def load_pool(file, stop_size=None):
    if file == "pool":
        filename = "\sport_pool_20190305_20190307"
    elif file == "common":
        filename = "\out_common"
    elif file == "allowed":
        filename = "\out_allowed"
    elif file == "shuffled":
        filename = "\shuf"
    else:
        print("filename error")

    df = pd.read_csv(r'C:\Users\Anastasiya\Desktop\диплом' + filename, delimiter='\t',
                     encoding='utf-8', nrows=300000, low_memory=False,
                     names=['query', 'factors', 'urls', 'target', 'clicks'])

    cnt = 0
    data = []
    target = []
    cnt0 = 0
    cnt1 = 0
    len_of_facts = len(list(str(df.values[0][1])[8:].split()))

    for ex in df.values:
        query = str(ex[0])[6:]

        if query == "":
            continue

        facts = list(str(ex[1])[8:].split())
        urls = list(str(ex[2])[5:].split())
        targ = int(str(ex[3])[7:])

        if len(facts) != len_of_facts:
            continue

        cur_list = [query]
        query_facts = facts[:]
        cur_list += list(map(float, query_facts))

        data.append(cur_list)
        if targ == 0:
            target.append(0)
            cnt0 += 1
        else:
            target.append(1)
            cnt1 += 1

        cnt += 1
        if cnt % 5000 == 0:
            print("total: {}, 0: {}, 1: {}".format(cnt, cnt0, cnt1))

        if stop_size is not None:
            if cnt >= stop_size:
                break

    print("data is loaded, cnt0 = {}, cnt1 = {}".format(cnt0, cnt1))
    return data, target