persistent-function-cache 0.1.0__tar.gz → 0.2.0__tar.gz

{persistent_function_cache-0.1.0/src/persistent_function_cache.egg-info → persistent_function_cache-0.2.0}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.2
 Name: persistent-function-cache
-Version: 0.1.0
+Version: 0.2.0
 Summary: Persistent cache for expensive functions
 Author-email: Quinten Roets <qdr2104@columbia.edu>
 License: MIT
@@ -22,7 +22,7 @@ Requires-Dist: torch<3,>=1.26.0; extra == "dev"
 ![PyPI downloads](https://img.shields.io/pypi/dm/persistent-function-cache)
 ![Python version](https://img.shields.io/badge/python-3.10--3.12-brightgreen)
 ![Operating system](https://img.shields.io/badge/os-linux%20%7c%20macOS%20%7c%20windows-brightgreen)
-![Coverage](https://img.shields.io/badge/coverage-88%25-brightgreen)
+![Coverage](https://img.shields.io/badge/coverage-100%25-brightgreen)
 
 ## Usage
 Use

{persistent_function_cache-0.1.0 → persistent_function_cache-0.2.0}/README.md

@@ -3,7 +3,7 @@
 ![PyPI downloads](https://img.shields.io/pypi/dm/persistent-function-cache)
 ![Python version](https://img.shields.io/badge/python-3.10--3.12-brightgreen)
 ![Operating system](https://img.shields.io/badge/os-linux%20%7c%20macOS%20%7c%20windows-brightgreen)
-![Coverage](https://img.shields.io/badge/coverage-88%25-brightgreen)
+![Coverage](https://img.shields.io/badge/coverage-100%25-brightgreen)
 
 ## Usage
 Use

{persistent_function_cache-0.1.0 → persistent_function_cache-0.2.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "persistent-function-cache"
-version = "0.1.0"
+version = "0.2.0"
 description = "Persistent cache for expensive functions"
 authors = [{name = "Quinten Roets", email = "qdr2104@columbia.edu"}]
 license = {text = "MIT"}
@@ -35,7 +35,7 @@ command_line = "-m pytest tests"
 
 [tool.coverage.report]
 precision = 4
-fail_under = 80
+fail_under = 100
 
 [tool.mypy]
 strict = true
@@ -52,8 +52,6 @@ fix = true
 [tool.ruff.lint]
 select = ["ALL"]
 ignore = [
-    "ANN101",   # annotate self
-    "ANN102",   # annotate cls
     "ANN401",   # annotated with Any
     "D",        # docstrings
     "G004",     # logging f-string

persistent_function_cache-0.2.0/src/persistent_cache/__init__.py

@@ -0,0 +1,2 @@
+from .caches import deep_learning, speedup_deep_learning
+from .main.decorator import cache

persistent_function_cache-0.2.0/src/persistent_cache/caches/deep_learning.py

@@ -0,0 +1,5 @@
+import functools
+
+from persistent_cache.main import decorator
+
+cache = functools.partial(decorator.cache, deep_learning=True)

persistent_function_cache-0.2.0/src/persistent_cache/caches/speedup_deep_learning.py

@@ -0,0 +1,5 @@
+import functools
+
+from persistent_cache.main import decorator
+
+cache = functools.partial(decorator.cache, speedup_deep_learning=True)

{persistent_function_cache-0.1.0 → persistent_function_cache-0.2.0}/src/persistent_cache/main/cacheslot.py

