前言
村里的老人说:“不会写缓存器的码农不是好程序员。”
今天我们就来讲讲如何编写一个简单又通用的缓存管理模块。
需求分析
根据以往经验,每一个缓存器,除了要保持被缓存数据/对象之外,还需要同时记录两个与之紧密相关的时间:
- 过期时间 expired —— 缓存超过了一定时间,需要更新了,但当前数据仍然有效
- 失效时间 deprecated —— 缓存超过了很长时间,当前数据已经无效,不能再用了
在过期时间 expired 之前,缓存数据/对象可直接使用,无需刷新;
超过了过期时间 expired,但未超过失效时间 deprecated,此时缓存数据仍然有效,可以继续使用,但需要委托一个线程去获取最新数据/对象,然后再更新本地缓存;
如果后台线程多次更新失败,当前缓存数据/对象已经严重超时,即超过了 deprecated,此时应该丢弃当前缓存数据/对象,返回空数据/对象给调用者。
模块设计
首先我们设计一个 ```CacheHolder``` 类来保存被缓存数据/对象,以及与之对应的时间信息:
import 'package:object_key/object_key.dart' show Time;/// Holder for cache value with times in seconds
class CacheHolder <V> {CacheHolder(V? cacheValue, double cacheLifeSpan, {double? now}): _value = cacheValue, _life = cacheLifeSpan {now ??= Time.currentTimestamp;_expired = now + cacheLifeSpan;_deprecated = now + cacheLifeSpan * 2;}V? _value;final double _life; // life span (in seconds)double _expired = 0; // time to expireddouble _deprecated = 0; // time to deprecatedV? get value => _value;/// update cache value with current time in secondsvoid update(V? newValue, {double? now}) {_value = newValue;now ??= Time.currentTimestamp;_expired = now + _life;_deprecated = now + _life * 2;}/// check whether cache is alive with current time in secondsbool isAlive({double? now}) {now ??= Time.currentTimestamp;return now < _expired;}/// check whether cache is deprecated with current time in secondsbool isDeprecated({double? now}) {now ??= Time.currentTimestamp;return now > _deprecated;}/// renewal cache with a temporary life span and current time in secondsvoid renewal(double? duration, {double? now}) {duration ??= 120;now ??= Time.currentTimestamp;_expired = now + duration;_deprecated = now + _life * 2;}}
该类提供 update() 和 renew() 两个函数来更新缓存信息,前者为获取到最新数据之后调用以更新数据及时间,后者仅刷新一下时间,用以推迟有效时间;
另外提供两个函数 isAlive() 和 isDeprecated(),分别用于判断是否需要更新,以及当前数据是否应该丢弃。
另外,为使 CacheHolder 能适用于任意类型数据/对象,这里使用了“泛型”类型定义。
缓存池
接下来我们需要设计一个缓冲池 ```CachePool```,用于保存同类型的 ```CacheHolder```:
import 'package:object_key/object_key.dart' show Time;
import 'holder.dart';class CachePair <V> {CachePair(this.value, this.holder);final V? value;final CacheHolder<V> holder;}/// Pool for cache holders with keys
class CachePool <K, V> {final Map<K, CacheHolder<V>> _holderMap = {};Iterable<K> get keys => _holderMap.keys;/// update cache holder for keyCacheHolder<V> update(K key, CacheHolder<V> holder) {_holderMap[key] = holder;return holder;}/// update cache value for key with timestamp in secondsCacheHolder<V> updateValue(K key, V? value, double life, {double? now}) =>update(key, CacheHolder(value, life, now: now));/// erase cache for keyCachePair<V>? erase(K key, {double? now}) {CachePair<V>? old;if (now != null) {// get exists value before erasingold = fetch(key, now: now);}_holderMap.remove(key);return old;}/// fetch cache value & its holderCachePair<V>? fetch(K key, {double? now}) {CacheHolder<V>? holder = _holderMap[key];if (holder == null) {// holder not foundreturn null;} else if (holder.isAlive(now: now)) {return CachePair(holder.value, holder);} else {// holder expiredreturn CachePair(null, holder);}}/// clear expired cache holdersint purge({double? now}) {now ??= Time.currentTimestamp;int count = 0;Iterable allKeys = keys;CacheHolder? holder;for (K key in allKeys) {holder = _holderMap[key];if (holder == null || holder.isDeprecated(now: now)) {// remove expired holders_holderMap.remove(key);++count;}}return count;}}
该缓冲池提供了 3 个接口给应用层使用:
- 更新缓存信息;
- 删除缓存信息;
- 获取缓存信息;
另外还提供一个 purge() 函数给缓存管理器调用,以清除已失效的 CacheHolder。
缓存管理器
最后,我们还需要设计一个缓存管理器 ```CacheManager```,去统一管理所有不同类型的 ```CachePool```:
