Configuration base = job.getConfiguration();
Map<FormatBundle, Map<Integer, List<Path>>> formatNodeMap = CrunchInputs.getFormatNodeMap(job);
// First, build a map of InputFormats to Paths
for (Map.Entry<FormatBundle, Map<Integer, List<Path>>> entry : formatNodeMap.entrySet()) {
FormatBundle inputBundle = entry.getKey();
Configuration conf = new Configuration(base);
inputBundle.configure(conf);
Job jobCopy = new Job(conf);
InputFormat<?, ?> format = (InputFormat<?, ?>) ReflectionUtils.newInstance(inputBundle.getFormatClass(),
jobCopy.getConfiguration());
for (Map.Entry<Integer, List<Path>> nodeEntry : entry.getValue().entrySet()) {
Integer nodeIndex = nodeEntry.getKey();
List<Path> paths = nodeEntry.getValue();
FileInputFormat.setInputPaths(jobCopy, paths.toArray(new Path[paths.size()]));
// Get splits for each input path and tag with InputFormat
// and Mapper types by wrapping in a TaggedInputSplit.
List<InputSplit> pathSplits = format.getSplits(jobCopy);
for (InputSplit pathSplit : pathSplits) {
splits.add(new CrunchInputSplit(pathSplit, inputBundle.getFormatClass(),
nodeIndex, jobCopy.getConfiguration()));
}
}
}
return splits;