comparison src/de/mpiwg/itgroup/eSciDoc/Tools/AddMPIWGIdentifiers.java @ 8:a844f6948dd8

?nderungen im Walker tools f?r pubman
author dwinter
date Mon, 14 May 2012 09:58:45 +0200
parents
children 9164f3f4b232
comparison
equal deleted inserted replaced
7:df8c62d84f8f 8:a844f6948dd8
1 package de.mpiwg.itgroup.eSciDoc.Tools;
2
3 import java.io.IOException;
4 import java.util.HashMap;
5 import java.util.List;
6
7 import org.apache.http.HttpEntity;
8 import org.apache.http.HttpResponse;
9 import org.apache.http.client.ClientProtocolException;
10 import org.apache.log4j.BasicConfigurator;
11 import org.apache.log4j.Level;
12 import org.apache.log4j.Logger;
13 import org.jdom.Attribute;
14 import org.jdom.Document;
15 import org.jdom.Element;
16 import org.jdom.JDOMException;
17 import org.jdom.input.SAXBuilder;
18 import org.jdom.xpath.XPath;
19
20 import sun.util.logging.resources.logging;
21
22 import com.sun.org.apache.xerces.internal.parsers.SAXParser;
23
24 import de.mpiwg.itgroup.eSciDoc.exceptions.ESciDocXmlObjectException;
25 import de.mpiwg.itgroup.eSciDoc.utils.eSciDocXmlObject;
26
27 public class AddMPIWGIdentifiers {
28
29 /**
30 * Erzeuge MPIWG identifier in den administrativen MPIWG Metadaten.
31 * Solange der Identifier nicht in den Metadaten (index.meta) steht, wird dazu der Filename genommen.
32 * @param args
33 * @throws ESciDocXmlObjectException
34 * @throws JDOMException
35 * @throws IOException
36 * @throws ClientProtocolException
37 * @throws IllegalStateException
38 */
39 public static void main(String[] args) throws IllegalStateException, ClientProtocolException, IOException, JDOMException, ESciDocXmlObjectException {
40 // TODO Auto-generated method stub
41 BasicConfigurator.configure();
42 Logger logger = Logger.getRootLogger();
43 logger.setLevel(Level.INFO);
44 EScidocBasicHandler connector = new EScidocBasicHandler(
45 "escidoc-test.mpiwg-berlin.mpg.de", 8080, "dwinter", "fl0rian");
46
47 if(args.length<2){
48 System.out.println("Usage: startrecord maximumrecords");
49 System.exit(-1);
50 }
51 String MAX_REC = args[1];
52 String start = args[0];
53 String objectXPath = "//escidocItem:item";
54
55 String query = "?maximumRecords=" + String.valueOf(MAX_REC)
56 + "&startRecord=" + String.valueOf(start);
57 String command = "/ir/context/escidoc:12001/resources/members";
58 for (eSciDocXmlObject obj : connector.getObjectsFromFilterResult(
59 command + query, objectXPath)) {
60
61 Document doc = obj.getDocument();
62 Boolean changed=false;
63 XPath archivePathXP = EScidocTools.getESciDocXpath("//mpiwg:admin/mpiwg:archivePath");
64 XPath mpiwgMDXP= EScidocTools.getESciDocXpath("//mpiwg:admin");
65 XPath mpiwgIdentifierXP= EScidocTools.getESciDocXpath("//mpiwg:admin/mpiwg:identifier");
66
67
68 Element archivePathNode = (Element) archivePathXP.selectSingleNode(doc);
69 if(archivePathNode==null) // kein Treffer
70 continue;
71
72 Element mpiwgIDNode = (Element) mpiwgIdentifierXP.selectSingleNode(doc);
73 if(mpiwgIDNode!=null){ // kein Treffer
74 System.out.println("ID already exists:"+mpiwgIDNode.getTextTrim());
75 continue;
76 }
77
78
79
80
81 String archivePath=archivePathNode.getTextTrim();
82 String[] pathElements=archivePath.split("/");
83 int length=pathElements.length;
84 String mpiwgID=pathElements[length-1];
85 Element mpiwgMDNode= (Element) mpiwgMDXP.selectSingleNode(doc);
86
87 if(mpiwgID.equals(""))
88 mpiwgID=handleEmptyID(doc,mpiwgMDNode);
89
90
91
92
93 mpiwgIDNode= new Element("identifier", EScidocTools.MPIWG);
94 mpiwgIDNode.setText(mpiwgID);
95 mpiwgMDNode.addContent(mpiwgIDNode);
96
97 //correct release number, some objecte don't have an release number, why??
