petite-difference-challenge2/run.py

men = ['samochód', 'auto', 'audi', 'bmw', 'mercedes', 'windows', 'silnika', 'mb', 'gb', 'mecz', 'legia', 'lech', 'real', 'komputer', 'telefon', 'apple', 'iphone', 'zwiastuny', 'hd',
       'windows', 'system', 'serwer', 'serwery', 'siłownia', 'biceps', 'triceps']
woman = ['zrobiłam', 'byłam', 'pojechałam','ciąży', 'miesiączki', 'makijaż', 'perfumy', 'kompleks', 'ciąża', 'ciazy', 'antykoncepcyjne', 'ginekologa',
         'tabletki', 'sukienka', 'spódniczka', 'mąż', 'miesiączkę', 'samotność', 'kolczyki']
def database(men, woman):
    men_base = []
    woman_base = []
    for i in men:
        men_base.append(i[:10].lower())
    for i in woman:
        woman_base.append(i[:10].lower())
    return men_base, woman_base
def predict(men_base, woman_base, data):
    res = []
    i = 0
    for slowo in data:
        men_count = 0
        woman_count = 0
        term = slowo.split()

        for w in term:
            if w[:10].lower() in men_base:
                men_count = men_count + 1
            elif w[:10].lower() in woman_base:
                woman_count = woman_count + 1
        if men_count >= woman_count:
            result.append('1')
        elif men_count < woman_count:
            result.append('0')
    return result


def out_file(result, name):
    with open(name, 'w') as file:
        for program in result:
            file.write(program + "\n")


men_dict, woman_dict = database(men, woman)
data = []
with open('dev-0/in.tsv') as data:
    for idx, line in enumerate(data.readlines()):
        data.append(line.replace('\n', ''))

result = predict(men_dict, woman_dict, data)
out_file(result, 'dev-0/out.tsv')

data2 = []
with open('dev-1/in.tsv') as data:
    for idx, line in enumerate(data.readlines()):
        data2.append(line.replace('\n', ''))

result = predict(men_dict, woman_dict, data2)
out_file(result, 'dev-1/out.tsv')

data3 = []
with open('test-A/in.tsv') as data:
    for idx, line in enumerate(data.readlines()):
        data3.append(line.replace('\n', ''))

result = predict(men_dict, woman_dict, data3)
out_file(result, 'test-A/out.tsv')