pgql 0.4.0

A lightweight Python GraphQL server framework with automatic resolver mapping and schema introspection

pgql

A lightweight Python GraphQL server framework with automatic resolver mapping, schema introspection, and built-in support for Starlette/Uvicorn.

Features

• 🚀 Automatic Resolver Mapping: Map Python class methods to GraphQL fields based on return types
• 📁 Recursive Schema Loading: Organize your .gql schema files in nested directories
• 🔍 Built-in Introspection: Full GraphQL introspection support out of the box
• 🎯 Instance-based Resolvers: Use class instances for stateful resolvers with dependency injection
• ⚡ Async Support: Built on Starlette and Uvicorn for high-performance async handling
• 🔧 YAML Configuration: Simple YAML-based server configuration
• 📦 Type Support: Full support for extend type, nested types, and GraphQL type modifiers
• 🔐 Authorization System: Intercept resolver calls with on_authorize function
• 🍪 Session Management: Built-in session store with automatic cookie handling

Installation

pip install pgql

Quick Start

1. Define Your GraphQL Schema

Create your schema files in a directory structure:

schema/
├── schema.gql
└── user/
    ├── types.gql
    └── queries.gql

schema/schema.gql:

schema {
    query: Query
}

schema/user/types.gql:

type User {
    id: ID!
    name: String!
    email: String!
}

schema/user/queries.gql:

extend type Query {
    getUser(id: ID!): User!
    getUsers: [User!]!
}

2. Create Resolver Classes

# resolvers/user.py
class User:
    def getUser(self, parent, info, id):
        # Your logic here
        return {'id': id, 'name': 'John Doe', 'email': '[e-mail]'}
    
    def getUsers(self, parent, info):
        return [
            {'id': 1, 'name': 'John', 'email': '[e-mail]'},
            {'id': 2, 'name': 'Jane', 'email': '[e-mail]'}
        ]

3. Configure Server

config.yml:

http_port: 8080
debug: true
server:
  host: localhost
  routes:
    - mode: gql
      endpoint: /graphql
      schema: schema  # Path to schema directory

4. Start Server

from pgql import HTTPServer
from resolvers.user import User

# Create resolver instances
user_resolver = User()

# Initialize server
server = HTTPServer('config.yml')

# Map resolvers to GraphQL types
server.gql({
    'User': user_resolver
})

# Start server
server.start()

5. Query Your API

curl -X POST http://localhost:8080/graphql \
  -H "Content-Type: application/json" \
  -d '{"query": "{ getUsers { id name email } }"}'

Response:

{
  "data": {
    "getUsers": [
      {"id": "1", "name": "John", "email": "[e-mail]"},
      {"id": "2", "name": "Jane", "email": "[e-mail]"}
    ]
  }
}

How It Works

Automatic Resolver Mapping

pgql automatically maps resolver methods to GraphQL fields based on return types:

1. If Query.getUser returns type User, pgql looks for a method named getUser in the User resolver class
2. The mapping works recursively for nested types (e.g., User.company → Company.company)

Example:

type User {
    id: ID!
    company: Company!
}

type Company {
    id: ID!
    name: String!
}

type Query {
    getUser: User!
}

class User:
    def getUser(self, parent, info):
        return {'id': 1, 'company': {'id': 1}}

class Company:
    def company(self, parent, info):
        # parent contains the User object
        company_id = parent['id']
        return {'id': company_id, 'name': 'Acme Corp'}

# Register both resolvers
server.gql({
    'User': User(),
    'Company': Company()
})

Resolver Arguments

All resolver methods receive:

• self: The resolver instance (for stateful resolvers)
• parent: The parent object from the previous resolver
• info: GraphQL execution info (field name, context, variables, etc.)
• **kwargs: Field arguments from the query

def getUser(self, parent, info, id):
    # id comes from query arguments
    return fetch_user(id)

Introspection

pgql supports full GraphQL introspection out of the box:

curl -X POST http://localhost:8080/graphql \
  -H "Content-Type: application/json" \
  -d '{"query": "{ __schema { queryType { name } } }"}'

This works with tools like:

