| 1 |
49643
|
alessia.ba
|
package eu.dnetlib.data.collector.plugins;
|
| 2 |
28959
|
michele.ar
|
|
| 3 |
29018
|
michele.ar
|
import java.io.IOException;
|
| 4 |
29169
|
andrea.man
|
import java.io.InputStream;
|
| 5 |
51556
|
claudio.at
|
import java.net.*;
|
| 6 |
28959
|
michele.ar
|
import java.security.GeneralSecurityException;
|
| 7 |
|
|
import java.security.cert.X509Certificate;
|
| 8 |
|
|
import java.util.List;
|
| 9 |
|
|
import java.util.Map;
|
| 10 |
|
|
import javax.net.ssl.HttpsURLConnection;
|
| 11 |
|
|
import javax.net.ssl.SSLContext;
|
| 12 |
|
|
import javax.net.ssl.TrustManager;
|
| 13 |
|
|
import javax.net.ssl.X509TrustManager;
|
| 14 |
|
|
|
| 15 |
51556
|
claudio.at
|
import eu.dnetlib.data.collector.plugin.CollectorPluginErrorLogList;
|
| 16 |
|
|
import eu.dnetlib.data.collector.rmi.CollectorServiceException;
|
| 17 |
29018
|
michele.ar
|
import org.apache.commons.io.IOUtils;
|
| 18 |
50665
|
jochen.sch
|
import org.apache.commons.lang3.math.NumberUtils;
|
| 19 |
28959
|
michele.ar
|
import org.apache.commons.logging.Log;
|
| 20 |
|
|
import org.apache.commons.logging.LogFactory;
|
| 21 |
|
|
|
| 22 |
|
|
/**
|
| 23 |
57866
|
alessia.ba
|
* @author jochen, michele, andrea, alessia
|
| 24 |
28959
|
michele.ar
|
*/
|
| 25 |
|
|
public class HttpConnector {
|
| 26 |
29020
|
michele.ar
|
|
| 27 |
28959
|
michele.ar
|
private static final Log log = LogFactory.getLog(HttpConnector.class);
|
| 28 |
|
|
|
| 29 |
34867
|
andrea.man
|
private int maxNumberOfRetry = 6;
|
| 30 |
|
|
private int defaultDelay = 120; // seconds
|
| 31 |
|
|
private int readTimeOut = 120; // seconds
|
| 32 |
51955
|
miriam.bag
|
|
| 33 |
53932
|
alessia.ba
|
private String responseType = null;
|
| 34 |
51955
|
miriam.bag
|
|
| 35 |
48804
|
jochen.sch
|
private String userAgent = "Mozilla/5.0 (compatible; OAI; +http://www.openaire.eu)";
|
| 36 |
29020
|
michele.ar
|
|
| 37 |
53932
|
alessia.ba
|
public HttpConnector() {
|
| 38 |
|
|
CookieHandler.setDefault(new CookieManager(null, CookiePolicy.ACCEPT_ALL));
|
| 39 |
|
|
}
|
| 40 |
|
|
|
| 41 |
28959
|
michele.ar
|
/**
|
| 42 |
51556
|
claudio.at
|
* Given the URL returns the content via HTTP GET
|
| 43 |
|
|
*
|
| 44 |
|
|
* @param requestUrl the URL
|
| 45 |
34867
|
andrea.man
|
* @return the content of the downloaded resource
|
| 46 |
51556
|
claudio.at
|
* @throws CollectorServiceException when retrying more than maxNumberOfRetry times
|
| 47 |
28959
|
michele.ar
|
*/
|
| 48 |
29018
|
michele.ar
|
public String getInputSource(final String requestUrl) throws CollectorServiceException {
|
| 49 |
53932
|
alessia.ba
|
return attemptDownlaodAsString(requestUrl, 1, new CollectorPluginErrorLogList());
|
| 50 |
|
|
}
|
| 51 |
|
|
|
| 52 |
|
|
/**
|
