用mapreduce实现从hbase导出到hdfs,实现一个工具类,能够支持任意表 任意列 任意路径导出,并且支持表头

分析:

1、由于是任意列 任意表 任意路径,我们很容易想到是参数传入,参数传入后怎么去获得参数,根据我们以往的经验就是通过args[]来获取,但是在mapper或者是reducer中,我们不能直接将参数传入,因为map 和reduce是通过反射机制来创建的,对于传入的参数我们不能直接使用;我们发现在map和reduce有一个参数context,此类中包含很多的信息,例如configuration,并且configuration 还有set()方法,因此 我们可以将参数传给conf,然后由context拿到conf,进而拿到参数/

2、表头在第一行,有且只有一行,因此可以将表头的信息写入mapper 的setup()方法,因为它只进行一次操作

代码:

定义最后的程序运行为 hadoop jar **.jar t1 /tt "f1:c1|f1:c2|f2:c3"

对"f1:c1|f1:c2|f2:c3"进行拆分,首先按照"|"进行拆分,在java中 split("\|"),因为|是转义字符,然后对每个f:c 再按照":"拆分,此时我们就拿到了单独的列族和列

定义自己的mapper函数(由于我们只是将数据读取出来,所以不用写reducer)

public static class MyMapper extends TableMapper<Text, Text> {

		@Override
		protected void setup(Context context) throws IOException,
				InterruptedException {
			String familyscolumns = context.getConfiguration().get(
					"familys:columns");
			String[] familycolumns = familyscolumns.split("\|");
			String familycolumn = "";
			for (String fc : familycolumns) {
				familycolumn += fc + "	";
			}
			// 在第一行增加header 行健 列族:列
			context.write(new Text("rowkey"), new Text(familycolumn));
		}

		Text k2 = new Text();
		Text v2 = new Text();

		@Override
		protected void map(ImmutableBytesWritable key, Result value,
				Context context) throws IOException, InterruptedException {
			k2.set(key.get());
			String familyscolumns = context.getConfiguration().get(
					"familys:columns");
			String[] familycolumns = familyscolumns.split("\|");
			String familycolumn="";
			Cell columnLatestCell1=null;
			for (String string : familycolumns) {
				String[] fc = string.split(":");
				 columnLatestCell1 = value.getColumnLatestCell(
						 fc[0].getBytes(), fc[1].getBytes());
				if (columnLatestCell1 != null) {
					familycolumn+=new String(columnLatestCell1.getValue())+"	";
				} else{
					familycolumn+="	";
				}
			
			}
			v2.set(new String(familycolumn));
			context.write(k2, v2);
		}

	}

  然后在客户端提交job

并把参数写入conf

Configuration conf = HBaseConfiguration.create();
		conf.set("table", args[0]);
		conf.set("hdfsPath", args[1]);
		conf.set("familys:columns", args[2]);

  

用mr实现将hadoop中的数据导入到hbase(我之前一直纠结是写在map还是reduce,其实无所谓啦)

public static void main(String[] args) throws Exception {
		PropertyConfigurator.configure(ClassLoader
				.getSystemResource("log4j.properties"));
		final Configuration configuration = new Configuration();
		// 设置zookeeper
		configuration.set("hbase.zookeeper.quorum", "hadoop");
		// 设置hbase表名称
		configuration.set(TableOutputFormat.OUTPUT_TABLE, "t2");
		// 将该值改大,防止hbase超时退出
		configuration.set("dfs.socket.timeout", "180000");

		final Job job = Job.getInstance(configuration,
				MR2HB.class.getSimpleName());
		TableMapReduceUtil.addDependencyJars(job);
		job.setJarByClass(MR2HB.class);

		job.setMapperClass(MyMapper.class);
		job.setNumReduceTasks(0);
		// 设置map的输出,不设置reduce的输出类型
		job.setMapOutputKeyClass(LongWritable.class);
		job.setMapOutputValueClass(Text.class);

		job.setInputFormatClass(TextInputFormat.class);
		// 不再设置输出路径,而是设置输出格式类型
		job.setOutputFormatClass(TableOutputFormat.class);

		FileInputFormat.setInputPaths(job, "hdfs://hadoop:9000/part-m-00000");

		job.waitForCompletion(true);

	}

	public static class MyMapper extends
			Mapper<LongWritable, Text, NullWritable, Put> {
		@Override
		protected void map(LongWritable key, Text value,
				org.apache.hadoop.mapreduce.Mapper.Context context)
				throws IOException, InterruptedException {
			String[] split = value.toString().split("	");
			Log.info(split[0]);

			Put put = new Put(Bytes.toBytes(split[0]));
			put.add(Bytes.toBytes("f1"), Bytes.toBytes("age"),
					Bytes.toBytes(split[1]));
			if (split.length == 4) {
				put.add(Bytes.toBytes("f2"), Bytes.toBytes("address"),
						Bytes.toBytes(split[2]));
				put.add(Bytes.toBytes("f1"), Bytes.toBytes("name"),
						Bytes.toBytes(split[3]));
			}
			context.write(NullWritable.get(), put);
		}

	}

  

原文地址:https://www.cnblogs.com/ggbond1988/p/4838129.html