• GraphiQL
• GraphQL Playground
• Apollo Studio
• Postman

Advanced Usage

Authorization Interceptor

pgql allows you to intercept every resolver call to implement authorization logic using on_authorize:

from pgql import HTTPServer, AuthorizeInfo

def on_authorize(auth_info: AuthorizeInfo) -> bool:
    """
    Intercept every resolver call for authorization
    
    Args:
        auth_info.operation: 'query', 'mutation', or 'subscription'
        auth_info.src_type: Parent GraphQL type invoking the resolver (e.g., 'User' for User.company)
        auth_info.dst_type: GraphQL type being executed (e.g., 'Company' for User.company)
        auth_info.resolver: Field/resolver name (e.g., 'getUser', 'company')
        auth_info.session_id: Session ID from cookie (None if not present)
    
    Returns:
        True to allow execution, False to deny
    """
    # Deny access if no session
    if not auth_info.session_id:
        return False
    
    # Restrict specific field access based on parent type
    if auth_info.src_type == "User" and auth_info.resolver == "company":
        return auth_info.session_id == "admin123"  # Only admin can access User.company
    
    return True

server = HTTPServer('config.yml')
server.on_authorize(on_authorize)  # Register authorization function
server.gql({...})

Session Management:

pgql extracts session_id from cookies automatically. Set the cookie in your client:

curl -X POST http://localhost:8080/graphql \
  -H "Content-Type: application/json" \
  -H "Cookie: session_id=abc123" \
  -d '{"query": "{ getUsers { id } }"}'

Authorization Flow Example:

When querying { getUser { id company { name } } }:

1. First call: Query.getUser → User (src_type='Query', dst_type='User', resolver='getUser')
2. Second call: User.company → Company (src_type='User', dst_type='Company', resolver='company')

Note: The on_authorize function is optional. If not set, all resolvers execute without authorization checks.

Session Management

pgql includes a built-in session store for managing user sessions:

from pgql import HTTPServer, Session

server = HTTPServer('config.yml')

# Create a new session
session = server.create_session(max_age=3600)  # 1 hour

# Store any data in the session
session.set('user_id', 123)
session.set('username', 'john')
session.set('roles', ['admin', 'user'])
session.set('preferences', {'theme': 'dark'})

# Retrieve session
session = server.get_session(session_id)
user_id = session.get('user_id')

# Delete session (logout)
server.delete_session(session_id)

Using Sessions in Resolvers:

class UserResolver:
    def __init__(self, server):
        self.server = server
    
    def login(self, parent, info, username, password):
        # Create session on successful login
        session = self.server.create_session(max_age=7200)
        session.set('user_id', 123)
        session.set('authenticated', True)
        
        # Mark session to set cookie in response
        info.context['new_session'] = session
        
        return {'success': True, 'session_id': session.session_id}
    
    def getUser(self, parent, info):
        # Access session data
        session = info.context.get('session')
        if session and session.get('authenticated'):
            return {'id': session.get('user_id'), 'name': 'John'}
        return None

Configure cookie name in YAML:

http_port: 8080
cookie_name: my_session_id  # Custom cookie name
server:
  host: localhost
  routes:
    - mode: gql
      endpoint: /graphql
      schema: schema

For complete session documentation, see SESSIONS.md.

Note: The on_authorize function is optional. If not set, all resolvers execute without authorization checks.

Nested Schema Organization

Organize your schemas by domain:

schema/
├── schema.gql
├── user/
│   ├── types.gql
│   ├── queries.gql
│   ├── mutations.gql
│   └── inputs.gql
└── company/
    ├── types.gql
    └── queries.gql

pgql recursively loads all .gql files.

Multiple Routes

Configure multiple GraphQL endpoints:

server:
  routes:
    - mode: gql
      endpoint: /graphql
      schema: schema
    - mode: gql
      endpoint: /admin/graphql
      schema: admin_schema

Requirements

• Python >= 3.8
• graphql-core >= 3.2.0
• starlette >= 0.27.0
• uvicorn >= 0.23.0
• pyyaml >= 6.0

License

MIT

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Links

• GitHub Repository
• Issue Tracker
• PyPI Package