Python实现GCS bucket断点续传功能，分块上传文件
环境：Python 3.6
我有一个关于使用断点续传到Google Cloud Storage的上传速度的问题。我已经编写了一个Python客户端，用于将大文件上传到GCS（它具有一些特殊功能，这就是为什么gsutil对我公司不适用的原因）。在大约2个月前运行的测试中，它很好地利用了可用的连接带宽，其中25Mbps连接中大约有20Mbps。该项目被冻结了将近2个月，现在，当重新打开该项目时，同一客户端以非常慢的速度上载，速度约为25Mbps的1.4Mbps。我已经编写了简单的Python脚本来检查它是否也会遇到相同的问题，并且速度稍快一些，但仍约为2Mbps。Gsutil工具的执行效果几乎与我的Python脚本相同。我还以超过50Mbps的上传速度在不同的网络基础架构上运行了该测试，效果非常好。
参考地址：https://googleapis.dev/python/google-resumable-media/latest/resumable_media/requests.html#resumable-uploads
import google.authimport google.auth.transport.requests as tr_requestsro_scope = u'/auth/devstorage.read_only'credentials, _ = google.auth.default(scopes=(ro_scope,))transport = tr_requests.AuthorizedSession(credentials)from google.resumable_media.requests import ResumableUploadimport iobucket_name='xxxxxxx' # 桶名csvfile_name = 'xxxxxxxxxxxxxxxxxxxx' # 文件名路径url_template = (u'/upload/storage/v1/b/'+ bucket_name +'/o?'u'uploadType=resumable')upload_url = url_template.format(bucket=bucket_name)# 分块传输的大小chunk_size = 1024 * 1024 * 33 # 33MB# 开始断点续传，并分块，意思是说，一个文件比如50M，33M每块要执行两次这个语句upload = ResumableUpload(upload_url, chunk_size)print(response)print(upload.resumable_url == response.headers[u'Location'])print(upload.total_bytes == len(data))upload_id = response.headers[u'X-GUploader-UploadID']print(upload_id)print(upload.resumable_url == upload_url + u'&upload_id=' + upload_id)response0 = upload.transmit_next_chunk(transport)print(response0)print(upload.finished)print(upload.bytes_uploaded == upload.chunk_size)response1 = upload.transmit_next_chunk(transport)print(response1)print(upload.finished)print(upload.bytes_uploaded == 2 * upload.chunk_size)response2 = upload.transmit_next_chunk(transport)print(response2)print(upload.finished)print(upload.bytes_uploaded == upload.total_bytes)json_response = response2.json()print(json_response[u'bucket'] == bucket)print(json_response[u'name'] == blob_name)
