跳转到主要内容

依赖项缓存管理器

项目描述

Build Status

CacheMan

管理依赖项缓存的Python接口。

‘Ba-Bop-Ba-Dop-Bop’

描述

此模块作为缓存依赖项管理器,适用于程序中有许多重复计算且可以安全持久化的实例。这通常涉及一个数据库层来存储键值对。然而,这样的层有时过于冗余,并且管理数据库与项目可能比它所值得的更多努力。这就是CacheMan发挥作用的地方,它通过一个接口提供定义保存器、加载器、构建器和依赖项的磁盘默认值。

默认情况下,所有缓存将在60秒内发生10k更改、300秒内发生10次更改(但在60秒后)或900秒内发生1次更改时自动保存。可以通过从autosync子模块实例化AutoSyncCache来更改此行为。

依赖项

psutil – 用于异步缓存保存

功能

  • 本地内存字典的即插即用替换

  • 默认持久化pickle缓存

  • 非持久化缓存

  • 缓存加载/保存/删除钩子(默认)

  • 缓存验证钩子

  • 缓存构建器钩子

  • 依赖项无效化

  • 自动同步缓存

如何使用

以下是一些使用存储库的简单示例。

设置简单的持久化缓存

from cacheman import cacher

manager = cacher.get_cache_manager() # Optional manager name argument can be used here
cache = manager.register_cache('my_simple_cache') # You now have a cache!
print cache.get('my_key') # `None` first run, 'my_value' if this code was executed earlier
cache['my_key'] = 'my_value'
cache.save() # Changes are now persisted to disk
manager.save_cache_contents('my_simple_cache') # Alternative way to save a cache

非持久化缓存

from cacheman import cacher

manager = cacher.get_cache_manager()
cache = manager.register_custom_cache('my_simple_cache', persistent=False) # You cache won't save to disk
cache.save() # This is a no-op

注册钩子

from cacheman import cacher
from cacheman import cachewrap

def my_saver(cache_name, contents):
    print("Save requested on {} cache content: {}".format(cache_name, contents))

def my_loader(cache_name):
    return { 'load': 'faked' }

manager = cacher.get_cache_manager()

cache = cachewrap.PersistentCache('my_cache', saver=my_saver, loader=my_loader)
# Can also use manager to set savers/loaders
#manager.retrieve_cache('my_cache')
#manager.register_saver('my_cache', my_saver)
#manager.register_loader('my_cache', my_loader)

cache.save() # Will print 'Save ... : { 'load': 'faked' }'
cache['new'] = 'real' # Add something to the cache
cache.save() # Will print 'Save ... : { 'load': 'faked', 'new': 'real' }'

依赖项缓存

from cacheman import cacher

manager = cacher.get_cache_manager()
edge_cache = manager.retrieve_cache('edge_cache')
root_cache = manager.register_cache('root_cache')
manager.register_dependent_cache('root_cache', 'edge_cache')

def set_processed_value():
    # Computes and caches 'processed' from root's 'raw' value
    processed = edge_cache.get('processed')
    if processed is None:
        processed = (root_cache.get('raw') or 0) * 5
        edge_cache['processed'] = processed
    return processed

# A common problem with caching computed or dependent values:
print set_processed_value() # 0 without raw value
root_cache['raw'] = 1
print set_processed_value() # still 0 because it's cache in edge

# Now we use cache invalidation to tell downstream caches they're no longer valid
root_cache.invalidate() # Invalidates dependent caches
print edge_cache # Prints {} even though we only invalidated the root_cache
root_cache['raw'] = 1
print set_processed_value() # Now 5 because the edge was cleared before the request
print edge_cache # Can see {'processed': 5} propogated

设置缓存目录

from cacheman import cacher

# Default cache directory is '/tmp/general_cacher' or 'user\appadata\local\temp\general_cache'
# All pickle caches now save to namespaced directories within the base_cache_directory directory
manager = cacher.get_cache_manager(base_cache_directory='secret/cache/location')

cache = manager.register_cache('my_cache')
cache['new'] = 'real' # Add something to the cache
cache.save('my_cache') # Will save contents to 'secret/cache/location/general_cache/my_cache.pkl'

语言偏好

  • Google 风格指南

  • 面向对象(少数例外)

待办事项

  • 更好的参数检查

  • 变更日志

作者

作者:Matthew Seal

项目详情


下载文件

下载适合您平台文件。如果您不确定选择哪个,请了解更多关于安装包的信息。

源代码分发

此版本没有提供源代码分发文件。请参阅生成分发存档的教程

构建分发

CacheMan-2.2.0-py2.py3-none-any.whl (13.2 kB 查看哈希值)

上传时间 Python 2 Python 3

由支持