class Quant_Func_Reducer:
def __call__(self, key, values):
F = values.next()
for v in values:
F += v
yield key, F
class Sum_Mapper:
def __call__(self, records):
F = records.next()[1]
for row in records:
F += row[1]
yield 'F', F
class Sum_Reducer:
def __call__(self, key, values):
F = values.next()
for v in values:
F += v
yield key, F.tolist()
if __name__ == '__main__':
import dumbo
job = dumbo.Job()
job.additer(Quant_Func_Mapper, Quant_Func_Reducer)
job.additer(Sum_Mapper, Sum_Reducer)
job.run()
def filterMain():
job = dumbo.Job()
job.additer(Mapper, reducer, combiner=reducer)
job.additer(FilterMapper, reducer, combiner=reducer)
job.run()