, and populated automatically when the child DataBaseModel is instantiated.
from uuid import uuid4
from datetime import datetime
from typing import List, Optional
from pydbantic import DataBaseModel, PrimaryKey
def time_now():
return datetime.now().isoformat()
def get_uuid4():
return str(uuid4())
class Coordinate(DataBaseModel):
id: str = PrimaryKey(default=get_uuid4)
lat_long: tuple
journeys: List[Optional["Journey"]] = []
class Journey(DataBaseModel):
trip_id: str = PrimaryKey(default=get_uuid4)
waypoints: List[Optional[Coordinate]] = []