平时我们都会封装一些处理缓存或其他的小工具。但每个人都封装一次,重复造轮子,有点费时间。有没有一些好的工具库推荐-guava。guava是谷歌基于java封装好的开源库,它的性能、实用性,比我们自己造的轮子更好,毕竟谷歌出品,下面介绍下几个常用的guava工具类
com.google.guava
guava
27.0-jre
CacheBuilder 方法参数 |
描述 |
---|---|
initialCapacity(int initialCapacity) |
缓存池的初始大小 |
concurrencyLevel(int concurrencyLevel) |
设置并发数 |
maximumSize(long maximumSize) |
缓存池大小,在缓存项接近该大小时, Guava开始回收旧的缓存项 |
weakValues() |
设置value的存储引用是虚引用 |
softValues() |
设置value的存储引用是软引用 |
expireAfterWrite(long duration, TimeUnit unit) |
设置时间对象没有被写则对象从内存中删除(在另外的线程里面不定期维护) |
expireAfterAccess(long duration, TimeUnit unit) |
设置时间对象没有被读/写访问则对象从内存中删除(在另外的线程里面不定期维护) |
refreshAfterWrite(long duration, TimeUnit unit) |
和expireAfterWrite类似,不过不立马移除key,而是在下次更新时刷新,这段时间可能会返回旧值 |
removalListener( RemovalListener super K1, ? super V1> listener) |
监听器,缓存项被移除时会触发 |
build(CacheLoader super K1, V1> loader) |
当数据不存在时,则使用loader加载数据 |
V get(K key)
, 获取缓存值,如果键不存在值,将调用CacheLoader的load方法加载新值到该键中 LoadingCache cacheMap = CacheBuilder.newBuilder().initialCapacity(10)
.concurrencyLevel(10)
.expireAfterAccess(Duration.ofSeconds(10))
.weakValues()
.recordStats()
.removalListener(new RemovalListener(){
@Override
public void onRemoval(RemovalNotification notification) {
System.out.println(notification.getValue());
}
})
.build(new CacheLoader(){
@Override
public Long load(Integer key) throws Exception {
return System.currentTimeMillis();
}
});
cacheMap.get(1);
//Multimap: key-value key可以重复,value也可重复
Multimap multimap = ArrayListMultimap.create();
multimap.put("csc","1");
multimap.put("lwl","1");
multimap.put("csc","1");
multimap.put("lwl","one");
System.out.println(multimap.get("csc"));
System.out.println(multimap.get("lwl"));
---------------------------
[1, 1]
[1, one]
//MultiSet: 无序+可重复 count()方法获取单词的次数 增强了可读性+操作简单
Multiset set = HashMultiset.create();
set.add("csc");
set.add("lwl");
set.add("csc");
System.out.println(set.size());
System.out.println(set.count("csc"));
---------------------------
3
2
BiMap的键必须唯一,值也必须唯一,可以实现value和key互转
BiMap biMap = HashBiMap.create();
biMap.put(1,"lwl");
biMap.put(2,"csc");
BiMap map = biMap.inverse(); // value和key互转
map.forEach((v, k) -> System.out.println(v + "-" + k));
Table table = HashBasedTable.create();
,由泛型可以看出,table由双主键R(行),C(列)共同决定,V是存储值table.put(R,C,V)
V v = table.get(R,C)
遍历数据: Set
// 双键的Map Map--> Table-->rowKey+columnKey+value
Table tables = HashBasedTable.create();
tables.put("csc", "lwl", 1);
//row+column对应的value
System.out.println(tables.get("csc","lwl"));
Sets和Maps
HashSet setA = newHashSet(1, 2, 3, 4, 5);
HashSet setB = newHashSet(4, 5, 6, 7, 8);
//并集
SetView union = Sets.union(setA, setB);
//差集 setA-setB
SetView difference = Sets.difference(setA, setB);
//交集
SetView intersection = Sets.intersection(setA, setB);
HashMap mapA = Maps.newHashMap();
mapA.put("a", 1);mapA.put("b", 2);mapA.put("c", 3);
HashMap mapB = Maps.newHashMap();
mapB.put("b", 20);mapB.put("c", 3);mapB.put("d", 4);
MapDifference mapDifference = Maps.difference(mapA, mapB);
//mapA 和 mapB 相同的 entry
System.out.println(mapDifference.entriesInCommon());
//mapA 和 mapB key相同的value不同的 entry
System.out.println(mapDifference.entriesDiffering());
//只存在 mapA 的 entry
System.out.println(mapDifference.entriesOnlyOnLeft());
//只存在 mapB 的 entry
System.out.println(mapDifference.entriesOnlyOnRight());;
-------------结果-------------
{c=3}
{b=(2, 20)}
{a=1}
{d=4}
EventBus内部实现原理不复杂,EventBus内部会维护一个Multimap
@Data
@AllArgsConstructor
public class OrderMessage {
String message;
}
//使用 @Subscribe 注解,表明使用dealWithEvent 方法处理 OrderMessage类型对应的消息
//可以注解多个方法,不同的方法 处理不同的对象消息
public class OrderEventListener {
@Subscribe
public void dealWithEvent(OrderMessage event) {
System.out.println("内容:" + event.getMessage());
}
}
-------------------------------------
