#!/usr/bin/env python
# 2021 - Advent Of Code - 8 part 2

# acedgfb cdfbe gcdfa fbcad dab cefabd cdfgeb eafb cagedb ab | cdfeb fcadb cdfeb cdbaf

def parse_file(file):
    numbers_list = []
    outputs = []
    with open(file, encoding="utf-8") as f:
        for line in f.readlines():
            numbers, output = line.strip().split(' | ')
            numbers_list.append(numbers)
            outputs.append(output)

    return numbers_list, outputs


def print_numbers(id_num):
    print('identified numbers:')
    for i in range(10):
        print(f'{i} is {id_num[i]}')


# entries, displays = parse_file('input_example1.txt')
entries, displays = parse_file('input.txt')

output_sum = 0
for entry, display in zip(entries, displays):
    identified_numbers = [-1] * 10
    words = entry.split(' ')

    # size of pattern for 1, 4, 7 and 8 is unique
    identified_numbers[1] = set([x for x in words if len(x) == 2][0])
    identified_numbers[7] = set([x for x in words if len(x) == 3][0])
    identified_numbers[4] = set([x for x in words if len(x) == 4][0])
    identified_numbers[8] = set([x for x in words if len(x) == 7][0])

    # size 6 pattern that doesn't include 4, nor 1 -> 6
    # size 6 pattern that includes 4 and 1 -> 9
    # the remaining size 6 pattern -> 0
    size_six = {x for x in words if len(x) == 6}
    for s in size_six:
        if identified_numbers[1].issubset(s) and identified_numbers[4].issubset(s):
            identified_numbers[9] = s
        if not identified_numbers[1].issubset(s) and not identified_numbers[4].issubset(s):
            identified_numbers[6] = s

    size_six.remove(identified_numbers[9])
    identified_numbers[9] = set(identified_numbers[9])
    size_six.remove(identified_numbers[6])
    identified_numbers[6] = set(identified_numbers[6])
    identified_numbers[0] = set(next(iter(size_six)))

    # size 5 pattern that includes 7 -> 3
    # size 5 pattern that only missed one element to be a 9 pattern -> 5
    # the remaining size 5 pattern -> 2
    size_five = {x for x in words if len(x) == 5}
    for s in size_five:
        if identified_numbers[7].issubset(s):
            identified_numbers[3] = s
    size_five.remove(identified_numbers[3])
    identified_numbers[3] = set(identified_numbers[3])

    for s in size_five:
        if len(identified_numbers[9] - set(s)) == 1:
            identified_numbers[5] = s
            break
    size_five.remove(identified_numbers[5])
    identified_numbers[5] = set(identified_numbers[5])

    identified_numbers[2] = set(next(iter(size_five)))
    # print_numbers(identified_numbers)

    words = display.split(' ')
    str_num = ""
    for d in words:
        n = identified_numbers.index(set(d))
        str_num += str(n)
    output_sum += int(str_num)

print(f'output_sum: {output_sum}')