Method keys_match() returns different keys types for Redis and memory

Method keys_match() returns keys in bytes type for Redis cache and in str type for memory cache. Is it a bug or feature?

Code example:

# setup two caches - for redis & memory
cache.setup(f'redis://{host}:{port}', password=pwd, prefix='test_redis')
cache.setup('mem://', prefix='test_mem')

# set values for memory and redis
await cache.set('test_redis:abc', 123)
await cache.set('test_mem:abc', 321)

# keys_match result
async def foo(mask):
  async for key in cache.keys_match(mask):
    print(key, type(key))

await foo('test_redis:*')
# b'test_redis:abc' <class 'bytes'>
await foo('test_mem:*')
# test_mem:abc <class 'str'>

[AIGeneratedUsername]

I’m affraid it will be a breaking change for some users in case of changing result type of keys_match by default.

That is why everybody must pin their requirements. And have a full tests coverage, of course. We can add a DeprecationWarning for the method, and update it on the next major release. Win-win, all will be happy. As a temporary workaround, another method may be added.

@AIGeneratedUsername Would you like to remake you MR with a new method that fix this issue?

@Krukov Do you mean creating a method with a new name? Will this method do the same as keys_match for the “Memory” and “Disk” backends, but return strings for the “Redis” backend? Simplified pseudo-code:

class MemoryBackend:
    keys = keys_match  # Or `scan = keys_match`

class DiskBackend:
    keys = keys_match  # Or `scan = keys_match`

class RedisBackend:
    async def keys():  # Or name the method as `scan` to be consistent with Redis terminology
           # yield keys as strings...
           # technically, we can call `keys_match` and yield from it `key.decode()`

[AIGeneratedUsername]

The “Memory” cache is a dictionary with strings as keys. The input key is a string, and it is stored as a string and returned as a string. It is never converted or stored as bytes. What you are describing is expected behavior.

Maybe adding type hinting for the whole package code will simplify the package usage. It may be more clear right away what type to expect before running the code.

Any better ideas or proposals are welcome.