comparison src/de/mpiwg/itgroup/eSciDoc/harvesting/ESciDocDataHarvester.java @ 10:ad5c0748bd08

minor
author dwinter
date Mon, 20 Aug 2012 12:53:45 +0200
parents b6cf6462d709
children
comparison
equal deleted inserted replaced
9:b6cf6462d709 10:ad5c0748bd08
238 try { 238 try {
239 logger.info(escidocItem.printXML()); 239 logger.info(escidocItem.printXML());
240 // TODO write PID to back to echo-obj 240 // TODO write PID to back to echo-obj
241 Boolean result = connector.createItem(escidocItem); 241 Boolean result = connector.createItem(escidocItem);
242 if (result) { 242 if (result) {
243 addedObjects.add(escidocItem.getESciDocId()); 243 String objID=escidocItem.getESciDocId();
244 addedFile.debug(escidocItem.getESciDocId() + "\n"); 244 addedObjects.add(objID);
245 245 addedFile.debug(objID + "\n");
246 ((ECHORessource)obj).writeEsciDocIDToIndexMeta(objID.replace("/ir/item/",""));
246 } else { 247 } else {
247 notAddedObjects.add(obj.echoUrl); 248 notAddedObjects.add(obj.echoUrl);
248 notAddedFile.debug(obj.echoUrl); 249 notAddedFile.debug(obj.echoUrl);
249 250
250 } 251 }
342 throws Exception { 343 throws Exception {
343 344
344 Integer numberOfHits = connector.getNumberOfHitsFromFilterResult( 345 Integer numberOfHits = connector.getNumberOfHitsFromFilterResult(
345 command, objectXPath,mode); 346 command, objectXPath,mode);
346 347
347 348 logger.info(String.format("%s objects found.", numberOfHits));
348 int tausend = ((numberOfHits-1) / MAX_REC); 349 int tausend = ((numberOfHits-1) / MAX_REC);
349 350
350 String queryRestrict=""; 351 String queryRestrict="";
351 if(mode==0 | mode==2){ 352 if(mode==0 | mode==2){
352 queryRestrict="query=%22/properties/version/status%22=pending"; 353 queryRestrict="query=%22/properties/version/status%22=pending";
353 } else { 354 } else {
354 queryRestrict="query=%22/properties/version/status%22=submitted"; 355 queryRestrict="query=%22/properties/version/status%22=submitted";
355 } 356 }
356 357 int counter=0;
357 for (int t = 0; t <= tausend; t++) { 358 for (int t = 0; t <= tausend; t++) {
358 int start = t * MAX_REC+1; 359 int start = t * MAX_REC+1;
359 // int max=Math.min((t+1)*1000, numberOfHits); 360 // int max=Math.min((t+1)*1000, numberOfHits);
360 String query = "?maximumRecords="+String.valueOf(MAX_REC)+"&startRecord=" 361 String query = "?maximumRecords="+String.valueOf(MAX_REC)+"&startRecord="
361 + String.valueOf(start)+"&"+queryRestrict; 362 + String.valueOf(start)+"&"+queryRestrict;
362 for (eSciDocXmlObject obj : connector 363 for (eSciDocXmlObject obj : connector
363 .getObjectsFromFilterResult(command+query, objectXPath)) { 364 .getObjectsFromFilterResult(command+query, objectXPath)) {
364 365
366 logger.info(String.format("%s/%s", counter,numberOfHits));
367 counter+=1;
368
365 //TODO is the following really necessary, currently the obj in the list is sometimes not the current one. 369 //TODO is the following really necessary, currently the obj in the list is sometimes not the current one.
366 try{ 370 try{
367 HttpResponse resObj = connector.eScidocGet(obj.getESciDocId()); 371 HttpResponse resObj = connector.eScidocGet(obj.getESciDocId());
368 HttpEntity ent = resObj.getEntity(); 372 HttpEntity ent = resObj.getEntity();
369 if (ent!=null){ 373 if (ent!=null){
434 "http://xserve09.mpiwg-berlin.mpg.de:19280/echo_nav/echo_pages/content/showRDF")); 438 "http://xserve09.mpiwg-berlin.mpg.de:19280/echo_nav/echo_pages/content/showRDF"));
435 ESciDocDataHarvester hv = new ESciDocDataHarvester(newimporter, 439 ESciDocDataHarvester hv = new ESciDocDataHarvester(newimporter,
436 new ECHOTransformer(), connector, "/ir/context/escidoc:1001"); 440 new ECHOTransformer(), connector, "/ir/context/escidoc:1001");
437 441
438 // hv.readObjectsFromInstance("ECHO_collection"); 442 // hv.readObjectsFromInstance("ECHO_collection");
439 hv.readObjectsFromInstance("ECHO_resource"); 443 //hv.readObjectsFromInstance("ECHO_resource");
440 444
441 hv.releaseAndSubmitObjects( 445 hv.releaseAndSubmitObjects(
442 "/ir/context/escidoc:1001/resources/members", 446 "/ir/context/escidoc:1001/resources/members",
443 "//escidocItem:item","first release",0); 447 "//escidocItem:item","first release",0);
444 hv.releaseAndSubmitObjects( 448 hv.releaseAndSubmitObjects(