annotate src/de/mpiwg/itgroup/nimanager/tools/StoreRDFToVirtuoso.java @ 0:1384a0d382fa

first input
author dwinter
date Thu, 30 Jun 2011 11:44:24 +0200
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
1384a0d382fa first input
dwinter
parents:
diff changeset
1 package de.mpiwg.itgroup.nimanager.tools;
1384a0d382fa first input
dwinter
parents:
diff changeset
2 // see http://docs.openlinksw.com/virtuoso/VirtuosoDriverJDBC.html
1384a0d382fa first input
dwinter
parents:
diff changeset
3 import java.io.BufferedReader;
1384a0d382fa first input
dwinter
parents:
diff changeset
4 import java.io.File;
1384a0d382fa first input
dwinter
parents:
diff changeset
5 import java.io.FileInputStream;
1384a0d382fa first input
dwinter
parents:
diff changeset
6 import java.io.FileNotFoundException;
1384a0d382fa first input
dwinter
parents:
diff changeset
7 import java.io.FileReader;
1384a0d382fa first input
dwinter
parents:
diff changeset
8 import java.io.IOException;
1384a0d382fa first input
dwinter
parents:
diff changeset
9 import java.io.InputStreamReader;
1384a0d382fa first input
dwinter
parents:
diff changeset
10 import java.sql.Connection;
1384a0d382fa first input
dwinter
parents:
diff changeset
11 import java.sql.DriverManager;
1384a0d382fa first input
dwinter
parents:
diff changeset
12 import java.sql.ResultSet;
1384a0d382fa first input
dwinter
parents:
diff changeset
13 import java.sql.ResultSetMetaData;
1384a0d382fa first input
dwinter
parents:
diff changeset
14 import java.sql.SQLException;
1384a0d382fa first input
dwinter
parents:
diff changeset
15 import java.sql.Statement;
1384a0d382fa first input
dwinter
parents:
diff changeset
16 import java.util.regex.Matcher;
1384a0d382fa first input
dwinter
parents:
diff changeset
17 import java.util.regex.Pattern;
1384a0d382fa first input
dwinter
parents:
diff changeset
18
1384a0d382fa first input
dwinter
parents:
diff changeset
19 public class StoreRDFToVirtuoso {
1384a0d382fa first input
dwinter
parents:
diff changeset
20
1384a0d382fa first input
dwinter
parents:
diff changeset
21 private Statement smt;
1384a0d382fa first input
dwinter
parents:
diff changeset
22 private Connection con;
1384a0d382fa first input
dwinter
parents:
diff changeset
23
1384a0d382fa first input
dwinter
parents:
diff changeset
24
1384a0d382fa first input
dwinter
parents:
diff changeset
25 public StoreRDFToVirtuoso(String username, String password) throws SQLException{
1384a0d382fa first input
dwinter
parents:
diff changeset
26 String connectString="jdbc:virtuoso://localhost:1111/charset=UTF-8";
1384a0d382fa first input
dwinter
parents:
diff changeset
27 con = DriverManager.getConnection(connectString,username,password);
1384a0d382fa first input
dwinter
parents:
diff changeset
28
1384a0d382fa first input
dwinter
parents:
diff changeset
29 smt = con.createStatement();
1384a0d382fa first input
dwinter
parents:
diff changeset
30
1384a0d382fa first input
dwinter
parents:
diff changeset
31 }
1384a0d382fa first input
dwinter
parents:
diff changeset
32
1384a0d382fa first input
dwinter
parents:
diff changeset
33 public void run(String filename) throws IOException, SQLException{
1384a0d382fa first input
dwinter
parents:
diff changeset
34
1384a0d382fa first input
dwinter
parents:
diff changeset
35
1384a0d382fa first input
dwinter
parents:
diff changeset
