Using Rest-Client to Download a File to Disk Without Loading It All in Memory First

Using rest-client to download a file to disk without loading it all in memory first

Another way is to use raw_response. This saves directly to a file, usually in /tmp and handles redirects without a problem.
See Streaming Responses. Here's their example:

>> raw = RestClient::Request.execute(
           method: :get,
           url: 'http://releases.ubuntu.com/16.04.2/ubuntu-16.04.2-desktop-amd64.iso',
           raw_response: true)
=> <RestClient::RawResponse @code=200, @file=#<Tempfile:/tmp/rest-client.20170522-5346-1pptjm1>, @request=<RestClient::Request @method="get", @url="http://releases.ubuntu.com/16.04.2/ubuntu-16.04.2-desktop-amd64.iso">>
>> raw.file.size
=> 1554186240
>> raw.file.path
=> "/tmp/rest-client.20170522-5346-1pptjm1"

How to use restsharp to download file

With RestSharp, it's right there in the readme:

var client = new RestClient("http://example.com");
client.DownloadData(request).SaveAs(path);

With HttpClient, it's a bit more involved. Have a look at this blog post.

Another option is Flurl.Http (disclaimer: I'm the author). It uses HttpClient under the hood and provides a fluent interface and lots of convenient helper methods, including:

await "http://example.com".DownloadFileAsync(folderPath, "foo.xml");

Get it on NuGet.

How to download a file using a Java REST service and a data stream

"How can I directly (without saving the file on 2nd server) download the file from 1st server to client's machine?"

Just use the Client API and get the InputStream from the response

Client client = ClientBuilder.newClient();
String url = "...";
final InputStream responseStream = client.target(url).request().get(InputStream.class);

There are two flavors to get the InputStream. You can also use

Response response = client.target(url).request().get();
InputStream is = (InputStream)response.getEntity();

Which one is the more efficient? I'm not sure, but the returned InputStreams are different classes, so you may want to look into that if you care to.

From 2nd server I can get a ByteArrayOutputStream to get the file from 1st server, can I pass this stream further to the client using the REST service?

So most of the answers you'll see in the link provided by @GradyGCooper seem to favor the use of StreamingOutput. An example implementation might be something like

final InputStream responseStream = client.target(url).request().get(InputStream.class);
System.out.println(responseStream.getClass());
StreamingOutput output = new StreamingOutput() {
    @Override
    public void write(OutputStream out) throws IOException, WebApplicationException {  
        int length;
        byte[] buffer = new byte[1024];
        while((length = responseStream.read(buffer)) != -1) {
            out.write(buffer, 0, length);
        }
        out.flush();
        responseStream.close();
    }   
};
return Response.ok(output).header(
        "Content-Disposition", "attachment, filename=\"...\"").build();

But if we look at the source code for StreamingOutputProvider, you'll see in the writeTo, that it simply writes the data from one stream to another. So with our implementation above, we have to write twice.

How can we get only one write? Simple return the InputStream as the Response

final InputStream responseStream = client.target(url).request().get(InputStream.class);
return Response.ok(responseStream).header(
        "Content-Disposition", "attachment, filename=\"...\"").build();

If we look at the source code for InputStreamProvider, it simply delegates to ReadWriter.writeTo(in, out), which simply does what we did above in the StreamingOutput implementation

 public static void writeTo(InputStream in, OutputStream out) throws IOException {
    int read;
    final byte[] data = new byte[BUFFER_SIZE];
    while ((read = in.read(data)) != -1) {
        out.write(data, 0, read);
    }
}

Asides:

• Client objects are expensive resources. You may want to reuse the same Client for request. You can extract a WebTarget from the client for each request.

  WebTarget target = client.target(url);
  InputStream is = target.request().get(InputStream.class);
  

  I think the WebTarget can even be shared. I can't find anything in the Jersey 2.x documentation (only because it is a larger document, and I'm too lazy to scan through it right now :-), but in the Jersey 1.x documentation, it says the Client and WebResource (which is equivalent to WebTarget in 2.x) can be shared between threads. So I'm guessing Jersey 2.x would be the same. but you may want to confirm for yourself.

• You don't have to make use of the Client API. A download can be easily achieved with the java.net package APIs. But since you're already using Jersey, it doesn't hurt to use its APIs

• The above is assuming Jersey 2.x. For Jersey 1.x, a simple Google search should get you a bunch of hits for working with the API (or the documentation I linked to above)

---

UPDATE

I'm such a dufus. While the OP and I are contemplating ways to turn a ByteArrayOutputStream to an InputStream, I missed the simplest solution, which is simply to write a MessageBodyWriter for the ByteArrayOutputStream

import java.io.ByteArrayOutputStream;
import java.io.IOException;
import java.io.OutputStream;
import java.lang.annotation.Annotation;
import java.lang.reflect.Type;
import javax.ws.rs.WebApplicationException;
import javax.ws.rs.core.MediaType;
import javax.ws.rs.core.MultivaluedMap;
import javax.ws.rs.ext.MessageBodyWriter;
import javax.ws.rs.ext.Provider;

@Provider
public class OutputStreamWriter implements MessageBodyWriter<ByteArrayOutputStream> {

    @Override
    public boolean isWriteable(Class<?> type, Type genericType,
            Annotation[] annotations, MediaType mediaType) {
        return ByteArrayOutputStream.class == type;
    }

    @Override
    public long getSize(ByteArrayOutputStream t, Class<?> type, Type genericType,
            Annotation[] annotations, MediaType mediaType) {
        return -1;
    }

    @Override
    public void writeTo(ByteArrayOutputStream t, Class<?> type, Type genericType,
            Annotation[] annotations, MediaType mediaType,
            MultivaluedMap<String, Object> httpHeaders, OutputStream entityStream)
            throws IOException, WebApplicationException {
        t.writeTo(entityStream);
    }
}

Then we can simply return the ByteArrayOutputStream in the response

return Response.ok(baos).build();

D'OH!

