encoding_city_land_usage.py

def encode_city_land_usage(x):
    # Encode {'其他': 0, '住': 1, '農': 2, '工': 3, '商': 4, 
    #         '住商': 5, '其他住宅': 6, nan: nan}

    # nan remains
    if not isinstance(x, str):
        return x

    if '非都市' in x:
        return ""

    if '其他' in x:
        end_idx = x.find('(') if '(' in x else len(x) + 1
        if '住商' in x[:end_idx] or '住宅商業' in x[:end_idx]:
            return 5
        if '住宅' in x[:end_idx] or '住' in x[:end_idx]:
            return 6
        return 0

    if '住' in x:
        return 1
    elif '農' in x:
        return 2
    elif '工' in x:
        return 3
    elif '商' in x:
        return 4
    else:
        # There is no such case in the data
        raise ValueError('Unexpected 都市使用分區:', x)