36 BufferedReader nr = new BufferedReader( new InputStreamReader(new FileInputStream(filename), "UTF8"));
1384a0d382fa first input
dwinter
parents:
diff changeset
37 String line;
1384a0d382fa first input
dwinter
parents:
diff changeset
38 String currentRDF="";
1384a0d382fa first input
dwinter
parents:
diff changeset
39 int counter=0;
1384a0d382fa first input
dwinter
parents:
diff changeset
40 while ((line = nr.readLine()) != null) {
1384a0d382fa first input
dwinter
parents:
diff changeset
41 if (line.startsWith("<rdf:RDF")){ // start a new file
1384a0d382fa first input
dwinter
parents:
diff changeset
42 currentRDF=line;
1384a0d382fa first input
dwinter
parents:
diff changeset
43 } else if (line.startsWith("</rdf:RDF>")) {
1384a0d382fa first input
dwinter
parents:
diff changeset
44 currentRDF+=line;
1384a0d382fa first input
dwinter
parents:
diff changeset
45 try {
1384a0d382fa first input
dwinter
parents:
diff changeset
46 process(currentRDF);
1384a0d382fa first input
dwinter
parents:
diff changeset
47 } catch (Exception e) {
1384a0d382fa first input
dwinter
parents:
diff changeset
48 // TODO Auto-generated catch block
1384a0d382fa first input
dwinter
parents:
diff changeset
49 e.printStackTrace();
1384a0d382fa first input
dwinter
parents:
diff changeset
50 }
1384a0d382fa first input
dwinter
parents:
diff changeset
51 counter+=1;
1384a0d382fa first input
dwinter
parents:
diff changeset
52 //if (counter>0)
1384a0d382fa first input
dwinter
parents:
diff changeset
53 // break;
1384a0d382fa first input
dwinter
parents:
diff changeset
54 } else {
1384a0d382fa first input
dwinter
parents:
diff changeset
55 currentRDF+=line;
1384a0d382fa first input
dwinter
parents:
diff changeset
56 }
1384a0d382fa first input
dwinter
parents:
diff changeset
57
1384a0d382fa first input
dwinter
parents:
diff changeset
58
1384a0d382fa first input
dwinter
parents:
diff changeset
59 }
1384a0d382fa first input
dwinter
parents:
diff changeset
60 }
1384a0d382fa first input
dwinter
parents:
diff changeset
61
1384a0d382fa first input
dwinter
parents:
diff changeset
62
1384a0d382fa first input
dwinter
parents:
diff changeset
63
1384a0d382fa first input
dwinter
parents:
diff changeset
64 private void process(String currentRDF) throws SQLException {
1384a0d382fa first input
dwinter
parents:
diff changeset
65
1384a0d382fa first input
dwinter
parents:
diff changeset
66 if (alreadyExists(currentRDF)){
1384a0d382fa first input
dwinter
parents:
diff changeset
67 return;
1384a0d382fa first input
dwinter
parents:
diff changeset
68 }
1384a0d382fa first input
dwinter
parents:
diff changeset
69 System.out.println("--Process:");
1384a0d382fa first input
dwinter
parents:
diff changeset
70 currentRDF = currentRDF.replace("'", "''");
1384a0d382fa first input
dwinter
parents:
diff changeset
71 String cmd = String.format("DB.DBA.RDF_LOAD_RDFXML ('%s', '', 'file:///GND.rdf')", currentRDF);
1384a0d382fa first input
dwinter
parents:
diff changeset
72 System.out.println(cmd);
1384a0d382fa first input
dwinter
parents:
diff changeset
73 boolean more = smt.execute(cmd);
1384a0d382fa first input
dwinter
parents:
diff changeset
74 ResultSetMetaData data = smt.getResultSet().getMetaData();
1384a0d382fa first input
dwinter
parents:
diff changeset
75 while(more)
1384a0d382fa first input
dwinter
parents:
diff changeset
76 {
1384a0d382fa first input
