RcFile存儲和讀取操作

工作中用到了RcFile來存儲和讀取RcFile格式的文件，記錄下。
RcFile是FaceBook開發的一個集行存儲和列存儲的優點於一身，壓縮比更高，讀取列更快，它在MapReduce環境中大規模數據處理中扮演着重要的角色。
讀取操作：

job信息：
	    Job job = new Job();
            job.setJarByClass(類.class);
		//設定輸入文件爲RcFile格式
            job.setInputFormatClass(RCFileInputFormat.class);  
		//普通輸出
            job.setOutputFormatClass(TextOutputFormat.class);
		//設置輸入路徑
            RCFileInputFormat.addInputPath(job, new Path(srcpath));
            //MultipleInputs.addInputPath(job, new Path(srcpath), RCFileInputFormat.class);
		// 輸出
            TextOutputFormat.setOutputPath(job, new Path(respath));
            // 輸出key格式
            job.setOutputKeyClass(Text.class);  
		//輸出value格式
            job.setOutputValueClass(NullWritable.class);  
		//設置mapper類
            job.setMapperClass(ReadTestMapper.class);
		//這裏沒設置reduce，reduce的操作就是讀Text類型文件，因爲mapper已經給轉換了。

            code = (job.waitForCompletion(true)) ? 0 : 1;


// mapper 類

pulic class ReadTestMapper extends Mapper<LongWritable, BytesRefArrayWritable, Text, NullWritable> {

        @Override
        protected void map(LongWritable key, BytesRefArrayWritable value, Context context) throws IOException, InterruptedException {
            // TODO Auto-generated method stub
            Text txt = new Text(); 
		//因爲RcFile行存儲和列存儲，所以每次進來的一行數據，Value是個列簇，遍歷，輸出。
            StringBuffer sb = new StringBuffer();
            for (int i = 0; i < value.size(); i++) {
                BytesRefWritable v = value.get(i);
                txt.set(v.getData(), v.getStart(), v.getLength());
                if(i==value.size()-1){
                    sb.append(txt.toString());
                }else{
                    sb.append(txt.toString()+"\t");
                }
            }
            context.write(new Text(sb.toString()),NullWritable.get());
            }
        }

輸出壓縮爲RcFile格式：

job信息：
            Job job = new Job();
            Configuration conf = job.getConfiguration();
		//設置每行的列簇數
            RCFileOutputFormat.setColumnNumber(conf, 4);
            job.setJarByClass(類.class);

            FileInputFormat.setInputPaths(job, new Path(srcpath));
            RCFileOutputFormat.setOutputPath(job, new Path(respath));

            job.setInputFormatClass(TextInputFormat.class);
            job.setOutputFormatClass(RCFileOutputFormat.class);

            job.setMapOutputKeyClass(LongWritable.class);
            job.setMapOutputValueClass(BytesRefArrayWritable.class);

            job.setMapperClass(OutPutTestMapper.class);

            conf.set("date", line.getOptionValue(DATE));
		//設置壓縮參數
            conf.setBoolean("mapred.output.compress", true);
            conf.set("mapred.output.compression.codec", "org.apache.hadoop.io.compress.GzipCodec");

            code = (job.waitForCompletion(true)) ? 0 : 1;

mapper類：
    public class OutPutTestMapper extends Mapper<LongWritable, Text, LongWritable, BytesRefArrayWritable> {
        @Override
        public void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException {
            String line = value.toString();
            String day = context.getConfiguration().get("date");
            if (!line.equals("")) {
                String[] lines = line.split(" ", -1);
                if (lines.length > 3) {
                    String time_temp = lines[1];
                    String times = timeStampDate(time_temp);
                    String d = times.substring(0, 10);
                    if (day.equals(d)) {
                        byte[][] record = {lines[0].getBytes("UTF-8"), lines[1].getBytes("UTF-8"),lines[2].getBytes("UTF-8"), lines[3].getBytes("UTF-8")};

                        BytesRefArrayWritable bytes = new BytesRefArrayWritable(record.length);

                        for (int i = 0; i < record.length; i++) {
                            BytesRefWritable cu = new BytesRefWritable(record[i], 0, record[i].length);
                            bytes.set(i, cu);
                        }
                        context.write(key, bytes);
                    }
                }
            }
        }

轉載，請超鏈接形式標明文章原始出處和作者。
永久鏈接: http://smallboby.iteye.com/blog/1592531。
感謝。不當之處，請指教。