Mercurial > hg > TripleStoreManager

comparison src/de/mpiwg/itgroup/indexMeta2RDF/TransformIndexMeta.java @ 4:e93de4e99b52 default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
indexMeta2rdf in dieses Projekt verschoben
author dwinter
date Thu, 21 Jun 2012 14:37:55 +0200
parents
children
comparison
equal deleted inserted replaced
3:6c8dac2c5214 4:e93de4e99b52
1 package de.mpiwg.itgroup.indexMeta2RDF;
2
3 import java.io.File;
4 import java.io.FileInputStream;
5 import java.io.FileWriter;
6 import java.io.IOException;
7 import java.io.InputStream;
8 import java.net.MalformedURLException;
9 import java.net.URL;
10 import java.util.Iterator;
11
12 import org.jdom.Attribute;
13 import org.jdom.Document;
14 import org.jdom.Element;
15 import org.jdom.JDOMException;
16 import org.jdom.input.SAXBuilder;
17 import org.jdom.xpath.XPath;
18 import org.openrdf.model.Statement;
19 import org.openrdf.model.impl.LiteralImpl;
20 import org.openrdf.model.impl.StatementImpl;
21 import org.openrdf.model.impl.URIImpl;
22 import org.openrdf.rio.RDFHandlerException;
23 import org.openrdf.rio.trig.TriGWriter;
24 import org.openrdf.rio.turtle.TurtleWriter;
25 import org.openrdf.model.Value;
26
27
28
29
30
31 public class TransformIndexMeta {
32 String OBJ_BASE_URL="http://echo.mpiwg-berlin.mpg.de/indexMeta/";
33 String ONT_BASE_URL="http://ontologies.mpiwg-berlin.mpg.de/general/MetaData/";
34 private String indexMetaType ="http://ontologies.mpiwg-berlin.mpg.de/general/IndexMeta";
35 private String bibObjType ="http://ontologies.mpiwg-berlin.mpg.de/general/BibData";
36 private FileWriter out;
37 private FileWriter error;
38 private TurtleWriter turtleWriter;
39
40
41 public TransformIndexMeta(FileWriter fw, FileWriter ew){
42 out=fw;
43 error=ew;
44 turtleWriter = new TurtleWriter(fw);
45 }
46 public void transform(String metaData) throws IOException, JDOMException, RDFHandlerException{
47 //URL url = new URL(metaData);
48 //InputStream is = url.openStream();
49
50 turtleWriter.startRDF();
51 FileInputStream is = new FileInputStream(metaData);
52 Document doc;
53 try {
54 doc = new SAXBuilder().build(is);
55 } catch (Exception e1) {
56 // TODO Auto-generated catch block
57 System.err.println("Cannot parse:"+metaData);
58 error.write("cannotparse:"+metaData+"\n");
59 return;
60 }
61
62 XPath xpDri = XPath.newInstance("//meta/dri[@type=\"escidoc-test\"]");
63
64 Element result = (Element)xpDri.selectSingleNode(doc);
65
66 if (result==null){
67 System.err.println("No dri for:"+metaData);
68 error.write("Non dri for:"+metaData+"\n");
69 return;
70 }
71
72 String dri=result.getTextTrim();
73
74 String objIdent=OBJ_BASE_URL+dri;
75
76
77 //out.write(String.format("<%s> rdf:type <%s>.\n", objIdent,indexMetaType));
78
79 Statement smt = new StatementImpl(new URIImpl(objIdent), new URIImpl("http://www.w3.org/1999/02/22-rdf-syntax-ns#type"), new URIImpl(indexMetaType));
80 turtleWriter.handleStatement(smt);
81
82 Element resElement = doc.getRootElement();
83
84 for (Object n: resElement.getChildren()){
85 if (Element.class.isInstance(n)) {
86 Element e = (Element)n;
87
88 if (!e.getTextTrim().equals("")){
89 String txt=e.getTextTrim();//.replace("\"","\\\"");
90
91 smt = new StatementImpl
92 (new URIImpl(objIdent), new URIImpl(ONT_BASE_URL+e.getName()), (Value)(new LiteralImpl(txt)));
93 turtleWriter.handleStatement(smt);
94
95
96 //out.write(String.format("<%s> <%s> \"%s\".\n", objIdent,ONT_BASE_URL+e.getName(),txt));
97 }
98 }
99
100
101 }
102
103 XPath bib = XPath.newInstance("//meta/bib");
104
105 Element bibElement = (Element)bib.selectSingleNode(doc);
106 if (bibElement==null){
107 System.err.println("No bibelement in:"+metaData);
108 error.write("No bibelement in:"+metaData+"\n");
109 return;
110 }
111 String bibIdent=objIdent+":bib";
112
113 smt = new StatementImpl(new URIImpl(bibIdent), new URIImpl("http://www.w3.org/1999/02/22-rdf-syntax-ns#type"), new URIImpl(bibObjType));
114 turtleWriter.handleStatement(smt);
115
116 //out.write(String.format("<%s> rdf:type <%s>.\n",bibIdent,bibObjType));
117
118 smt = new StatementImpl(new URIImpl(objIdent), new URIImpl(ONT_BASE_URL+"has_bibl_metaData"), new URIImpl(bibIdent));
119 turtleWriter.handleStatement(smt);
120
121 //out.write(String.format("<%s> <%s> <%s>.\n", objIdent,ONT_BASE_URL+"has_bibl_metaData",bibIdent));
122
123 Attribute bibType = bibElement.getAttribute("type");
124 if (bibType==null){
125 System.err.println("No bibtype in:"+metaData);
126 error.write("No bibtype in:"+metaData+"\n");
127 return;
128 }
129 String type=bibType.getValue();
130
131 smt = new StatementImpl
132 (new URIImpl(bibIdent), new URIImpl(ONT_BASE_URL+"is_of_type"), (Value)(new LiteralImpl(type)));
133 turtleWriter.handleStatement(smt);
134
135 //out.write(String.format("<%s> <%s> \"%s\".\n", bibIdent,ONT_BASE_URL+"is_of_type",type));
136
137 for (Object n: bibElement.getChildren()){
138 if (Element.class.isInstance(n)) {
139 Element e = (Element)n;
140 String txt=e.getTextTrim();//.replace("\"","\\\"");
141 smt = new StatementImpl
142 (new URIImpl(bibIdent), new URIImpl(ONT_BASE_URL+e.getName()), (Value)(new LiteralImpl(txt)));
143 turtleWriter.handleStatement(smt);
144
145 //out.write(String.format("<%s> <%s> \"%s\".\n", bibIdent,ONT_BASE_URL+e.getName(),txt));
146 }
147
148
149 }
150 turtleWriter.endRDF();
151
152 }
153
154 static public void main(String[] args) throws IOException, RDFHandlerException{
155 if (args.length!=1){
156 System.out.println("Usage: transfom path");
157 System.exit(1);
158 }
159 String root = args[0];
160 FileWriter fw = new FileWriter("/tmp/out.rdf");
161 FileWriter ew = new FileWriter("/tmp/errors_transform.txt");
162 TransformIndexMeta tim = new TransformIndexMeta(fw,ew);
163
164 Iterator<String> it = new IndexMetaIterator(new File(root));
165 while (it.hasNext()){
166 String nx = it.next();
167 try {
168 if(nx==null){
169 continue; //weiss noch nicht warum das passiert.
170 }
171 tim.transform(nx);
172 } catch (JDOMException e) {
173 System.out.println("JDOM exception:"+nx);
174 //e.printStackTrace();
175 }
176 fw.flush();
177 ew.flush();
178 }
179 fw.close();
180 ew.close();
181 }
182 }