Mercurial > hg > eSciDocImport

comparison src/de/mpiwg/itgroup/eSciDoc/Tools/ingestLib.java @ 0:c6929e63b0b8

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
first import
author dwinter
date Wed, 24 Nov 2010 16:52:07 +0100
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:c6929e63b0b8
1 package de.mpiwg.itgroup.eSciDoc.Tools;
2 import java.io.IOException;
3 import java.io.StringReader;
4 import java.net.MalformedURLException;
5 import java.net.URL;
6 import java.rmi.RemoteException;
7 import java.text.SimpleDateFormat;
8 import java.util.Date;
9 import java.util.regex.Matcher;
10 import java.util.regex.Pattern;
11
12 import javax.xml.parsers.DocumentBuilder;
13 import javax.xml.parsers.DocumentBuilderFactory;
14 import javax.xml.rpc.ServiceException;
15
16 import org.apache.axis.types.NonNegativeInteger;
17 import org.apache.xmlrpc.XmlRpcException;
18 import org.apache.xmlrpc.client.XmlRpcClient;
19 import org.apache.xmlrpc.client.XmlRpcClientConfigImpl;
20 import org.w3c.dom.Document;
21 import org.xml.sax.InputSource;
22
23
24 public class ingestLib extends IngestECHO{
25
26 /**
27 * @param args
28 * @throws Exception
29 * @throws Exception
30 */
31
32 private static String ZOPEPROVIDER = "http://127.0.0.1:18080";
33 private static String createFoxml(String litid) throws Exception{
34
35 String pid =
36 eSciDocXmlObject obj = new eSciDocXmlObject("mpiwg:"+pid,"vlpRessourceTemplate.xml");
37 //obj.setTitle("lit11111 - title");
38
39
40 //get DC Metadata from the Ressource
41 XmlRpcClientConfigImpl config = new XmlRpcClientConfigImpl();
42 XmlRpcClient client = new XmlRpcClient();
43 Object[] params = new Object[]{};
44 config.setServerURL(new URL(ZOPEPROVIDER+"/metadataMain"));
45 client.setConfig(config);
46 params = new Object[]{"http://vlp.mpiwg-berlin.mpg.de/library/data/"+litid+"/index_meta"};
47 String result = (String) client.execute("getDCFormatted", params);
48 System.out.println("dC:"+result);
49
50
51 DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
52 factory.setNamespaceAware(true);
53 DocumentBuilder db =factory.newDocumentBuilder();
54 InputSource resultStream = new InputSource(new StringReader(result));
55 Document dc = db.parse(resultStream);
56 obj.insertDC(dc);
57
58 obj.setReferenceFolder("/mpiwg/online/permanent/vlp/"+litid);
59 //obj.generateXMLIndex(new File("/mpiwg/online/permanent/vlp/"+litid));
60 //long date = new Date().getTime();
61 SimpleDateFormat dateformat = new SimpleDateFormat("yyyy-MM-dd'T'hh:mm:ss.S'Z'");
62 String dateStr = dateformat.format(new Date());
63 obj.addWebUrl("http://vlp.mpiwg-berlin.mpg.de/references?id="+litid, litid, dateStr);
64 obj.addIndexMetaUrl("http://vlp.mpiwg-berlin.mpg.de/library/data/"+litid+"/index_meta", litid, dateStr);
65 obj.setRelationship("info:fedora/vlp:col1");
66 obj.setVLPId(litid);
67 return obj.printXML();
68
69 }
70
71
72
73
74
75
76 private static void ingestAllVLPObjects() throws MalformedURLException,
77 XmlRpcException {
78 XmlRpcClientConfigImpl config = new XmlRpcClientConfigImpl();
79 XmlRpcClient client = new XmlRpcClient();
80 config.setServerURL(new URL("http://127.0.0.1:18080/vlp/vlp_coll/library/data"));
81 client.setConfig(config);
82 Object[] params = new Object[]{};
83 Object[] res = (Object[]) client.execute("getAllRessources", params);
84
85 String string = "";
86 for (Object re: res){
87 Object[] rA = (Object [])re;
88 String[] splitted = ((String) rA[0]).split("/");
89 String id = splitted[splitted.length-1];
90
91 try {
92 String xml = createFoxml(id);
93
94 String ret = ingest(xml);
95
96 config.setServerURL(new URL((String)rA[1]));
97 client.setConfig(config);
98 params = new Object[]{ret};
99 client.execute("setPID", params);
100
