Better support for sub-classed storage engines

[anentropic]

We wanted to make a Storage sub-class that uses Django cache backend rather than in-memory dict.

So we wanted StorageBase to actually implement the token bucket algorithm so as not to just copy and paste all the code from MemoryStorage into our own sub-class.

The changes in this PR allowed us to make a Django cache storage class that looks like:

from django.core.cache import cache
from token_bucket import StorageBase


class DjangoCacheBucketProvider(object):

    def __init__(self, key_template='token_bucket_{key}'):
        self.key_template = key_template

    def _key(self, key):
        return self.key_template.format(key=key)

    def __getitem__(self, key):
        val = cache.get(self._key(key))
        if val is None:
            raise KeyError(key)
        return val

    def __setitem__(self, key, val):
        cache.set(
            self._key(key),
            val
        )


class DjangoCacheStorage(StorageBase):

    bucket_provider_cls = DjangoCacheBucketProvider

[codecov[bot]]

Codecov Report

:exclamation: No coverage uploaded for pull request base (master@61f4e63). Click here to learn what that means. The diff coverage is 100%.

@@            Coverage Diff            @@
##             master      #10   +/-   ##
=========================================
  Coverage          ?   98.57%           
=========================================
  Files             ?        8           
  Lines             ?      211           
  Branches          ?        0           
=========================================
  Hits              ?      208           
  Misses            ?        3           
  Partials          ?        0

Impacted Files	Coverage Δ
token_bucket/storage_base.py	100% <100%> (ø)
token_bucket/storage.py	100% <100%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 61f4e63...158324f. Read the comment docs.

[vytas7]

Hi @anentropic ! And sorry that we're not checking this repository too often.

You bring a very valid point here; indeed, it would make sense to decouple storage from the algorithm itself.

I'm planning to release 0.3.0 featuring other cleanup and minor fixes/improvements as is, and then circle back on decoupling storage. Ideally, it would be good to get an async story for token_bucket too while at it :smiling_imp: