Mercurial > hg > eSciDocImport

comparison src/de/mpiwg/itgroup/eSciDoc/Tools/EScidocBasicHandler.java @ 3:58b52df9763c

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
added update functionality if index.meta has changed
author dwinter
date Wed, 12 Jan 2011 11:00:14 +0100
parents fab8e78184fa
children cb5668b07bfc
comparison
equal deleted inserted replaced
2:fab8e78184fa 3:58b52df9763c
9 import java.io.UnsupportedEncodingException; 9 import java.io.UnsupportedEncodingException;
10 import java.net.URI; 10 import java.net.URI;
11 import java.net.URL; 11 import java.net.URL;
12 import java.net.URLEncoder; 12 import java.net.URLEncoder;
13 import java.util.ArrayList; 13 import java.util.ArrayList;
14 import java.util.HashMap;
14 import java.util.List; 15 import java.util.List;
16 import java.util.Map;
15 import java.util.StringTokenizer; 17 import java.util.StringTokenizer;
16 import java.util.regex.Matcher; 18 import java.util.regex.Matcher;
17 import java.util.regex.Pattern; 19 import java.util.regex.Pattern;
18 20
19 import javax.swing.text.html.HTMLDocument.HTMLReader.IsindexAction; 21 import javax.swing.text.html.HTMLDocument.HTMLReader.IsindexAction;
20 22
21 23
24 import org.apache.commons.codec.EncoderException;
22 import org.apache.http.HttpEntity; 25 import org.apache.http.HttpEntity;
23 import org.apache.http.HttpResponse; 26 import org.apache.http.HttpResponse;
24 import org.apache.http.client.ClientProtocolException; 27 import org.apache.http.client.ClientProtocolException;
25 import org.apache.http.client.HttpClient; 28 import org.apache.http.client.HttpClient;
26 import org.apache.http.client.methods.HttpDelete; 29 import org.apache.http.client.methods.HttpDelete;
42 import org.jdom.Element; 45 import org.jdom.Element;
43 import org.jdom.JDOMException; 46 import org.jdom.JDOMException;
44 import org.jdom.Text; 47 import org.jdom.Text;
45 import org.jdom.input.SAXBuilder; 48 import org.jdom.input.SAXBuilder;
46 import org.jdom.xpath.XPath; 49 import org.jdom.xpath.XPath;
47 50 import org.w3c.dom.Node;
51
52 import de.mpiwg.itgroup.eSciDoc.echoObjects.ECHOObject;
53 import de.mpiwg.itgroup.eSciDoc.echoObjects.ECHORessource;
54 import de.mpiwg.itgroup.eSciDoc.exceptions.ConnectorException;
55 import de.mpiwg.itgroup.eSciDoc.exceptions.ESciDocXmlObjectException;
56 import de.mpiwg.itgroup.eSciDoc.exceptions.ObjectNotUniqueError;
48 import de.mpiwg.itgroup.eSciDoc.utils.eSciDocXmlObject; 57 import de.mpiwg.itgroup.eSciDoc.utils.eSciDocXmlObject;
49 58
50 59
51 60
52 61
475 484
476 } 485 }
477 486
478 487
479 488
480 public String getIDfromPID(String pid, String context) throws ClientProtocolException, IOException, IllegalStateException, JDOMException { 489 public String getIDfromPID(String pid, String context) throws ConnectorException {
481 490
482 491
483 String filter = "<param><filter name=\"http://escidoc.de/core/01/properties/pid\">"; 492 String filter = "<param><filter name=\"http://escidoc.de/core/01/properties/pid\">";
484 493
485 filter += pid; 494 filter += pid;
486 filter += "</filter></param>"; 495 filter += "</filter></param>";
487 496
488 String command = context 497 String command = context
489 + "/resources/members/filter"; 498 + "/resources/members/filter";
490 HttpResponse result =eScidocPost(command, 499 HttpResponse result;
491 new ByteArrayInputStream(filter.getBytes())); 500 try {
492 501 result = eScidocPost(command,
493 Document dom = new SAXBuilder().build(result.getEntity().getContent()); 502 new ByteArrayInputStream(filter.getBytes()));
494 503 } catch (IOException e) {
