基于Pandas JOIN实现DataFrame版本控制与差异对比

DevOps 小满

2026-8-2

问题：用Pandas JOIN操作对比DataFrame版本变更

我正在对比两个DataFrame以检查二者间的变更情况，这是版本控制脚本的一部分，为此我编写了简化示例来寻找解决方案：

import pandas as pd

data = {'ID':  ['1', '2', '3', '4'],
        'Date':  ['23-01-2023', '01-12-1995', '03-07-2013', '05-09-2013'],
        'Time':  ['01:45:08', '02:15:21', '23:57:14', '03:57:15'],
        'Path':  ['//server/test/File1.txt', '//server/test/File2.txt', '//server/test/File3.txt', '//server/test/File4.txt'],
        }
data2 = {'ID':  ['1', '2', '3', '4'],
        'Date':  ['23-01-2023', '03-07-2013', '01-12-1995', '21-11-1991'],
        'Time':  ['01:45:08', '23:57:14', '02:17:21', '03:18:31'],
        'Path':  ['//server/test/File1.txt', '//server/test/File3.txt', '//server/test/File2.txt', '//server/test/File5.txt'],
        }

df = pd.DataFrame(data)
df2 = pd.DataFrame(data2)

生成的两个DataFrame如下：

DataFrame 1

| ID | Date       | Time      | Path                       |
 |  1 | 23-01-2023 | 01:45:08  | //server/test/File1.txt    |
 |  2 | 01-12-1995 | 02:15:21  | //server/test/File2.txt    |
 |  3 | 03-07-2013 | 23:57:14  | //server/test/File3.txt    |
 |  4 | 05-09-2013 | 03:57:15  | //server/test/File4.txt    |

DataFrame 2

| ID | Date       | Time      | Path                       |
 |  1 | 23-01-2023 | 01:45:08  | //server/test/File1.txt    |
 |  2 | 03-07-2013 | 23:57:14  | //server/test/File3.txt    |
 |  3 | 01-12-1995 | 02:17:21  | //server/test/File2.txt    |
 |  4 | 21-11-1991 | 03:18:31  | //server/test/File5.txt    |

以第一个DataFrame为基准，变更情况如下：

ID为4的文件已删除
ID为2的文件已修改
新增了一个文件（DataFrame2中的ID4）

最终希望得到如下输出：

| ID | Date       | Time      | Path                       |  Status  |
 |  1 | 23-01-2023 | 01:45:08  | //server/test/File1.txt    |     -    |
 |  2 | 01-12-1995 | 02:15:21  | //server/test/File2.txt    |  UPDATED |
 |  3 | 03-07-2013 | 23:57:14  | //server/test/File3.txt    |     -    |
 |  4 | 05-09-2013 | 03:57:15  | //server/test/File4.txt    |  DELETED |
 |  5 | 21-11-1991 | 03:18:31  | //server/test/File5.txt    |   ADDED  |

能否仅通过Pandas的JOIN操作实现该需求？

内容的提问来源于stack exchange，提问作者Ralk