如果你想减少依赖，这是你如何使用标准csv库：

import csv

# collect the corresponding items by username
user_details_managers = {}

with open("./user_details.csv") as details_fp:
    reader = csv.reader(details_fp)
    # read first row to skip headers
    next(reader)
    for row in reader:
        user, details = row
        user_details_managers[user] = [user, details]

with open("./user_managers.csv") as managers_fp:
    reader = csv.reader(managers_fp)
    # read first row to skip headers
    next(reader)
    for row in reader:
        username, firstname, lastname = row
        if username in user_details_managers:
            user_details_managers[username].extend([firstname, lastname])

with open("./user_details_managers.csv", "w") as details_managers_fp:
    writer = csv.writer(details_managers_fp)

    # write headers
    writer.writerow(["user", "details", "manager firstname", "manager lastname"])

    # write combined rows
    writer.writerows(user_details_managers.values())

字符串

可以用pandas轻松完成：

import pandas as pd

user_details = pd.read_csv('user_details.csv')
user_managers = pd.read_csv('user_mangers.csv')
merged_df = user_details.merge(user_managers, how='left', left_on='user', right_on='username')
merged_df.to_csv('user_details_managers.csv')

字符串
这里发生了什么？我使用pandas将两个csv文件读入到多个框架中。然后我们将它们合并到一个框架中：在“左边”的框架（这里是user_details）中，我们合并了“user”列，在“右边”的框架（user_magers）中，它接受了“username”列。这意味着它将合并从左边合并每个“user”和从右边合并“username”。
结果是一个单一的框架，我使用.to_csv()将其保存为csv。