一起聊聊redis的scan操作

• redis的发布订阅功能
• redis消息队列
• redis的pipeline
• redis的scan操作

序

在redis的db存在大量key或者db里头的某个set、zset、hash里头的元素非常多的话，用普通的get all操作很可能导致redis因为这个操作阻塞了，导致不能响应其他操作，特别是在高并发、海量数据的背景下，这个问题显得尤其严重。那么能不能像数据库那样有个分页的功能呢，答案就是scan操作。本文主要展示怎么在redis-cli以及SpringDataRedis中的使用。【推荐：redis视频教程】

scan语法

scan之后返回两部分，第一部分是下次scan的参数，第二部分就是scan出来的项

作用对象(db、set、zset、hash)

• db(key)

127.0.0.1:6379> scan 0
1) "120"
2)  1) "articleMap:63"
    2) "articleMap:37"
    3) "counter:__rand_int__"
    4) "articleMap:60"
    5) "tagSet:tag5"
    6) "articleMap:80"
    7) "messageCache~keys"
    8) "mymap"
    9) "articleMap:46"
   10) "articleMap:55"
127.0.0.1:6379> scan 120
1) "28"
2)  1) "articleMap:17"
    2) "tagSet:tag1"
    3) "articleMap:18"
    4) "articleMap:81"
    5) "\xac\xed\x00\x05t\x00\btest-cas"
    6) "articleMap:51"
    7) "articleMap:94"
    8) "articleMap:26"
    9) "articleMap:71"
   10) "user-abcde"

• set(value)

127.0.0.1:6379> sscan myset 0
1) "3"
2)  1) "m"
    2) "j"
    3) "c"
    4) "h"
    5) "f"
    6) "i"
    7) "a"
    8) "g"
    9) "n"
   10) "e"
   11) "b"
127.0.0.1:6379> sscan myset 3
1) "0"
2) 1) "l"
   2) "k"
   3) "d"

• zset(value & score)

127.0.0.1:6379> zscan sortset 0
1) "0"
2) 1) "tom"
   2) "89"
   3) "jim"
   4) "90"
   5) "david"
   6) "100"

• hash(key & value)

127.0.0.1:6379> hscan mymap 0
1) "0"
2)  1) "name"
    2) "codecraft"
    3) "email"
    4) "pt@g.cn"
    5) "age"
    6) "20"
    7) "desc"
    8) "hello"
    9) "sex"
   10) "male"

SCAN的额外参数

• count(指定每次取多少条)

127.0.0.1:6379> scan 0 count 5
1) "240"
2) 1) "articleMap:63"
   2) "articleMap:37"
   3) "counter:__rand_int__"
   4) "articleMap:60"
   5) "tagSet:tag5"

• match(匹配key)

127.0.0.1:6379> scan 0 match article*
1) "120"
2) 1) "articleMap:63"
   2) "articleMap:37"
   3) "articleMap:60"
   4) "articleMap:80"
   5) "articleMap:46"
   6) "articleMap:55"

RedisTemplate操作

遍历数据库key

@Test
    public void scanDbKeys(){
        template.execute(new RedisCallback<Iterable<byte[]>>() {
            @Override
            public Iterable<byte[]> doInRedis(RedisConnection connection) throws DataAccessException {

                List<byte[]> binaryKeys = new ArrayList<byte[]>();

                Cursor<byte[]> cursor = connection.scan(ScanOptions.scanOptions().count(5).build());
                while (cursor.hasNext()) {
                    byte[] key = cursor.next();
                    binaryKeys.add(key);
                    System.out.println(new String(key, StandardCharsets.UTF_8));
                }

                try {
                    cursor.close();
                } catch (IOException e) {
                    // do something meaningful
                }

                return binaryKeys;
            }
        });
    }

遍历set

/**
     * sadd myset a b c d e f g h i j k l m n
     */
    @Test
    public void scanSet(){
        Cursor<String> cursor = template.opsForSet().scan("myset",ScanOptions.NONE);
        while (cursor.hasNext()){
            System.out.println(cursor.next());
        }
    }

遍历zset

/**
     * zadd sortset 89 tom 90 jim 100 david
     */
    @Test
    public void scanZSet(){
        Cursor<ZSetOperations.TypedTuple<String>> cursor = template.opsForZSet().scan("sortset",ScanOptions.NONE);
        while (cursor.hasNext()){
            ZSetOperations.TypedTuple<String> item = cursor.next();
            System.out.println(item.getValue() + ":" + item.getScore());
        }
    }

遍历hash

/**
     *  hset mymap name "codecraft"
     *  hset mymap email "pt@g.cn"
     *  hset mymap age 20
     *  hset mymap desc "hello"
     *  hset mymap sex "male"
     */
    @Test
    public void scanHash(){
        Cursor<Map.Entry<Object, Object>> curosr = template.opsForHash().scan("mymap", ScanOptions.NONE);
        while(curosr.hasNext()){
            Map.Entry<Object, Object> entry = curosr.next();
            System.out.println(entry.getKey()+":"+entry.getValue());
        }
    }