pydantic-resolve 1.9.1

create nested data structure easily

If you are using pydantic v2, please use pydantic2-resolve instead.

A small yet powerful tool to extend your pydantic schema, and then resolve all descendants automatically.

It's also the key to the realm of composition-oriented-development-pattern (wip)

What is composable pattarn? https://github.com/allmonday/composable-development-pattern

Change Logs

discord

Install

pip install pydantic-resolve

Code snippets

1. basic usage, resolve your fields.

import asyncio
from pydantic import BaseModel
from pydantic_resolve import Resolver

async def query_age(name):
    print(f'query {name}')
    await asyncio.sleep(1)
    _map = {
        'kikodo': 21,
        'John': 14,
        '老王': 40,
    }
    return _map.get(name)

class Person(BaseModel):
    name: str

    age: int = 0
    async def resolve_age(self):
        return await query_age(self.name)

    is_adult: bool = False
    def post_is_adult(self):
        return self.age > 18

async def simple():
    p = Person(name='kikodo')
    p = await Resolver().resolve(p)
    print(p)
    # query kikodo
    # Person(name='kikodo', age=21, is_adult=True)

    people = [Person(name=n) for n in ['kikodo', 'John', '老王']]
    people = await Resolver().resolve(people)
    print(people)
    # Oops!! the issue of N+1 query happens
    # query kikodo
    # query John
    # query 老王
    # [Person(name='kikodo', age=21, is_adult=True), Person(name='John', age=14, is_adult=False), Person(name='老王', age=40, is_adult=True)]

asyncio.run(simple())

2. optimize N+1 with dataloader

import asyncio
from typing import List
from pydantic import BaseModel
from pydantic_resolve import Resolver, LoaderDepend as LD

async def batch_person_age_loader(names: List[str]):
    print(names)
    _map = {
        'kikodo': 21,
        'John': 14,
        '老王': 40,
    }
    return [_map.get(n) for n in names]

class Person(BaseModel):
    name: str

    age: int = 0
    def resolve_age(self, loader=LD(batch_person_age_loader)):
        return loader.load(self.name)

    is_adult: bool = False
    def post_is_adult(self):
        return self.age > 18

async def simple():
    people = [Person(name=n) for n in ['kikodo', 'John', '老王']]
    people = await Resolver().resolve(people)
    print(people)

    # query query kikodo,John,老王 (N+1 query fixed)
    # [Person(name='kikodo', age=21, is_adult=True), Person(name='John', age=14, is_adult=False), Person(name='老王', age=40, is_adult=True)]

asyncio.run(simple())

More examples:

cd examples

python -m readme_demo.0_basic
python -m readme_demo.1_filter
python -m readme_demo.2_post_methods
python -m readme_demo.3_context
python -m readme_demo.4_loader_instance
python -m readme_demo.5_subset
python -m readme_demo.6_mapper
python -m readme_demo.7_single

API

Resolver(loader_filters, global_loader_filter, loader_instances, context)

• loader_filters: dict

  provide extra query filters along with loader key.

  reference: 6_sqlalchemy_loaderdepend_global_filter.py L55, L59

• global_loader_filter: dict

  provide global filter config for all dataloader instances

  it will raise exception if some fields are duplicated with specific loader filter config in loader_filters

  reference: test_33_global_loader_filter.py L47, L49

• loader_instances: dict

  provide pre-created loader instance, with can prime data into loader cache.

  reference: test_20_loader_instance.py, L62, L63

• context: dict

  context can carry setting into each single resolver methods.

  class Earth(BaseModel):
      humans: List[Human] = []
      def resolve_humans(self, context):
          return [dict(name=f'man-{i}') for i in range(context['count'])]
  
  earth = await Resolver(context={'count': 10}).resolve(earth)
  

LoaderDepend(loader_fn)

• loader_fn: subclass of DataLoader or batch_load_fn. detail

  declare dataloader dependency, pydantic-resolve will take the care of lifecycle of dataloader.

build_list(rows, keys, fn), build_object(rows, keys, fn)

• rows: list, query result

• keys: list, batch_load_fn:keys

• fn: lambda, define the way to get primary key

  helper function to generate return value required by batch_load_fn. read the code for details.

  reference: test_utils.py, L32

mapper(param)

• param: class of pydantic or dataclass, or a lambda

  pydantic-resolve will trigger the fn in mapper after inner future is resolved. it exposes an interface to change return schema even from the same dataloader. if param is a class, it will try to automatically transform it.

  reference: test_16_mapper.py

  you may need it if there has some reuseable transforming params.

ensure_subset(target_model)

• target_model: class

  it will raise exception if fields of decorated class has field not existed in base_class.

  this provide a validation to ensure your schema's field is a subset of targe schema.

  reference: test_2_ensure_subset.py

model_config(hidden_fields: list[str], default_required: bool) (new in v1.9.1)

• hidden_fields: list the field names you don't want to expose.

• default_required: if True, fields with default values will also in schema['required']

  reference: test_schema_config.py

Run FastAPI example

poetry shell
cd examples
uvicorn fastapi_demo.main:app
# http://localhost:8000/docs#/default/get_tasks_tasks_get

Unittest

poetry run python -m unittest  # or
poetry run pytest  # or
poetry run tox

Coverage

poetry run coverage run -m pytest
poetry run coverage report -m