布隆过滤器用Redisson实现的例子
背景介绍
之前这边有写过关于布隆过滤器过滤器的安装使用，但是这次使用使用jrebloom来实现，
jrebloom目前只提供一个maven依赖供我们使用。
考虑到后续代码维护，bug的修复。使用redisson来做布隆过滤器是更加安全做法。
下面是关于布隆过滤器的使用：
布隆过滤器的原理、redis布隆过滤器的安装和使用
代码实现
下面使用Redisson对布隆过滤器进行的封装。
1、引入maven依赖
org.redisson
redisson
3.13.1
2、关于redis的配置
redis配置：
#客户端超时时间单位是毫秒 默认是2000
redis.timeout=10000
#最大空闲数
redis.maxIdle=300
#连接池的最大数据库连接数。设为0表示无限制,如果是jedis 2.4以后用redis.maxTotal
#redis.maxActive=600
#控制一个pool可分配多少个jedis实例,用来替换上面的redis.maxActive,如果是jedis 2.4以后用该属性
redis.maxTotal=2000
#最大建立连接等待时间。如果超过此时间将接到异常。设为-1表示无限制。
redis.maxWaitMillis=1000
redis.nodes=192.168.25.128:7000,192.168.25.128:7001,192.168.25.128:7002,192.168.25.128:7003,192.168.25.128:7004,192.168.25.128:7005,192.168.25.128:7006,192.168.25.128:7007
3、配置文件的书写
@Configuration
@PropertySource("classpath:conf/redis.properties")
public class RedisConfig {
@Value("${redis.nodes}")
private String clusterNodes;
@Bean
public RedissonClient getBloomFilter(){
Config config = new Config();
ClusterServersConfig clusterServersConfig = config.useClusterServers();
String[] cNodes = clusterNodes.split(",");
//分割出集群节点
for (String node : cNodes) {
clusterServersConfig.addNodeAddress("redis://" + node);
}
return Redisson.create(config);
}
}
4、service的的代码
public interface RedissonBloomFilterService {
/**
* 初始化一个布隆过滤器
*
* @param key key
* @param expireDate 过期时间长度
* @param timeUnit 过期时间单位
* @return 是否创建成功
*/
boolean initNewFilter(String key, Long expireDate, TimeUnit timeUnit, long expectedInsertions, double falseProbability);
/**
* 添加value到今天布隆过滤器
* @param key key
* @param value 需要添加到过滤器的值
* @return 是否添加成功
*/
boolean add(final String key, final String value);
/**
* 添加value到今天布隆过滤器
* @param key key
* @param list 需要添加到过滤器的值
* @return 是否添加成功
*/
boolean add(final String key, final List list);
/**
* 判断对应key是否有value值
* @param key key
* @param value 对应key的value值
* @return 应key是否有value值
*/
boolean containsValue(final String key, final String value);
/**
* 查看是否存在key值
* @param key key
* @return 是否有key值
*/
boolean containKey(final String key);
}
实现类：
@Slf4j
@Service
public class RedissonBloomServiceImpl implements RedissonBloomFilterService {
public static String prefix = "redission_bf_";
private static DateTimeFormatter formatter = DateTimeFormatter.ofPattern("yyyyMMdd");
private static final int TIME_FORMAT_LENGTH = 8;
@Resource
private RedissonClient redissonClient;
/**
* 1、第一层的key为传入key值
* 2、第二次key为 prefix + key+20200718
*/
private Map> bloomFilterMap = new ConcurrentHashMap<>();
@Override
public synchronized boolean initNewFilter(String key, Long expireDate, TimeUnit timeUnit, long expectedInsertions, double falseProbability) {
String timeFormatter = formatter.format(LocalDateTime.now());
RBloomFilter bloomFilter = redissonClient.getBloomFilter(prefix + timeFormatter + key);
//初始化布隆过滤器
boolean isSuccess = bloomFilter.tryInit(expectedInsertions, falseProbability);
if (isSuccess) {
//设置过期时间
bloomFilter.expire(expireDate, timeUnit);
//设置过期监听器
bloomFilter.addListener(new ExpiredObjectListener() {
@Override
public void onExpired(String key) {
log.info("onExpired callback running key is {}", key);
if (key.startsWith(prefix)) {
String firstKey = key.substring(prefix.length() + TIME_FORMAT_LENGTH);
ConcurrentHashMap map = bloomFilterMap.get(firstKey);
if (Objects.isNull(map)) {
log.warn("listener callback key is empty key is {}", key);
return;
}
map.remove(key);
}
}
});
ConcurrentHashMap map = bloomFilterMap.computeIfAbsent(key, (k) -> {
return new ConcurrentHashMap(4);
});
map.put(prefix + timeFormatter + key, bloomFilter);
isSuccess = true;
}else {
log.error("initNewFilter fail key is {}", key);
}
return isSuccess;
}
@Override
public boolean add(String key, String value) {
String timeFormatter = formatter.format(LocalDateTime.now());
ConcurrentHashMap map = bloomFilterMap.get(key);
if (Objects.isNull(map)) {
log.error("add name key one value is null, name is {},value is {}", key, value);
return false;
}
RBloomFilter rBloomFilter = map.get(prefix + timeFormatter + key);
if (Objects.isNull(rBloomFilter)) {
log.error("add name key one value is null, name is {},redis key is {},value is {}", key, prefix + timeFormatter + key, value);
return false;
}
return rBloomFilter.add(value);
}
@Override
public boolean add(String key, List list) {
String timeFormatter = formatter.format(LocalDateTime.now());
ConcurrentHashMap map = bloomFilterMap.get(key);
if (Objects.isNull(map)) {
log.error("add name key list value is null, name is {}", key);
return false;
}
RBloomFilter rBloomFilter = map.get(prefix + timeFormatter + key);
if (Objects.isNull(rBloomFilter)) {
log.error("add name key list value is null, name is {},redis key is {}", key,prefix + timeFormatter + key);
return false;
}
list.forEach(value -> rBloomFilter.add(value));
return true;
}
@Override
public boolean containsValue(String key, String value) {
ConcurrentHashMap map = bloomFilterMap.get(key);
if (Objects.isNull(map)) {
log.error("containsValue key is exist, key is {}", key);
return false;
}
for (Map.Entry entry : map.entrySet()) {
if (entry.getValue().contains(value)) {
return true;
}
}
return false;
}
@Override
public boolean containKey(String key) {
return bloomFilterMap.containsKey(key);
}
}
测试类：（这里的测试，有关于已经存在值的判断）
@Test
public void testRedissonBloom() throws InterruptedException {
//创建过滤器
redissonBloomFilterService.initNewFilter("bloomkey1", 60L, SECONDS, 50, 0.1);
redissonBloomFilterService.initNewFilter("bloomkey2", 60L, SECONDS, 50, 0.1);
for (int i = 0; i < 50; i++) {
redissonBloomFilterService.add("bloomkey1", "" + i);
if (i > 30) {
redissonBloomFilterService.add("bloomkey2", "" + i);
}
}
int j1 = 0;
int j2 = 0;
for (int i = 0; i < 100; i++) {
boolean exists = redissonBloomFilterService.containsValue("bloomkey1", "" + i);
if (exists) {
System.out.println("bloomkey1有误判" + (++j1) + "i是" + i);
}
boolean exists2 = redissonBloomFilterService.containsValue("bloomkey2", "" + i);
if (exists2) {
System.out.println("bloomkey2" +
"有误判" + (++j2) + "i是" + i);
}
}
CountDownLatch beginCount = new CountDownLatch(1);
beginCount.await();
}