UPDATE 2

Here are the tests I used (

Resource class

@Path("test")
public class TestResource {

    final String path = "some_150_mb_file";

    @GET
    @Produces(MediaType.APPLICATION_OCTET_STREAM)
    public Response doTest() throws Exception {
        InputStream is = new FileInputStream(path);
        ByteArrayOutputStream baos = new ByteArrayOutputStream();
        int len;
        byte[] buffer = new byte[4096];
        while ((len = is.read(buffer, 0, buffer.length)) != -1) {
            baos.write(buffer, 0, len);
        }
        System.out.println("Server size: " + baos.size());
        return Response.ok(baos).build();
    }
}

Client test

public class Main {
    public static void main(String[] args) throws Exception {
        Client client = ClientBuilder.newClient();
        String url = "http://localhost:8080/api/test";
        Response response = client.target(url).request().get();
        String location = "some_location";
        FileOutputStream out = new FileOutputStream(location);
        InputStream is = (InputStream)response.getEntity();
        int len = 0;
        byte[] buffer = new byte[4096];
        while((len = is.read(buffer)) != -1) {
            out.write(buffer, 0, len);
        }
        out.flush();
        out.close();
        is.close();
    }
}

UPDATE 3

So the final solution for this particular use case was for the OP to simply pass the OutputStream from the StreamingOutput's write method. Seems the third-party API, required a OutputStream as an argument.

StreamingOutput output = new StreamingOutput() {
    @Override
    public void write(OutputStream out) {
        thirdPartyApi.downloadFile(.., .., .., out);
    }
}
return Response.ok(output).build();

Not quite sure, but seems the reading/writing within the resource method, using ByteArrayOutputStream`, realized something into memory.

The point of the downloadFile method accepting an OutputStream is so that it can write the result directly to the OutputStream provided. For instance a FileOutputStream, if you wrote it to file, while the download is coming in, it would get directly streamed to the file.

It's not meant for us to keep a reference to the OutputStream, as you were trying to do with the baos, which is where the memory realization comes in.

So with the way that works, we are writing directly to the response stream provided for us. The method write doesn't actually get called until the writeTo method (in the MessageBodyWriter), where the OutputStream is passed to it.

You can get a better picture looking at the MessageBodyWriter I wrote. Basically in the writeTo method, replace the ByteArrayOutputStream with StreamingOutput, then inside the method, call streamingOutput.write(entityStream). You can see the link I provided in the earlier part of the answer, where I link to the StreamingOutputProvider. This is exactly what happens

Scala & Finatra: send file server response directly from disk to network w/o loading into memory

It's unfortunately not possible with Finatra 1.6. Streaming seems to be on the roadmap for 2.0, but there is no official word on its release. Right now, Request => Response relies on memorized inputs and outputs. The most efficient way to deal with bodies in Finatra is to keep them in ChannelBuffers so that there is at least only one instance of the bytes materialized.

Download Large file from server using REST template Java Spring MVC

Here is how I do it. Based on hints from this Spring Jira issue.

RestTemplate restTemplate // = ...;

// Optional Accept header
RequestCallback requestCallback = request -> request.getHeaders()
        .setAccept(Arrays.asList(MediaType.APPLICATION_OCTET_STREAM, MediaType.ALL));

// Streams the response instead of loading it all in memory
ResponseExtractor<Void> responseExtractor = response -> {
    // Here I write the response to a file but do what you like
    Path path = Paths.get("some/path");
    Files.copy(response.getBody(), path);
    return null;
};
restTemplate.execute(URI.create("www.something.com"), HttpMethod.GET, requestCallback, responseExtractor);

From the aforementioned Jira issue:

Note that you cannot simply return the InputStream from the extractor, because by the time the execute method returns, the underlying connection and stream are already closed.

Update for Spring 5

Spring 5 introduced the WebClient class which allows asynchronous (e.g. non-blocking) http requests. From the doc:

By comparison to the RestTemplate, the WebClient is:

• non-blocking, reactive, and supports higher concurrency with less hardware resources.
• provides a functional API that takes advantage of Java 8 lambdas.
• supports both synchronous and asynchronous scenarios.
• supports streaming up or down from a server.

To get WebClient in Spring Boot, you need this dependency:

<dependency>
    <groupId>org.springframework.boot</groupId>
    <artifactId>spring-boot-starter-webflux</artifactId>
</dependency>

For the moment, I'm sticking with RestTemplate because I don't want to pull in another dependency only to get access to WebClient.

c# Dropbox RestSharp download file to stream

I found the best answer in link:

 string url = String.Format("https://content.dropboxapi.com/1/files/auto{0}?oauth_consumer_key={1}&oauth_token={2}&oauth_signature={3}%26{4}", path, app-key, access-token, app-secret, access-token-secret);
 HttpWebRequest webRequest = (HttpWebRequest)WebRequest.Create(url);
 webRequest.Method = "Get";
 WebResponse webResponse = null;
  try
  {
     webResponse = webRequest.GetResponse();
     return webResponse.GetResponseStream();
  }
  catch (Exception ex)
  {
    throw ex;       
  }

---

Related Topics