Q2 popular products - Moloco

#use: 
#get_product_ranking('sample_data.tsv', 1)

import json
from collections import defaultdict

def get_product_ranking(file, rank_method=1):
    # Outputs most popular product(s) based on rank method
    # rank_method = 1 is based on the unique number of users who purchased each product
    # rank_method = 2 is based on the total quantity of each product sold
    rank_method_function_map = {
        1: product_unique_purchases,
        2: product_sold
    }
    method = rank_method_function_map[rank_method]
    if not method:
        return Error('choose a rank_method that is supported')
    with open(file, 'r') as fd:
        return method(fd)

def product_unique_purchases(fd):
    product_to_unique_purchases = defaultdict(set)
    top_products = set()
    top_product_count = 0
    for line in fd:
        data = json.loads(line.rstrip())
        current_product = data['product_id']
        product_to_unique_purchases[current_product].add(data['user_id'])
        current_product_unique_purchases = len(product_to_unique_purchases[current_product])
        if current_product_unique_purchases == top_product_count:
            top_products.add(current_product)
        elif current_product_unique_purchases > top_product_count:
            top_products = {current_product}
            top_product_count = current_product_unique_purchases
    print("Most popular product(s) based on the number of purchasers: {}".format(top_products))
        
def product_sold(fd):
    product_to_purchase_count = defaultdict(lambda :0)
    top_products = set()
    top_product_count = 0
    for line in fd:
        data = json.loads(line.rstrip())
        current_product = data['product_id']
        quantity = data['quantity']
        product_to_purchase_count[current_product] += quantity
        current_product_purchase_count = product_to_purchase_count[current_product]
        if current_product_purchase_count == top_product_count:
            top_products.add(current_product)
        elif current_product_purchase_count > top_product_count:
            top_product_count = current_product_purchase_count
            top_products = {current_product}
    print("Most popular product(s) based on the number of goods sold: {}".format(top_products))