本文为大家分享了Redis支持多人多聊天室功能的设计代码,供大家参考,具体内容如下
设计原理
左边的一个数据域,代表两个聊天室,聊天室id分别是827,729
在聊天室827里,有2个人,分别是jason22,jeff24他们分别已经阅读过聊天室内的id为5和6的消息
右边的一个数据域,代表了用户在不同的聊天室,jason22参与了827与729聊天室,在这两个聊天室里,他分别阅读到了id为5和id为10的消息
另外827聊天室内id为5的消息与729聊天室内id为5的消息不一样。
同时还有三个域
msgs:chatid
这是一个zset,有序集合,member是消息体,score是消息id
代表的是某个聊天室内已经发出的消息
另外 这里面存的是有用的消息,已经被所有人都阅读的消息就会被删除
ids:chatid
是一个String型的数据,里面放的是最新的消息的编号(发消息时,自增这个字段,即可获得最新的值)
ids:chat:
是一个String型的数据,里面放的是最新的聊天室的编号(创建聊天室时,自增这个字段)
代码
OK 开始看代码
public String createChat(Jedis conn, String sender, Set<String> recipients, String message) {
//启动的时候redis里是没有ids:chat:这个键的
//自增之后返回1
String chatId = String.valueOf(conn.incr("ids:chat:"));
return createChat(conn, sender, recipients, message, chatId);
}
public String createChat( Jedis conn, String sender,
Set<String> recipients, String message, String chatId){
//自己发的消息 自己也能接受到
recipients.add(sender);
Transaction trans = conn.multi();
for (String recipient : recipients){
//聊天室的成员 最开始时 都阅读的是0号信息
trans.zadd("chat:" + chatId, 0, recipient);
//记录每个人参加的聊天室
trans.zadd("seen:" + recipient, 0, chatId);
}
trans.exec();
return sendMessage(conn, chatId, sender, message);
}
public String sendMessage(Jedis conn, String chatId, String sender, String message) {
//锁住聊天室 为啥? 人员变动了咋办
//这个acquireLock见上一章
String identifier = acquireLock(conn, "chat:" + chatId);
if (identifier == null){
throw new RuntimeException("Couldn't get the lock");
}
try {
//给要发布的消息设定一个最新的编号 第一次时 返回的是1
long messageId = conn.incr("ids:" + chatId);
HashMap<String,Object> values = new HashMap<String,Object>();
values.put("id", messageId);
values.put("ts", System.currentTimeMillis());
values.put("sender", sender);
values.put("message", message);
String packed = new Gson().toJson(values);
//某个聊天室的消息列表
//最旧的消息----消息json
//默认的zset是按照score的值从小到大排序
conn.zadd("msgs:" + chatId, messageId, packed);
}finally{
releaseLock(conn, "chat:" + chatId, identifier);
}
return chatId;
}
发消息现在就OK了,剩下的就是用户去拉取未读的消息了。这个比较麻烦,恩,相当的麻烦
@SuppressWarnings("unchecked")
public List<ChatMessages> fetchPendingMessages(Jedis conn, String recipient) {
//获得用户在各个聊天室 已经看到的最新消息的id
//有几个聊天室 seenSet的size就是几
Set<Tuple> seenSet = conn.zrangeWithScores("seen:" + recipient, 0, -1);
List<Tuple> seenList = new ArrayList<Tuple>(seenSet);
Transaction trans = conn.multi();
for (Tuple tuple : seenList){
String chatId = tuple.getElement();
int seenId = (int)tuple.getScore();
//获取每个聊天室里 未读的所有消息
//min 和 max 可以是 -inf 和 +inf
trans.zrangeByScore("msgs:" + chatId, String.valueOf(seenId + 1), "inf");
}
//我参加了几个聊天室 results的长度就是几
List<Object> results = trans.exec();
//com.google.gson.Gson jar包自己下载吧
Gson gson = new Gson();
Iterator<Tuple> seenIterator = seenList.iterator();
Iterator<Object> resultsIterator = results.iterator();
//用户最后成功拉取的未读消息 存放在chatMessages
List<ChatMessages> chatMessages = new ArrayList<ChatMessages>();
List<Object[]> seenUpdates = new ArrayList<Object[]>();
List<Object[]> msgRemoves = new ArrayList<Object[]>();
//这个大的while循环 用户参与了几个聊天室 就循环几次
while (seenIterator.hasNext()){
Tuple seen = seenIterator.next();
Set<String> messageStrings = (Set<String>)resultsIterator.next();
if (messageStrings.size() == 0){
//没有未读的消息
continue;
}
//代码运行到这里
//说明 我在某个聊天室 还有未读的消息
//seedid记录我已经拉取到的消息 初始为0
int seenId = 0;
//当前处理的是哪个聊天室
String chatId = seen.getElement();
List<Map<String,Object>> messages = new ArrayList<Map<String,Object>>();
//我在聊天室未读的消息列表
for (String messageJson : messageStrings){
Map<String,Object> message = (Map<String,Object>)gson.fromJson(
messageJson, new TypeToken<Map<String,Object>>(){}.getType());
int messageId = ((Double)message.get("id")).intValue();
if (messageId > seenId){
seenId = messageId;
}
message.put("id", messageId);
//加入到成功拉取的列表里
messages.add(message);
}
//更新我在这个聊天室读到的最新消息
conn.zadd("chat:" + chatId, seenId, recipient);
//记录我在某个聊天室读到的最新记录
seenUpdates.add(new Object[]{"seen:" + recipient, seenId, chatId});
//取出第0个member-score
Set<Tuple> minIdSet = conn.zrangeWithScores("chat:" + chatId, 0, 0);
//为啥删除呢? 每个聊天室是一个zset表 第一条记录代表的就是 所有用户至少都读了的消息