@@ -15,12 +15,11 @@ class CacheSlot(Generic[T]):
         self,
         function: Callable[..., T],
         args: tuple[Any, ...],
-        kwargs: Any,
         key_reducer: type[Reducer] = Reducer,
         cache_path: Path = Path.cache,
     ) -> None:
         # change cache key when implementation changes
-        cache_keys = (function, args, kwargs)
+        cache_keys = (function, args)
         self.location = (
             cache_path
             / function.__module__.replace(".", "_")

persistent_function_cache-0.2.0/src/persistent_cache/main/decorator.py

@@ -0,0 +1,122 @@
+import inspect
+from collections.abc import Callable, Iterator
+from functools import wraps
+from typing import Any, cast, overload
+
+from persistent_cache.models import F, Path
+from persistent_cache.reducers.base import Reducer
+
+from .cacheslot import CacheSlot
+
+
+@overload
+def cache(
+    function: F,
+    *,
+    cache_path: Path = Path.cache,
+    cache_key_arguments: tuple[str, ...] | str | None = None,
+    key_reducer: type[Reducer] = Reducer,
+    deep_learning: bool = False,
+    speedup_deep_learning: bool = False,
+) -> F: ...
+
+
+@overload
+def cache(
+    function: None = None,
+    *,
+    cache_path: Path = Path.cache,
+    cache_key_arguments: tuple[str, ...] | str | None = None,
+    key_reducer: type[Reducer] = Reducer,
+    deep_learning: bool = False,
+    speedup_deep_learning: bool = False,
+) -> Callable[[F], F]: ...
+
+
+def cache(  # noqa: PLR0913
+    function: F | None = None,
+    *,
+    cache_path: Path = Path.cache,
+    cache_key_arguments: tuple[str, ...] | str | None = None,
+    key_reducer: type[Reducer] | None = None,
+    deep_learning: bool = False,
+    speedup_deep_learning: bool = False,
+) -> F | Callable[[F], F]:
+    """A decorator to cache function results. Decorated functions are only executed if
+    result is not present in cache. The arguments of the function can be any nested
+    complex object.
+
+    Use as:
+
+    from persistent_cache import cache
+
+    @cache
+    def long_function(complex_object):
+        ...
+    """
+
+    reducer = extract_reducer(
+        key_reducer,
+        deep_learning=deep_learning,
+        speedup_deep_learning=speedup_deep_learning,
+    )
+
+    def cache_decorator(function: F) -> F:
+        @wraps(function)
+        def wrapped_function(*args: Any, **kwargs: Any) -> Any:
+            arguments = tuple(
+                extract_argument_values(function, args, kwargs, cache_key_arguments),
+            )
+            cache_slot = CacheSlot(function, arguments, reducer, cache_path)
+            try:
+                value = cache_slot.value
+            except KeyError:
+                value = function(*args, **kwargs)
+                cache_slot.value = value
+            return value
+
+        return cast(F, wrapped_function)
+
+    if function is not None:
+        cache_decorator = cache_decorator(function)
+    return cache_decorator
+
+
+def extract_argument_values(
+    function: F,
+    args: tuple[Any, ...],
+    kwargs: dict[str, Any],
+    cache_key_arguments: tuple[str, ...] | str | None = None,
+) -> Iterator[Any]:
+    if cache_key_arguments is None:
+        yield from (args, kwargs)
+    else:
+        arguments = inspect.signature(function).bind(*args, **kwargs)
+        arguments.apply_defaults()
+        if isinstance(cache_key_arguments, str):
+            yield arguments.arguments.get(cache_key_arguments)
+        else:
+            for name in cache_key_arguments:
+                yield arguments.arguments.get(name)
+
+
+def extract_reducer(
+    reducer: type[Reducer] | None,
+    *,
+    deep_learning: bool,
+    speedup_deep_learning: bool,
+) -> type[Reducer]:
+    if reducer is None:
+        if deep_learning:
+            from persistent_cache.reducers.deep_learning import Reducer as Reducer_
+
+            reducer = Reducer_
+        elif speedup_deep_learning:
+            from persistent_cache.reducers.speedup_deep_learning import (
+                Reducer as Reducer_,
+            )
+
+            reducer = Reducer_
+        else:
+            reducer = Reducer
+    return reducer

{persistent_function_cache-0.1.0 → persistent_function_cache-0.2.0}/src/persistent_cache/main/hashing.py