101 System.out.println(ret);
102
103
104 } catch (Exception e) {
105 System.err.println("cannot get:"+(String) rA[0]);
106
107 }
108 }
109 }
110
111 private static void modifyDCSet(String litid, String PID) throws XmlRpcException, ServiceException, IOException
112 {
113 XmlRpcClientConfigImpl config = new XmlRpcClientConfigImpl();
114 XmlRpcClient client = new XmlRpcClient();
115 config.setServerURL(new URL(ZOPEPROVIDER+"/metadataMain"));
116
117
118 client.setConfig(config);
119 Object[] params = new Object[]{"http://vlp.mpiwg-berlin.mpg.de/library/data/"+litid+"/index_meta"};
120 String result = (String) client.execute("getDCFormatted", params);
121 //System.out.println("dC:"+result);
122
123 FedoraAPIM APIM;
124 FedoraAPIA APIA;
125
126 System.setProperty("javax.net.ssl.trustStore", "/usr/local/fedora/tomcat/conf/keystore");
127
128
129 String baseURL = "https://127.0.0.1:8443/fedora";
130 FedoraClient fc = new FedoraClient(baseURL, "fedoraAdmin", "XXXX");
131 APIA=fc.getAPIA();
132 APIM=fc.getAPIM();
133 String content = "<oai_dc:dc xmlns:oai_dc='http://www.openarchives.org/OAI/2.0/oai_dc/' xmlns:dc='http://purl.org/dc/elements/1.1/'><dc:description>VLP Literature Object</dc:description><dc:publisher>MPIWG</dc:publisher>";
134 content += result;
135 content += "</oai_dc:dc>";
136 //content="";
137
138 //System.out.println(content);
139
140
141 String chksum = MD5.asHex(new MD5(content).Final());
142
143 //System.out.println(chksum);
144 String[] em= new String[]{};
145 String pid = APIM.modifyDatastreamByValue(PID, "DC",em , "Dublin Core Record for this object", "text/xml", "",content.getBytes("utf-8"), null, null,"metadata changed",false);
146 System.out.println(pid);
147
148 }
149
150
151 public static void main(String[] args) throws Exception {
152
153 //ingestAllVLPObjects();
154 changeDCMetadata();
155 //modifyDCSet("lit14191","mpiwg:PR9MPM4E");
156
157 }
158
159 private static void changeDCMetadata() throws MalformedURLException,
160 ServiceException, IOException, RemoteException, XmlRpcException {
161 System.setProperty("javax.net.ssl.trustStore", "/usr/local/fedora/tomcat/conf/keystore");
162
163 String baseURL = "https://127.0.0.1:8443/fedora";
164 FedoraClient fc = new FedoraClient(baseURL, "fedoraAdmin", "XXX");
165 FedoraAPIA APIA;
166 APIA=fc.getAPIA();
167 Condition[] condition =
168 {new Condition("pid", ComparisonOperator.has, "mpiwg:*")};
169 FieldSearchQuery query = new FieldSearchQuery(condition, null);
170
171 NonNegativeInteger maxResults = new NonNegativeInteger("" + 200);
172
173 FieldSearchResult res = APIA.findObjects(new String[]{"pid"}, maxResults, query);
174 processResults(APIA, res);
175 while (true)
176 {
177 ListSession ses = res.getListSession();
178 if (ses == null)
179 break;
180 String tok = ses.getToken();
181 if (tok == null)
182 break;
183
184 res = APIA.resumeFindObjects(res.getListSession().getToken());
185 if (res.getResultList().length>0)
186 processResults(APIA, res);
187 else
188 break;
189 }
190 }
191
192 private static void processResults(FedoraAPIA APIA, FieldSearchResult res)
193 throws XmlRpcException, ServiceException, IOException {
194 ObjectFields[] fields = res.getResultList();
195
196 System.out.println("found:"+fields.length);
197 for (ObjectFields field: fields){
198 String pid = field.getPid();
199 MIMETypedStream ds = APIA.getDatastreamDissemination(pid, "vlp-admin", null);
200 byte[] x = ds.getStream();
201 String s = new String(x);
202 //System.err.println(s);
203
204 Pattern p = Pattern.compile("<vlp:identifier>(lit.*)</vlp:identifier>");
205 Matcher m = p.matcher(s);
206 m.find();
207 String lit = m.group(1);
208 System.out.println(lit);
209 modifyDCSet(lit, pid);
210 }
211 }
212 }