从 Google Cloud Storage Bucket 下载文件夹

Question

我是 Google Cloud Platform 的新手。我已经在 datalab 上训练了我的模型，并将模型文件夹保存在我的存储桶中的云存储中。通过右键单击文件->另存为链接，我可以将存储桶中的现有文件下载到本地计算机。但是当我尝试通过与上述相同的过程下载文件夹时，我得到的不是文件夹而是它的图像。无论如何我可以下载整个文件夹及其内容吗？是否有任何 gsutil 命令可以将文件夹从云存储复制到本地目录？

Answer 1

您可以在 gsutil 工具 here 上找到文档，更具体地说是在 here 上找到您的问题。

你要使用的命令是：

gsutil cp -r gs://bucket/folder .

Answer 2

先决条件： 已安装并初始化 Google Cloud SDK ($ glcoud init)

命令：

gsutil -m cp -r  gs://bucket-name .

这将使用更快的多线程复制所有文件。我发现官方 Gsutil Docs 中指示使用的“dir”命令不起作用。

Answer 3

这是从 Google Cloud Storage Bucket 下载文件夹的方法

运行以下命令将其从存储桶存储下载到您的 Google Cloud Console 本地路径

gsutil -m cp -r gs://{bucketname}/{folderPath} {localpath}

运行该命令后，通过运行ls 命令列出本地路径上的文件和目录，确认您的文件夹位于本地路径中

现在通过运行以下命令压缩您的文件夹

zip -r foldername.zp yourfolder/*

压缩过程完成后，点击 Google Cloud Console 右侧的更多下拉菜单，

然后选择“下载文件”选项。系统会提示你输入要下载的文件名，输入zip文件名-“foldername.zp”

Answer 4

如果您使用 python 从谷歌云存储中下载数据并希望保持相同的文件夹结构，请按照我在 python 中编写的这段代码。

选项 1

from google.cloud import storage

def findOccurrences(s, ch): # to find position of '/' in blob path ,used to create folders in local storage
    return [i for i, letter in enumerate(s) if letter == ch]

def download_from_bucket(bucket_name, blob_path, local_path):    
    # Create this folder locally
    if not os.path.exists(local_path):
        os.makedirs(local_path)        

    storage_client = storage.Client()
    bucket = storage_client.get_bucket(bucket_name)
    blobs=list(bucket.list_blobs(prefix=blob_path))

    startloc = 0
    for blob in blobs:
        startloc = 0
        folderloc = findOccurrences(blob.name.replace(blob_path, ''), '/') 
        if(not blob.name.endswith("/")):
            if(blob.name.replace(blob_path, '').find("/") == -1):
                downloadpath=local_path + '/' + blob.name.replace(blob_path, '')
                logging.info(downloadpath)
                blob.download_to_filename(downloadpath)
            else:
                for folder in folderloc:
                    
                    if not os.path.exists(local_path + '/' + blob.name.replace(blob_path, '')[startloc:folder]):
                        create_folder=local_path + '/' +blob.name.replace(blob_path, '')[0:startloc]+ '/' +blob.name.replace(blob_path, '')[startloc:folder]
                        startloc = folder + 1
                        os.makedirs(create_folder)
                    
                downloadpath=local_path + '/' + blob.name.replace(blob_path, '')

                blob.download_to_filename(downloadpath)
                logging.info(blob.name.replace(blob_path, '')[0:blob.name.replace(blob_path, '').find("/")])

    logging.info('Blob {} downloaded to {}.'.format(blob_path, local_path))


bucket_name = 'google-cloud-storage-bucket-name' # do not use gs://
blob_path = 'training/data' # blob path in bucket where data is stored 
local_dir = 'local-folder name' #trainingData folder in local
download_from_bucket(bucket_name, blob_path, local_dir)

选项 2：使用 gsutil sdk 下面是通过 python 程序执行此操作的另一种选择。

def download_bucket_objects(bucket_name, blob_path, local_path):
    # blob path is bucket folder name
    command = "gsutil cp -r gs://{bucketname}/{blobpath} {localpath}".format(bucketname = bucket_name, blobpath = blob_path, localpath = local_path)
    os.system(command)
    return command

选项 3 - 没有 python，直接使用终端和谷歌 SDK 先决条件：Google Cloud SDK 已安装并初始化（$ glcoud init） 命令参考以下链接：

https://cloud.google.com/storage/docs/gsutil/commands/cp

Answer 5

gsutil -m cp -r gs://bucket-name "{本地现有文件夹的路径}"

确实有效。

Answer 6

这是我编写的代码。 这会将完整的目录结构下载到您的虚拟机/本地存储中。

from google.cloud import storage
import os
bucket_name = "ar-data"
    
storage_client = storage.Client()
bucket = storage_client.get_bucket(bucket_name)

dirName='Data_03_09/' #***folder in bucket whose content you want to download
blobs = bucket.list_blobs(prefix = dirName)#, delimiter = '/')
destpath=r'/home/jupyter/DATA_test/' #***path on your vm/local where you want to download the bucket directory
for blob in blobs:
    #print(blob.name.lstrip(dirName).split('/'))
    currpath=destpath
    if not os.path.exists(os.path.join(destpath,'/'.join(blob.name.lstrip(dirName)).split('/')[:-1])):
        for n in blob.name.lstrip(dirName).split('/')[:-1]:
            currpath=os.path.join(currpath,n)
            if not os.path.exists(currpath):
                print('creating directory- ', n , 'On path-', currpath)
                os.mkdir(currpath)
    print("downloading ... ",blob.name.lstrip(dirName))
    blob.download_to_filename(os.path.join(destpath,blob.name.lstrip(dirName)))

或者只是在终端中使用：

gsutil -m cp -r gs://{bucketname}/{folderPath} {localpath}