clientity 0.1.6

rapid http clients

clientity

Rapid HTTP clients with operator-based endpoint definitions.

Installation

pip install clientity

With your preferred HTTP library:

pip install clientity[httpx]    # or requests, aiohttp

Quick Start

import httpx
from dataclasses import dataclass
from clientity import client, endpoint, Resource

# Define response models
@dataclass
class User:
    id: int
    name: str
    email: str

@dataclass
class Status:
    ok: bool
    version: str

# Create client
api = client(httpx.AsyncClient()) @ "https://api.example.com"

# Define endpoints with operators
api.status = endpoint.get @ "/status" >> Status
api.user = endpoint.get @ "/users/{id}" >> User
api.users = endpoint.get @ "/users" >> list[User]

# Use it
async def main():
    status = await api.status()
    user = await api.user(id=123)
    users = await api.users()

Operators

Operator	Method	Description
@	.at(path)	Set endpoint path
%	.queries(model)	Set query parameter model
<<	.requests(model)	Set request body model
>>	.responds(model)	Set response model
&	.prehook(fn)	Add pre-request hook
|	.posthook(fn)	Add post-response hook

# Full example
api.search = (
    endpoint.post 
    @ "/search" 
    % SearchQuery      # query params
    << SearchBody      # request body
    >> list[Result]    # response model
    & log_request      # pre-hook [(Request) -> Request]
    | log_response     # post-hook [(Response) -> Response]
)

results = await api.search(q="python", limit=10, filters={"type": "code"})

Path Parameters

Path parameters are extracted from {param} syntax and can be passed positionally or by name:

api.user = endpoint.get @ "/users/{user_id}/posts/{post_id}"

# Both work:
post = await api.user(123, 456)
post = await api.user(user_id=123, post_id=456)

Query & Body Models

Use dataclasses or Pydantic models for query parameters and request bodies:

from dataclasses import dataclass

@dataclass
class SearchQuery:
    q: str
    limit: int = 10
    offset: int = 0

@dataclass
class CreateUser:
    name: str
    email: str

api.search = endpoint.get @ "/search" % SearchQuery
api.create_user = endpoint.post @ "/users" << CreateUser

# Query params from model fields
results = await api.search(q="python", limit=20)

# Body from model fields  
user = await api.create_user(name="Joel", email="[e-mail]")

Response Models

Single Model

@dataclass
class User:
    id: int
    name: str

api.user = endpoint.get @ "/users/{id}" >> User
user = await api.user(id=1)  # Returns User instance

List Response

api.users = endpoint.get @ "/users" >> list[User]
users = await api.users()  # Returns list[User]

Custom Response Handling

Implement __respond__ for custom single-item parsing:

@dataclass
class User:
    id: int
    name: str
    
    @classmethod
    def __respond__(cls, response) -> 'User':
        data = response.json()["data"]["user"]
        return cls(**data)

Implement __respondall__ for custom list parsing:

@dataclass  
class User:
    id: int
    name: str
    
    @classmethod
    def __respondall__(cls, response) -> list['User']:
        data = response.json()
        return [cls(**u) for u in data["results"]]

Resources

Group related endpoints under a path prefix:

from clientity import Resource

users = Resource("users")
users.list = endpoint.get @ "" >> list[User]
users.get = endpoint.get @ "{id}" >> User
users.create = endpoint.post @ "" << CreateUser >> User
users.delete = endpoint.delete @ "{id}"

api.users = users

# Usage
all_users = await api.users.list()
user = await api.users.get(id=123)
new_user = await api.users.create(name="Joel", email="[e-mail]")
await api.users.delete(id=123)

Nested Resources

api_resource = Resource("api")
v1 = Resource("v1")
v1.users = endpoint.get @ "users" >> list[User]
v1.posts = endpoint.get @ "posts" >> list[Post]
api_resource.v1 = v1

api.api = api_resource

# Resolves to /api/v1/users
users = await api.api.v1.users()

Namespaces

For endpoints with different base URLs or independent HTTP clients:

from clientity import Namespace

# Independent namespace with own client
search = Namespace(
    base="https://search.example.com",
    interface=httpx.AsyncClient()
)
search.query = endpoint.post @ "/query" >> list[Result]

api.search = search

# Uses search.example.com, not api.example.com
results = await api.search.query(q="test")

Dependent namespace (uses parent client):

auth = Namespace(name="auth")
auth.login = endpoint.post @ "/auth/login" << Credentials >> Token

api.auth = auth

# Uses api.example.com/auth/login
token = await api.auth.login(username="joel", password="secret")

Hooks

Pre-hooks modify the request before sending:

def add_auth(request):
    request.headers["Authorization"] = "Bearer token123"
    return request

api.secure = endpoint.get @ "/secure" & add_auth

Post-hooks modify the response after receiving:

def log_response(response):
    print(f"Status: {response.status_code}")
    return response

api.logged = endpoint.get @ "/data" | log_response

Async hooks work too:

async def refresh_token(request):
    token = await get_fresh_token()
    request.headers["Authorization"] = f"Bearer {token}"
    return request

api.secure = endpoint.get @ "/secure" & refresh_token

Supported HTTP Libraries

clientity adapts to your preferred HTTP library:

# httpx (sync or async)
import httpx
api = client(httpx.AsyncClient()) @ "https://api.example.com"
api = client(httpx.Client()) @ "https://api.example.com"

# requests
import requests
api = client(requests.Session()) @ "https://api.example.com"

# aiohttp
import aiohttp
api = client(aiohttp.ClientSession()) @ "https://api.example.com"

Lazy Interface

Pass a callable to defer client creation:

def make_client():
    return httpx.AsyncClient(headers={"X-API-Key": os.environ["API_KEY"]})

api = client(make_client) @ "https://api.example.com"

License

MIT