AWS S3 大文件分片上传

import os
import redis
import boto3


################################## ##################################
# 把图片数据上传到本地 redis 中

conn = redis.Redis()


def store_img2redis():
    """
    把图片数据存入到 redis 的列表中
    """

    with open(‘/Users/zxk/Desktop/1min1sec.mp4‘, ‘rb‘) as f:
        while True:
            bytes_data = f.read(1024 * 1024 * 6)   # 每次读取 6 M的数据
            if not bytes_data:
                return
            conn.rpush(‘gopher_img‘, bytes_data)


################################# #################################
# 把图片数据从 redis 中迭代读取出来，并分片上传到 S3 中

def list_iter(name, count=2):
    """
    迭代获取 redis 列表中的数据
    """
    idx = 0
    while True:
        data = conn.lrange(name, idx, idx + count - 1)
        if not data:
            return

        idx += count

        for item in data:
            yield item


client = boto3.client(
    ‘s3‘,
    aws_access_key_id=os.getenv(
        ‘AWS_S3_ACCESS_KEY_ID‘, ‘xxxxxxxxxxxxxxxxxxxx‘),
    aws_secret_access_key=os.getenv(
        ‘AWS_S3_SECRET_ACCESS_KEY‘, ‘xxyxyxxyxyxxyxyxyxxy‘),
    region_name=os.getenv(‘AWS_S3_REGION_NAME‘, ‘cn-northwest-1‘)
)


part_info = {
    ‘Parts‘: []
}


def create_multipart_upld():
    """
    分片上传功能的函数
    """
    response = client.create_multipart_upload(
        Bucket=‘bucket-test‘, Key=‘static/gopher.mp4‘)   # This action initiates a multipart upload and returns an upload ID
    print(response)

    part_no = 1     # 用于标识分片上传时的数据块号，即 PartNumber
    for part_bytes in list_iter(‘gopher_img‘):
        print(‘upload ID:‘, response[‘UploadId‘])
        part = client.upload_part(
            Body=part_bytes,
            Bucket=‘bucket-test‘,
            Key=‘static/gopher.mp4‘,
            PartNumber=part_no,     # 这个号要唯一，不唯一的话会把前面相同的PartNumber数据覆盖掉
            UploadId=response[‘UploadId‘],
        )

        # 每个 PartNumber 和 ETag 要对应起来
        part_info[‘Parts‘].append({
            ‘PartNumber‘: part_no,
            ‘ETag‘: part[‘ETag‘]
        })
        part_no += 1
    print(‘part_info:‘, part_info)

    # Completes a multipart upload by assembling previously uploaded parts. 告诉 AWS S3 已经分片上传完了
    complete_rsp = client.complete_multipart_upload(Bucket=‘bucket-test‘,
                                                    Key=‘static/gopher.mp4‘,
                                                    UploadId=response[‘UploadId‘],
                                                    MultipartUpload=part_info)
    print(‘complete_rsp:‘, complete_rsp)


create_multipart_upld()