| 53 |
|
|
* Given the URL returns the content as a stream via HTTP GET
|
| 54 |
|
|
*
|
| 55 |
|
|
* @param requestUrl the URL
|
| 56 |
|
|
* @return the content of the downloaded resource as InputStream
|
| 57 |
|
|
* @throws CollectorServiceException when retrying more than maxNumberOfRetry times
|
| 58 |
|
|
*/
|
| 59 |
|
|
public InputStream getInputSourceAsStream(final String requestUrl) throws CollectorServiceException {
|
| 60 |
38175
|
andrea.man
|
return attemptDownload(requestUrl, 1, new CollectorPluginErrorLogList());
|
| 61 |
29020
|
michele.ar
|
}
|
| 62 |
29018
|
michele.ar
|
|
| 63 |
53932
|
alessia.ba
|
private String attemptDownlaodAsString(final String requestUrl, final int retryNumber, final CollectorPluginErrorLogList errorList)
|
| 64 |
|
|
throws CollectorServiceException {
|
| 65 |
|
|
try {
|
| 66 |
|
|
InputStream s = attemptDownload(requestUrl, 1, new CollectorPluginErrorLogList());
|
| 67 |
|
|
try {
|
| 68 |
|
|
return IOUtils.toString(s);
|
| 69 |
|
|
} catch (IOException e) {
|
| 70 |
|
|
log.error("error while retrieving from http-connection occured: " + requestUrl, e);
|
| 71 |
|
|
Thread.sleep(defaultDelay * 1000);
|
| 72 |
|
|
errorList.add(e.getMessage());
|
| 73 |
|
|
return attemptDownlaodAsString(requestUrl, retryNumber + 1, errorList);
|
| 74 |
|
|
}
|
| 75 |
|
|
finally{
|
| 76 |
|
|
IOUtils.closeQuietly(s);
|
| 77 |
|
|
}
|
| 78 |
|
|
} catch (InterruptedException e) {
|
| 79 |
|
|
throw new CollectorServiceException(e);
|
| 80 |
|
|
}
|
| 81 |
|
|
}
|
| 82 |
51955
|
miriam.bag
|
|
| 83 |
53932
|
alessia.ba
|
private InputStream attemptDownload(final String requestUrl, final int retryNumber, final CollectorPluginErrorLogList errorList)
|
| 84 |
38175
|
andrea.man
|
throws CollectorServiceException {
|
| 85 |
29020
|
michele.ar
|
|
| 86 |
38175
|
andrea.man
|
if (retryNumber > maxNumberOfRetry) { throw new CollectorServiceException("Max number of retries exceeded. Cause: \n " + errorList); }
|
| 87 |
34867
|
andrea.man
|
|
| 88 |
29020
|
michele.ar
|
log.debug("Downloading " + requestUrl + " - try: " + retryNumber);
|
| 89 |
28959
|
michele.ar
|
try {
|
| 90 |
29169
|
andrea.man
|
InputStream input = null;
|
| 91 |
34867
|
andrea.man
|
|
| 92 |
29020
|
michele.ar
|
try {
|
| 93 |
37627
|
michele.ar
|
final HttpURLConnection urlConn = (HttpURLConnection) new URL(requestUrl).openConnection();
|
| 94 |
34867
|
andrea.man
|
urlConn.setInstanceFollowRedirects(false);
|
| 95 |
31724
|
michele.ar
|
urlConn.setReadTimeout(readTimeOut * 1000);
|
| 96 |
51556
|
claudio.at
|
urlConn.addRequestProperty("User-Agent", userAgent);
|
| 97 |
28959
|
michele.ar
|
|
| 98 |
29020
|
michele.ar
|
if (log.isDebugEnabled()) {
|
| 99 |
|
|
logHeaderFields(urlConn);
|
| 100 |
|