@@ -4,13 +4,28 @@ import hashlib
 import inspect
 import io
 import pickle
-from typing import TYPE_CHECKING, Any, get_args, get_type_hints
+from types import UnionType
+from typing import TYPE_CHECKING, Any, get_args, get_origin, get_type_hints
 
 from persistent_cache.reducers.base import Reducer
 
 if TYPE_CHECKING:
-    from collections.abc import Callable, Iterator
-    from typing import BinaryIO
+    from collections.abc import Callable, Iterator  # pragma: nocover
+    from typing import BinaryIO  # pragma: nocover
+
+
+def extract_types(method: Callable[[Any], Any]) -> Iterator[type]:
+    type_hints = get_type_hints(method).values()
+    if type_hints:
+        argument_type = next(iter(type_hints))
+        origin = get_origin(argument_type)
+        arguments = get_args(argument_type)
+        if origin is UnionType:
+            yield from arguments
+        elif origin is not None:
+            yield origin
+        else:
+            yield argument_type
 
 
 class HashPickler(pickle.Pickler):
@@ -23,13 +38,9 @@ class HashPickler(pickle.Pickler):
         self.reducer = reducer
         self.reducers = {}
         for _, method in inspect.getmembers(reducer, predicate=inspect.ismethod):
-            type_hints = get_type_hints(method).values()
-            if type_hints:
-                argument_type = next(iter(type_hints))
-                argument_types = get_args(argument_type) or (argument_type,)
-                for argument_type in argument_types:
-                    self.reducers[argument_type] = method
-        self.reducers.pop(Any, None)
+            argument_types = extract_types(method)
+            for argument_type in argument_types:
+                self.reducers[argument_type] = method
 
     def reducer_override(self, obj: Any) -> Any:
         """The goal of this pickler is to create hashes of complex objects, not to

persistent_function_cache-0.2.0/src/persistent_cache/models/__init__.py

@@ -0,0 +1,2 @@
+from .function import F
+from .path import Path

persistent_function_cache-0.2.0/src/persistent_cache/models/function.py

@@ -0,0 +1,4 @@
+from collections.abc import Callable
+from typing import Any, TypeVar
+
+F = TypeVar("F", bound=Callable[..., Any])

persistent_function_cache-0.2.0/src/persistent_cache/reducers/__init__.py

@@ -0,0 +1 @@
+from .base import Reducer

{persistent_function_cache-0.1.0 → persistent_function_cache-0.2.0}/src/persistent_cache/reducers/speedup_deep_learning.py

@@ -51,7 +51,7 @@ class Reducer(deep_learning.Reducer):
 
         # only use part of dataset for speedup
         data = dataset[13**17 % length] if length > 0 else []
-        if isinstance(data, tuple):
+        if isinstance(data, tuple) and len(data) == 2:  # noqa: PLR2004
             data, label = data
         else:
             label = None

{persistent_function_cache-0.1.0 → persistent_function_cache-0.2.0/src/persistent_function_cache.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.2
 Name: persistent-function-cache
-Version: 0.1.0
+Version: 0.2.0
 Summary: Persistent cache for expensive functions
 Author-email: Quinten Roets <qdr2104@columbia.edu>
 License: MIT
@@ -22,7 +22,7 @@ Requires-Dist: torch<3,>=1.26.0; extra == "dev"
 ![PyPI downloads](https://img.shields.io/pypi/dm/persistent-function-cache)
 ![Python version](https://img.shields.io/badge/python-3.10--3.12-brightgreen)
 ![Operating system](https://img.shields.io/badge/os-linux%20%7c%20macOS%20%7c%20windows-brightgreen)
-![Coverage](https://img.shields.io/badge/coverage-88%25-brightgreen)
+![Coverage](https://img.shields.io/badge/coverage-100%25-brightgreen)
 
 ## Usage
 Use

{persistent_function_cache-0.1.0 → persistent_function_cache-0.2.0}/src/persistent_function_cache.egg-info/SOURCES.txt

