Revision 52235
Added by Miriam Baglioni over 6 years ago
| modules/dnet-collector-plugins/trunk/src/main/java/eu/dnetlib/data/collector/plugins/HTTPWithFileName/HTTPWithFileNameCollectorIterable.java | ||
|---|---|---|
| 1 | 1 |
package eu.dnetlib.data.collector.plugins.HTTPWithFileName; |
| 2 | 2 |
|
| 3 |
import eu.dnetlib.data.collector.plugins.projects.gtr2.Gtr2ProjectsIterable; |
|
| 3 | 4 |
import eu.dnetlib.data.collector.rmi.CollectorServiceException; |
| 4 | 5 |
import org.apache.commons.logging.Log; |
| 5 | 6 |
import org.apache.commons.logging.LogFactory; |
| ... | ... | |
| 21 | 22 |
*/ |
| 22 | 23 |
public class HTTPWithFileNameCollectorIterable implements Iterable<String> {
|
| 23 | 24 |
private static final Log log = LogFactory.getLog(HTTPWithFileNameCollectorIterable.class); |
| 25 |
private final String TERMINATOR = "FINITO"; |
|
| 26 |
final ArrayBlockingQueue<String> queue = new ArrayBlockingQueue<String>(100); |
|
| 24 | 27 |
|
| 25 | 28 |
private final ArrayList<String> urls = new ArrayList<>(); |
| 26 | 29 |
private final ArrayList<String> metas = new ArrayList<String>(); |
| 27 |
private String filter; |
|
| 30 |
private String filterParam;
|
|
| 28 | 31 |
|
| 32 |
int total = 0; |
|
| 33 |
int filtered = 0; |
|
| 34 |
|
|
| 29 | 35 |
public HTTPWithFileNameCollectorIterable(String startUrl, String filter){
|
| 30 |
|
|
| 31 |
urls.add(startUrl); |
|
| 32 |
this.filter = filter; |
|
| 36 |
if (!startUrl.isEmpty()) |
|
| 37 |
urls.add(startUrl); |
|
| 38 |
this.filterParam = filter; |
|
| 39 |
Thread ft = new Thread(new FillMetaQueue()); |
|
| 40 |
ft.start(); |
|
| 33 | 41 |
} |
| 34 | 42 |
|
| 35 |
private boolean containsFilter(String meta){
|
|
| 43 |
/* private boolean containsFilter(String meta){
|
|
| 36 | 44 |
if (filter == null || filter.isEmpty()) |
| 37 | 45 |
return false; |
| 38 | 46 |
String[] filter = this.filter.split(";");
|
| ... | ... | |
| 92 | 100 |
urls.add(url+file); |
| 93 | 101 |
} |
| 94 | 102 |
} |
| 95 |
} |
|
| 103 |
}*/
|
|
| 96 | 104 |
|
| 97 | 105 |
|
| 98 | 106 |
@Override |
| 99 | 107 |
public Iterator<String> iterator() {
|
| 100 |
final ArrayBlockingQueue<String> queue = new ArrayBlockingQueue<String>(100); |
|
| 101 | 108 |
|
| 102 | 109 |
|
| 110 |
|
|
| 103 | 111 |
return new Iterator<String>(){
|
| 104 |
int total = 0; |
|
| 112 |
/* int total = 0;
|
|
| 105 | 113 |
int filtered = 0; |
| 106 | 114 |
public void fillQueue() {
|
| 107 | 115 |
Connector c = new Connector(); |
| ... | ... | |
| 155 | 163 |
|
| 156 | 164 |
} |
| 157 | 165 |
|
| 158 |
} |
|
| 166 |
}*/
|
|
| 159 | 167 |
@Override |
| 160 | 168 |
public boolean hasNext() {
|
| 161 |
if (queue.isEmpty()){
|
|
| 169 |
while (queue.isEmpty()); |
|
| 170 |
if (queue.peek().equals(TERMINATOR)) |
|
| 171 |
log.info(String.format("Total number of metadata %d, Number of metadata filtered %d", total, filtered));
|
|
| 172 |
