process the lax.json File

process the lax.json File

import json
import sys
import math


def convert_item(item):
    year = str(item['year']).split('-')[0]
    terminal = item['terminal']
    if terminal == 'Terminal 1':
        terminal = 't1'
    elif terminal == 'Terminal 2':
        terminal = 't2'
    elif terminal == 'Terminal 3':
        terminal = 't3'
    elif terminal == 'Terminal 4':
        terminal = 't4'
    elif terminal == 'Terminal 5':
        terminal = 't5'
    elif terminal == 'Terminal 6':
        terminal = 't6'
    elif terminal == 'Tom Bradley International Terminal':
        terminal = 'tbi'
    type = str(item['type']).lower()
    count = int(item['count'])
    return {'year': year, 'terminal': terminal, 'type': type, 'count': count}


if __name__ == "__main__":
    json_file = sys.argv[1]
    conditions = sys.argv[2]
    conditions = conditions.lower()

    arrival = False
    departure = False
    years = set()
    terminals = set()

    conditions = conditions.split(' ')
    for condition in conditions:
        if condition.startswith('t'):
            terminals.add(condition)
        elif condition == 'arrival':
            arrival = True
        elif condition == 'departure':
            departure = True
        else:
            years.add(condition)

    if not arrival and not departure:
        arrival = True
        departure = True

    if len(years) == 0:
        years = {'2006', '2007', '2008', '2009', '2010', '2011', '2012', '2013', '2014', '2015', '2016'}

    if len(terminals) == 0:
        terminals = {'t1', 't2', 't3', 't4', 't5', 't6', 'tbi'}

    query = {
        'arrival': arrival,
        'departure': departure,
        'years': years,
        'terminals': terminals
    }

    print(query)

    json_data = json.load(open(json_file))['data']
    data = []

    for item in json_data:
        projectedItem = {'year': item[9], 'terminal': item[10], 'type': item[11], 'count': item[13]}
        convertedItem = convert_item(projectedItem)
        if convertedItem['terminal'] in ['t1', 't2', 't3', 't4', 't5', 't6', 'tbi']:
            data.append(convertedItem)

    counts = []

    sum = 0

    for item in data:
        type_flag = (item['type'] == 'arrival' and query['arrival']) or (item['type'] == 'departure' and query['departure'])
        year_flag = item['year'] in query['years']
        terminal_flag = item['terminal'] in query['terminals']
        if type_flag and year_flag and terminal_flag:
            counts.append(item['count'])
            sum += item['count']

    counts = sorted(counts)

    size = len(counts)

    min = counts[0]
    max = counts[size-1]
    avg = sum*1.0 / size

    sqrsum = 0
    for count in counts:
        diff = avg - count
        sqrsum += diff * diff

    median = counts[size/2]

    if size % 2 == 0:
        median = (counts[size/2] + counts[size/2-1]) / 2.0

    standard_deviation = math.sqrt(sqrsum*1.0 / size)

    print(str(min) + ',' + str(max) + ',' + '%.1f' % median + ','
          + '%.2f' % avg + ',' + '%.2f' % standard_deviation)