operate_hash.py 文件源码

def district_hash_map(data_frame):
    district_map_f = "cluster_map.pickle"
    district_map_f_path = os.path.join(DATA_DIR, CONCRETE_DIR, CLUSTER_MAP_SHEET_DIR, 
                                        district_map_f)
    if not os.path.exists(district_map_f_path):
        create_hash_district_map_dict()
    # load the needed map file
    with open(district_map_f_path, "rb") as f:
        map_rule = pickle.load(f)

    # map the needed cols..
    for i in range(len(data_frame.columns)):
        if "district_hash" in data_frame.columns[i]:
            # map the hash according to the map rule
            district_hash_col = data_frame.columns[i]
            data_frame[district_hash_col] = data_frame[district_hash_col].replace(map_rule)

            # after mapping, delete its hash str
            new_name = re.sub("_hash","",district_hash_col)
            data_frame.rename(columns={district_hash_col: new_name}, inplace = True)

    return data_frame


## input the dir you want to map the hash