Graphene

3.4.3 · active · verified Sun Mar 29

Graphene is an opinionated Python library for building GraphQL APIs easily. It provides a simple yet extendable API, with built-in support for Relay and integrations for popular web frameworks like Django and SQLAlchemy. The current stable version is 3.4.3, with active development and frequent releases.

Warnings

breaking In Graphene v3.3.0 and later, if an optional field in an `InputObjectType` is omitted in a GraphQL query, it is now passed as `None` to the Python input. This makes it indistinguishable from a field explicitly passed with `null`, requiring changes to how `None` values are interpreted in your resolvers.
Fix: Review `InputObjectType` usage and resolvers to correctly differentiate between omitted fields and explicitly `null` fields. Consider using `default_value` in `InputField` if `None` implies a default rather than `null`.
breaking Graphene v3.0.0 involved a significant upgrade to `graphql-core` v3, dropping support for Python 2. Major changes include modifications to schema types, removal of the 'backends' concept, and renaming the `type` argument to `type_` in various Graphene constructs (e.g., `Field`, `Argument`) to avoid clashes with the Python built-in `type` function.
Fix: Ensure your project is running Python 3.x. Update argument names from `type` to `type_` where applicable. Review schema definitions and custom backend integrations as they may require refactoring due to underlying `graphql-core` changes.
breaking The import path for the `to_global_id` utility function, frequently used with Relay, changed in Graphene v3.0.0. Code importing it from `graphene.relay.node.to_global_id` will now raise an `ImportError`.
Fix: Update imports from `from graphene.relay.node import to_global_id` to `from graphene.relay import to_global_id`.
gotcha Graphene v3.4.0 removed the `aniso8601` dependency, which caused a regression in `DateTime` scalar parsing for Python versions prior to 3.11. This issue was resolved in v3.4.1 by introducing `python-dateutil` for `DateTime` parsing on Python < 3.11.
Fix: If you are using Python versions less than 3.11 and the `DateTime` scalar, ensure you upgrade Graphene to v3.4.1 or later to avoid parsing errors.
gotcha When using `graphene-sqlalchemy`, automatically generated connection classes (e.g., `UserConnection` for a `User` model) can conflict with custom `Connection` classes you define if they share the same name. This can lead to unexpected errors during schema generation or execution.
Fix: Ensure that any custom `Connection` classes you define for your SQLAlchemy models have unique names that do not clash with those `graphene-sqlalchemy` might generate automatically.

Install

pip install graphene Install latest version
pip install 'graphene>=3.1' Install Graphene v3.1 or newer (for v3 features)

Imports

ObjectType
```
from graphene import ObjectType
```
String
```
from graphene import String
```
Schema
```
from graphene import Schema
```
Field
```
from graphene import Field
```
to_global_id
```
from graphene.relay import to_global_id
```
The import path for `to_global_id` changed in Graphene v3.0.0.

Quickstart

This quickstart defines a simple GraphQL schema with a single 'hello' field that returns 'World'. It demonstrates basic schema definition and execution.

import graphene

class Query(graphene.ObjectType):
    hello = graphene.String(description='A typical hello world')

    def resolve_hello(self, info):
        return 'World'

schema = graphene.Schema(query=Query)

# Example usage:
query = '''
    query SayHello {
        hello
    }
'''
result = schema.execute(query)
print(result.data['hello'])
# Expected output: World

view raw JSON →