495 XPath xp = EScidocTools.getESciDocXpath("//escidocItem:item/@xlink:href"); 504 // TODO Auto-generated catch block
496 505 e.printStackTrace();
497 Attribute attr = (Attribute)xp.selectSingleNode(dom); 506 throw new ConnectorException();
498
499 if (attr!=null){
500 return attr.getValue();
501 } 507 }
502 508
503 return null; 509 try {
504 //return convertStreamToString(result.getEntity().getContent()); 510 Document dom = new SAXBuilder().build(result.getEntity().getContent());
511
512 XPath xp = EScidocTools.getESciDocXpath("//escidocItem:item/@xlink:href");
513
514 Attribute attr = (Attribute)xp.selectSingleNode(dom);
515
516 if (attr!=null){
517 return attr.getValue();
518 }
519
520 return null;
521 //return convertStreamToString(result.getEntity().getContent());
522 } catch (IllegalStateException e) {
523 // TODO Auto-generated catch block
524 e.printStackTrace();
525 throw new ConnectorException();
526 } catch (JDOMException e) {
527 // TODO Auto-generated catch block
528 e.printStackTrace();
529 throw new ConnectorException();
530 } catch (IOException e) {
531 // TODO Auto-generated catch block
532 e.printStackTrace();
533 throw new ConnectorException();
534 }
505 535
506 536
507 } 537 }
508 538
509 539
527 557
528 return result; 558 return result;
529 559
530 } 560 }
531 561
532 public List<eSciDocXmlObject> getObjectListFromFilterResult(String command, String objectXpath) throws IOException, IllegalStateException, JDOMException { 562 public List<eSciDocXmlObject> getObjectListFromFilterResult(String command, String objectXpath) throws IOException, IllegalStateException, JDOMException,ESciDocXmlObjectException {
533 //String filter = "<param><filter></filter></param>"; 563 //String filter = "<param><filter></filter></param>";
534 // 564 //
535 //String command = context 565 //String command = context
536 // + "/resources/members/filter"; 566 // + "/resources/members/filter";
537 //HttpResponse result =eScidocPost(command, 567 //HttpResponse result =eScidocPost(command,
644 return true; 674 return true;
645 } 675 }
646 676
647 677
648 678
649 public boolean alreadyExists(String indexField, String testString, String context) throws Exception { 679 public ECHOObject alreadyExists(String indexField, String testString, String context) throws ConnectorException, ObjectNotUniqueError {
650 680
651 String[] ct = context.split("/"); // gebraucht wird hier nur die id, dh ohne /ir/... 681 String[] ct = context.split("/"); // gebraucht wird hier nur die id, dh ohne /ir/...
652 682
653 String contextId=ct[ct.length-1]; 683 String contextId=ct[ct.length-1];
654 684
655 685
656 String searchString = String.format("\"%s\"=\"%s\"",indexField,testString); 686 String searchString = String.format("\"%s\"=\"%s\"",indexField,testString);
657 searchString += " and "+String.format("\"%s\"=\"%s\"","/properties/context/id",contextId); 687 searchString += " and "+String.format("\"%s\"=\"%s\"","/properties/context/id",contextId);
658 688
689 HttpResponse ret;
690 try{
659 searchString = URLEncoder.encode(searchString,"utf-8"); 691 searchString = URLEncoder.encode(searchString,"utf-8");
660 HttpResponse ret = eScidocGet("/ir/items?operation=searchRetrieve&version=1.1&query="+searchString); 692 ret = eScidocGet("/ir/items?operation=searchRetrieve&version=1.1&query="+searchString);
693 } catch (UnsupportedEncodingException e) {
694 throw new ConnectorException();
695 } catch (IOException e) {
696 // TODO Auto-generated catch block
697 e.printStackTrace();
698 throw new ConnectorException();
699 }
661 700
662 if (ret.getStatusLine().getStatusCode()!=200) 701 if (ret.getStatusLine().getStatusCode()!=200)