|
}
|
| 101 |
|
|
|
| 102 |
|
|
int retryAfter = obtainRetryAfter(urlConn.getHeaderFields());
|
| 103 |
57866
|
alessia.ba
|
if (is2xx(urlConn.getResponseCode())) {
|
| 104 |
29169
|
andrea.man
|
input = urlConn.getInputStream();
|
| 105 |
51955
|
miriam.bag
|
responseType = urlConn.getContentType();
|
| 106 |
53932
|
alessia.ba
|
return input;
|
| 107 |
29020
|
michele.ar
|
}
|
| 108 |
57866
|
alessia.ba
|
if (is3xx(urlConn.getResponseCode())) {
|
| 109 |
|
|
//REDIRECTS
|
| 110 |
|
|
final String newUrl = obtainNewLocation(urlConn.getHeaderFields());
|
| 111 |
|
|
log.debug(String.format("The requested url %s has been moved to %s", requestUrl, newUrl));
|
| 112 |
|
|
errorList.add(String.format("%s %s %s. Moved to: %s", requestUrl, urlConn.getResponseCode(), urlConn.getResponseMessage(), newUrl));
|
| 113 |
|
|
urlConn.disconnect();
|
| 114 |
|
|
if (retryAfter > 0) Thread.sleep(retryAfter * 1000);
|
| 115 |
|
|
return attemptDownload(newUrl, retryNumber + 1, errorList);
|
| 116 |
|
|
}
|
| 117 |
|
|
if (is4xx(urlConn.getResponseCode())) {
|
| 118 |
|
|
//CLIENT ERROR, DO NOT RETRY
|
| 119 |
|
|
errorList.add(String.format("%s error %s: %s", requestUrl, urlConn.getResponseCode(), urlConn.getResponseMessage()));
|
| 120 |
|
|
throw new CollectorServiceException("4xx error: request will not be repeated." + errorList);
|
| 121 |
|
|
}
|
| 122 |
|
|
if (is5xx(urlConn.getResponseCode())) {
|
| 123 |
|
|
//SERVER SIDE ERRORS RETRY ONLY on 503
|
| 124 |
|
|
switch (urlConn.getResponseCode()) {
|
| 125 |
|
|
case HttpURLConnection.HTTP_UNAVAILABLE:
|
| 126 |
|
|
if (retryAfter > 0) {
|
| 127 |
|
|
log.warn(requestUrl+" - waiting and repeating request after suggested retry-after" + retryAfter + " sec.");
|
| 128 |
|
|
Thread.sleep(retryAfter * 1000);
|
| 129 |
|
|
}
|
| 130 |
|
|
else {
|
| 131 |
|
|
log.warn(requestUrl+" - waiting and repeating request after default delay of " + defaultDelay + " sec.");
|
| 132 |
|
|
Thread.sleep(defaultDelay * 1000);
|
| 133 |
|
|
}
|
| 134 |
|
|
errorList.add(requestUrl+" 503 Service Unavailable");
|
| 135 |
|
|
urlConn.disconnect();
|
| 136 |
|
|
return attemptDownload(requestUrl, retryNumber + 1, errorList);
|
| 137 |
|
|
default:
|
| 138 |
|
|
errorList.add(String.format("%s Error %s: %s", requestUrl, urlConn.getResponseCode(), urlConn.getResponseMessage()));
|
| 139 |
|
|
throw new CollectorServiceException(urlConn.getResponseCode() + "error" + errorList);
|
| 140 |
|
|
}
|
| 141 |
|
|
}
|
| 142 |
|
|
throw new CollectorServiceException("Unexpected status code: " + urlConn.getResponseCode() + "error" + errorList);
|
| 143 |
|
|
}catch(MalformedURLException | NoRouteToHostException e){
|
| 144 |
|
|
errorList.add(String.format("Error: %s for request url: %s", e.getCause(), requestUrl));
