废话不多说，直接贴代码，注释很详细，代码也可以直接用。

"""
此脚本只解决文件源下载慢的问题，并不会提升带宽。
"""
import os
import httpx
from tqdm import tqdm
from threading import Thread

# 开辟线程数量，建议1-10个以内
THREAD_NUM = 5
tqdm_obj = None
# 下载文件地址
download_url = "https://acvrpublicycchen.blob.core.windows.net/dialogpt/keys-full.tar"


def download_file(file_path, thread_index, etag, start_index, stop_index):
    sub_path_file = "{}_{}".format(file_path, thread_index)
    if os.path.exists(sub_path_file):
        temp_size = os.path.getsize(sub_path_file)  # 本地已经下载的文件大小
        tqdm_obj.update(temp_size)  # 更新下载进度条
    else:
        temp_size = 0
    if stop_index == '-': stop_index = ""
    headers = {'Range': 'bytes={}-{}'.format(start_index + temp_size, stop_index),
               'ETag': etag, 'if-Range': etag,
               }
    with open(sub_path_file, 'ab') as down_file:
        with httpx.stream("GET", download_url, headers=headers) as response:
            num_bytes_downloaded = response.num_bytes_downloaded
            for chunk in response.iter_bytes():
                if chunk:
                    down_file.write(chunk)
                    tqdm_obj.update(response.num_bytes_downloaded - num_bytes_downloaded)
                    num_bytes_downloaded = response.num_bytes_downloaded
    return


def get_file_size():
    """
    获取预下载文件大小和文件etag
    :return:
    """
    with httpx.stream("GET", download_url) as response2:
        etag = ''
        total_size = int(response2.headers["Content-Length"])
        for tltle in response2.headers.raw:
            if tltle[0].decode() == "ETag":
                etag = tltle[1].decode()
                break
    return total_size, etag


def cutting(file_size, thread_num):
    """
    切割成若干份
    :param file_size: 下载文件大小
    :param thread_num: 线程数量
    :return:
    """
    cut_info = {}
    cut_size = file_size // thread_num
    for num in range(1, thread_num + 1):
        if num != 1:
            cut_info[num] = [cut_size, cut_size * (num - 1) + 1, cut_size * num]
        else:
            cut_info[num] = [cut_size, cut_size * (num - 1), cut_size * num]
        if num == thread_num:
            cut_info[num][2] = '-'
    return cut_info


def write_file(file_path, file_size):
    """
    合并分段下载的文件
    :param file_path:
    :return:
    """
    if os.path.exists(file_path):
        if len(file_path) >= file_size:
            return
    with open(file_path, 'ab') as f_count:
        for thread_index in range(1, THREAD_NUM + 1):
            with open("{}_{}".format(file_path, thread_index), 'rb') as sub_write:
                f_count.write(sub_write.read())
            # 合并完成删除子文件
            os.remove("{}_{}".format(file_path, thread_index))
    return


def create_thread(file_path, etag, cut_info):
    """
    开辟多线程下载
    :param file_path: 文件存储路径
    :param etag: headers校验
    :param cut_info:
    :return:
    """
    thread_list = []
    for thread_index in range(1, THREAD_NUM + 1):
        thread_list.append(Thread(target=download_file, args=(
            file_path, thread_index, etag, cut_info[thread_index][1], cut_info[thread_index][2])))
    for t in thread_list:
        t.setDaemon(True)
        t.start()
    for t in thread_list:
        t.join()
    return


def main():
    # 平分几份
    global tqdm_obj
    file_size, etag = get_file_size()
    # 按线程数量均匀切割下载文件
    cut_info = cutting(file_size, THREAD_NUM)
    # 下载文件名称
    data_folder = os.path.join(os.path.dirname(os.path.realpath(__file__)), 'Data')
    if not os.path.exists(data_folder):
        os.mkdir(data_folder)
    file_path = os.path.join(data_folder, download_url.split('/')[-1])
    # 创建下载进度条
    tqdm_obj = tqdm(total=file_size, unit_scale=True, desc=file_path.split('/')[-1], unit_divisor=1024, unit="B",
                    )
    # 开始多线程下载
    create_thread(file_path, etag, cut_info)
    # 合并多线程下载文件
    write_file(file_path, file_size)
    return


if __name__ == '__main__':
    main()