'''
    normalize_cities.py

    A throwaway script for turning the populations table (city, year, population)
    into two tables:

       pops: (city_id, year, population)
       cities: (id, name)
'''

import csv

cities = {}
populations = {}

# Collect the data
reader = csv.reader(open('poprows.csv'))
for row in reader:
    (city, year, population) = row
    if city not in cities:
        cities[city] = len(cities)
    assert (city, year) not in populations
    populations[(city, year)] = population

# Print it out, using city i
cityRows = [(cities[city] + 1, city) for city in cities]
cityRows.sort()
writer = csv.writer(open('cities.csv', 'w'))
for row in cityRows:
    writer.writerow(row)

populationRows = [(cities[city], year, populations[(city, year)]) for (city, year) in populations]
populationRows.sort()
writer = csv.writer(open('pops.csv', 'w'))
for row in populationRows:
    writer.writerow(row)