// new AsyncEventBus(String identifier, Executor executor);
EventBus eventBus = new EventBus("lwl");
eventBus.register(new OrderEventListener());
// 发布消息
eventBus.post(new OrderMessage("csc"));
Stopwatch stopwatch = Stopwatch.createStarted();
for(int i=0; i<100000; i++){
// do some thing
}
long nanos = stopwatch.elapsed(TimeUnit.MILLISECONDS);
System.out.println("逻辑代码运行耗时:"+nanos);
File newFile = new File("D:/text.txt");
Files.write("this is a test".getBytes(), newFile);
//再次写入会把之前的内容冲掉
Files.write("csc".getBytes(), newFile);
//追加写
Files.append("lwl", newFile, Charset.defaultCharset());
File newFile = new File("E:/text.txt");
List lines = Files.readLines(newFile, Charset.defaultCharset());
方法 |
描述 |
---|---|
Files.copy(File from, File to) |
复制文件 |
Files.deleteDirectoryContents(File directory) |
删除文件夹下的内容(包括文件与子文件夹) |
Files.deleteRecursively(File file) |
删除文件或者文件夹 |
Files.move(File from, File to) |
移动文件 |
Files.touch(File file) |
创建或者更新文件的时间戳 |
Files.getFileExtension(String file) |
获得文件的扩展名 |
Files.getNameWithoutExtension(String file) |
获得不带扩展名的文件名 |
Files.map(File file, MapMode mode) |
获取内存映射buffer |
//RateLimiter 构造方法,每秒限流permitsPerSecond
public static RateLimiter create(double permitsPerSecond)
//每秒限流 permitsPerSecond,warmupPeriod 则是数据初始预热时间,从第一次acquire 或 tryAcquire 执行开时计算
public static RateLimiter create(double permitsPerSecond, Duration warmupPeriod)
//获取一个令牌,阻塞,返回阻塞时间
public double acquire()
//获取 permits 个令牌,阻塞,返回阻塞时间
public double acquire(int permits)
//获取一个令牌,超时返回
public boolean tryAcquire(Duration timeout)
获取 permits 个令牌,超时返回
public boolean tryAcquire(int permits, Duration timeout)
复制
RateLimiter limiter = RateLimiter.create(2, 3, TimeUnit.SECONDS);
System.out.println("get one permit cost time: " + limiter.acquire(1) + "s");
System.out.println("get one permit cost time: " + limiter.acquire(1) + "s");
System.out.println("get one permit cost time: " + limiter.acquire(1) + "s");
System.out.println("get one permit cost time: " + limiter.acquire(1) + "s");
System.out.println("get one permit cost time: " + limiter.acquire(1) + "s");
System.out.println("get one permit cost time: " + limiter.acquire(1) + "s");
System.out.println("get one permit cost time: " + limiter.acquire(1) + "s");
System.out.println("get one permit cost time: " + limiter.acquire(1) + "s");
--------------- 结果 -------------------------
get one permit cost time: 0.0s
get one permit cost time: 1.331672s
get one permit cost time: 0.998392s
get one permit cost time: 0.666014s
get one permit cost time: 0.498514s
get one permit cost time: 0.498918s
get one permit cost time: 0.499151s
get one permit cost time: 0.488548s
复制
com.github.rholder
guava-retrying
2.0.0
复制
RetryerBuilder方法 |
描述 |
---|---|
withRetryListener |
重试监听器 |
withWaitStrategy |
失败后重试间隔时间 |
withStopStrategy |
停止策略 |
withBlockStrategy |
阻塞策略BlockStrategy |
withAttemptTimeLimiter |
执行时间限制策略 |
retryIfException |
发生异常,则重试 |
retryIfRuntimeException |
发生RuntimeException异常,则重试 |
retryIfExceptionOfType(Class extends Throwable> ex) |
发生ex异常,则重试 |
retryIfException(Predicate |
对异常判断,是否重试 |
retryIfResult(Predicate |
对返回结果判断,是否重试 |
Retryer retryer = RetryerBuilder.newBuilder()
.retryIfException()
.retryIfResult(Predicates.equalTo(false))
.withAttemptTimeLimiter(AttemptTimeLimiters.fixedTimeLimit(1, TimeUnit.SECONDS))
.withStopStrategy(StopStrategies.stopAfterAttempt(5))
.build();
//Retryer调用
retryer.call(() -> true);
复制
[1]
重试框架Guava-Retry和spring-Retry: https://blog.csdn.net/zzzgd_666/article/details/84377962
[2]
Google guava工具类的介绍和使用: https://blog.csdn.net/wwwdc1012/article/details/82228458
[3]
重试框架Guava-Retry和spring-Retry: https://blog.csdn.net/zzzgd_666/article/details/84377962
[4]
超详细的Guava RateLimiter限流原理解析: https://zhuanlan.zhihu.com/p/60979444