Hadoop的SequenceFile读写实例

    SequenceFile可以处理hdfs上大量小文件，它可以作为大量小文件的容器。HDFS和MapReduce是针对大文件优化的，所以通过SequenceFile类型将小文件包装起来可以获得更高效的存储和处理。存储
在SequenceFile中的键和值并不一定是Writable类型，只要能被Serialization序列化和反序列化，任何类型都可以。

　 SequenceFile的优点是：以键值对方式存储、支持压缩、合并大量小文件。

   
        Configuration conf = new Configuration();
        FileSystem fs = FileSystem.get(new URI("hdfs://single32:9000"), conf);
        Path targetPath = new Path("/sfs");
        
        //在HDFS上创建不使用压缩的SequenceFile
        final Option optPath = SequenceFile.Writer.file(targetPath);
        final Option optKeyClass = SequenceFile.Writer.keyClass(Text.class);
        final Option optValueClass = SequenceFile.Writer.valueClass(BytesWritable.class);
        final SequenceFile.Writer writer = SequenceFile.createWriter(conf, optPath, optKeyClass, optValueClass);
        final Collection<File> listFiles = FileUtils.listFiles(new File("/usr/local/"), new String[]{"txt"}, false);
        Text key = null;
        BytesWritable value = null;
        for (File file : listFiles) {
            key = new Text(file.getPath());
            value = new BytesWritable(FileUtils.readFileToByteArray(file));
            writer.append(key, value);
        }
        IOUtils.closeStream(writer);

        //读取HDFS上指定目录下的SequenceFile文件
        final SequenceFile.Reader reader = new SequenceFile.Reader(fs, targetPath, conf);
        final Text outputKey = new Text();
        final BytesWritable outputValue = new BytesWritable();
        while(reader.next(outputKey, outputValue)){
            final File file = new File("/usr/"+outputKey.toString());
            FileUtils.writeByteArrayToFile(file, outputValue.getBytes());
        }
        IOUtils.closeStream(reader);