if (minIdSet.size() > 0){
Tuple tuple=minIdSet.iterator().next();
System.out.println("要删除的 tuple:"+tuple.getElement()+"--"+tuple.getScore());
msgRemoves.add(new Object[]{"msgs:" + chatId, tuple.getScore()});
}
chatMessages.add(new ChatMessages(chatId, messages));
}
trans = conn.multi();
for (Object[] seenUpdate : seenUpdates){
trans.zadd(
(String)seenUpdate[0],
(Integer)seenUpdate[1],
(String)seenUpdate[2]);
}
for (Object[] msgRemove : msgRemoves){
trans.zremrangeByScore(
(String)msgRemove[0], 0, ((Double)msgRemove[1]).intValue());
}
trans.exec();
//返回的是我这次拉取获得的 最新的消息
return chatMessages;
}
OK,咱们看看测试代码:
package redisinaction;
import java.util.Arrays;
import java.util.HashSet;
import java.util.Iterator;
import java.util.List;
import java.util.Map;
import java.util.Set;
import org.junit.BeforeClass;
import org.junit.Test;
import jedis.redis_in_action.Chapter06;
import jedis.redis_in_action.Chapter06.ChatMessages;
import redis.clients.jedis.Jedis;
import redis.clients.jedis.Tuple;
public class Chapter06Test {
static Jedis conn = null;
static Chapter06 c=null;
@BeforeClass
public static void initConn(){
System.out.println("test before");
conn = new Jedis("10.150.0.80");
conn.auth("dlf123123");
c=new Chapter06();
}
@Test
public void testMultiRecipientMessaging() {
System.out.println("n----- testMultiRecipientMessaging -----");
conn.del("ids:chat:", "msgs:1", "ids:1", "seen:joe", "seen:jeff", "seen:jenny");
System.out.println("Let's create a new chat session with some recipients...");
Set<String> recipients = new HashSet<String>();
recipients.add("jeff");
recipients.add("jenny");
String chatId = c.createChat(conn, "joe", recipients, "message 1");
System.out.println("Now let's send a few messages...");
for (int i = 2; i < 5; i++){
c.sendMessage(conn, chatId, "joe", "message " + i);
}
System.out.println();
System.out.println("看看消息库");
//消息库里的所有消息
Set<Tuple> messageFromBase=conn.zrangeWithScores("msgs:"+chatId, 0, -1);
Iterator<Tuple> iterator=messageFromBase.iterator();
while(iterator.hasNext()){
Tuple tuple=iterator.next();
System.out.println(tuple.getElement()+" -- "+tuple.getScore());
}
System.out.println("And let's get the messages that are waiting for jeff and jenny...");
List<ChatMessages> r1 = c.fetchPendingMessages(conn, "jeff");
List<ChatMessages> r2 = c.fetchPendingMessages(conn, "jenny");
//当我拉取了joe的未读信息后 就会删除msgs:1里面的信息
//为什么?想明白了么?
List<ChatMessages> r3 = c.fetchPendingMessages(conn, "joe");
System.out.println("They are the same? " + r1.equals(r2));
System.out.println("Those messages are:");
for(ChatMessages chat : r1){
System.out.println(" chatId: " + chat.chatId);
System.out.println(" messages:");
for(Map<String,Object> message : chat.messages){
System.out.println(" " + message);
}
}
System.out.println("看看还有没");
messageFromBase=conn.zrangeWithScores("msgs:"+chatId, 0, -1);
iterator=messageFromBase.iterator();
while(iterator.hasNext()){
Tuple tuple=iterator.next();
System.out.println(tuple.getElement()+" -- "+tuple.getScore());
}
conn.del("ids:chat:", "msgs:1", "ids:1", "seen:joe", "seen:jeff", "seen:jenny");
}
}
搞定了,大家不妨把代码复制一份,自己看看
下面的是测试的结果
test before
----- testMultiRecipientMessaging -----
Let's create a new chat session with some recipients...
Now let's send a few messages...
看看消息库
{"sender":"joe","id":1,"message":"message 1","ts":1477276890018} -- 1.0
{"sender":"joe","id":2,"message":"message 2","ts":1477276890113} -- 2.0
{"sender":"joe","id":3,"message":"message 3","ts":1477276890115} -- 3.0
{"sender":"joe","id":4,"message":"message 4","ts":1477276890116} -- 4.0
And let's get the messages that are waiting for jeff and jenny...
要删除的 tuple:jenny--0.0
要删除的 tuple:joe--0.0
要删除的 tuple:jeff--4.0
They are the same? true
Those messages are:
chatId: 1
messages:
{sender=joe, id=1, message=message 1, ts=1.477276890018E12}
{sender=joe, id=2, message=message 2, ts=1.477276890113E12}
{sender=joe, id=3, message=message 3, ts=1.477276890115E12}
{sender=joe, id=4, message=message 4, ts=1.477276890116E12}
看看还有没
以上就是本文的全部内容,希望对大家的学习有所帮助,也希望大家多多支持编程网。