将java对象转换为csv并上传到谷歌云存储

Question

我有一个休息端点，它接受一个对象（requestbody）作为输入。我需要读取该对象，将其转换为 csv 并将 csv 上传到谷歌云存储。

我在下面分享我迄今为止所取得的成就：

@GetMapping("/upload")
public void getTranscriptsByDate(@RequestBody SearchRequest searchRequest) throws Exception{
    log.info("Search Request {}", searchRequest);

    createFileWithSerachCriteria(searchRequest);
    uploadFileToGoogleCloudStorage(searchRequest);

}

private void createFileWithSerachCriteria(SearchRequest searchRequest) throws IOException {
    String filename =  "myFile" + searchRequest.getConversationDate() + ".csv";
    String filePath = "C://Users//abc//" + filename;

    CSVWriter writer = new CSVWriter(new FileWriter(filePath));
    String[] header = { "ID", "NAME" };
    writer.writeNext(header);

    String[] data = new String[2];
    data[0] = searchRequest.getId();
    data[1] = searchRequest.getName();
    writer.writeNext(data);

    writer.close();
}

private void uploadFileToGoogleCloudStorage(SearchRequest searchRequest) throws Exception{
    tring filename =  "myFile" + searchRequest.getConversationDate() + ".csv";
    String filePath = "C://Users//abc//" + filename;
    myService.uploadFile(filePath);
}

下面是来自MyService.java的代码

public void uploadFile(String localFilePath) throws Exception{
    BlobId blobId = BlobId.of(bucketName, localFilePath);
    BlobInfo blobInfo = BlobInfo.newBuilder(blobId).build();
    storage.createFrom(blobInfo, Paths.get(localFilePath));
}

这段代码运行良好。

我不想在本地写入文件然后读回并上传到云端。我想直接将我的请求对象转换为 csv 并将其上传到云上，而不需要两步（读/写）过程。在官方GCP文档上，我找到了一个流式传输的小示例，但我无法理解它。

是否有可能实现我想要做的事情，即避免读/写并一次性完成。请帮忙。

Answer 1

您可以使用字节流在内存中而不是文件系统中读取和写入文件数据：

ByteArrayOutputStream stream = new ByteArrayOutputStream();
OutputStreamWriter streamWriter = new OutputStreamWriter(stream);
CSVWriter writer = new CSVWriter(streamWriter);

// add data to writer and close

BlobId blobId = BlobId.of(bucketName, someDerivedFilePath);
BlobInfo blobInfo = BlobInfo.newBuilder(blobId).build();
storage.createFrom(blobInfo, new ByteArrayInputStream(stream.toByteArray()));

仅供参考，我还没有测试过这个，但看起来应该可行。