I think you will need to implement a RecordReader/InputFormat of your own for this and use it with a LoadFunc. Not sure if Hadoop has a Reader that you could re-use for this.
How do you handle the case when a file exceeds block size? On Thu, Jun 21, 2012 at 2:34 PM, Jonathan Coveney <jcove...@gmail.com>wrote: > It can even be a bytearray. Basically I have a bunch of files, and I want > one file -> one row. Is there an easy way to do this? Or will I need to > provide a special fileinputformat etc? >