movies count

movies_count.py

#-*-coding: utf-8 -*-

'''
This module computes the number of movies rated by each
user.

'''

__author__ = 'Marcel Caraciolo <[email protected]>'

from mrjob.job import MRJob


class MoviesCount(MRJob):

    def mapper(self, key, line):
        """
        Mapper: send score from a single movie to
        other movies
        """
        #user_id, item_id, rating = line.split('|')
        #yield item_id, (user_id, rating, 1)
        user_id, item_id, rating = line.split('|')
        yield item_id, (user_id, float(rating))

    def reducer(self, movie, values):
        #yield(movie, sum(values))
        total = 0
        final = []
        for user_id, rating in values:
            total += 1
            final.append((user_id, rating))

        for user_id, rating in final:
            yield '%s|%s|%.2f|%d' % (user_id, movie, rating, total), None

if __name__ == '__main__':
    MoviesCount.run()