Caching¶

DRF Toolkit provides a flexible caching system with robust cache key generation and support for cache directives.

Enabling Cache¶

Enable caching in your viewsets by using the pre-built cached viewsets from drf_kit:

from drf_kit.viewsets import CachedModelViewSet

class UserViewSet(CachedModelViewSet):
    queryset = User.objects.all()
    serializer_class = UserSerializer
    cache_timeout = 300  # Cache for 5 minutes

Cache Key Generation¶

The toolkit uses a robust CacheKeyConstructor that generates unique cache keys based on: - Unique view method identifier - URL arguments - URL keyword arguments - Query parameters (including handling of multiple values) - Request's accepted media type

Configuration¶

You can configure the cache key generation function globally in your settings:

REST_FRAMEWORK_EXTENSIONS = {
    'DEFAULT_CACHE_KEY_FUNC':
        'drf_kit.cache.cache_key_constructor'
}

Cache Directives¶

The caching system respects HTTP cache control directives. Currently supported:

no-cache: Forces a fresh response, bypassing the cache
When cache is used, responses include:
Expires header with the expiration timestamp
Cache-Control: max-age=<timeout> header
X-Cache: HIT/MISS header indicating cache status

Advanced Usage¶

Custom Cache Timeout¶

You can customize the cache timeout per viewset:

class UserViewSet(CachedModelViewSet):
    cache_timeout = 3600  # Cache for 1 hour

Custom Cache Key Constructor¶

While the default CacheKeyConstructor is suitable for most cases, you can create your own key constructor by extending it:

from drf_kit.cache import CacheKeyConstructor
from rest_framework_extensions.key_constructor import bits

class CustomCacheKeyConstructor(CacheKeyConstructor):
    # Add custom bits for key generation
    user_id = bits.UserKeyBit()

custom_cache_key_constructor = CustomCacheKeyConstructor()

class UserViewSet(CachedModelViewSet):
    cache_key_constructor = custom_cache_key_constructor

Best Practices¶

Choose appropriate cache timeouts:
Short for frequently changing data
Longer for stable data
Consider user-specific needs