#!/usr/bin/python3
# Released in public domain under CC0 license:
# <https://creativecommons.org/publicdomain/zero/1.0/legalcode>

'''Analysis of GUADEC accommodation survey results.'''

import numpy
import pandas

IN_FILE = 'GUADEC 2017 accommodation survey.csv'


ACCOMMODATION_OPTION_LABELS = [
    'hostel',
    'uni',
    'hotel_sharing',
    'hotel_single',
    'own'
]

NIGHT_OPTION_LABELS = [
    'thu27',
    'fri28',
    'sat29',
    'sun30',
    'mon31',
    'tue1',
    'wed2',
]


COLUMN_NAMES = [
    'TIMESTAMP',     # Timestamp
    'ACCOMMODATION', # Which accommodation option(s) would you choose for GUADEC 2017?
    'REGISTRATION',  # How early could you register for the conference & pay for accommodation?
    'NIGHTS',        # Which nights would you expect to stay at GUADEC 2017?
    'BREAKFAST'      # Would you pay £8.50 extra per day for a buffet breakfast?
]


def expand_multiple_choice_column(data, source_column, option_strings, option_labels):
    '''Parse string of option names, turn into additional true/false columns.'''
    to_append = []
    for option_number, full_option_string in enumerate(option_strings):
        new_column = []

        for column_number, response in enumerate(source_column):
            chosen = response.split(';')
            new_column.append(full_option_string in chosen)

        new_column_series = pandas.Series(new_column, name=option_labels[option_number])
        to_append.append(new_column_series)

    new_data = pandas.concat([data] + to_append, axis=1)
    return new_data


def expand_accommodation_column(data):
    accommodation_option_strings = [
        'Youth hostel (4-bed dorm, £33 per night)',
        'University Townhouse (single room, £43 or more per night)',
        'Jury\'s Inn Hotel (sharing a twin room, ~£51 per night)',
        'Jury\'s Inn Hotel (single room, ~£94 per night)',
        'I would organise my own accommodation'
    ]

    return expand_multiple_choice_column(data,
                                         source_column=data.ACCOMMODATION,
                                         option_strings=accommodation_option_strings,
                                         option_labels=ACCOMMODATION_OPTION_LABELS)


def expand_nights_column(data):
    nights_option_strings = [
        'Thursday 27th July 2017',
        'Friday 28th July 2017 (core day #1)',
        'Saturday 29th July 2017 (core day #2)',
        'Sunday 30th July 2017 (core day #3)',
        'Monday 31st July 2017 (BOF/workshop day #1)',
        'Tuesday 1st August 2017 (BOF/workshop day #2)',
        'Wednesday 2nd August 2017 (BOF/workshop day #3)'
    ]

    return expand_multiple_choice_column(data,
                                         source_column=data.NIGHTS,
                                         option_strings=nights_option_strings,
                                         option_labels=NIGHT_OPTION_LABELS)


def number_and_percentage(number, total):
    return '%i of %i (%i%%)' % (number, total, (number / total) * 100.0)


def main():
    everyone = pandas.read_csv(IN_FILE, names=COLUMN_NAMES, header=0)
    print("Total responses: %i" % len(everyone))

    everyone = expand_accommodation_column(everyone)
    everyone = expand_nights_column(everyone)

    print("How many people would consider staying at the MMU Townhouses?")
    townhouse_guests = everyone.query('uni == True')
    print(number_and_percentage(len(townhouse_guests), len(everyone)))

    print("How many people would only consider staying at the MMU Townhouses?")
    answer = len(townhouse_guests.query('hostel == False and hotel_single == False and hotel_sharing == False and own == False'))
    print(number_and_percentage(answer, len(everyone)))

    print("How many people who would stay at the Townhouses could register "
          "before March 2017?")
    answer = len(townhouse_guests.query('REGISTRATION == "By 1st March 2017"'))
    print(number_and_percentage(answer, len(townhouse_guests)))

    print("How many people who would stay at the Townhouses would pay "
          "£8.50 extra for breakfast")
    answer = len(townhouse_guests.query('BREAKFAST == "Yes"'))
    print(number_and_percentage(answer, len(townhouse_guests)))

    print("What nights would Townhouses guests stay?")
    for date in NIGHT_OPTION_LABELS:
        answer = len(townhouse_guests.query('%s == True' % date))
        print("  %s: %s, or %i%% of total attendees" % (
            date, number_and_percentage(answer, len(townhouse_guests)),
            (answer / len(everyone) * 100.0)))

    print("How many people would only stay in the youth hostel?")
    answer = len(everyone.query('hostel == True and uni == False and hotel_single == False and hotel_sharing == False'))
    print(number_and_percentage(answer, len(everyone)))

    print("How many people would only stay the hotel?")
    answer = len(everyone.query('hostel == False and uni == False'))
    print(number_and_percentage(answer, len(everyone)))


main()