|
| 145 |
|
|
throw new CollectorServiceException(e+ "error "+errorList);
|
| 146 |
29020
|
michele.ar
|
} catch (IOException e) {
|
| 147 |
31724
|
michele.ar
|
Thread.sleep(defaultDelay * 1000);
|
| 148 |
57866
|
alessia.ba
|
errorList.add(requestUrl+ " "+e.getMessage());
|
| 149 |
38175
|
andrea.man
|
return attemptDownload(requestUrl, retryNumber + 1, errorList);
|
| 150 |
29018
|
michele.ar
|
}
|
| 151 |
|
|
} catch (InterruptedException e) {
|
| 152 |
|
|
throw new CollectorServiceException(e);
|
| 153 |
28959
|
michele.ar
|
}
|
| 154 |
|
|
}
|
| 155 |
|
|
|
| 156 |
29018
|
michele.ar
|
private void logHeaderFields(final HttpURLConnection urlConn) throws IOException {
|
| 157 |
|
|
log.debug("StatusCode: " + urlConn.getResponseMessage());
|
| 158 |
29020
|
michele.ar
|
|
| 159 |
34867
|
andrea.man
|
for (Map.Entry<String, List<String>> e : urlConn.getHeaderFields().entrySet()) {
|
| 160 |
29018
|
michele.ar
|
if (e.getKey() != null) {
|
| 161 |
|
|
for (String v : e.getValue()) {
|
| 162 |
29020
|
michele.ar
|
log.debug(" key: " + e.getKey() + " - value: " + v);
|
| 163 |
29018
|
michele.ar
|
}
|
| 164 |
|
|
}
|
| 165 |
|
|
}
|
| 166 |
28959
|
michele.ar
|
}
|
| 167 |
29020
|
michele.ar
|
|
| 168 |
29018
|
michele.ar
|
private int obtainRetryAfter(final Map<String, List<String>> headerMap) {
|
| 169 |
34867
|
andrea.man
|
for (String key : headerMap.keySet()) {
|
| 170 |
53932
|
alessia.ba
|
if ((key != null) && key.toLowerCase().equals("retry-after") && (headerMap.get(key).size() > 0) && NumberUtils.isCreatable(headerMap.get(key).get(0))) {
|
| 171 |
|
|
return Integer
|
| 172 |
|
|
.parseInt(headerMap.get(key).get(0)) + 10;
|
| 173 |
|
|
}
|
| 174 |
28959
|
michele.ar
|
}
|
| 175 |
29018
|
michele.ar
|
return -1;
|
| 176 |
28959
|
michele.ar
|
}
|
| 177 |
29020
|
michele.ar
|
|
| 178 |
29018
|
michele.ar
|
private String obtainNewLocation(final Map<String, List<String>> headerMap) throws CollectorServiceException {
|
| 179 |
34867
|
andrea.man
|
for (String key : headerMap.keySet()) {
|
| 180 |
38105
|
andrea.man
|
if ((key != null) && key.toLowerCase().equals("location") && (headerMap.get(key).size() > 0)) { return headerMap.get(key).get(0); }
|
| 181 |
29018
|
michele.ar
|
}
|
| 182 |
|
|
throw new CollectorServiceException("The requested url has been MOVED, but 'location' param is MISSING");
|
| 183 |
|
|
}
|
| 184 |
28959
|
michele.ar
|
|
| 185 |
|
|
/**
|
| 186 |
|
|
* register for https scheme; this is a workaround and not intended for the use in trusted environments
|
| 187 |
|
|
*/
|
| 188 |
31724
|
michele.ar
|
public void initTrustManager() {
|
| 189 |
28959
|
michele.ar
|
final X509TrustManager tm = new X509TrustManager() {
|
| 190 |
34867
|