@@ -4,7 +4,6 @@ pyproject.toml
 src/persistent_cache/__init__.py
 src/persistent_cache/py.typed
 src/persistent_cache/caches/__init__.py
-src/persistent_cache/caches/base.py
 src/persistent_cache/caches/deep_learning.py
 src/persistent_cache/caches/speedup_deep_learning.py
 src/persistent_cache/cli/__init__.py
@@ -14,6 +13,7 @@ src/persistent_cache/main/cacheslot.py
 src/persistent_cache/main/decorator.py
 src/persistent_cache/main/hashing.py
 src/persistent_cache/models/__init__.py
+src/persistent_cache/models/function.py
 src/persistent_cache/models/path.py
 src/persistent_cache/reducers/__init__.py
 src/persistent_cache/reducers/base.py
@@ -26,4 +26,5 @@ src/persistent_function_cache.egg-info/entry_points.txt
 src/persistent_function_cache.egg-info/requires.txt
 src/persistent_function_cache.egg-info/top_level.txt
 tests/test_cache.py
-tests/test_clear_cache.py
+tests/test_clear_cache.py
+tests/test_deep_learning_cache.py

persistent_function_cache-0.2.0/tests/test_cache.py

@@ -0,0 +1,55 @@
+import io
+import math
+from collections.abc import Callable
+from typing import Any
+
+import cli
+import pytest
+
+from persistent_cache import cache, deep_learning, speedup_deep_learning
+
+caches = [cache, deep_learning.cache, speedup_deep_learning.cache]
+
+
+def calculate_with_name(value: str) -> None:
+    cli.console.print(value)
+
+
+def calculate(*args: Any, **kwargs: Any) -> None:
+    cli.console.print("calculation started")
+    cli.console.print(args, kwargs)
+
+
+def verify_cached_function(function: Callable[..., Any]) -> None:
+    with io.BytesIO() as fp:
+        function(fp, lambda x: x, math, {})
+
+
+@pytest.mark.parametrize("cache_decorator", caches)
+def test_cache_with_argument_combination(cache_decorator: Callable[..., Any]) -> None:
+    cached_function = cache_decorator(calculate)
+    verify_cached_function(cached_function)
+
+
+@pytest.mark.parametrize("cache_decorator", caches)
+def test_cache_as_function(cache_decorator: Callable[..., Any]) -> None:
+    cached_function = cache_decorator()(calculate)
+    verify_cached_function(cached_function)
+
+
+@pytest.mark.parametrize("cache_decorator", caches)
+def test_cache_with_argument_name(
+    cache_decorator: Callable[..., Any],
+) -> None:
+    cached_function = cache_decorator(calculate_with_name, cache_key_arguments="value")
+    cached_function("test")
+
+
+@pytest.mark.parametrize("cache_decorator", caches)
+def test_cache_with_argument_names(cache_decorator: Callable[..., Any]) -> None:
+    cache_key_arguments = ("value",)
+    cached_function = cache_decorator(
+        calculate_with_name,
+        cache_key_arguments=cache_key_arguments,
+    )
+    cached_function("test")

persistent_function_cache-0.2.0/tests/test_deep_learning_cache.py

@@ -0,0 +1,34 @@
+from collections.abc import Callable
+from typing import Any
+
+import cli
+import numpy as np
+import pytest
+import torch
+from torch.utils.data import TensorDataset
+
+from persistent_cache import deep_learning, speedup_deep_learning
+from persistent_cache.reducers.speedup_deep_learning import LARGE_DIMENSION
+
+caches = [deep_learning.cache, speedup_deep_learning.cache]
+
+
+def calculate(*args: Any, **kwargs: Any) -> None:
+    cli.console.print("calculation started")
+    cli.console.print(args, kwargs)
+
+
+def verify_cached_function(function: Callable[..., Any]) -> None:
+    array = np.zeros(10)
+    large_array = np.zeros((10, LARGE_DIMENSION))
+    tensor = torch.tensor(array)
+    module = torch.nn.Linear(10, 2)
+    dataset = TensorDataset(tensor)
+    labeled_dataset = TensorDataset(tensor, tensor)
+    function(array, tensor, module, dataset, large_array, labeled_dataset)
+
+
+@pytest.mark.parametrize("cache_decorator", caches)
+def test_cache_with_argument_combination(cache_decorator: Callable[..., Any]) -> None:
+    cached_function = cache_decorator(calculate)
+    verify_cached_function(cached_function)

