A small python package for persistent caching of function results. Main features:
- Supports both sync and async functions.
- Properly typed, so your decorated functions will have the same type hints as the original function.
- Customizable serializers and storage backends.
Unsupported, todo:
- TTL/expiry
- Class/instance methods
Examples
Sync
from miniperscache import cached
@cached(tag="expensive_calculation")
def expensive_calculation(x: int, y: int) -> float:
return x * y
Async
from miniperscache import cached_async
@cached_async(tag="expensive_calculation")
async def expensive_calculation(x: int, y: int) -> float:
result = await fetch_expensive_data(x, y)
return result
Custom serializers and storage
from miniperscache import cached, FileStorage, JsonSerializer
@cached(tag="expensive_calculation", storage=FileStorage(), value_serializer=JsonSerializer())
def expensive_calculation(x: int, y: int) -> float:
...
Skipping unhashable arguments
The cache works by hashing the arguments of the function call. If some of the arguments, e.g. random Python objects, are unhashable, you can skip them by passing a custom arg hasher set to DefaultArgHasher(skip_args=["arg_names", ...]):
from miniperscache import cached_async, DefaultArgHasher
@cached_async(tag="expensive_calculation", arg_hasher=DefaultArgHasher(skip_args=["client"]))
async def model_call(client: ClientObject, prompt: str) -> float:
res = await client.messages.create(
prompt=prompt
)
return res.content
Or, you can provide your own arg hasher
from miniperscache import cached_async, default_raw_arg_hasher
def model_call_arg_hasher(client: ClientObject, prompt: str) -> bytes:
return default_raw_arg_hasher(client.model, prompt)
@cached_async(tag="expensive_calculation", arg_hasher=model_call_arg_hasher)
async def model_call(client: ClientObject, prompt: str) -> float:
res = await client.messages.create(
...
prompt=prompt
)
return res.content
Deletion
TTL / more systematic expiration isn't supported yet, but what you can do is:
from miniperscache import SqliteStorage
SqliteStorage().delete_tag("expensive_calculation")
Customization
Serializers
Currently:
- DillSerializer (default)
- PickleSerializer
- JsonSerializer
Storage
- SqliteStorage (default) -- Note: serializing to/from disk should be very fast so we don't even provide AsyncSqliteStorage - if writing to disk becomes a bottleneck for you, you can implement it using aiosqlite.
- FileStorage
- AsyncFileStorage
Release files for miniperscache 0.1.3
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| miniperscache-0.1.3.tar.gz | 21.5 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| miniperscache-0.1.3-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 35.6 kB
Release files / miniperscache-0.1.3.tar.gz
| Download URL | miniperscache-0.1.3.tar.gz |
|---|---|
| Size | 21.5 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
6a4cb1972f3cc189d72c8e292948483492503fa25462e022919e490a9392f4c5
|
|
BLAKE2b-256 checksum How to use checksums |
b9eaf0f103812f66323a56aa0da7aa204b3aff2cfe0b693b8c93155bf878a315
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
uv/0.5.5
|
Release files / miniperscache-0.1.3-py3-none-any.whl
| Download URL | miniperscache-0.1.3-py3-none-any.whl |
|---|---|
| Size | 14.1 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
10a7d5f7bbd4c02373af3f17f7b0a7b954ded661ee8ff239717f1a03ba36d4cb
|
|
BLAKE2b-256 checksum How to use checksums |
e5a2077612343a5784e3c64bd486b058e19069ee55f4a38615c03253172f9e2a
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
uv/0.5.5
|