98
99
100 Element latestRelease = (Element)EScidocBasicHandler.getXPath(doc.getRootElement(),"//escidocItem:properties/prop:latest-release/release:number",true);
101 if(latestRelease==null){
102 logger.info("not released yet:"+obj.getESciDocId());
103 } else {
104 String rn= latestRelease.getTextTrim();
105 if (rn.equals("")){
106 String latestVersionNumber = ((Element)EScidocBasicHandler.getXPath(doc.getRootElement(),"//escidocItem:properties/prop:version/version:number",true)).getTextTrim();
107 logger.info("have to add relase number!");
108 logger.info("Will add:"+latestVersionNumber);
109
110 latestRelease.setText(latestVersionNumber);
111 }
112 }
113 System.out.println("Adding:"+mpiwgID);
114 changed=true;
115
116
117 //System.out.println(obj.printXML());
118 if (changed){
119
120 Boolean retVal = connector.updateItem(obj);
121 System.out.println("Replaced:"+obj.getESciDocId());
122 HttpResponse retValu = connector.submitAnObject(obj, "adding identifier");
123
124 System.out.println(EScidocBasicHandler.convertStreamToString(retValu.getEntity().getContent()));
125 HttpResponse resObj = connector.eScidocGet(obj.getESciDocId());
126 HttpEntity ent = resObj.getEntity();
127 if (ent!=null){
128 obj= new eSciDocXmlObject(ent.getContent());
129 } else {
130 System.out.println("Can not retrieve:" + obj.getESciDocId());
131 continue;
132 }
133
134 HttpResponse reValue2 = connector.releaseAnObject(obj, "repairing publishing info");
135 System.out.println(EScidocBasicHandler.convertStreamToString(reValue2.getEntity().getContent()));
136
137 }
138 }
139
140 }
141
142 private static String handleEmptyID(Document doc, Element mpiwgMDNode) throws JDOMException, IOException {
143 XPath xlinkXP = EScidocTools.getESciDocXpath("//escidocComponents:component[escidocComponents:properties/prop:content-category/text()='index_meta']/@xlink:href");
144
145 SAXBuilder sb = new SAXBuilder();
146 Object node = xlinkXP.selectSingleNode(doc);
147 Attribute attr =(Attribute) node;
148 String md = attr.getValue();
149
150 Document doc2 = sb.build("http://escidoc-test.mpiwg-berlin.mpg.de:8080"+md+"/content");
151 XPath apXP= EScidocTools.getESciDocXpath("//resource/archive-path");
152 XPath imXP= EScidocTools.getESciDocXpath("//texttool/image");
153
154 Element apNode = (Element)apXP.selectSingleNode(doc2);
155 String ap="";
156 if(apNode!=null)
157 {
158 ap = apNode.getTextTrim();
159 XPath apNewXP= EScidocTools.getESciDocXpath(".//mpiwg:archivePath");
160 Element apNewNode = (Element) apNewXP.selectSingleNode(mpiwgMDNode);
161 apNewNode.setText(ap);
162 }
163 Element imNode = (Element)imXP.selectSingleNode(doc2);
164
165 if(imNode!=null)
166 {
167 String im = imNode.getTextTrim();
168 XPath imNewXP= EScidocTools.getESciDocXpath(".//mpiwg:imageFolder");
169 Element imNewNode = (Element) imNewXP.selectSingleNode(mpiwgMDNode);
170 imNewNode.setText(ap+"/"+im);
171 }
172
173 String[] pathElements=ap.split("/");
174 int length=pathElements.length;
175 return pathElements[length-1];
176
177
178 }
179 }