听起来，也许这种特殊类型的任务更适合“数据库”。
以duckdb为例，因为它也可以从 Dataframe 读取：

# probably better to write out the full query
query = "\nand\n".join(
    f"(df2.{column} = '*' or df1.{column} = df2.{column})" 
    for column in df2.columns
)

duckdb.sql(f"""
from df1, df2
select df1.*
where
   {query}
""")

（尽管在提到大量数据时直接使用它可能有意义）

┌───────────┬──────────┬────────────────┬─────────┐
│ FirstName │ LastName │   Occupation   │ Gender  │
│  varchar  │ varchar  │    varchar     │ varchar │
├───────────┼──────────┼────────────────┼─────────┤
│ Angela    │ Norris   │ Firefighter    │ Female  │
│ Angela    │ Thompson │ Teacher        │ Female  │
│ Ben       │ Johnson  │ Police Officer │ Male    │
│ Debbie    │ Smith    │ Lawyer         │ Female  │
└───────────┴──────────┴────────────────┴─────────┘

使用的代码示例：

import duckdb
import pandas as pd

df1 = pd.read_csv(io.BytesIO(b"""
FirstName,LastName,Occupation,Gender
Angela,Norris,Firefighter,Female
Angela,Thompson,Teacher,Female
Ben,Johnson,Police Officer,Male
Ben,Peterson,Solicitor,Male
Charlie,Davies,Nurse,Male
Debbie,Smith,Lawyer,Female"""))

df2 = pd.read_csv(io.BytesIO(b"""
FirstName,LastName,Occupation,Gender
Angela,*,*,*
Ben,Johnson,*,Male
Debbie,*,*,Female"""))

query = "\nand\n".join(
    f"(df2.{column} = '*' or df1.{column} = df2.{column})" 
    for column in df2.columns
)

duckdb.sql(f"""
from df1, df2
select df1.*
where
   {query}
""")

假设协调密钥中始终存在FirstName，则这应该可以实现：
设置：

import numpy as np
import pandas as pd

master_df = pd.DataFrame({
    'FirstName': ['Angela', 'Angela', 'Ben', 'Ben', 'Charlie', 'Debbie'],
    'LastName': ['N', 'T', 'J', 'P', 'D', 'S'],
    'Occupation': ['Firefighter', 'Teacher', 'Police', 'Solicitor', 'Nurse', 'Lawyer'],
    'Gender': ['Female', 'Female', 'Male', 'Male', 'Male', 'Female'],
})

recon_df = pd.DataFrame({
    'FirstName': ['Angela', 'Ben', 'Debbie'],
    'LastName': ['*', 'J', '*'],
    'Occupation': ['*', '*', '*'],
    'Gender': ['*', 'Male', 'Female'],
})

解决方案：

temp_df = master_df.merge(
    recon_df,
    how='left',
    on='FirstName',
)

update_cols = [col for col in temp_df.columns if col[-2:] == '_x']
for col in update_cols:
    y_col = col[:-2] + '_y'
    temp_df[col] = np.where(
        temp_df[y_col] == '*',
        temp_df[y_col],
        temp_df[col],
    )
temp_df = temp_df.rename(columns={col: col[:-2] for col in update_cols})
temp_df = temp_df.drop(columns=[col for col in temp_df.columns if col[-2:] == '_y'])

keys = list(recon_df.columns.values)
index1 = temp_df.set_index(keys).index
index2 = recon_df.set_index(keys).index

output_df = master_df.loc[index1.isin(index2)]

输出：

FirstName LastName   Occupation  Gender
0    Angela        N  Firefighter  Female
1    Angela        T      Teacher  Female
2       Ben        J       Police    Male
5    Debbie        S       Lawyer  Female

（基于solution here构建）