Lesson 5 - MapReduce. Mapper and reducer with aadhaar data.

aadhaar_generated_mapper.py

import csv
import sys
import string

def mapper(input=sys.stdin):
    reader = csv.DictReader(input)
    for row in reader: #cycle through lines of code
        print "%s\t%s" % (row["District"], row["Aadhaar generated"])
        
mapper()

aadhaar_generated_reducer.py

import sys

def reducer(input=sys.stdin):
    elems = (line.strip().split("\t") for line in input if len(line.strip()) > 0)
    aadhaar_generated = 0
    old_key = None
    for (key, value) in elems:    
        if old_key != key:
            if old_key is not None:
                print "%s\t%.1f" % (old_key, aadhaar_generated)
            old_key = key
            aadhaar_generated = 0
        aadhaar_generated += int(value)
    if old_key is not None:
        print "%s\t%.1f" % (old_key, aadhaar_generated)
    
reducer()