redis-tagged-cache 0.10.1

No project description provided

redis-tagged-cache

What is it?

redis-tagged-cache is a Python 3.7+ cache library backed with Redis with O(1) tags-based invalidation system.

Low level example

Installation: pip install redis-tagged-cache

Usage:

from rtc import CacheMiss, RedisTaggedCache

cache = RedisTaggedCache(
    namespace="foo",
    host="localhost",
    port=6379,
)

invalidation_tags = ["tag1", "tag2"]  # tags are only strings of your choice

# Let's store something in the cache under the key "key1"
# (with a 60s lifetime)
cache.set("key1", "my value", tags=invalidation_tags, lifetime=60)

# it will output "my value" (cache hit!)
print(cache.get("key1", tags=invalidation_tags))

# Let's invalidate a tag (O(1) operation)
cache.invalidate("tag2")

# As the "key1" entry is tagged with "tag1" and "tag2"...
# ...the entry is invalidated (because we just invalidated "tag2")

# It will print "cache miss"
try:
    cache.get("key1", tags=invalidation_tags)
except CacheMiss:
    print("cache miss")

High level example

from rtc import RedisTaggedCache

cache = RedisTaggedCache(
    namespace="foo",
    host="localhost",
    port=6379,
)


class A:
    @cache.method_decorator(lifetime=60, tags=["tag1", "tag2"])
    def slow_method(self, arg1: str, arg2: str = "foo"):
        print("called")
        return arg1 + arg2


if __name__ == "__main__":
    a = A()

    # It will output "called" and "foobar" (cache miss)
    print(a.slow_method("foo", arg2="bar"))

    # It will output "foobar" (cache hit)
    print(a.slow_method("foo", arg2="bar"))

    # It will output "called" and "foo2bar" (cache miss)
    print(a.slow_method("foo2", arg2="bar"))

# Note: for plain functions, you can use @cache.function_decorator
#       that works the same way

Full API

You will find the full API in this reference documentation.

Pros & Cons

Pros

All methods have a O(1) complexity regarding the number of keys. The invalidation is synchronous and very fast event if you invalidate millions of keys.

Note: complexity is O(n) regarding the number of tags.

Cons

The invalidation system does not really remove keys from redis. Invalidated entries are inaccessible (from the API) but they are not removed when the invalidation occurred. They are going to expire by themselves.

Be sure to configure your redis instance as a cache with capped memory (see maxmemory configuration parameter) and maxmemory-policy allkeys-lru settings about keys automatic eviction

Dev

As we support Python 3.7+ for the runtime, the dev environment requires Python 3.9+.

• make lint: for linting the code
• make test: for executing test

(the poetry env will be automatically created)