persistent_function_cache-0.1.0/src/persistent_cache/__init__.py

@@ -1 +0,0 @@
-from .caches.base import cache

persistent_function_cache-0.1.0/src/persistent_cache/caches/base.py

@@ -1,4 +0,0 @@
-from persistent_cache.main import decorator
-from persistent_cache.reducers.base import Reducer
-
-cache = decorator.cache(Reducer)

persistent_function_cache-0.1.0/src/persistent_cache/caches/deep_learning.py

@@ -1,4 +0,0 @@
-from persistent_cache.main import decorator
-from persistent_cache.reducers.deep_learning import Reducer
-
-cache = decorator.cache(Reducer)

persistent_function_cache-0.1.0/src/persistent_cache/caches/speedup_deep_learning.py

@@ -1,4 +0,0 @@
-from persistent_cache.main import decorator
-from persistent_cache.reducers.speedup_deep_learning import Reducer
-
-cache = decorator.cache(Reducer)

persistent_function_cache-0.1.0/src/persistent_cache/main/decorator.py

@@ -1,44 +0,0 @@
-from collections.abc import Callable
-from functools import wraps
-from typing import Any, TypeVar, cast
-
-from persistent_cache.models import Path
-from persistent_cache.reducers.base import Reducer
-
-from .cacheslot import CacheSlot
-
-T = TypeVar("T")
-F = TypeVar("F", bound=Callable[..., Any])
-
-
-def cache(
-    key_reducer: type[Reducer] = Reducer,
-    cache_path: Path = Path.cache,
-) -> Callable[[F], F]:
-    """A decorator to cache function results. Decorated functions are only executed if
-    result is not present in cache. The arguments of the function can be any nested
-    complex object.
-
-    Use as:
-
-    from persistent_cache import cache
-
-    @cache
-    def long_function(complex_object):
-        ...
-    """
-
-    def cache_decorator(function: F) -> F:
-        @wraps(function)
-        def wrapped_function(*args: Any, **kwargs: Any) -> Any:
-            cache_slot = CacheSlot(function, args, kwargs, key_reducer, cache_path)
-            try:
-                value = cache_slot.value
-            except KeyError:
-                value = function(*args, **kwargs)
-                cache_slot.value = value
-            return value
-
-        return cast(F, wrapped_function)
-
-    return cache_decorator

persistent_function_cache-0.1.0/src/persistent_cache/models/__init__.py

@@ -1 +0,0 @@
-from .path import Path

persistent_function_cache-0.1.0/tests/test_cache.py

@@ -1,26 +0,0 @@
-import io
-import math
-from collections.abc import Callable
-from typing import Any, TypeVar
-
-import cli
-import pytest
-
-from persistent_cache.caches import base, deep_learning, speedup_deep_learning
-
-F = TypeVar("F", bound=Callable)  # type: ignore[type-arg]
-
-
-def calculate(*args: Any, **kwargs: Any) -> None:
-    cli.console.print("calculation started")
-    cli.console.print(args, kwargs)
-
-
-caches = [base.cache, deep_learning.cache, speedup_deep_learning.cache]
-
-
-@pytest.mark.parametrize("cache", caches)
-def test_cache_with_argument_combination(cache: Callable) -> None:  # type: ignore[type-arg]
-    cached_function = cache(calculate)
-    with io.BytesIO() as fp:
-        cached_function(fp, lambda x: x, math, {})