{"id":13725117,"url":"https://github.com/lonelyenvoy/python-memoization","last_synced_at":"2025-10-21T20:39:50.173Z","repository":{"id":57440792,"uuid":"144746016","full_name":"lonelyenvoy/python-memoization","owner":"lonelyenvoy","description":"A powerful caching library for Python, with TTL support and multiple algorithm options.","archived":false,"fork":false,"pushed_at":"2021-08-01T19:17:13.000Z","size":212,"stargazers_count":231,"open_issues_count":14,"forks_count":15,"subscribers_count":4,"default_branch":"master","last_synced_at":"2024-11-14T12:56:36.208Z","etag":null,"topics":["algorithm","cache","cache-python","decorator","extendable","extensible","fifo","fifo-cache","functional-programming","lfu","lfu-cache","lru","lru-cache","memoization","memoization-library","memoize-decorator","python-memoization","ttl","ttl-cache","ttl-support"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/lonelyenvoy.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null}},"created_at":"2018-08-14T16:31:14.000Z","updated_at":"2024-10-23T13:18:20.000Z","dependencies_parsed_at":"2022-09-02T06:43:21.316Z","dependency_job_id":null,"html_url":"https://github.com/lonelyenvoy/python-memoization","commit_stats":null,"previous_names":[],"tags_count":12,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lonelyenvoy%2Fpython-memoization","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lonelyenvoy%2Fpython-memoization/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lonelyenvoy%2Fpython-memoization/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lonelyenvoy%2Fpython-memoization/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/lonelyenvoy","download_url":"https://codeload.github.com/lonelyenvoy/python-memoization/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":224645223,"owners_count":17346102,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["algorithm","cache","cache-python","decorator","extendable","extensible","fifo","fifo-cache","functional-programming","lfu","lfu-cache","lru","lru-cache","memoization","memoization-library","memoize-decorator","python-memoization","ttl","ttl-cache","ttl-support"],"created_at":"2024-08-03T01:02:13.456Z","updated_at":"2025-10-21T20:39:50.070Z","avatar_url":"https://github.com/lonelyenvoy.png","language":"Python","readme":"# python-memoization\n\n[![Repository][repositorysvg]][repository] [![Build Status][travismaster]][travis] [![Codacy Badge][codacysvg]][codacy]\n[![Coverage Status][coverallssvg]][coveralls] [![Downloads][downloadssvg]][repository]\n\u003cbr\u003e\n[![PRs welcome][prsvg]][pr] [![License][licensesvg]][license] [![Supports Python][pythonsvg]][python]\n\n\nA powerful caching library for Python, with TTL support and multiple algorithm options.\n\nIf you like this work, please [star](https://github.com/lonelyenvoy/python-memoization) it on GitHub.\n\n## Why choose this library?\n\nPerhaps you know about [```functools.lru_cache```](https://docs.python.org/3/library/functools.html#functools.lru_cache)\nin Python 3, and you may be wondering why we are reinventing the wheel.\n\nWell, actually not. This lib is based on ```functools```. Please find below the comparison with ```lru_cache```.\n\n|Features|```functools.lru_cache```|```memoization```|\n|--------|-------------------|-----------|\n|Configurable max size|✔️|✔️|\n|Thread safety|✔️|✔️|\n|Flexible argument typing (typed \u0026 untyped)|✔️|Always typed|\n|Cache statistics|✔️|✔️|\n|LRU (Least Recently Used) as caching algorithm|✔️|✔️|\n|LFU (Least Frequently Used) as caching algorithm|No support|✔️|\n|FIFO (First In First Out) as caching algorithm|No support|✔️|\n|Extensibility for new caching algorithms|No support|✔️|\n|TTL (Time-To-Live) support|No support|✔️|\n|Support for unhashable arguments (dict, list, etc.)|No support|✔️|\n|Custom cache keys|No support|✔️|\n|On-demand partial cache clearing|No support|✔️|\n|Iterating through the cache|No support|✔️|\n|Python version|3.2+|3.4+|\n\n```memoization``` solves some drawbacks of ```functools.lru_cache```:\n\n1. ```lru_cache``` does not support __unhashable types__, which means function arguments cannot contain dict or list.