Python 如何按顺序合并多个数据帧

虽然我认为这个问题应该重复，但我找不到合适的答案.

我在按顺序合并多个数据帧时遇到一些问题.

例如，我有以下四个数据帧:

df1 = pd.DataFrame({'source': ['A', 'A', 'A', 'B', 'B', 'C', 'C'],
       'target': ['1', '2', '3', '4', '5', '6', '7']})
df2 = pd.DataFrame({'source': ['A', 'A'],
       'temp': ['a', 'b']})
df3 = pd.DataFrame({'source': ['B', 'B'],
       'temp': ['c', 'd']})
df4 = pd.DataFrame({'source': ['C'],
       'temp': ['e']})

我想将数据帧合并如下:

#   source  target  temp
#0  A   1   a
#1  A   1   b
#2  A   2   a
#3  A   2   b
#4  A   3   a
#5  A   3   b
#6  B   4   c
#7  B   4   d
#8  B   5   c
#9  B   5   d
#10 C   6   e
#11 C   7   e

为此，我try 运行代码，但它返回了意外的结果.

#Trial 1
dfs = pd.merge(df1, df2, on='source', how='left')
dfs = pd.merge(dfs, df3, on='source', how='left') # new column was created with prefix, but I want to keep the three columns; source, target, temp

#Trial 2
dfs = pd.merge(df1, df2, on='source', how='left')
dfs['temp']=dfs.set_index('source')['temp'].fillna(df3.set_index('source')['temp'].to_dict()).values # it only fills the fixed number of NaN value, but there are some exception; one NaN in dfs, multiple values in other df3 or df4

#Trial 3
dfs = pd.merge(df1, df2, on='source', how='left')
dfs[dfs['source']=='B']['temp']=pd.merge(df1, df3, on='source', how='left')['temp'].dropna() # it didn't change the dfs