mapper.py

#!/usr/bin/env python

import sys

def read_input(file):
    for line in file:
        yield float(line)

def main(separator='\t'):
    data = read_input(sys.stdin)
    for num in data:
        print '%f%s%d' % (num, separator, 1)

if __name__ == "__main__":
    main()

reducer.py

#!/usr/bin/env python

import sys

def read_mapper_output(file, separator='\t'):
    for line in file:
        value_text, multi_text = line.rstrip().split(separator, 1)
        yield float(value_text), int(multi_text)

def read_by_two(iterable):
    buf = None
    for v in iterable:
        if buf is None:
            buf = v
        else:
            yield buf, v
            buf = None
    if buf is not None:
        yield buf, None

def main(separator='\t'):
    data = read_mapper_output(sys.stdin, separator=separator)

    for (v1, m1), item2 in read_by_two(data):
        if item2 is None:
            avg, total_m = v1, m1
        else:
            v2, m2 = item2
            avg = float(m1) / (m1 + m2) * v1 + float(m2) / (m1 + m2) * v2
            total_m = m1 + m2
        print "%f%s%d" % (avg, separator, total_m)
        
if __name__ == "__main__":
    main()