andrea.man
|
|
| 191 |
|
|
@Override
|
| 192 |
53932
|
alessia.ba
|
public void checkClientTrusted(final X509Certificate[] xcs, final String string) {
|
| 193 |
|
|
}
|
| 194 |
34867
|
andrea.man
|
|
| 195 |
|
|
@Override
|
| 196 |
53932
|
alessia.ba
|
public void checkServerTrusted(final X509Certificate[] xcs, final String string) {
|
| 197 |
|
|
}
|
| 198 |
34867
|
andrea.man
|
|
| 199 |
|
|
@Override
|
| 200 |
|
|
public X509Certificate[] getAcceptedIssuers() {
|
| 201 |
28959
|
michele.ar
|
return null;
|
| 202 |
|
|
}
|
| 203 |
|
|
};
|
| 204 |
|
|
try {
|
| 205 |
|
|
final SSLContext ctx = SSLContext.getInstance("TLS");
|
| 206 |
34867
|
andrea.man
|
ctx.init(null, new TrustManager[] { tm }, null);
|
| 207 |
28959
|
michele.ar
|
HttpsURLConnection.setDefaultSSLSocketFactory(ctx.getSocketFactory());
|
| 208 |
|
|
} catch (GeneralSecurityException e) {
|
| 209 |
|
|
log.fatal(e);
|
| 210 |
|
|
throw new IllegalStateException(e);
|
| 211 |
|
|
}
|
| 212 |
|
|
}
|
| 213 |
31724
|
michele.ar
|
|
| 214 |
57866
|
alessia.ba
|
private boolean is2xx(final int statusCode) {
|
| 215 |
|
|
return statusCode >= 200 && statusCode <=299;
|
| 216 |
|
|
}
|
| 217 |
|
|
|
| 218 |
|
|
private boolean is4xx(final int statusCode) {
|
| 219 |
|
|
return statusCode >= 400 && statusCode <=499;
|
| 220 |
|
|
}
|
| 221 |
|
|
|
| 222 |
|
|
private boolean is3xx(final int statusCode) {
|
| 223 |
|
|
return statusCode >= 300 && statusCode <=399;
|
| 224 |
|
|
}
|
| 225 |
|
|
|
| 226 |
|
|
private boolean is5xx(final int statusCode) {
|
| 227 |
|
|
return statusCode >= 500 && statusCode <=599;
|
| 228 |
|
|
}
|
| 229 |
|
|
|
| 230 |
|
|
|
| 231 |
31724
|
michele.ar
|
public int getMaxNumberOfRetry() {
|
| 232 |
|
|
return maxNumberOfRetry;
|
| 233 |
|
|
}
|
| 234 |
|
|
|
| 235 |
34867
|
andrea.man
|
public void setMaxNumberOfRetry(final int maxNumberOfRetry) {
|
| 236 |
31724
|
michele.ar
|
this.maxNumberOfRetry = maxNumberOfRetry;
|
| 237 |
|
|
}
|
| 238 |
|
|
|
| 239 |
|
|
public int getDefaultDelay() {
|
| 240 |
|
|
return defaultDelay;
|
| 241 |
|
|
}
|
| 242 |
|
|
|
| 243 |
34867
|
andrea.man
|
public void setDefaultDelay(final int defaultDelay) {
|
| 244 |
31724
|
michele.ar
|
this.defaultDelay = defaultDelay;
|
| 245 |
|
|
}
|
| 246 |
|
|
|
| 247 |
|
|
public int getReadTimeOut() {
|
| 248 |
|
|
return readTimeOut;
|
| 249 |
|
|
}
|
| 250 |
|
|
|
| 251 |
34867
|
andrea.man
|
public void setReadTimeOut(final int readTimeOut) {
|
| 252 |
31724
|
michele.ar
|
this.readTimeOut = readTimeOut;
|
| 253 |
|
|
}
|
| 254 |
29169
|
andrea.man
|
|
| 255 |
53932
|
alessia.ba
|
public String getResponseType() {
|
| 256 |
|
|
return responseType;
|
| 257 |
|
|
}
|
| 258 |
51955
|
miriam.bag
|
|
| 259 |
34867
|
andrea.man
|
}
|