Simple redis cache for Python functions

• Redis 5+
• Python 3.8+ (should work in Python 3.6+, but not tested)

pip install python-redis-cache

from redis import StrictRedis
from redis_cache import RedisCache

client = StrictRedis(host="redis", decode_responses=True)
cache = RedisCache(redis_client=client)


@cache.cache()
def my_func(arg1, arg2):
    return some_expensive_operation()

# Use the function
my_func(1, 2)

# Call it again with the same arguments and it will use cache
my_func(1, 2)

# Invalidate a single value
my_func.invalidate(1, 2)

# Invalidate all values for function
my_func.invalidate_all()

# Create the redis cache
cache = RedisCache(redis_client, prefix="rc", serializer=dumps, deserializer=loads, key_serializer=None, support_cluster=True, exception_handler=None)

# Cache decorator to go on functions, see above
cache.cache(ttl=..., limit=..., namespace=...) -> Callable[[Callable], Callable]

# Get multiple values from the cache
cache.mget([{"fn": my_func, "args": [1,2], "kwargs": {}}, ...]) -> List[Any]

Redis

# Cached function API

# Returns a cached value, if it exists in cache. Saves value in cache if it doesn't exist
cached_func(*args, *kwargs)

# Invalidates a single value
cached_func.invalidate(*args, **kwargs)

# Invalidates all values for cached function
cached_func.invalidate_all()

• prefix - The string to prefix the redis keys with
• serializer/deserializer - functions to convert arguments and return value to a string (user JSON by default)
• ttl - The time in seconds to cache the return value
• namespace - The string namespace of the cache. This is useful for allowing multiple functions to use the same cache. By default its f'{function.__module__}.{function.__file__}'
• exception_handler - Function to handle Redis cache exceptions. This allows you to fall back to calling the original function or logging exceptions. Function has the following signature exception_handler(exception: Exception, function: Callable, args: List, kwargs: Dict) -> Any. If using this handler, reraise the exception in the handler to stop execution of the function. All return results will be used even if None. If handler not defined, it will raise the exception and not call the original function.
• support_cluster - Set to False to disable the { prefix on the keys. This is NOT recommended. See below for more info.
• active - Optional flag to disable the caching completly for troubleshooting/lower environments

Arguments and return types must be JSON serializable by default. You can override the serializer, but be careful with using Pickle. Make sure you understand the security risks. Pickle should not be used with untrusted values. https://security.stackexchange.com/questions/183966/safely-load-a-pickle-file decode_responses parameter must be False in redis client if you use pickle.

• ttl - is based on the time from when it's first inserted in the cache, not based on the last access
• limit - The limit will revoke keys (once it hits the limit) based on FIFO, not based on LRU

The key names by default are as follows:

from base64 import b64encode

key = f"{{rc:{fn.__module__}.{fn.__qualname__}}}:{b64encode(function_args).decode('utf-8')}"

The cache key names start with {, which can be confusing, but is required for redis clusters to place the keys in the correct slots.

NOTE: It is NOT recommended to use any of the options below. The key name generation by default handles all use cases.

cache = RedisCache(redis_client, prefix="custom_prefix")

# Changes keys to the following
key = f"{{custom_prefix:{fn.__module__}.{fn.__qualname__}}}:{b64encode(function_args).decode('utf-8')}"

cache = RedisCache(redis_client)

@cache.cache(namespace="custom_func_name")
def my_func(arg1, arg2):
    pass

# Changes keys to the following
key = f"{{rc:custom_func_name}}:{b64encode(function_args).decode('utf-8')}"

def custom_key_serializer(fn_args):
    ## Do something with fn_args and return a string. For instance
    return my_custom_serializer(fn_args)

cache = RedisCache(redis_client, key_serializer=custom_key_serializer)

# Changes keys to the following
key = f"{{rc:{fn.__module__}.{fn.__qualname__}}}:{b64encode(custom_serialized_args).decode('utf-8')}"

This option is NOT recommended because this library will no longer work with redis clusters. Often times people/companies will start not using cluster mode and then will migrate to using cluster. This option will make that migration require a lot of work. If you know for sure you will never use a redis cluster, then you can enable this option. If you are unsure, don't use this option. There is not any benefit.

cache = RedisCache(redis_client, support_cluster=False)

# Changes keys to the following
key = f"rc:{fn.__module__}.{fn.__qualname__}:{b64encode(custom_serialized_args).decode('utf-8')}"