如何使用 pandas 读取 json 字典类型的文件？

Posted 2023-02-23

【中文标题】如何使用 pandas 读取 json 字典类型的文件？

【英文标题】：How to read a json-dictionary type file with pandas?

【发布时间】：2015-04-07 00:55:46

【问题描述】：

我有一个像这样的长 json：http://pastebin.com/gzhHEYGy

我想将它放入 pandas datframe 中以便使用它，因此根据文档我执行以下操作：

df = pd.read_json('/user/file.json')
print df

我得到了这个回溯：

  File "/Users/user/PycharmProjects/PAN-pruebas/json_2_dataframe.py", line 6, in <module>
    df = pd.read_json('/Users/user/Downloads/54db3923f033e1dd6a82222aa2604ab9.json')
  File "/usr/local/lib/python2.7/site-packages/pandas/io/json.py", line 198, in read_json
    date_unit).parse()
  File "/usr/local/lib/python2.7/site-packages/pandas/io/json.py", line 266, in parse
    self._parse_no_numpy()
  File "/usr/local/lib/python2.7/site-packages/pandas/io/json.py", line 483, in _parse_no_numpy
    loads(json, precise_float=self.precise_float), dtype=None)
  File "/usr/local/lib/python2.7/site-packages/pandas/core/frame.py", line 203, in __init__
    mgr = self._init_dict(data, index, columns, dtype=dtype)
  File "/usr/local/lib/python2.7/site-packages/pandas/core/frame.py", line 327, in _init_dict
    dtype=dtype)
  File "/usr/local/lib/python2.7/site-packages/pandas/core/frame.py", line 4620, in _arrays_to_mgr
    index = extract_index(arrays)
  File "/usr/local/lib/python2.7/site-packages/pandas/core/frame.py", line 4668, in extract_index
    raise ValueError('arrays must all be same length')
ValueError: arrays must all be same length

然后从上一个问题中我发现我需要做这样的事情：

d = dict( A = np.array([1,2]), B = np.array([1,2,3,4]) )

但我不明白我应该如何获取像 numpy 数组这样的内容。如何在这样的大文件中保留数组的长度？提前致谢。

【问题讨论】：

好像字典一样。

【参考方案1】：

json 方法不起作用，因为 json 文件不是它期望的格式。由于我们可以轻松地将 json 作为 dict 加载，让我们尝试这种方式：

import pandas as pd
import json
import os

os.chdir('/Users/nicolas/Downloads')

# Reading the json as a dict
with open('json_example.json') as json_data:
    data = json.load(json_data)

# using the from_dict load function. Note that the 'orient' parameter 
#is not using the default value (or it will give the same error that you got before)
# We transpose the resulting df and set index column as its index to get this result
pd.DataFrame.from_dict(data, orient='index').T.set_index('index')   

输出：

                                                                 data columns
index                                                                        
311210177061863424  [25-34\n, FEMALE, @bikewa absolutely the best....     age
310912785183813632  [25-34\n, FEMALE, Photo: I love the Burke-Gilm...  gender
311290293871849472  [25-34\n, FEMALE, Photo: Inhaled! #fitfoodie h...    text
309386414548717569  [25-34\n, FEMALE, Facebook Is Making The Most ...    None
312327801187495936  [25-34\n, FEMALE, Still upset about this >&...    None
312249421079400449  [25-34\n, FEMALE, @JoeM_PM_UK @JonAntoine I've...    None
308692673194246145  [25-34\n, FEMALE, @Social_Freedom_ actually, t...    None
308995226633129984  [25-34\n, FEMALE, @seattleweekly that's more t...    None
308660851219501056  [25-34\n, FEMALE, @adamholdenbache I noticed 1...    None
308658690528014337  [25-34\n, FEMALE, @CEM_Social I am waiting pat...    None
309719798001070080  [25-34\n, FEMALE, Going to be watching Faceboo...    None
312349448049152002  [25-34\n, FEMALE, @anikamarketer I applied for...    None
312325152698404864  [25-34\n, FEMALE, @_chrisrojas_ wow, that's so...    None
310546490844135425  [25-34\n, FEMALE, Photo: Feeling like a bit of...    None

【讨论】：

非常感谢@knightofni 的帮助

如果有人收到此错误：the JSON object must be str, bytes or bytearray, not TextIOWrapper，请注意它使用的是json.load()而不是json.loads()，它需要是json.loads(json_data.read())

【参考方案2】：

pandas 模块而不是 json 模块应该是答案： pandas 本身具有 read_json 功能，问题的根源一定是您没有以正确的方向读取 json。 您必须首先传递创建 json 变量时使用的确切 orient 参数

例如：

df_json = globals()['df'].to_json(orient='split')

然后：

read_to_json = pd.read_json(df_json, orient='split')