map获取输入文件名称和GenericOptionsParser使用

该文章中实现业务需求的方法不可取(其实可以使用hadoop提供的多输入来实现)，但是该例子去能很好的展现在map函数中怎么拿到输入文件的名称：

原文地址：http://datamachine.iteye.com/blog/1963256

选择Hadoop，低成本和高扩展性是主要原因，但但它的开发效率实在无法让人满意。 
    以关联计算为例。 
    假设：HDFS上有2个文件，分别是客户信息和订单信息，customerID是它们之间的关联字段。如何进行关联计算，以便将客户名称添加到订单列表中？ 
    一般方法是：输入2个源文件。根据文件名在Map中处理每条数据，如果是Order，则在foreign key上加标记”O”，形成combined key；如果是Customer则做标记”C”。Map之后的数据按照key分区，再按照combined key分组排序。最后在reduce中合并结果再输出。 
实现代码： 
public static class JMapper extends Mapper<LongWritable, Text, TextPair, Text> { 
    //mark every row with "O" or "C" according to file name 
    @Override 
    protected void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException { 
    String pathName = ((FileSplit) context.getInputSplit()).getPath().toString(); 
    if (pathName.contains("order.txt")) {//identify order by file name 
            String values[] = value.toString().split("\t"); 
            TextPair tp = new TextPair(new Text(values[1]), new Text("O"));//mark with "O" 
            context.write(tp, new Text(values[0] + "\t" + values[2])); 
        } 
   if (pathName.contains("customer.txt")) {//identify customer by file name 
           String values[] = value.toString().split("\t"); 
           TextPair tp = new TextPair(new Text(values[0]), new Text("C"));//mark with "C" 
           context.write(tp, new Text(values[1])); 
        } 
    } 
} 
public static class JPartitioner extends Partitioner<TextPair, Text> { 
    //partition by key, i.e. customerID 
    @Override 
    public int getPartition(TextPair key, Text value, int numParititon) { 
        return Math.abs(key.getFirst().hashCode() * 127) % numParititon; 
    } 
} 
public static class JComparator extends WritableComparator { 
    //group by muti-key 
    public JComparator() { 
        super(TextPair.class, true); 
    } 
    @SuppressWarnings("unchecked") 
    public int compare(WritableComparable a, WritableComparable b) { 
        TextPair t1 = (TextPair) a; 
        TextPair t2 = (TextPair) b; 
        return t1.getFirst().compareTo(t2.getFirst()); 
    } 
} 
public static class JReduce extends Reducer<TextPair, Text, Text, Text> { 
    //merge and output 
    protected void reduce(TextPair key, Iterable<Text> values, Context context) throws IOException,InterruptedException { 
    Text pid = key.getFirst(); 
    String desc = values.iterator().next().toString(); 
    while (values.iterator().hasNext()) { 
        context.write(pid, new Text(values.iterator().next().toString() + "\t" + desc)); 
   } 
    } 
} 
public class TextPair implements WritableComparable<TextPair> { 
    //make muti-key 
    private Text first; 
    private Text second; 
    public TextPair() { 
        set(new Text(), new Text()); 
    } 
    public TextPair(String first, String second) { 
        set(new Text(first), new Text(second)); 
    } 
    public TextPair(Text first, Text second) { 
        set(first, second); 
    } 
    public void set(Text first, Text second) { 
  this.first = first; 
  this.second = second; 
    } 
    public Text getFirst() { 
  return first; 
    } 
    public Text getSecond() { 
  return second; 
    } 
    public void write(DataOutput out) throws IOException { 
  first.write(out); 
  second.write(out); 
    } 
    public void readFields(DataInput in) throws IOException { 
  first.readFields(in); 
  second.readFields(in); 
    } 
    public int compareTo(TextPair tp) { 
  int cmp = first.compareTo(tp.first); 
  if (cmp != 0) { 
       return cmp; 
  } 
    return second.compareTo(tp.second); 
    } 
} 
public static void main(String agrs[]) throws IOException, InterruptedException, ClassNotFoundException { 
    //job entrance 
    Configuration conf = new Configuration(); 
    GenericOptionsParser parser = new GenericOptionsParser(conf, agrs); 
    String[] otherArgs = parser.getRemainingArgs(); 
    if (agrs.length < 3) { 
   System.err.println("Usage: J <in_path_one> <in_path_two> <output>"); 
   System.exit(2); 
    } 
    Job job = new Job(conf, "J"); 
    job.setJarByClass(J.class);//Join class 
    job.setMapperClass(JMapper.class);//Map class 
    job.setMapOutputKeyClass(TextPair.class);//Map output key class 
    job.setMapOutputValueClass(Text.class);//Map output value class 
    job.setPartitionerClass(JPartitioner.class);//partition class 
    job.setGroupingComparatorClass(JComparator.class);//condition group class after partition 
    job.setReducerClass(Example_Join_01_Reduce.class);//reduce class 
    job.setOutputKeyClass(Text.class);//reduce output key class 
    job.setOutputValueClass(Text.class);//reduce ouput value class 
    FileInputFormat.addInputPath(job, new Path(otherArgs[0]));//one of source files 
    FileInputFormat.addInputPath(job, new Path(otherArgs[1]));//another file 
    FileOutputFormat.setOutputPath(job, new Path(otherArgs[2]));//output path 
    System.exit(job.waitForCompletion(true) ? 0 : 1);//run untill job ends 
} 

    不能直接使用原始数据，而是要搞一堆代码处理标记，并绕过MapReduce原本的架构，最后从底层设计并计算数据之间的关联关系。这还是最简单的关联计算，如果用MapReduce进行多表关联或逻辑更复杂的关联计算，复杂度会呈几何级数递增。