def parse_events(path_to_logs):
    list_of_events_dicts = []
    count_of_decode_errors = 0

    with open(path_to_logs, 'r') as logs:
        rows = logs.readlines()
        for i in range(len(rows)):
            if re.search('Account order event', rows[i]):
                if i + 1 < len(rows):
                    try:
                        event = json.loads(re.sub(r'(LIMIT_MAKER")', r'\1,', rows[i+1][1:].strip()))
                        if type(event) is dict:
                            list_of_events_dicts.append(event)
                    except json.JSONDecodeError:
                        count_of_decode_errors += 1

    return (list_of_events_dicts, count_of_decode_errors)


def create_events_csv(path_to_logs, name='mm_events.csv'):
    events, errors_count = parse_events(path_to_logs)
    print(f'events rows = {len(events)}; errors count = {errors_count}')

    df = pd.DataFrame(events)
    df.drop(['symbol', 'type', 'is_maker', 'average_price', 'filled_quantity'], axis=1, inplace=True)

    for float_column in ['amount', 'price', 'quantity']:
        df[float_column] = df[float_column].astype(float)

    for str_column in ['status', 'side']:
        df[str_column] = df[str_column].astype('string')

    df['create_time'] = pd.to_datetime(df['create_time'] * 1e6).dt.round('1s')
    df['update_time'] = pd.to_datetime(df['update_time'] * 1e6).dt.round('1s')

    df = df.sort_values('update_time')
    df.to_csv(name, index=False)

    return df