python - pd.update 有两个匹配的行

Question

我想做的是从 df2 更新空的 df1，它是在通过 API 请求数据的 while 循环中创建的。我想保留 df1 中的所有行及其顺序。

df1:

df = pd.DataFrame({'A': ['c1', 'c1', 'c2','c2', 'c3', 'c3'], 'B': ['y1', 'y2', 'y1', 'y2', 'y1', 'y2'], 'C': ["","","","","",""], 'D': ["","","","","",""]})

    A   B   C   D
0   c1  y1      
1   c1  y2      
2   c2  y1      
3   c2  y2      
4   c3  y1      
5   c3  y2  

df2:

values_for_df = pd.DataFrame({'A': ['c1', 'c1', 'c2', 'c3'], 'B': ['y1', 'y2', 'y1', 'y2'], 'C': [4, 5, 4, 6], 'D': [7, 8, 9,""]})

    A   B   C   D
0   c1  y1  4   7
1   c1  y2  5   8
2   c2  y1  4   9
3   c3  y2  6   


Output:
    A   B   C   D
0   c1  y1  4   7
1   c1  y2  5   8
2   c2  y1  4   9
3   c3  y2  6   
4   c3  y1      
5   c3  y2      

Wanted output:

    A   B   C   D
0   c1  y1  4   7   
1   c1  y2  5   8   
2   c2  y1  4   9   
3   c2  y2      
4   c3  y1      
5   c3  y2      6   

此过程将重复数千次。有人可以帮助我解决这个问题，分享他的想法/替代方法，或者解释一下为什么实际输出与我的预期输出不同？

Answer 1

尝试:

df = df.set_index(['A','B'])
values_for_df = values_for_df.set_index(['A','B'])
df.update(values_for_df, filter_func=lambda x: x=='')
df.reset_index()

    A   B  C  D
0  c1  y1  4  7
1  c1  y2  5  8
2  c2  y1  4  9
3  c2  y2      
4  c3  y1      
5  c3  y2  6