【Hbase】之好友关系（20）

tech2022-09-06 127

文章目录

一、需求二、实现思路三、实现代码四、实验结果

一、需求

需求:

在社交网站，社交 APP 上会存储有大量的用户数据以及用户之间的关系数据。比如： A用户的好友列表会展示出他所有的好友，现有一张 Hbase 表，存储就是当前注册用户的好友关系数据。

使用 Hbase 相关 API 创建一张结构如上的表

删除好友操作实现（好友关系双向，一方删除好友，另一方也会被迫删除好友）

例如：uid1 用户执行删除 uid2 这个好友，则 uid2 的好友列表中也必须删除 uid1

二、实现思路

# rowkey 为用户Id # friends 列族 rowkey friends uid1 uid2 uid3 uid4 uid2 uid1 uid3 创建表和基本数据 # 版本为 1 $ create 'relations', {NAME => 'friends', VERSIONS => '1'} # 插入记录 uid1 ：uid2 uid3 uid4 $ put 'relations', 'uid1', 'friends:uid2', 'uid2' $ put 'relations', 'uid1', 'friends:uid3', 'uid3' $ put 'relations', 'uid1', 'friends:uid4', 'uid4' # 插入记录 uid2 ：uid1 uid3 $ put 'relations', 'uid2', 'friends:uid1', 'uid1' $ put 'relations', 'uid2', 'friends:uid3', 'uid3' hbase(main):015:0> scan 'relations' ROW COLUMN+CELL uid1 column=friends:uid2, timestamp=1598881670153, value=uid2 uid1 column=friends:uid3, timestamp=1598881677381, value=uid3 uid1 column=friends:uid4, timestamp=1598881683645, value=uid4 uid2 column=friends:uid1, timestamp=1598881698725, value=uid1 uid2 column=friends:uid3, timestamp=1598881690459, value=uid3 2 row(s) in 0.0520 seconds 编写 Observer 协处理器

通过 Observer 协处理器捕捉到 relations 删除数据时, 对应的 rowkey也删除。

上传 HDFS $ scp -P 22 hadoop-practice-1.0-SNAPSHOT.jar root@172.16.64.121:/root $ mv hadoop-practice-1.0-SNAPSHOT.jar processor4.jar [root@linux121 ~]# hadoop fs -mkdir /processor [root@linux121 ~]# hadoop fs -put processor4.jar /processor 挂载协处理器 $ alter 'relations',METHOD =>'table_att','Coprocessor'=>'hdfs://linux121:9000/processor/processor5.jar|com.donaldy.hbase.homework.DeleteRelationsProcessor|1001|' $ desc 'relations' hbase(main):024:0> desc 'relations' Table relations is ENABLED relations, {TABLE_ATTRIBUTES => {coprocessor$1 => 'hdfs://linux121:9000/processor/processor5.jar|com.donaldy.hbase.homework.DeleteRelationsProcessor|1001|'} COLUMN FAMILIES DESCRIPTION {NAME => 'friends', BLOOMFILTER => 'ROW', VERSIONS => '1', IN_MEMORY => 'false', KEEP_DELETED_CELLS => 'FALSE', DATA_BLOCK_ENCODING => 'NONE', TTL => 'FOREVER', COMPRESSION => 'NONE', MIN_VERSIONS => '0', BLOCKCACHE => 'true', BLOCKSIZE => '65536', REPLICATION_SCOPE => '0'} 1 row(s) in 0.0470 seconds 执行删除操作 $ delete 'relations', 'uid1', 'friends:uid2'

三、实现代码

API 创建 @Test public void createTable() throws IOException { admin = (HBaseAdmin) conn.getAdmin(); HTableDescriptor relations = new HTableDescriptor(TableName.valueOf("relations")); relations.addFamily(new HColumnDescriptor("friends")); admin.createTable(relations); } @Test public void putRelationData() throws IOException { Table table = conn.getTable(TableName.valueOf("relations")); List<Put> puts = new ArrayList<Put>(); // rowkey user1 Put put = new Put(Bytes.toBytes("user1")); put.addColumn(Bytes.toBytes("friends"),Bytes.toBytes("user2"),Bytes.toBytes("user2")); put.addColumn(Bytes.toBytes("friends"),Bytes.toBytes("user3"),Bytes.toBytes("user3")); put.addColumn(Bytes.toBytes("friends"),Bytes.toBytes("user4"),Bytes.toBytes("user4")); puts.add(put); // rowkey user2 Put put2 = new Put(Bytes.toBytes("user2")); put.addColumn(Bytes.toBytes("friends"),Bytes.toBytes("user1"),Bytes.toBytes("user1")); put.addColumn(Bytes.toBytes("friends"),Bytes.toBytes("user3"),Bytes.toBytes("user3")); puts.add(put2); table.put(puts); table.close(); } 协处理器 package com.donaldy.hbase.homework; import org.apache.hadoop.hbase.Cell; import org.apache.hadoop.hbase.CellUtil; import org.apache.hadoop.hbase.TableName; import org.apache.hadoop.hbase.client.Delete; import org.apache.hadoop.hbase.client.Durability; import org.apache.hadoop.hbase.client.HTableInterface; import org.apache.hadoop.hbase.coprocessor.BaseRegionObserver; import org.apache.hadoop.hbase.coprocessor.ObserverContext; import org.apache.hadoop.hbase.coprocessor.RegionCoprocessorEnvironment; import org.apache.hadoop.hbase.regionserver.wal.WALEdit; import org.apache.hadoop.hbase.util.Bytes; import org.apache.hadoop.hbase.util.CollectionUtils; import java.io.IOException; import java.util.List; /** * uid1 解除 uid2关系， uid2 同时解除 uid1 关系 * * uid1 ： currUser * uid2 ： otherUser * * @author donald * @date 2020/09/01 */ public class DeleteRelationsProcessor extends BaseRegionObserver { @Override public void postDelete(ObserverContext<RegionCoprocessorEnvironment> e, Delete delete, WALEdit edit, Durability durability) throws IOException { final HTableInterface relations = e.getEnvironment().getTable(TableName.valueOf("relations")); List<Cell> cells = delete.getFamilyCellMap().get(Bytes.toBytes("friends")); if (CollectionUtils.isEmpty(cells)) { relations.close(); return; } // 获取 uid1 第一个 column Cell cell = cells.get(0); // 创建 uid2，并设置需要删除的 column Delete otherUserDelete = new Delete(CellUtil.cloneQualifier(cell)); otherUserDelete.addColumns(Bytes.toBytes("friends"), CellUtil.cloneRow(cell)); relations.delete(otherUserDelete); // 关闭 table 对象 relations.close(); } }

四、实验结果

hbase(main):026:0> scan 'relations' ROW COLUMN+CELL uid1 column=friends:uid2, timestamp=1598885427534, value=uid2 uid1 column=friends:uid3, timestamp=1598881677381, value=uid3 uid1 column=friends:uid4, timestamp=1598881683645, value=uid4 uid2 column=friends:uid1, timestamp=1598881698725, value=uid1 uid2 column=friends:uid3, timestamp=1598881690459, value=uid3 2 row(s) in 0.0280 seconds hbase(main):0027:0> scan 'relations' ROW COLUMN+CELL uid1 column=friends:uid3, timestamp=1598956561339, value=uid3 uid1 column=friends:uid4, timestamp=1598956567193, value=uid4 uid2 column=friends:uid3, timestamp=1598956599905, value=uid3 2 row(s) in 0.1150 seconds

截图如下：

最新回复(0)