"""
输出结果：
{‘ResponseMetadata‘: {‘RequestId‘: ‘NVDSJHCASJD‘, ‘HostId‘: ‘MCDCSNDSNAnvdsahnvdasjk‘, ‘HTTPStatusCode‘: 200, ‘HTTPHeaders‘: {‘x-amz-id-2‘: ‘aJGYwJy7UhP/PtFW/OprgKvLxDMi9fUZJuvR2925pnzvk=‘, ‘x-amz-request-id‘: ‘NVDSJHCASJD‘, ‘date‘: ‘Fri, 02 Jul 2021 15:22:52 GMT‘, ‘transfer-encoding‘: ‘chunked‘, ‘server‘: ‘AmazonS3‘}, ‘RetryAttempts‘: 0}, ‘Bucket‘: ‘bucket-test‘, ‘Key‘: ‘static/gopher.mp4‘, ‘UploadId‘: ‘5GuYO83ZlEkhPzApxM49nxvdpvNGF3AQkG0FZ..ALMMpoZGm8KaFBVZOf2qZWY3PGBRgDXZDZXUy_tiNNYeBUOpuUbIm_meupByUIZTdzy4CYkiy0CXKrDf1_lKeGoiv‘}
upload ID: 5GuYO83ZlEkhPzApxM49nxvdpvNGF3AQkG0FZ..ALMMpoZGm8KaFBVZOf2qZWY3PGBRgDXZDZXUy_tiNNYeBUOpuUbIm_meupByUIZTdzy4CYkiy0CXKrDf1_lKeGoiv
upload ID: 5GuYO83ZlEkhPzApxM49nxvdpvNGF3AQkG0FZ..ALMMpoZGm8KaFBVZOf2qZWY3PGBRgDXZDZXUy_tiNNYeBUOpuUbIm_meupByUIZTdzy4CYkiy0CXKrDf1_lKeGoiv
upload ID: 5GuYO83ZlEkhPzApxM49nxvdpvNGF3AQkG0FZ..ALMMpoZGm8KaFBVZOf2qZWY3PGBRgDXZDZXUy_tiNNYeBUOpuUbIm_meupByUIZTdzy4CYkiy0CXKrDf1_lKeGoiv
upload ID: 5GuYO83ZlEkhPzApxM49nxvdpvNGF3AQkG0FZ..ALMMpoZGm8KaFBVZOf2qZWY3PGBRgDXZDZXUy_tiNNYeBUOpuUbIm_meupByUIZTdzy4CYkiy0CXKrDf1_lKeGoiv
upload ID: 5GuYO83ZlEkhPzApxM49nxvdpvNGF3AQkG0FZ..ALMMpoZGm8KaFBVZOf2qZWY3PGBRgDXZDZXUy_tiNNYeBUOpuUbIm_meupByUIZTdzy4CYkiy0CXKrDf1_lKeGoiv
part_info: {‘Parts‘: [{‘PartNumber‘: 1, ‘ETag‘: ‘"3e3d3bae6867b8d207cae1974be3e"‘}, {‘PartNumber‘: 2, ‘ETag‘: ‘"92720cb8294ee04961345f3b664d7"‘}, {‘PartNumber‘: 3, ‘ETag‘: ‘"cc20868db2f8c8c4a6aa499340e0c"‘}, {‘PartNumber‘: 4, ‘ETag‘: ‘"e1bee9034f163677e5ac917576dc7"‘}, {‘PartNumber‘: 5, ‘ETag‘: ‘"b3cd45be65e828debf0d635e75a46"‘}]}
complete_rsp: {‘ResponseMetadata‘: {‘RequestId‘: ‘B9TJF3EZZ8GSDV2V‘, ‘HostId‘: ‘g7r9/b1OQ8kPLoiDMlKbDREkoRapcDD+zsUJSf4NvlchxT2sS2xHxc42MyJO5ZlcAQ=‘, ‘HTTPStatusCode‘: 200, ‘HTTPHeaders‘: {‘x-amz-id-2‘: ‘g7r9/b1OQ8kPLoiDMlKbqbE51X9iJDRapcDD+zsUJSf4NvlchxT2sS2xHxc42MyJO5ZlcAQ=‘, ‘x-amz-request-id‘: ‘B9TJF3EZZ8GSDV2V‘, ‘date‘: ‘Fri, 02 Jul 2021 15:23:08 GMT‘, ‘content-type‘: ‘application/xml‘, ‘transfer-encoding‘: ‘chunked‘, ‘server‘: ‘AmazonS3‘}, ‘RetryAttempts‘: 0}, ‘Location‘: ‘https://bucket-test.s3.cn-northwest-1.amazonaws.com.cn/static%2Fgopher.mp4‘, ‘Bucket‘: ‘bucket-test‘, ‘Key‘: ‘static/gopher.mp4‘, ‘ETag‘: ‘"2b287762a4300abda18268e1151c83e9-5"‘}
"""