fastapi-cachekit

A high-performance, flexible caching solution for FastAPI applications. fastapi-cachekit supports both synchronous and asynchronous operations with a clean API and multiple backend options.

Features

• ✅ Full async/sync support for all operations
• ✅ Multiple backend Support So you can use the same tech stack as your app
• ✅ Function result caching with decorator syntax
• ✅ FastAPI dependency injection support
• ✅ Namespace support for isolating cache entries
• ✅ Customizable key generation
• ✅ Type hinting throughout the codebase
• ✅ Expiration time support (seconds or timedelta)

📦 Backends & Sync/Async Support

🛠️ Installation

Base (in-memory only):

pip install fastapi-cachekit

With Redis:

pip install fastapi-cachekit[redis]

With Postgres:

pip install fastapi-cachekit[postgres]

With Memcached:

pip install fastapi-cachekit[memcached]

With MongoDB:

pip install fastapi-cachekit[mongodb]

With FireStore:

pip install fastapi-cachekit[firestore]

With DynamoDB:

pip install fastapi-cachekit[dynamodb]

All backends:

pip install fastapi-cachekit[all]

Quick Start

from fastapi import FastAPI, Depends
from fast_cache import cache, RedisBackend
from typing import Annotated

app = FastAPI()

# Initialize cache with Redis backend
cache.init_app(
    app=app,
    backend=RedisBackend(redis_url="redis://localhost:6379/0", namespace="myapp"),
    default_expire=300  # 5 minutes default expiration
)


# Use function caching decorator
@app.get("/items/{item_id}")
@cache.cached(expire=60)  # Cache for 60 seconds
async def read_item(item_id: int):
    # Expensive operation simulation
    return {"item_id": item_id, "name": f"Item {item_id}"}

# Use cache backend directly with dependency injection
@app.get("/manual-cache")
async def manual_cache_example(cache_backend: Annotated[RedisBackend, Depends(cache.get_cache)]):
    # Check if key exists
    has_key = await cache_backend.ahas("my-key")

    if not has_key:
        # Set a value in the cache
        await cache_backend.aset("my-key", {"data": "cached value"}, expire=30)
        return {"cache_set": True}

    # Get the value from cache
    value = await cache_backend.aget("my-key")
    return {"cached_value": value}

Now You Can use the cache from other Sub Routes by importing from from fast_cache import cache

Detailed Usage

Initializing the Cache

Before using the cache, you need to initialize it with a backend:

from fastapi import FastAPI
from fast_cache import cache, RedisBackend
from datetime import timedelta

app = FastAPI()

cache.init_app(
    app=app,
    backend=RedisBackend(
        redis_url="redis://localhost:6379/0",
        namespace="myapp",
        max_connections=20
    ),
    default_expire=timedelta(minutes=5)
)

Method1: Caching a Function Result Using A Cache Decorator

The @cache.cached() decorator is the simplest way to cache function results:

from fast_cache import cache

# Cache with default expiration time
@cache.cached()
def get_user_data(user_id: int):
    # Expensive database query
    return {"user_id": user_id, "name": "John Doe"}

# Cache with custom namespace and expiration
@cache.cached(namespace="users", expire=300)
async def get_user_profile(user_id: int):
    # Async expensive operation
    return {"user_id": user_id, "profile": "..."}

# Cache with custom key builder
@cache.cached(key_builder=lambda user_id, **kwargs: f"user:{user_id}")
def get_user_permissions(user_id: int):
    # Complex permission calculation
    return ["read", "write"]

# Skip Cache for Specific Calls

Sometimes you need to bypass the cache for certain requests:

@cache.cached()
async def get_weather(city: str, skip_cache: bool = False):
    # Function will be called directly if skip_cache is True
    return await fetch_weather_data(city)

# Usage:
weather = await get_weather("New York", skip_cache=True)  # Bypasses cache

Method 2: Using Via Dependency Injection

You can access the cache backend directly for more control:

from fastapi import Depends
from fast_cache import cache, CacheBackend
from typing import Annotated

@app.get("/api/data")
async def get_data(cache_backend: Annotated[CacheBackend, Depends(cache.get_cache)]):
    # Try to get from cache
    cached_data = await cache_backend.aget("api:data")
    if cached_data:
        return cached_data

    # Generate new data
    data = await fetch_expensive_api_data()

    # Store in cache for 1 hour
    await cache_backend.aset("api:data", data, expire=3600)

    return data

Advanced: Implementing Custom Backends

You can create your own cache backend by implementing the CacheBackend abstract class:

from fast_cache.backends.backend import CacheBackend
from typing import Any, Optional, Union
from datetime import timedelta

class MyCustomBackend(CacheBackend):
    # Implement all required methods
    async def aget(self, key: str) -> Any:
        # Your implementation here
        ...

    def get(self, key: str) -> Any:
        # Your implementation here
        ...

    # ... implement all other required methods

API Reference

Cache Instance

• cache.init_app(app, backend, default_expire=None) - Initialize cache with FastAPI app
• cache.get_cache() - Get cache backend instance (for dependency injection)
• cache.cached(expire=None, key_builder=None, namespace=None, stampede_protection=True, lock_timeout=30, lock_wait=5.0) - Caching decorator
• cache.lifespan_handler - FastAPI lifespan that closes the backend on shutdown (FastAPI(lifespan=cache.lifespan_handler))

CacheBackend Interface

All backends implement these methods in both sync and async versions:

• get(key) / aget(key) - Retrieve a value
• set(key, value, expire) / aset(key, value, expire) - Store a value
• delete(key) / adelete(key) - Delete a value
• clear() / aclear() - Clear all values
• has(key) / ahas(key) - Check if key exists

RedisBackend Configuration

• redis_url - Redis connection string (required)
• namespace - Key prefix (default: "fastapi-cache")
• pool_size - Minimum pool connections (default: 10)
• max_connections - Maximum pool connections (default: 20)

Contributing

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

License

This project is licensed under the MIT License - see the LICENSE file for details.