https://medium.com/@mmas/hadoop-streaming-practical-introduction-to-mapreduce-with-python-772bf08070c5

word_counter_mapper.py

#!/usr/bin/env python

import re
import sys


started = False

for line in sys.stdin:
    if started:
        if line.startswith('*** END OF THIS PROJECT'):
            break
        # Filter out some punctuation marks and set to lowercase.
        line = re.sub(r'["?!.,;:()-]', '', line).strip().lower()
        for word in line.split():
            print '%s\t1' % word
    elif line.startswith('*** START OF THIS PROJECT'):
        started = True