\n\n```python\n\u003e\u003e\u003e from functools import lru_cache\n\u003e\u003e\u003e @lru_cache()\n... def f(x): return x\n... \n\u003e\u003e\u003e f([1, 2])  # unsupported\nTraceback (most recent call last):\n  File \"\u003cstdin\u003e\", line 1, in \u003cmodule\u003e\nTypeError: unhashable type: 'list'\n```\n\n2. ```lru_cache``` is vulnerable to [__hash collision attack__](https://learncryptography.com/hash-functions/hash-collision-attack)\n   and can be hacked or compromised. Using this technique, attackers can make your program __unexpectedly slow__ by\n   feeding the cached function with certain cleverly designed inputs. However, in ```memoization```, caching is always\n   typed, which means ```f(3)``` and ```f(3.0)``` will be treated as different calls and cached separately. Also,\n   you can build your own cache key with a unique hashing strategy. These measures __prevents the attack__ from\n   happening (or at least makes it a lot harder).\n\n```python\n\u003e\u003e\u003e hash((1,))\n3430019387558\n\u003e\u003e\u003e hash(3430019387558.0)  # two different arguments with an identical hash value\n3430019387558\n```\n\n3. Unlike `lru_cache`, `memoization` is designed to be highly extensible, which make it easy for developers to add and integrate\n__any caching algorithms__ (beyond FIFO, LRU and LFU) into this library. See [Contributing Guidance](https://github.com/lonelyenvoy/python-memoization/blob/master/CONTRIBUTING.md) for further detail.\n\n\n## Installation\n\n```bash\npip install -U memoization\n```\n\n\n## 1-Minute Tutorial\n\n```python\nfrom memoization import cached\n\n@cached\ndef func(arg):\n    ...  # do something slow\n```\n\nSimple enough - the results of ```func()``` are cached. \nRepetitive calls to ```func()``` with the same arguments run ```func()``` only once, enhancing performance.\n\n\u003e:warning:__WARNING:__ for functions with unhashable arguments, the default setting may not enable `memoization` to work properly. See [custom cache keys](https://github.com/lonelyenvoy/python-memoization#custom-cache-keys) section below for details.\n\n## 15-Minute Tutorial\n\nYou will learn about the advanced features in the following tutorial, which enable you to customize `memoization` .\n\nConfigurable options include `ttl`, `max_size`, `algorithm`, `thread_safe`, `order_independent` and `custom_key_maker`.\n\n### TTL (Time-To-Live)\n\n```python\n@cached(ttl=5)  # the cache expires after 5 seconds\ndef expensive_db_query(user_id):\n    ...\n```\n\nFor impure functions, TTL (in second) will be a solution. This will be useful when the function returns resources that is valid only for a short time, e.g. fetching something from databases.\n\n### Limited cache capacity\n \n```python\n@cached(max_size=128)  # the cache holds no more than 128 items\ndef get_a_very_large_object(filename):\n    ...\n```\n\nBy default, if you don't specify ```max_size```, the cache can hold unlimited number of items.\nWhen the cache is fully occupied, the former data will be overwritten by a certain algorithm described below.\n\n### Choosing your caching algorithm\n\n```python\nfrom memoization import cached, CachingAlgorithmFlag\n\n@cached(max_size=128, algorithm=CachingAlgorithmFlag.LFU)  # the cache overwrites items using the LFU algorithm\ndef func(arg):\n    ...\n```\n\nPossible values for ```algorithm``` are:\n\n- `CachingAlgorithmFlag.LRU`: _Least Recently Used_  (default)\n- `CachingAlgorithmFlag.LFU`: _Least Frequently Used_ \n- `CachingAlgorithmFlag.FIFO`: _First In First Out_ \n\nThis option is valid only when a ```max_size``` is explicitly specified.\n\n### Thread safe?\n\n```python\n@cached(thread_safe=False)\ndef func(arg):\n    ...\n```\n\n```thread_safe``` is ```True``` by default. Setting it to ```False``` enhances performance.\n\n### Order-independent cache key\n\nBy default, the following function calls will be treated differently and cached twice, which means the cache misses at the second call.\n\n```python\nfunc(a=1, b=1)\nfunc(b=1, a=1)\n```\n\nYou can avoid this behavior by passing an `order_independent` argument to the decorator, although it will slow down the performance a little bit. \n\n```python\n@cached(order_independent=True)\ndef func(**kwargs):\n    ...