Apache Http Client slower than browser download

2020-07-30 02:51发布

问题:

I am downloading a large repository from github of size 300M. It takes 10-15sec when I download from my browser. On the same machine, it takes 110-120sec when I use below code to download. I am wondering if I am doing wrong. Please suggest me to get the same speed(10-15sec) using apache http client. Or is there anything better than http client ?

Apache httpclient = 4.5

java - 8

Code that I used:

import java.io.File;
import java.io.IOException;
import java.io.InputStream;
import java.net.URL;

import org.apache.commons.io.FileUtils;
import org.apache.commons.io.IOUtils;
import org.apache.http.HttpResponse;
import org.apache.http.client.ClientProtocolException;
import org.apache.http.client.ResponseHandler;
import org.apache.http.client.methods.HttpGet;
import org.apache.http.impl.client.CloseableHttpClient;
import org.apache.http.impl.client.HttpClientBuilder;
import org.apache.http.impl.client.LaxRedirectStrategy;
import org.apache.http.impl.conn.PoolingHttpClientConnectionManager;

public class Downloader {

    public File download(URL url, File dstFile) {
        PoolingHttpClientConnectionManager manager = new PoolingHttpClientConnectionManager();
        manager.setDefaultMaxPerRoute(20);
        manager.setMaxTotal(200);
        CloseableHttpClient httpclient = HttpClientBuilder.create()
                .setConnectionManager(manager)
                .build();

//  Second option: it also takes same time.
//        .setRedirectStrategy(new LaxRedirectStrategy())
//        .setMaxConnTotal(2 * 50)
//        .setMaxConnPerRoute(50)
//        .build();
//      CloseableHttpClient httpclient = HttpClients.custom()
//              .setRedirectStrategy(new LaxRedirectStrategy()) // adds HTTP REDIRECT support to GET and POST methods 
//              .build();
        try {
            HttpGet get = new HttpGet(url.toURI()); // we're using GET but it could be via POST as well
            File downloaded = httpclient.execute(get, new FileDownloadResponseHandler(dstFile));
            return downloaded;
        } catch (Exception e) {
            throw new IllegalStateException(e);
        } finally {
            IOUtils.closeQuietly(httpclient);
        }
    }

    static class FileDownloadResponseHandler implements ResponseHandler<File> {

        private final File target;

        public FileDownloadResponseHandler(File target) {
            this.target = target;
        }

        @Override
        public File handleResponse(HttpResponse response) throws ClientProtocolException, IOException {
            InputStream source = response.getEntity().getContent();
            FileUtils.copyInputStreamToFile(source, this.target);
            return this.target;
        }

    }

}

回答1:

I had the same problem; the fix was implementing the suggestion from here: https://stackoverflow.com/a/35458078/9204142, i.e. use disableContentCompression() or contentCompressionEnabled=false at the RequestConfig level. After the fix the speed matched the one of curl.

As I've used Apache Camel the fix implied adding ?httpClient.contentCompressionEnabled=false in the URI of the Endpoint.