return !queue.peek().equals(TERMINATOR); |
|
| 173 |
/*if (queue.isEmpty()){
|
|
| 162 | 174 |
fillQueue(); |
| 163 |
} |
|
| 164 |
if(queue.isEmpty()){
|
|
| 165 |
log.info(String.format("Total number of metadata %d, Number of metadata filtered %d", total, filtered));
|
|
| 166 |
return false; |
|
| 167 |
} |
|
| 175 |
}*/
|
|
| 176 |
// if(queue.isEmpty()){
|
|
| 177 |
//log.info(String.format("Total number of metadata %d, Number of metadata filtered %d", total, filtered));
|
|
| 178 |
// return false;
|
|
| 179 |
//}
|
|
| 168 | 180 |
|
| 169 |
return true; |
|
| 181 |
//return true;
|
|
| 170 | 182 |
} |
| 171 | 183 |
|
| 172 | 184 |
@Override |
| ... | ... | |
| 174 | 186 |
return queue.poll(); |
| 175 | 187 |
} |
| 176 | 188 |
|
| 177 |
// @Override |
|
| 178 |
// public void remove() {
|
|
| 179 |
// |
|
| 180 |
// } |
|
| 181 |
// |
|
| 182 |
// @Override |
|
| 183 |
// public void forEachRemaining(Consumer<? super String> action) {
|
|
| 184 |
// |
|
| 185 |
// } |
|
| 186 | 189 |
}; |
| 187 | 190 |
} |
| 188 | 191 |
|
| 192 |
private class FillMetaQueue implements Runnable{
|
|
| 189 | 193 |
|
| 194 |
|
|
| 195 |
Connector c = new Connector(); |
|
| 196 |
|
|
| 197 |
public void fillQueue() {
|
|
| 198 |
String url; |
|
| 199 |
while((metas.size()>0 || urls.size() > 0 )){
|
|
| 200 |
log.debug("metas.size() = " + metas.size() + " urls.size() = " + urls.size() + " queue.size() = " +queue.size());
|
|
| 201 |
if (metas.size() > 0){
|
|
| 202 |
url = metas.remove(0); |
|
| 203 |
try {
|
|
| 204 |
c.get(url); |
|
| 205 |
} catch (CollectorServiceException e) {
|
|
| 206 |
log.info("Impossible to collect url: " + url + " error: " + e.getMessage());
|
|
| 207 |
} |
|
| 208 |
if(c.isStatusOk()){
|
|
| 209 |
try {
|
|
| 210 |
String ret = c.getResponse(); |
|
| 211 |
if (ret != null && ret.length()>0) {
|
|
| 212 |
if (!containsFilter(ret)) |
|
| 213 |
queue.put(addFilePath(ret, url, url.endsWith(".json")));
|
|
| 214 |
else |
|
| 215 |
filtered++; |
|
| 216 |
total++; |
|
| 217 |
} |
|
| 218 |
} catch (InterruptedException e) {
|
|
| 219 |
log.info("not inserted in queue element associate to url " + url + " error: " + e.getMessage() );
|
|
| 220 |
|
|
| 221 |
|
|
| 222 |
} |
|
| 223 |
} |
|
| 224 |
}else{
|
|
| 225 |
url = urls.remove(0); |
|
| 226 |
try {
|
|
| 227 |
c.get(url); |
|
| 228 |
} catch (CollectorServiceException e) {
|
|
| 229 |
log.info("Impossible to collect url: " + url + " error: " + e.getMessage());
|
|
| 230 |
} |
|
| 231 |
if(c.isStatusOk()){
|
|
| 232 |
if (c.responseTypeContains("text/html")){
|
|
| 233 |
recurFolder(c.getResponse(),url); |
|
| 234 |
} |
|
| 235 |
else if(c.responseTypeContains("application/json") || c.responseTypeContains("application/xml")){
|
|
| 236 |
try {
|
|
| 237 |
queue.put(addFilePath(c.getResponse(),url, c.responseTypeContains("application/json")));
|
|
| 238 |
} catch (InterruptedException e) {
