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。
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:
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。
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 expired
- double _deprecated = 0; // time to deprecated
-
- V? get value => _value;
-
- /// update cache value with current time in seconds
- void 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 seconds
- bool isAlive({double? now}) {
- now ??= Time.currentTimestamp;
- return now < _expired;
- }
-
- /// check whether cache is deprecated with current time in seconds
- bool isDeprecated({double? now}) {
- now ??= Time.currentTimestamp;
- return now > _deprecated;
- }
-
- /// renewal cache with a temporary life span and current time in seconds
- void 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。
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
holder; -
- }
-
-
- /// Pool for cache holders with keys
- class CachePool <K, V> {
-
- final Map
> _holderMap = {}; -
- Iterable
get keys => _holderMap.keys; -
- /// update cache holder for key
- CacheHolder
update(K key, CacheHolder holder) { - _holderMap[key] = holder;
- return holder;
- }
-
- /// update cache value for key with timestamp in seconds
- CacheHolder
updateValue(K key, V? value, double life, {double? now}) => - update(key, CacheHolder(value, life, now: now));
-
- /// erase cache for key
- CachePair
? erase(K key, {double? now}) { - CachePair
? old; - if (now != null) {
- // get exists value before erasing
- old = fetch(key, now: now);
- }
- _holderMap.remove(key);
- return old;
- }
-
- /// fetch cache value & its holder
- CachePair
? fetch(K key, {double? now}) { - CacheHolder
? holder = _holderMap[key]; - if (holder == null) {
- // holder not found
- return null;
- } else if (holder.isAlive(now: now)) {
- return CachePair(holder.value, holder);
- } else {
- // holder expired
- return CachePair(null, holder);
- }
- }
-
- /// clear expired cache holders
- int 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:
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。
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 CachePool
- CachePool
getPool(String name) { - CachePool
? pool = _poolMap[name]; - if (pool == null) {
- pool = CachePool();
- _poolMap[name] = pool;
- }
- return pool;
- }
-
- /// Purge all pools
- ///
- /// @param now - current time
- int 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:
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。
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 {
-
- @override
- Future getMeta(ID entity) async {
- // 从数据库中获取 meta 信息
- }
-
- @override
- Future<bool> saveMeta(Meta meta, ID entity) async {
- // 保存 meta 信息到数据库
- }
-
- }
-
- class MetaCache extends MetaTable {
-
- final CachePool
_cache = CacheManager().getPool('meta'); -
- @override
- Future getMeta(ID entity) async {
- CachePair? pair;
- CacheHolder? holder;
- Meta? value;
- double now = Time.currentTimeSeconds;
- await lock();
- try {
- // 1. check memory cache
- pair = _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 exists
- return null;
- } else {
- // cache expired, wait to reload
- holder.renewal(128, now: now);
- }
- // 2. load from database
- value = await super.getMeta(entity);
- // update cache
- _cache.updateValue(entity, value, 36000, now: now);
- }
- } finally {
- unlock();
- }
- // OK, return cache now
- return value;
- }
-
- @override
- Future<bool> saveMeta(Meta meta, ID entity) async {
- _cache.updateValue(entity, meta, 36000, now: Time.currentTimeSeconds);
- return await super.saveMeta(meta, entity);
- }
-
- }
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。
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。
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';
- /* license: https://mit-license.org
- *
- * ObjectKey : Object & Key kits
- *
- * Written in 2023 by Moky
- *
- * =============================================================================
- * 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 expired
- double _deprecated = 0; // time to deprecated
-
- V? get value => _value;
-
- /// update cache value with current time in seconds
- void 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 seconds
- bool isAlive({double? now}) {
- now ??= Time.currentTimestamp;
- return now < _expired;
- }
-
- /// check whether cache is deprecated with current time in seconds
- bool isDeprecated({double? now}) {
- now ??= Time.currentTimestamp;
- return now > _deprecated;
- }
-
- /// renewal cache with a temporary life span and current time in seconds
- void 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
holder; -
- }
-
-
- /// Pool for cache holders with keys
- class CachePool <K, V> {
-
- final Map
> _holderMap = {}; -
- Iterable
get keys => _holderMap.keys; -
- /// update cache holder for key
- CacheHolder
update(K key, CacheHolder holder) { - _holderMap[key] = holder;
- return holder;
- }
-
- /// update cache value for key with timestamp in seconds
- CacheHolder
updateValue(K key, V? value, double life, {double? now}) => - update(key, CacheHolder(value, life, now: now));
-
- /// erase cache for key
- CachePair
? erase(K key, {double? now}) { - CachePair
? old; - if (now != null) {
- // get exists value before erasing
- old = fetch(key, now: now);
- }
- _holderMap.remove(key);
- return old;
- }
-
- /// fetch cache value & its holder
- CachePair
? fetch(K key, {double? now}) { - CacheHolder
? holder = _holderMap[key]; - if (holder == null) {
- // holder not found
- return null;
- } else if (holder.isAlive(now: now)) {
- return CachePair(holder.value, holder);
- } else {
- // holder expired
- return CachePair(null, holder);
- }
- }
-
- /// clear expired cache holders
- int 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 CachePool
- CachePool
getPool(String name) { - CachePool
? pool = _poolMap[name]; - if (pool == null) {
- pool = CachePool();
- _poolMap[name] = pool;
- }
- return pool;
- }
-
- /// Purge all pools
- ///
- /// @param now - current time
- int 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
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)