663 { 702 {
664 logger.debug("alreadyExists: error searchstring:"+searchString); 703 logger.debug("alreadyExists: error searchstring:"+searchString);
665 HttpEntity ent = ret.getEntity(); 704 HttpEntity ent = ret.getEntity();
666 if (ent!=null) 705 if (ent!=null) {
667 ent.consumeContent(); 706 try {
668 throw new Exception(); 707 ent.consumeContent();
669 } 708 } catch (IOException e) {
709 // TODO Auto-generated catch block
710 e.printStackTrace();
711 throw new ConnectorException();
712 }
713 throw new ConnectorException();
714 }
715 }
716
717 try{
670 Document doc = new SAXBuilder().build(ret.getEntity().getContent()); 718 Document doc = new SAXBuilder().build(ret.getEntity().getContent());
671 719
672 XPath xp = EScidocTools.getESciDocXpath("//zs:numberOfRecords/text()"); 720 XPath xp = EScidocTools.getESciDocXpath("//zs:numberOfRecords/text()");
673 String hitsStr = ((Text)xp.selectSingleNode(doc)).getText(); 721 String hitsStr = ((Text)xp.selectSingleNode(doc)).getText();
674 Integer hits = Integer.valueOf(hitsStr); 722 Integer hits = Integer.valueOf(hitsStr);
675 if (hits>0) 723 if (hits>0){
676 return true; 724 if (hits>1)
677 return false; 725 throw new ObjectNotUniqueError();
726 return getOldObjectFromESciDoc(doc);
727 }
728 return null;
729 } catch (IOException e) {
730 e.printStackTrace();
731 throw new ConnectorException();
732 } catch (IllegalStateException e) {
733 // TODO Auto-generated catch block
734 e.printStackTrace();
735 throw new ConnectorException();
736 } catch (JDOMException e) {
737 // TODO Auto-generated catch block
738 e.printStackTrace();
739 throw new ConnectorException();
740 }
741 }
742
743
744
745 private ECHOObject getOldObjectFromESciDoc(Document doc) {
746 Map<String,String>retMap = new HashMap<String,String>();
747 XPath md5Nodes;
748 XPath itemId;
749 XPath lastModificationDate;
750 try {
751 md5Nodes= EScidocTools.getESciDocXpath(".//escidocComponents:component/escidocComponents:properties[prop:content-category[text()='index_meta']]/prop:checksum");
752 itemId= EScidocTools.getESciDocXpath(".//escidocItem:item/@xlink:href");
753 lastModificationDate = EScidocTools.getESciDocXpath(".//escidocItem:item/@last-modification-date");
754 } catch (JDOMException e) {
755 // TODO Auto-generated catch block
756 e.printStackTrace();
757 return null;
758 }
759 Element node;
760 Attribute idNode;
761 Attribute lastModificationDateNode;
762 try {
763 node = (Element)md5Nodes.selectSingleNode(doc);
764 idNode = (Attribute)itemId.selectSingleNode(doc);
765 lastModificationDateNode =(Attribute)lastModificationDate.selectSingleNode(doc);
766
767 } catch (JDOMException e) {
768 // TODO Auto-generated catch block
769 e.printStackTrace();
770 return null;
771 }
772 String md5= node.getTextTrim();
773 String escidocId= idNode.getValue();
774 ECHORessource er;
775 try {
776 er = new ECHORessource();
777 er.eScidocId=escidocId;
778 er.indexMetaMD5stored=md5;
779 er.lastModificationDate= lastModificationDateNode.getValue();
780 } catch (IOException e) {
781 // TODO Auto-generated catch block
782 e.printStackTrace();
783 return null;
784 }
785 return er;
678 } 786 }
679 787
680 788
681 789
682 public ArrayList<String> getAllLinksOfContext(String string, String context) throws IOException, IllegalStateException, JDOMException { 790 public ArrayList<String> getAllLinksOfContext(String string, String context) throws IOException, IllegalStateException, JDOMException {
709 return ret; 817 return ret;
710 } 818 }
711 819
712 820
713 821
822
823
824
714 825
715 } 826 }