Python中修改csv.DictReader字段名后如何正确生成新CSV

原始CSV文件（oldFile.csv） #

oldHeader1,oldHeader2,oldHeader3
data01,data02,data03
data11,data12,data13
data21,data22,data23

尝试的代码 #

import csv

with open("oldFile.csv", "r") as input:
    reader = csv.DictReader(input, delimiter=",")
    reader.fieldnames = ["newHeader1", "newHeader2", "newHeader3"]
    rows = list(reader)
    with open("newFile.csv", "w", newline="") as output:
        writer = csv.DictWriter(output, fieldnames=reader.fieldnames, delimiter=",")
        writer.writeheader()
        for row in rows:
            writer.writerow(row)

问题现象 #

运行后生成的newFile.csv把旧表头当成了数据行：

newHeader1,newHeader2,newHeader3
oldHeader1,oldHeader2,oldHeader3
data01,data02,data03
data11,data12,data13
data21,data22,data23

测试验证 #

单独测试csv.DictReader能正确识别旧表头：

import csv

with open("oldFile.csv", "r") as input:
    reader = csv.DictReader(input, delimiter=",")
    print(reader.fieldnames)

输出：

['oldHeader1', 'oldHeader2', 'oldHeader3']

疑问：明明修改了fieldnames，为什么旧表头还是会被写入？不想用rows = list(reader)[1:]这种切片方式，求正确实现方法。

问题原因 #

csv.DictReader初始化时已经自动读取了文件第一行作为表头，后续修改reader.fieldnames只是改变了后续数据行的键名映射，但已经被读取的旧表头行，会被当成第一行数据来处理——因为DictReader的表头读取步骤已经完成，修改fieldnames不会回溯跳过这一行。

正确实现方法 #

方法一：手动跳过旧表头，用新表头初始化DictReader

import csv

with open("oldFile.csv", "r") as input:
    # 先读取并跳过旧表头行
    input.readline()
    # 直接用新表头创建DictReader，后续行自动按新表头映射
    reader = csv.DictReader(input, fieldnames=["newHeader1", "newHeader2", "newHeader3"], delimiter=",")
    rows = list(reader)
    
    with open("newFile.csv", "w", newline="") as output:
        writer = csv.DictWriter(output, fieldnames=["newHeader1", "newHeader2", "newHeader3"], delimiter=",")
        writer.writeheader()
        writer.writerows(rows)

方法二：用普通csv.reader处理（更直观）

import csv

with open("oldFile.csv", "r") as input:
    reader = csv.reader(input, delimiter=",")
    # 跳过旧表头
    next(reader)
    rows = list(reader)
    
    with open("newFile.csv", "w", newline="") as output:
        writer = csv.writer(output, delimiter=",")
        # 写入新表头
        writer.writerow(["newHeader1", "newHeader2", "newHeader3"])
        # 写入所有数据行
        writer.writerows(rows)

这两种方法都能直接跳过旧表头，不需要切片操作，确保新文件只包含新表头和原始数据行。

内容的提问来源于stack exchange，提问作者deltonio2