将 JSON 行解包到 pandas 数据框

Posted 2023-02-16

【中文标题】将 JSON 行解包到 pandas 数据框

【英文标题】：Unpack JSON lines to pandas dataframe

【发布时间】：2021-04-14 17:32:31

【问题描述】：

我正在处理 JSON 行格式并尝试将字典对象“解包”到单个列表中。由于它使用列表来保存字典对象，因此我之前没有找到任何处理该问题的帖子。数据看起来是这样的，其中有一堆嵌套字典在一个列表对象中：

0        ['created_at': 'Sun Jun 14 20:20:28 +0000 202...
1        ['created_at': 'Sat Jul 25 22:30:14 +0000 202...
2        ['created_at': 'Sat May 30 02:22:04 +0000 202...
3        ['created_at': 'Tue May 05 16:54:05 +0000 202...
4        ['created_at': 'Sat Jun 20 13:50:23 +0000 202...
                               ...                        
17453    ['created_at': 'Mon Apr 13 01:01:10 +0000 202...
17454    ['created_at': 'Fri Jul 17 09:00:50 +0000 202...
17455    ['created_at': 'Sun Jun 21 00:51:54 +0000 202...
17456    ['created_at': 'Tue Jun 02 18:23:49 +0000 202...
17457    ['created_at': 'Thu May 28 00:27:01 +0000 202...

我现在尝试的是：

with open('data') as file:
    lines = file.read().splitlines()
df_inter = pd.DataFrame(lines)
df_inter.columns = ['json_element']

对于嵌套字典，我会使用此post 提供的pd.json_normalize(df_inter['json_element'].apply(json.loads))。但是，如何将多个字典对象解压缩到一行中？

编辑

由于数据量大，我提供部分单行数据：

['created_at': 'Sun Jun 14 20:20:28 +0000 2020', 'id': 1272262651100434433, 'id_str': '1272262651100434433', 'truncated': False, 'display_text_range': [0, 243], 'entities': 'hashtags': ['text': 'Tenet', 'indices': [82, 88]], 'symbols': [], 'user_mentions': [], 'urls': [], 'media': ['id': 1272262640753094656, 'id_str': '1272262640753094656', 'indices': [244, 267], 'media_url': 'http://pbs.twimg.com/media/Eaf8IYsWsAAHVHV.jpg'...]

【问题讨论】：

这取决于你的字典的结构，列表的每个元素是否具有相同的结构以及有多少嵌套字典。提供更全面的样本数据

谢谢，我已经在帖子中放了一个示例数据。

@Kapocsi 你是对的。我已经编辑了帖子。

【参考方案1】：

如果您的data 文件如下所示：

["created_at": "Sun Jun 14 20:20:28 +0000 2020", "id": 1272262651100434433, "id_str": "1272262651100434433", "truncated": false, "display_text_range": [0, 243], "entities": "hashtags": ["text": "Tenet", "indices": [82, 88]], "symbols": [], "user_mentions": [], "urls": [], "media": ["id": 1272262640753094656, "id_str": "1272262640753094656", "indices": [244, 267], "media_url": "http://pbs.twimg.com/media/Eaf8IYsWsAAHVHV.jpg"]]
["created_at": "Sun Jun 14 20:20:28 +0000 2020", "id": 1272262651100434433, "id_str": "1272262651100434433", "truncated": false, "display_text_range": [0, 243], "entities": "hashtags": ["text": "Tenet", "indices": [82, 88]], "symbols": [], "user_mentions": [], "urls": [], "media": ["id": 1272262640753094656, "id_str": "1272262640753094656", "indices": [244, 267], "media_url": "http://pbs.twimg.com/media/Eaf8IYsWsAAHVHV.jpg"]]
["created_at": "Sun Jun 14 20:20:28 +0000 2020", "id": 1272262651100434433, "id_str": "1272262651100434433", "truncated": false, "display_text_range": [0, 243], "entities": "hashtags": ["text": "Tenet", "indices": [82, 88]], "symbols": [], "user_mentions": [], "urls": [], "media": ["id": 1272262640753094656, "id_str": "1272262640753094656", "indices": [244, 267], "media_url": "http://pbs.twimg.com/media/Eaf8IYsWsAAHVHV.jpg"]]
["created_at": "Sun Jun 14 20:20:28 +0000 2020", "id": 1272262651100434433, "id_str": "1272262651100434433", "truncated": false, "display_text_range": [0, 243], "entities": "hashtags": ["text": "Tenet", "indices": [82, 88]], "symbols": [], "user_mentions": [], "urls": [], "media": ["id": 1272262640753094656, "id_str": "1272262640753094656", "indices": [244, 267], "media_url": "http://pbs.twimg.com/media/Eaf8IYsWsAAHVHV.jpg"]]
["created_at": "Sun Jun 14 20:20:28 +0000 2020", "id": 1272262651100434433, "id_str": "1272262651100434433", "truncated": false, "display_text_range": [0, 243], "entities": "hashtags": ["text": "Tenet", "indices": [82, 88]], "symbols": [], "user_mentions": [], "urls": [], "media": ["id": 1272262640753094656, "id_str": "1272262640753094656", "indices": [244, 267], "media_url": "http://pbs.twimg.com/media/Eaf8IYsWsAAHVHV.jpg"]]

您可以使用以下代码在 jsonl 文件中每行获取一个数据帧行。

import json
import pandas as pd

with open('data') as f:
    df = pd.DataFrame(json.loads(line)[0] for line in f)

您的 df 将如下所示：

                       created_at                   id               id_str  truncated display_text_range                                           entities
0  Sun Jun 14 20:20:28 +0000 2020  1272262651100434433  1272262651100434433      False           [0, 243]  'hashtags': ['text': 'Tenet', 'indices': [82...
1  Sun Jun 14 20:20:28 +0000 2020  1272262651100434433  1272262651100434433      False           [0, 243]  'hashtags': ['text': 'Tenet', 'indices': [82...
2  Sun Jun 14 20:20:28 +0000 2020  1272262651100434433  1272262651100434433      False           [0, 243]  'hashtags': ['text': 'Tenet', 'indices': [82...
3  Sun Jun 14 20:20:28 +0000 2020  1272262651100434433  1272262651100434433      False           [0, 243]  'hashtags': ['text': 'Tenet', 'indices': [82...
4  Sun Jun 14 20:20:28 +0000 2020  1272262651100434433  1272262651100434433      False           [0, 243]  'hashtags': ['text': 'Tenet', 'indices': [82...

>>> df.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5 entries, 0 to 4
Data columns (total 6 columns):
 #   Column              Non-Null Count  Dtype
---  ------              --------------  -----
 0   created_at          5 non-null      object
 1   id                  5 non-null      int64
 2   id_str              5 non-null      object
 3   truncated           5 non-null      bool
 4   display_text_range  5 non-null      object
 5   entities            5 non-null      object
dtypes: bool(1), int64(1), object(4)
memory usage: 333.0+ bytes

【讨论】：

谢谢您，先生。这是惊人而简单的代码，无需解压列表并将其转换为数据框。