1
|
package eu.dnetlib.data.collector.plugins;
|
2
|
|
3
|
import java.io.BufferedInputStream;
|
4
|
import java.io.ByteArrayInputStream;
|
5
|
import java.io.IOException;
|
6
|
import java.io.InputStream;
|
7
|
|
8
|
import eu.dnetlib.rmi.data.CollectorServiceException;
|
9
|
import org.apache.commons.io.IOUtils;
|
10
|
import org.apache.http.HttpStatus;
|
11
|
import org.apache.http.client.methods.CloseableHttpResponse;
|
12
|
import org.apache.http.client.methods.HttpGet;
|
13
|
import org.apache.http.impl.client.CloseableHttpClient;
|
14
|
import org.apache.http.impl.client.HttpClients;
|
15
|
|
16
|
public class HttpCollectorPlugin extends AbstractSplittedRecordPlugin {
|
17
|
|
18
|
@Override
|
19
|
protected BufferedInputStream getBufferedInputStream(final String baseUrl) throws CollectorServiceException {
|
20
|
final HttpGet method = new HttpGet(baseUrl);
|
21
|
|
22
|
try(CloseableHttpResponse response = HttpClients.createDefault().execute(method)) {
|
23
|
|
24
|
int responseCode = response.getStatusLine().getStatusCode();
|
25
|
|
26
|
if (HttpStatus.SC_OK != responseCode) {
|
27
|
throw new CollectorServiceException("Error " + responseCode + " downloading url: " + baseUrl);
|
28
|
}
|
29
|
|
30
|
byte[] content = IOUtils.toByteArray(response.getEntity().getContent());
|
31
|
|
32
|
try(InputStream in = new ByteArrayInputStream(content)) {
|
33
|
return new BufferedInputStream(in);
|
34
|
}
|
35
|
} catch (IOException e) {
|
36
|
throw new CollectorServiceException("Error downloading url: " + baseUrl);
|
37
|
}
|
38
|
}
|
39
|
|
40
|
}
|