If you use the new API, you can access the MapContext object in the setup method of the mapper. Then, you can get the input split with MapContext#getInputSplit(), cast it to FileSplit and obtain the path of the file the current split is part of through the FileSplit#getPath() method. All records of the mapper instance will be part of that split so you only need to get the information once per map task in the setup method. Am 14.12.2012 19:46 schrieb "Xiaowei Li" <[EMAIL PROTECTED]>:
> hi, > > my MR job has multiple inputs, and I wanna how to differentiate which > input directory current row/record comes from in my mapper? > > thanks! > -xw >
All projects made searchable here are trademarks of the Apache Software Foundation.
Service operated by Sematext