Skip to content

README.md

Latest commit

 

History

History
297 lines (222 loc) · 8.79 KB

README.md

File metadata and controls

297 lines (222 loc) · 8.79 KB

Asynchronous GraphQL Client

PyPI version Python Versions License: MIT Documentation Status

An asynchronous GraphQL client built on top of aiohttp and graphql-core-next. It supports multiple transports, including aiohttp (default) and httpx. The client by default introspects schemas and validates all queries prior to dispatching to the server.

Documentation

For the most recent project documentation, you can visit https://aiographql-client.readthedocs.io/.

Installation

To install the client with the default aiohttp transport:

pip install aiographql-client[aiohttp]

To use httpx as the transport (includes websockets for subscriptions):

pip install aiographql-client[httpx]

To install the client with the default aiohttp transport and pydantic support:

pip install aiographql-client[aiohttp,pydantic]

Transports & Auto-detection

The client supports multiple HTTP backends. You can choose which one to use during installation via extras.

  • aiohttp: The default asynchronous HTTP client for the library. Supports both queries and subscriptions.
  • httpx: A modern, feature-rich HTTP client. Supports queries and subscriptions (via websockets).

Auto-detection

When you initialize a GraphQLClient without specifying a transport, it will try to automatically detect an available backend:

HTTP Transport (Queries/Mutations)

  1. It first checks if aiohttp is installed.
  2. If not, it checks if httpx is installed.
  3. If neither is found, it raises a RuntimeError.

Subscription Transport

  1. It first checks if aiohttp is installed.
  2. If not, it checks if websockets is installed.
  3. If neither is found, it raises a RuntimeError.

Note that websockets is installed automatically when the httpx extra is used.

from aiographql.client import GraphQLClient

# Auto-detects available transport (aiohttp preferred if both installed)
client = GraphQLClient(endpoint="https://api.github.com/graphql")

Manual Selection

You can also explicitly specify which transport to use by providing a transport instance:

from aiographql.client import GraphQLClient
from aiographql.client.transport import HttpxTransport, AiohttpTransport, WebsocketSubscriptionTransport

# Explicitly use httpx
transport = HttpxTransport(endpoint="https://api.github.com/graphql")
client = GraphQLClient(
    endpoint="https://api.github.com/graphql",
    transport=transport
)

# Explicitly use aiohttp
transport = AiohttpTransport(endpoint="https://api.github.com/graphql")
client = GraphQLClient(
    endpoint="https://api.github.com/graphql",
    transport=transport
)

# Explicitly use websockets for subscriptions
from aiographql.client.transport.websocket import WebsocketSubscriptionTransport

subscription_transport = WebsocketSubscriptionTransport(endpoint="wss://your-api.com/graphql")
client = GraphQLClient(
    endpoint="https://your-api.com/graphql",
    subscription_transport=subscription_transport
)

Development

To run the full test suite with all supported transport combinations and Python versions, you can use tox:

# Install tox
pip install tox

# Run all environments
tox

# Run a specific environment (e.g., Python 3.10 with aiohttp and pydantic)
tox -e py310-aiohttp-pydantic

# Run with only aiohttp
tox -e py310-aiohttp

# Run with only httpx
tox -e py310-httpx

See pyproject.toml for all available environment factors and combinations.

Example Usage

Here are some example usages of this client implementation. For more examples, and advanced scenarios, see Usage Examples section in the documentation or the examples directory in this repository.

Examples

We provide several well-documented, scenario-based examples using a Strawberry GraphQL server. You can find them in the examples/ directory.

To run the examples, first start the Strawberry server using podman compose:

podman compose run --build strawberry-server

Then you can run any of the scenario scripts:

# Basic Queries
poetry run python examples/basic_queries.py

# Pydantic Data Models
poetry run python examples/data_models.py

# Dataclass Models
poetry run python examples/dataclass_models.py

# Authenticated Requests
poetry run python examples/authenticated_requests.py

# Mutations with Variables
poetry run python examples/mutations.py

# Real-time Subscriptions
poetry run python examples/subscriptions.py

# Httpx Transport & Shared Sessions
poetry run python examples/httpx_transport.py

# Custom Serialization & Codecs
poetry run python examples/custom_serialization.py

See examples/README.md for more detailed descriptions of each scenario.

Simple Query

async def get_logged_in_username(token: str) -> GraphQLResponse:
    client = GraphQLClient(
        endpoint="https://api.github.com/graphql",
        headers={"Authorization": f"Bearer {token}"},
    )
    request = GraphQLRequest(
        query="""
            query {
              viewer {
                login
              }
            }
        """
    )
    return await client.query(request=request)
>>> import asyncio
>>> response = asyncio.run(get_logged_in_username("<TOKEN FROM GITHUB GRAPHQL API>"))
>>> response.data
{'viewer': {'login': 'username'}}

Query Subscription

async def print_city_updates(client: GraphQLClient, city: str) -> None:
    request = GraphQLRequest(
        query="""
            subscription ($city:String!) {
              city(where: {name: {_eq: $city}}) {
                description
                id
              }
            }
        """,
        variables={"city": city},
    )
    # subscribe to data and error events, and print them
    await client.subscribe(
        request=request, on_data=print, on_error=print, wait=True
    )

For custom event specific callback registration, see Callback Registry Documentation.

Query Validation Failures

If your query is invalid, thanks to graphql-core-next, we get a detailed exception in the traceback.

aiographql.client.exceptions.GraphQLClientValidationException: Query validation failed

Cannot query field 'ids' on type 'chatbot'. Did you mean 'id'?

GraphQL request (4:13)
3:           chatbot {
4:             ids, bot_names
               ^
5:           }

Cannot query field 'bot_names' on type 'chatbot'. Did you mean 'bot_name' or 'bot_language'?

GraphQL request (4:18)
3:           chatbot {
4:             ids, bot_names
                    ^
5:           }

Query Variables & Operations

Support for multi-operation requests and variables is available via the client. For example, the following request contains multiple operations. The instance specifies default values to use.

request = GraphQLRequest(
    query="""
    query get_bot_created($id: Int) {
      chatbot(where: {id: {_eq: $id}}) {
        id, created
      }
    }
    query get_bot_name($id: Int) {
      chatbot(where: {id: {_eq: $id}}) {
        id, bot_name
      }
    }
    """,
    variables={"id": 109},
    operation="get_bot_name"
)

The default values can be overridden at the time of making the request if required.

await client.query(request=request, variables={"id": 20}, operation="get_bot_created")

Typed Usage (Bring Your Own Models)

The client supports explicit decoding of results into your own models, like dataclasses or Pydantic (v2) models.

from pydantic import BaseModel
from aiographql.client import GraphQLClient

class User(BaseModel):
    id: int
    name: str

client = GraphQLClient(endpoint="http://localhost/graphql")

# Explicitly decode into a Pydantic model
user = await client.query_data_as(
    "{ user(id: 1) { id name } }",
    User,
    path="user"
)

print(user.name) # Alice

Pass models directly into variables:

# Model as variable
user_input = CreateUserInput(name="Alice")
response = await client.query(query, variables={"input": user_input})

See Data Models Documentation for more information.