如何将pickle文件转换为JSON格式？

Question

我有一个pickle文件需要转换为json格式，我使用了这段代码

    import pickle
    import pandas as pd
    from sys import argv
    
    script, filename = argv
    input_file = open('f.pkl', 'rb')
    new_dict = pickle.load(input_file)
    
    df = pd.DataFrame.from_dict(new_dict, orient='index')
    df['index1'] = df.index
    index1 = df['index1']
    df.drop(labels=['index1'], axis=1, inplace=True)
    df.insert(0, 'index1', index1)
    json_df = df.to_json(orient='values', date_format='iso', date_unit='s')
    with open('data.json', 'w') as js_file:
        js_file.write(json_df)

但是得到了

 df = pd.DataFrame.from_dict(new_dict, orient='index')
  File "pandas/core/frame.py", line 1127, in from_dict
    if isinstance(list(data.values())[0], (Series, dict)):
AttributeError: 'list' object has no attribute 'values'

Answer 1

行 df = pd.DataFrame.from_dict(new_dict, orient='index') 是错误。根据错误，您尝试从 pickle 文件加载，该文件是一个列表，而不是字典：

new_dict = pickle.load(input_file)

的结果是一个列表（它可以是任何类型的数据结构）。尝试添加

isinstance()

来检查 unpickled 的数据是列表还是字典。

例如我的是 dataFrame，为此我在 Python 中使用了以下代码：

import json
import pickle
import pandas as pd

# opens the pickle file
with open('changewithyours.pickle', 'rb') as input_file:
    # loads the pickle file into a pandas DataFrame
    data = pd.read_pickle(input_file)

# resets the index of the DataFrame
data.reset_index(drop=True, inplace=True)

# converts the DataFrame to JSON and write it to a file
data.to_json('data.json', orient='columns')