\n```\n\n### Custom cache keys\n\nPrior to memorize your function inputs and outputs (i.e. putting them into a cache), `memoization` needs to\nbuild a __cache key__ using the inputs, so that the outputs can be retrieved later.\n\n\u003e By default, `memoization` tries to combine all your function\narguments and calculate its hash value using `hash()`. If it turns out that parts of your arguments are\nunhashable, `memoization` will fall back to turning them into a string using `str()`. This behavior relies\non the assumption that the string exactly represents the internal state of the arguments, which is true for\nbuilt-in types.\n\nHowever, this is not true for all objects. __If you pass objects which are\ninstances of non-built-in classes, sometimes you will need to override the default key-making procedure__,\nbecause the `str()` function on these objects may not hold the correct information about their states.\n\nHere are some suggestions. __Implementations of a valid key maker__:\n\n- MUST be a function with the same signature as the cached function.\n- MUST produce unique keys, which means two sets of different arguments always map to two different keys.\n- MUST produce hashable keys, and a key is comparable with another key (`memoization` only needs to check for their equality).\n- should compute keys efficiently and produce small objects as keys.\n\nExample:\n\n```python\ndef get_employee_id(employee):\n    return employee.id  # returns a string or a integer\n\n@cached(custom_key_maker=get_employee_id)\ndef calculate_performance(employee):\n    ...\n```\n\nNote that writing a robust key maker function can be challenging in some situations. If you find it difficult,\nfeel free to ask for help by submitting an [issue](https://github.com/lonelyenvoy/python-memoization/issues).\n\n\n### Knowing how well the cache is behaving\n\n```python\n\u003e\u003e\u003e @cached\n... def f(x): return x\n... \n\u003e\u003e\u003e f.cache_info()\nCacheInfo(hits=0, misses=0, current_size=0, max_size=None, algorithm=\u003cCachingAlgorithmFlag.LRU: 2\u003e, ttl=None, thread_safe=True, order_independent=False, use_custom_key=False)\n```\n\nWith ```cache_info```, you can retrieve the number of ```hits``` and ```misses``` of the cache, and other information indicating the caching status.\n\n- `hits`: the number of cache hits\n- `misses`: the number of cache misses\n- `current_size`: the number of items that were cached\n- `max_size`: the maximum number of items that can be cached (user-specified)\n- `algorithm`: caching algorithm (user-specified)\n- `ttl`: Time-To-Live value (user-specified)\n- `thread_safe`: whether the cache is thread safe (user-specified)\n- `order_independent`: whether the cache is kwarg-order-independent (user-specified)\n- `use_custom_key`: whether a custom key maker is used\n\n### Other APIs\n\n- Access the original undecorated function `f` by `f.__wrapped__`.\n- Clear the cache by `f.cache_clear()`.\n- Check whether the cache is empty by `f.cache_is_empty()`.\n- Check whether the cache is full by `f.cache_is_full()`.\n- Disable `SyntaxWarning` by `memoization.suppress_warnings()`.\n\n## Advanced API References\n\n\u003cdetails\u003e\n\u003csummary\u003eDetails\u003c/summary\u003e\n\n### Checking whether the cache contains something\n\n#### cache_contains_argument(function_arguments, alive_only)\n\n```\nReturn True if the cache contains a cached item with the specified function call arguments\n\n:param function_arguments:  Can be a list, a tuple or a dict.