组合两个以这种格式分块的文件XXXXX.csv.gz_1_2.tar & XXXXX.csv.gz_2_2.tar (使用python或pyspark)

import tarfile
import shutil

def extract_tar_file(file_path, extract_path):
    with tarfile.open(file_path, 'r') as tar:
        tar.extractall(extract_path)

def merge_files(file1, file2, output_file):
    with open(output_file, 'wb') as output:
        with open(file1, 'rb') as f1:
            shutil.copyfileobj(f1, output)
        with open(file2, 'rb') as f2:
            shutil.copyfileobj(f2, output)

# 解压缩第一个文件
extract_tar_file('XXXXX.csv.gz_1_2.tar', './extracted_files')

# 解压缩第二个文件
extract_tar_file('XXXXX.csv.gz_2_2.tar', './extracted_files')

# 合并文件
merge_files('./extracted_files/XXXXX.csv.gz_1_2', './extracted_files/XXXXX.csv.gz_2_2', 'merged_file.csv.gz')