|
| 1 | +/* |
| 2 | + * Copyright (c) Meta Platforms, Inc. and affiliates. |
| 3 | + * |
| 4 | + * This source code is licensed under the MIT license found in the |
| 5 | + * LICENSE file in the root directory of this source tree. |
| 6 | + */ |
| 7 | + |
| 8 | +#include "fbpcf/io/cloud_util/GCSFileUploader.h" |
| 9 | +#include <aws/s3/model/AbortMultipartUploadRequest.h> |
| 10 | +#include <aws/s3/model/CompleteMultipartUploadRequest.h> |
| 11 | +#include <aws/s3/model/CompletedMultipartUpload.h> |
| 12 | +#include <aws/s3/model/CreateMultipartUploadRequest.h> |
| 13 | +#include <aws/s3/model/UploadPartRequest.h> |
| 14 | +#include <folly/logging/xlog.h> |
| 15 | +#include "fbpcf/aws/S3Util.h" |
| 16 | +#include "fbpcf/exception/AwsException.h" |
| 17 | +#include "fbpcf/gcp/GCSUtil.h" |
| 18 | + |
| 19 | +namespace fbpcf::cloudio { |
| 20 | + |
| 21 | +static const std::string FILE_TYPE = "text/csv"; |
| 22 | +static const int MAX_RETRY_COUNT = 3; |
| 23 | + |
| 24 | +void GCSFileUploader::init() { |
| 25 | + XLOG(INFO) << "Start multipart upload initialization. "; |
| 26 | + const auto& ref = fbpcf::gcp::uriToObjectReference(filePath_); |
| 27 | + bucket_ = ref.bucket; |
| 28 | + key_ = ref.key; |
| 29 | + Aws::S3::Model::CreateMultipartUploadRequest request; |
| 30 | + request.SetBucket(bucket_); |
| 31 | + request.SetKey(key_); |
| 32 | + request.SetContentType(FILE_TYPE); |
| 33 | + |
| 34 | + XLOG(INFO) << "Bucket: " << bucket_ << ", Key: " << key_; |
| 35 | + |
| 36 | + auto createMultipartUploadOutcome = |
| 37 | + gcsClient_->CreateMultipartUpload(request); |
| 38 | + |
| 39 | + if (createMultipartUploadOutcome.IsSuccess()) { |
| 40 | + uploadId_ = createMultipartUploadOutcome.GetResult().GetUploadId(); |
| 41 | + XLOG(INFO) << "Multipart upload initialization succeed. Upload id is: " |
| 42 | + << uploadId_; |
| 43 | + } else { |
| 44 | + XLOG(ERR) << createMultipartUploadOutcome.GetError(); |
| 45 | + throw AwsException{ |
| 46 | + "Multipart upload initialization failed: " + |
| 47 | + createMultipartUploadOutcome.GetError().GetMessage()}; |
| 48 | + } |
| 49 | +} |
| 50 | + |
| 51 | +int GCSFileUploader::upload(std::vector<char>& buf) { |
| 52 | + XLOG(INFO) << "Start uploading part:" |
| 53 | + << "Part number: " << partNumber_ << "\nBucket: " << bucket_ |
| 54 | + << "\nKey: " << key_; |
| 55 | + Aws::S3::Model::UploadPartRequest request; |
| 56 | + request.SetBucket(bucket_); |
| 57 | + request.SetKey(key_); |
| 58 | + request.SetUploadId(uploadId_); |
| 59 | + request.SetPartNumber(partNumber_); |
| 60 | + request.SetContentLength(buf.size()); |
| 61 | + |
| 62 | + Aws::String str(buf.begin(), buf.end()); |
| 63 | + auto inputData = Aws::MakeShared<Aws::StringStream>("UploadPartStream", str); |
| 64 | + request.SetBody(inputData); |
| 65 | + XLOG(INFO) << "Upload stream size: " << str.size(); |
| 66 | + |
| 67 | + auto uploadPartResult = gcsClient_->UploadPart(request); |
| 68 | + int retryCount = 0; |
| 69 | + while (!uploadPartResult.IsSuccess() && retryCount < MAX_RETRY_COUNT) { |
| 70 | + XLOG(INFO) << "Upload part " << partNumber_ << " failed. Retrying..."; |
| 71 | + uploadPartResult = gcsClient_->UploadPart(request); |
| 72 | + retryCount++; |
| 73 | + } |
| 74 | + |
| 75 | + if (uploadPartResult.IsSuccess()) { |
| 76 | + XLOG(INFO) << "Upload part " << partNumber_ << " succeeed."; |
| 77 | + Aws::S3::Model::CompletedPart part; |
| 78 | + part.SetPartNumber(request.GetPartNumber()); |
| 79 | + part.SetETag(uploadPartResult.GetResult().GetETag()); |
| 80 | + completedParts_.push_back(part); |
| 81 | + partNumber_++; |
| 82 | + return str.size(); |
| 83 | + } else { |
| 84 | + XLOG(INFO) << "Upload part " << partNumber_ << " failed. Aborting..."; |
| 85 | + abortUpload(); |
| 86 | + return 0; |
| 87 | + } |
| 88 | +} |
| 89 | + |
| 90 | +int GCSFileUploader::complete() { |
| 91 | + Aws::S3::Model::CompleteMultipartUploadRequest request; |
| 92 | + request.SetBucket(bucket_); |
| 93 | + request.SetKey(key_); |
| 94 | + request.SetUploadId(uploadId_); |
| 95 | + request.SetMultipartUpload( |
| 96 | + Aws::S3::Model::CompletedMultipartUpload().WithParts(completedParts_)); |
| 97 | + |
| 98 | + auto completeMultipartUploadResult = |
| 99 | + gcsClient_->CompleteMultipartUpload(request); |
| 100 | + if (completeMultipartUploadResult.IsSuccess()) { |
| 101 | + XLOG(INFO) << "File " << filePath_ << " uploaded successfully."; |
| 102 | + return 0; |
| 103 | + } else { |
| 104 | + XLOG(ERR) << "File " << filePath_ << " failed to upload."; |
| 105 | + XLOG(ERR) << "Error: " << completeMultipartUploadResult.GetError(); |
| 106 | + abortUpload(); |
| 107 | + return -1; |
| 108 | + } |
| 109 | +} |
| 110 | + |
| 111 | +void GCSFileUploader::abortUpload() { |
| 112 | + Aws::S3::Model::AbortMultipartUploadRequest abortRequest; |
| 113 | + abortRequest.SetBucket(bucket_); |
| 114 | + abortRequest.SetKey(key_); |
| 115 | + abortRequest.SetUploadId(uploadId_); |
| 116 | + auto abortMultipartUploadResult = |
| 117 | + gcsClient_->AbortMultipartUpload(abortRequest); |
| 118 | + if (abortMultipartUploadResult.IsSuccess()) { |
| 119 | + XLOG(INFO) << "Abort upload successed. "; |
| 120 | + } else { |
| 121 | + XLOG(ERR) << "Abort upload failed. Upload ID: " + uploadId_; |
| 122 | + } |
| 123 | +} |
| 124 | + |
| 125 | +} // namespace fbpcf::cloudio |
0 commit comments