\n                            - Full arguments: use a list to represent both positional arguments and keyword\n                              arguments. The list contains two elements, a tuple (positional arguments) and\n                              a dict (keyword arguments). For example,\n                                f(1, 2, 3, a=4, b=5, c=6)\n                              can be represented by:\n                                [(1, 2, 3), {'a': 4, 'b': 5, 'c': 6}]\n                            - Positional arguments only: when the arguments does not include keyword arguments,\n                              a tuple can be used to represent positional arguments. For example,\n                                f(1, 2, 3)\n                              can be represented by:\n                                (1, 2, 3)\n                            - Keyword arguments only: when the arguments does not include positional arguments,\n                              a dict can be used to represent keyword arguments. For example,\n                                f(a=4, b=5, c=6)\n                              can be represented by:\n                                {'a': 4, 'b': 5, 'c': 6}\n\n:param alive_only:          Whether to check alive cache item only (default to True).\n\n:return:                    True if the desired cached item is present, False otherwise.\n```\n\n#### cache_contains_result(return_value, alive_only)\n\n```\nReturn True if the cache contains a cache item with the specified user function return value. O(n) time\ncomplexity.\n\n:param return_value:        A return value coming from the user function.\n\n:param alive_only:          Whether to check alive cache item only (default to True).\n\n:return:                    True if the desired cached item is present, False otherwise.\n```\n\n### Iterating through the cache\n\n#### cache_arguments()\n\n```\nGet user function arguments of all alive cache elements\n\nsee also: cache_items()\n\nExample:\n   @cached\n   def f(a, b, c, d):\n       ...\n   f(1, 2, c=3, d=4)\n   for argument in f.cache_arguments():\n       print(argument)  # ((1, 2), {'c': 3, 'd': 4})\n\n:return: an iterable which iterates through a list of a tuple containing a tuple (positional arguments) and\n        a dict (keyword arguments)\n```\n\n#### cache_results()\n\n```\nGet user function return values of all alive cache elements\n\nsee also: cache_items()\n\nExample:\n   @cached\n   def f(a):\n       return a\n   f('hello')\n   for result in f.cache_results():\n       print(result)  # 'hello'\n\n:return: an iterable which iterates through a list of user function result (of any type)\n```\n\n#### cache_items()\n\n```\nGet cache items, i.e. entries of all alive cache elements, in the form of (argument, result).\n\nargument: a tuple containing a tuple (positional arguments) and a dict (keyword arguments).\nresult: a user function return value of any type.\n\nsee also: cache_arguments(), cache_results().\n\nExample:\n   @cached\n   def f(a, b, c, d):\n       return 'the answer is ' + str(a)\n   f(1, 2, c=3, d=4)\n   for argument, result in f.cache_items():\n       print(argument)  # ((1, 2), {'c': 3, 'd': 4})\n       print(result)    # 'the answer is 1'\n\n:return: an iterable which iterates through a list of (argument, result) entries\n```\n\n#### cache_for_each()\n\n```\nPerform the given action for each cache element in an order determined by the algorithm until all\nelements have been processed or the action throws an error\n\n:param consumer:           an action function to process the cache elements. Must have 3 arguments:\n                             def consumer(user_function_arguments, user_function_result, is_alive): ...\n                           user_function_arguments is a tuple holding arguments in the form of (args, kwargs).\n                             args is a tuple holding positional arguments.\n                             kwargs is a dict holding keyword arguments.\n                             for example, for a function: foo(a, b, c, d), calling it by: foo(1, 2, c=3, d=4)\n                             user_function_arguments == ((1, 2), {'c': 3, 'd': 4})\n                           user_function_result is a return value coming from the user function.\n                           is_alive is a boolean value indicating whether the cache is still alive\n                           (if a TTL is given).