|
|
| 239 |
log.info("not inserted in queue element associate to url " + url + " error: " + e.getMessage() );
|
|
| 240 |
} |
|
| 241 |
} |
|
| 242 |
} |
|
| 243 |
|
|
| 244 |
} |
|
| 245 |
|
|
| 246 |
} |
|
| 247 |
try {
|
|
| 248 |
queue.put(TERMINATOR); |
|
| 249 |
} catch (InterruptedException e) {
|
|
| 250 |
e.printStackTrace(); |
|
| 251 |
} |
|
| 252 |
|
|
| 253 |
} |
|
| 254 |
|
|
| 255 |
private boolean containsFilter(String meta){
|
|
| 256 |
if (filterParam == null || filterParam.isEmpty()) |
|
| 257 |
return false; |
|
| 258 |
String[] filter = filterParam.split(";");
|
|
| 259 |
for(String item:filter){
|
|
| 260 |
if (meta.contains(item)) |
|
| 261 |
return true; |
|
| 262 |
} |
|
| 263 |
return false; |
|
| 264 |
} |
|
| 265 |
|
|
| 266 |
private String addFilePath(String meta,String url, boolean isJson){
|
|
| 267 |
String path = url.replace("metadata", "pdf");
|
|
| 268 |
|
|
| 269 |
try {
|
|
| 270 |
if(isJson) |
|
| 271 |
meta = meta.substring(0, meta.length() - 1) + ",'downloadFileUrl':'" + path.substring(0, path.indexOf(".json")) + ".pdf'}";
|
|
| 272 |
else{
|
|
| 273 |
|
|
| 274 |
if (meta.contains("<!DOCTYPE")) {
|
|
| 275 |
meta = meta.substring(meta.indexOf("<!DOCTYPE"));
|
|
| 276 |
meta = meta.substring(meta.indexOf(">") + 1);
|
|
| 277 |
} |
|
| 278 |
int index = meta.lastIndexOf("</");
|
|
| 279 |
meta = meta.substring(0, index) + "<downloadFileUrl>" + path.substring(0, path.indexOf(".xml")) + ".pdf</downloadFileUrl>" + meta.substring(index);
|
|
| 280 |
|
|
| 281 |
|
|
| 282 |
} |
|
| 283 |
|
|
| 284 |
}catch(Exception ex){
|
|
| 285 |
log.info("not file with extension .json or .xml");
|
|
| 286 |
} |
|
| 287 |
|
|
| 288 |
|
|
| 289 |
if(isJson){
|
|
| 290 |
JSONObject jsonobj = null; |
|
| 291 |
try {
|
|
| 292 |
jsonobj = new JSONObject("{'resource':" + meta + "}");
|
|
| 293 |
|
|
| 294 |
return XML.toString(jsonobj); |
|
| 295 |
}catch(Exception e){
|
|
| 296 |
log.fatal("Impossible to transform json object to xml \n" + jsonobj + "\n " + e.getMessage() + "\n" + url);
|
|
| 297 |
throw new RuntimeException(); |
|
| 298 |
} |
|
| 299 |
} |
|
| 300 |
return meta; |
|
| 301 |
} |
|
| 302 |
|
|
| 303 |
private void recurFolder(String text, String url){
|
|
| 304 |
Document doc = Jsoup.parse(text); |
|
| 305 |
Elements links = doc.select("a");
|
|
| 306 |
for(Element e:links){
|
|
| 307 |
if (!e.text().equals("../")){
|
|
| 308 |
String file = e.attr("href");
|
|
| 309 |
if(file.endsWith(".json") || file.endsWith(".xml"))
|
|
| 310 |
metas.add(url+file); |
|
| 311 |
else |
|
| 312 |
urls.add(url+file); |
|
| 313 |
} |
|
| 314 |
} |
|
| 315 |
} |
|
| 316 |
|
|
| 317 |
|
|
| 318 |
@Override |
|
| 319 |
public void run() {
|
|
| 320 |
fillQueue(); |
|
| 321 |
} |
|
| 322 |
} |
|
| 323 |
|
|
| 324 |
|
|
| 325 |
|
|
| 190 | 326 |
} |
Also available in: Unified diff
changed implementation of data gathering