def runner(job): multimap = MultiMapper() multimap.add("hdi", primary(parse_hdi)) multimap.add("disasters", secondary(parse_disasters)) opts = [("inputformat", "text"), ("outputformat", "text")] o1 = job.additer(multimap, Reducer, opts=opts)
def runner(job): multimap = MultiMapper() multimap.add("hdi", primary(parse_hdi)) multimap.add("disasters", secondary(parse_disasters)) opts = [ ("inputformat", "text"), ("outputformat", "text"), ] o1 = job.additer(multimap, Reducer, opts=opts)
def runner(job): opts = [("inputformat", "text"), ("outputformat", "text"), ] multimap = MultiMapper() multimap.add("outputgrid", primary(mapper_locations)) multimap.add("metropolitan", secondary(mapper_footfalls)) o1 = job.additer(multimap, Joiner, opts=opts)
def runner(job): opts = [ ("inputformat", "text"), ("outputformat", "text"), ] multimap = MultiMapper() multimap.add("outputgrid", primary(mapper_locations)) multimap.add("metropolitan", secondary(mapper_footfalls)) o1 = job.additer(multimap, Joiner, opts=opts)
def runner(job): multimapper = dumbo.MultiMapper() multimapper.add("hostnames", primary(mapper1)) multimapper.add("logs", secondary(mapper1)) job.additer(multimapper, Reducer1) job.additer(mapper2, reducer2, combiner=reducer2)