\n```\n\n### Removing something from the cache\n\n#### cache_clear()\n\n```\nClear the cache and its statistics information\n```\n\n#### cache_remove_if(predicate)\n\n```\nRemove all cache elements that satisfy the given predicate\n\n:param predicate:           a predicate function to judge whether the cache elements should be removed. Must\n                            have 3 arguments, and returns True or False:\n                              def consumer(user_function_arguments, user_function_result, is_alive): ...\n                            user_function_arguments is a tuple holding arguments in the form of (args, kwargs).\n                              args is a tuple holding positional arguments.\n                              kwargs is a dict holding keyword arguments.\n                              for example, for a function: foo(a, b, c, d), calling it by: foo(1, 2, c=3, d=4)\n                              user_function_arguments == ((1, 2), {'c': 3, 'd': 4})\n                            user_function_result is a return value coming from the user function.\n                            is_alive is a boolean value indicating whether the cache is still alive\n                            (if a TTL is given).\n\n:return:                    True if at least one element is removed, False otherwise.\n```\n\n\u003c/details\u003e\n\n## Q\u0026A\n\n1. **Q: There are duplicated code in `memoization` and most of them can be eliminated by using another level of\nabstraction (e.g. classes and multiple inheritance). Why not refactor?**\n\n   A: We would like to keep the code in a proper level of abstraction. However, these abstractions make it run slower.\nAs this is a caching library focusing on speed, we have to give up some elegance for better performance. Refactoring\nis our future work.\n\n\n2. **Q: I have submitted an issue and not received a reply for a long time. Anyone can help me?**\n\n   A: Sorry! We are not working full-time, but working voluntarily on this project, so you might experience some delay.\nWe appreciate your patience.\n\n\n## Contributing\n\nThis project welcomes contributions from anyone.\n- [Read Contributing Guidance](https://github.com/lonelyenvoy/python-memoization/blob/master/CONTRIBUTING.md) first.\n- [Submit bugs](https://github.com/lonelyenvoy/python-memoization/issues) and help us verify fixes.\n- [Submit pull requests](https://github.com/lonelyenvoy/python-memoization/pulls) for bug fixes and features and discuss existing proposals. Please make sure that your PR passes the tests in ```test.py```.\n- [See contributors](https://github.com/lonelyenvoy/python-memoization/blob/master/CONTRIBUTORS.md) of this project.\n\n\n## License\n\n[The MIT License](https://github.com/lonelyenvoy/python-memoization/blob/master/LICENSE)\n\n\n[pythonsvg]: https://img.shields.io/pypi/pyversions/memoization.svg\n[python]: https://www.python.org\n\n[travismaster]: https://travis-ci.com/lonelyenvoy/python-memoization.svg?branch=master\n[travis]: https://travis-ci.com/lonelyenvoy/python-memoization\n\n[coverallssvg]: https://coveralls.io/repos/github/lonelyenvoy/python-memoization/badge.svg?branch=master\n[coveralls]: https://coveralls.io/github/lonelyenvoy/python-memoization?branch=master\n\n[repositorysvg]: https://img.shields.io/pypi/v/memoization\n[repository]: https://pypi.org/project/memoization\n\n[downloadssvg]: https://img.shields.io/pypi/dm/memoization\n\n[prsvg]: https://img.shields.io/badge/pull_requests-welcome-blue.svg\n[pr]: https://github.com/lonelyenvoy/python-memoization#contributing\n\n[licensesvg]: https://img.shields.io/badge/license-MIT-blue.svg\n[license]: https://github.com/lonelyenvoy/python-memoization/blob/master/LICENSE\n\n[codacysvg]: https://api.codacy.com/project/badge/Grade/52c68fb9de6b4b149e77e8e173616db6\n[codacy]: https://www.codacy.com/manual/petrinchor/python-memoization?utm_source=github.com\u0026amp;utm_medium=referral\u0026amp;utm_content=lonelyenvoy/python-memoization\u0026amp;utm_campaign=Badge_Grade\n","funding_links":[],"categories":["Python"],"sub_categories":[],"project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Flonelyenvoy%2Fpython-memoization","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Flonelyenvoy%2Fpython-memoization","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Flonelyenvoy%2Fpython-memoization/lists"}