import 'package:object_key/object_key.dart' show Time;
import 'pool.dart';class CacheManager {factory CacheManager() => _instance;static final CacheManager _instance = CacheManager._internal();CacheManager._internal();final Map<String, dynamic> _poolMap = {};/// Get pool with name////// @param name - pool name/// @param <K> - key type/// @param <V> - value type/// @return CachePoolCachePool<K, V> getPool<K, V>(String name) {CachePool<K, V>? pool = _poolMap[name];if (pool == null) {pool = CachePool();_poolMap[name] = pool;}return pool;}/// Purge all pools////// @param now - current timeint purge(double? now) {now ??= Time.currentTimestamp;int count = 0;CachePool? pool;Iterable allKeys = _poolMap.keys;for (var key in allKeys) {pool = _poolMap[key];if (pool != null) {count += pool.purge(now: now);}}return count;}}
我们这个缓存管理包括两个接口:
- 一个工厂方法 getPool(),用于获取/创建缓存池;
- 一个清除接口 purge(),供系统在适当的时候(例如系统内存不足时)调用以释放缓存空间。
至此,一个简单高效的本地缓存管理模块就写好了,下面我们来看看怎么用。
应用示例
假设我们有一个类 MetaTable,其作用是从数据库或者网络中获取 meta 信息,考虑到 I/O 的时间,以及数据解析为对象所消耗的 CPU 时间等,如果该类信息访问十分频繁,我们就需要为它加上一层缓存管理。
先来看看代码:
class MetaTable implements MetaDBI {@overrideFuture<Meta?> getMeta(ID entity) async {// 从数据库中获取 meta 信息}@overrideFuture<bool> saveMeta(Meta meta, ID entity) async {// 保存 meta 信息到数据库}}class MetaCache extends MetaTable {final CachePool<ID, Meta> _cache = CacheManager().getPool('meta');@overrideFuture<Meta?> getMeta(ID entity) async {CachePair<Meta>? pair;CacheHolder<Meta>? holder;Meta? value;double now = Time.currentTimeSeconds;await lock();try {// 1. check memory cachepair = _cache.fetch(entity, now: now);holder = pair?.holder;value = pair?.value;if (value == null) {if (holder == null) {// not load yet, wait to load} else if (holder.isAlive(now: now)) {// value not existsreturn null;} else {// cache expired, wait to reloadholder.renewal(128, now: now);}// 2. load from databasevalue = await super.getMeta(entity);// update cache_cache.updateValue(entity, value, 36000, now: now);}} finally {unlock();}// OK, return cache nowreturn value;}@overrideFuture<bool> saveMeta(Meta meta, ID entity) async {_cache.updateValue(entity, meta, 36000, now: Time.currentTimeSeconds);return await super.saveMeta(meta, entity);}}
带缓存读数据
当需要读取数据时,先通过 ```_cache.fetch()``` 检查当前缓存池中是否存在有效的值:
如果 (值存在),则 {
直接返回该值;
}
否则检查 holder;
如果 (holder 存在且未过期),则 {
说明确实不存在该数据,返回空值;
}
否则调用父类接口获取最新数据;
然后再更新本地缓存。
带缓存写数据
写数据就简单了,只需要在调用父类接口写数据库的同时刷新一下缓存即可。
代码引用
由于我已将这部分代码提交到了 pub.dev,所以在实际应用中,你只需要在项目工程文件 ```pubspec.yaml``` 中添加:
dependencies:
object_key: ^0.1.1
然后在需要使用的 dart 文件头引入即可:
import 'package:object_key/object_key.dart';
全部源码
/* license: https://mit-license.org** ObjectKey : Object & Key kits** Written in 2023 by Moky <albert.moky@gmail.com>** =============================================================================* The MIT License (MIT)** Copyright (c) 2023 Albert Moky** Permission is hereby granted, free of charge, to any person obtaining a copy* of this software and associated documentation files (the "Software"), to deal* in the Software without restriction, including without limitation the rights* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell* copies of the Software, and to permit persons to whom the Software is* furnished to do so, subject to the following conditions:** The above copyright notice and this permission notice shall be included in all* copies or substantial portions of the Software.** THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE* SOFTWARE.* =============================================================================*/
