refactor: Use functools.lru_cache instead of the stale memoization library (#1981)

Author: edgarrmondragon

docs/code_samples.md

@@ -240,16 +240,15 @@ class SingletonAuthStream(RESTStream):
 ### Make a stream reuse the same authenticator instance for all requests
 
 ```python
-from memoization import cached
+from functools import cached_property
 
 from singer_sdk.authenticators import APIAuthenticatorBase
 from singer_sdk.streams import RESTStream
 
 class CachedAuthStream(RESTStream):
     """A stream with singleton authenticator."""
 
-    @property
-    @cached
+    @cached_property
     def authenticator(self) -> APIAuthenticatorBase:
         """Stream authenticator."""
         return APIAuthenticatorBase(stream=self)

poetry.lock

@@ -47,7 +47,6 @@ inflection = ">=0.5.1"
 joblib = ">=1.0.1"
 jsonpath-ng = ">=1.5.3"
 jsonschema = ">=4.16.0"
-memoization = ">=0.3.2,<0.5.0"
 packaging = ">=23.1"
 pendulum = ">=2.1.0"
 PyJWT = "~=2.4"

pyproject.toml

@@ -5,19 +5,19 @@
 import typing as t
 from copy import deepcopy
 
-from memoization import cached
-
 from singer_sdk.helpers._typing import is_object_type
 
 if t.TYPE_CHECKING:
     from logging import Logger
 
     from singer_sdk._singerlib import Catalog, SelectionMask
 
-_MAX_LRU_CACHE = 500
-
 
-@cached(max_size=_MAX_LRU_CACHE)
+# TODO: this was previously cached using the `memoization` library. However, the
+# `functools.lru_cache` decorator does not support non-hashable arguments.
+# It is possible that this function is not a bottleneck, but if it is, we should
+# consider implementing a custom LRU cache decorator that supports non-hashable
+# arguments.
 def get_selected_schema(
     stream_name: str,
     schema: dict,

singer_sdk/helpers/_catalog.py

@@ -4,8 +4,8 @@
 
 import logging
 import typing as t
+from functools import lru_cache
 
-import memoization
 from jsonpath_ng.ext import parse
 
 if t.TYPE_CHECKING:
@@ -39,7 +39,7 @@ def extract_jsonpath(
         yield match.value
 
 
-@memoization.cached
+@lru_cache
 def _compile_jsonpath(expression: str) -> jsonpath_ng.JSONPath:
     """Parse a JSONPath expression and cache the result.
 

singer_sdk/helpers/jsonpath.py

@@ -3,9 +3,9 @@
 from __future__ import annotations
 
 import typing as t
+from functools import cached_property
 
 import pytest
-from memoization.memoization import cached
 from requests.auth import HTTPProxyAuth
 
 from singer_sdk.authenticators import APIAuthenticatorBase, SingletonMeta
@@ -49,8 +49,7 @@ class NaiveAuthenticator(APIAuthenticatorBase):
 class CachedAuthStream(SimpleRESTStream):
     """A stream with Naive authentication."""
 
-    @property
-    @cached
+    @cached_property
     def authenticator(self) -> NaiveAuthenticator:
         """Stream authenticator."""
         return NaiveAuthenticator(stream=self)

tests/core/rest/conftest.py

@@ -16,7 +16,6 @@
 
 from singer_sdk._singerlib import Catalog
 from singer_sdk.exceptions import MapExpressionError
-from singer_sdk.helpers._catalog import get_selected_schema
 from singer_sdk.mapper import PluginMapper, RemoveRecordTransform, md5
 from singer_sdk.streams.core import Stream
 from singer_sdk.tap_base import Tap
@@ -563,19 +562,11 @@ def discover_streams(self):
         return [MappedStream(self)]
 
 
-@pytest.fixture
-def _clear_schema_cache() -> None:
-    """Schemas are cached, so the cache needs to be cleared between test invocations."""
-    yield
-    get_selected_schema.cache_clear()
-
-
 @time_machine.travel(
     datetime.datetime(2022, 1, 1, tzinfo=datetime.timezone.utc),
     tick=False,
 )
 @pytest.mark.snapshot()
-@pytest.mark.usefixtures("_clear_schema_cache")
 @pytest.mark.parametrize(
     "stream_maps,flatten,flatten_max_depth,snapshot_name",
     [