Mapreduce程序无法读取输入文件

Question

I have a small csv file on hdfs in this location: "hdfs://csehadoop/user/department.csv" . I use the following code in my mapreduce program to read the contents of the file and store it in a hashmap .

brReader = new BufferedReader(new FileReader("hdfs://csehadoop/user/department.csv"));
while ((strLineRead = brReader.readLine()) != null) {
        String deptFieldArray[] = strLineRead.split(",");
        System.out.println(deptFieldArray[2].toString());
        String businessId = deptFieldArray[2].toString();
        String fields = deptFieldArray[3].toString() + "\t" + deptFieldArray[10].toString();
        DepartmentMap.put(businessId, fields);
}

However, the resultant hashmap ( DepartmentMap ) is always null . How can I rectify this?

Answer 1

You cannot read Hdfs files that way. You must open a FSDataInputStream at the indicated Path. For instance:

Path pt = new Path("hdfs://csehadoop/user/department.csv");
FileSystem fs = FileSystem.get(context.getConfiguration());
BufferedReader br = new BufferedReader(new InputStreamReader(fs.open(pt)));