import 'package:object_key/object_key.dart' show Time;/// Holder for cache value with times in seconds
class CacheHolder <V> {CacheHolder(V? cacheValue, double cacheLifeSpan, {double? now}): _value = cacheValue, _life = cacheLifeSpan {now ??= Time.currentTimestamp;_expired = now + cacheLifeSpan;_deprecated = now + cacheLifeSpan * 2;}V? _value;final double _life; // life span (in seconds)double _expired = 0; // time to expireddouble _deprecated = 0; // time to deprecatedV? get value => _value;/// update cache value with current time in secondsvoid update(V? newValue, {double? now}) {_value = newValue;now ??= Time.currentTimestamp;_expired = now + _life;_deprecated = now + _life * 2;}/// check whether cache is alive with current time in secondsbool isAlive({double? now}) {now ??= Time.currentTimestamp;return now < _expired;}/// check whether cache is deprecated with current time in secondsbool isDeprecated({double? now}) {now ??= Time.currentTimestamp;return now > _deprecated;}/// renewal cache with a temporary life span and current time in secondsvoid renewal(double? duration, {double? now}) {duration ??= 120;now ??= Time.currentTimestamp;_expired = now + duration;_deprecated = now + _life * 2;}}class CachePair <V> {CachePair(this.value, this.holder);final V? value;final CacheHolder<V> holder;}/// Pool for cache holders with keys
class CachePool <K, V> {final Map<K, CacheHolder<V>> _holderMap = {};Iterable<K> get keys => _holderMap.keys;/// update cache holder for keyCacheHolder<V> update(K key, CacheHolder<V> holder) {_holderMap[key] = holder;return holder;}/// update cache value for key with timestamp in secondsCacheHolder<V> updateValue(K key, V? value, double life, {double? now}) =>update(key, CacheHolder(value, life, now: now));/// erase cache for keyCachePair<V>? erase(K key, {double? now}) {CachePair<V>? old;if (now != null) {// get exists value before erasingold = fetch(key, now: now);}_holderMap.remove(key);return old;}/// fetch cache value & its holderCachePair<V>? fetch(K key, {double? now}) {CacheHolder<V>? holder = _holderMap[key];if (holder == null) {// holder not foundreturn null;} else if (holder.isAlive(now: now)) {return CachePair(holder.value, holder);} else {// holder expiredreturn CachePair(null, holder);}}/// clear expired cache holdersint purge({double? now}) {now ??= Time.currentTimestamp;int count = 0;Iterable allKeys = keys;CacheHolder? holder;for (K key in allKeys) {holder = _holderMap[key];if (holder == null || holder.isDeprecated(now: now)) {// remove expired holders_holderMap.remove(key);++count;}}return count;}}class CacheManager {factory CacheManager() => _instance;static final CacheManager _instance = CacheManager._internal();CacheManager._internal();final Map<String, dynamic> _poolMap = {};/// Get pool with name////// @param name - pool name/// @param <K> - key type/// @param <V> - value type/// @return CachePoolCachePool<K, V> getPool<K, V>(String name) {CachePool<K, V>? pool = _poolMap[name];if (pool == null) {pool = CachePool();_poolMap[name] = pool;}return pool;}/// Purge all pools////// @param now - current timeint purge(double? now) {now ??= Time.currentTimestamp;int count = 0;CachePool? pool;Iterable allKeys = _poolMap.keys;for (var key in allKeys) {pool = _poolMap[key];if (pool != null) {count += pool.purge(now: now);}}return count;}}
GitHub 地址:
https://github.com/moky/ObjectKey/tree/main/object_key/lib/src/mem
结语
这里向大家展示了一个简单高效的本地缓存管理模块,该模块能有效避免重复创建相同对象,同时也可避免内存泄漏等问题。
合理使用该模块,可以令你的应用程序访问数据的平均速度大幅提升,特别是在重复滚动展示大量数据的列表时,能让你的应用体验更加丝滑。
如有其他问题,可以下载登录 Tarsier 与我交流(默认通讯录i找 Albert Moky)