dwinter
parents:
diff changeset
77
1384a0d382fa first input
dwinter
parents:
diff changeset
78 ResultSet rs = smt.getResultSet();
1384a0d382fa first input
dwinter
parents:
diff changeset
79 while(rs.next())
1384a0d382fa first input
dwinter
parents:
diff changeset
80 {
1384a0d382fa first input
dwinter
parents:
diff changeset
81 for(int i = 1;i <= data.getColumnCount();i++)
1384a0d382fa first input
dwinter
parents:
diff changeset
82 {
1384a0d382fa first input
dwinter
parents:
diff changeset
83 String s = rs.getString(i);
1384a0d382fa first input
dwinter
parents:
diff changeset
84 System.out.println(s);
1384a0d382fa first input
dwinter
parents:
diff changeset
85 }
1384a0d382fa first input
dwinter
parents:
diff changeset
86 }
1384a0d382fa first input
dwinter
parents:
diff changeset
87 more = smt.getMoreResults();
1384a0d382fa first input
dwinter
parents:
diff changeset
88 }
1384a0d382fa first input
dwinter
parents:
diff changeset
89
1384a0d382fa first input
dwinter
parents:
diff changeset
90 System.out.println(currentRDF);
1384a0d382fa first input
dwinter
parents:
diff changeset
91 System.out.println("--End");
1384a0d382fa first input
dwinter
parents:
diff changeset
92 con.commit();
1384a0d382fa first input
dwinter
parents:
diff changeset
93 }
1384a0d382fa first input
dwinter
parents:
diff changeset
94
1384a0d382fa first input
dwinter
parents:
diff changeset
95
1384a0d382fa first input
dwinter
parents:
diff changeset
96 private boolean alreadyExists(String currentRDF) throws SQLException {
1384a0d382fa first input
dwinter
parents:
diff changeset
97 //String cmdString="sparql select * from <file:///GND.rdf> where {<http://d-nb.info/gnd/100004776> ?x ?y}";
1384a0d382fa first input
dwinter
parents:
diff changeset
98 // find entries like <rdf:Description rdf:about="http://d-nb.info/gnd/100004776"> in currentRDF
1384a0d382fa first input
dwinter
parents:
diff changeset
99 Pattern p = Pattern.compile("<rdf:Description rdf:about=\"(.*?)\">", Pattern.MULTILINE | Pattern.DOTALL);
1384a0d382fa first input
dwinter
parents:
diff changeset
100 Matcher m = p.matcher(currentRDF);
1384a0d382fa first input
dwinter
parents:
diff changeset
101 m.find();
1384a0d382fa first input
dwinter
parents:
diff changeset
102 String gndID = m.group(1);
1384a0d382fa first input
dwinter
parents:
diff changeset
103 //String gndID="XSDGG";
1384a0d382fa first input
dwinter
parents:
diff changeset
104 String cmdString=String.format("sparql select count(*) from <file:///GND.rdf> where {<%s> ?x ?y}",gndID);
1384a0d382fa first input
dwinter
parents:
diff changeset
105 smt.execute(cmdString);
1384a0d382fa first input
dwinter
parents:
diff changeset
106 ResultSet rs = smt.getResultSet();
1384a0d382fa first input
dwinter
parents:
diff changeset
107 rs.next();
1384a0d382fa first input
dwinter
parents:
diff changeset
108 int count=rs.getInt(1);
1384a0d382fa first input
dwinter
parents:
diff changeset
109 if (count>0){
1384a0d382fa first input
dwinter
parents:
diff changeset
110 System.out.println("Already in:"+gndID);
1384a0d382fa first input
dwinter
parents:
diff changeset
111 return true;}
1384a0d382fa first input
dwinter
parents:
diff changeset
112 else
1384a0d382fa first input
dwinter
parents:
diff changeset
113 return false;
1384a0d382fa first input
dwinter
parents:
diff changeset
114 }
1384a0d382fa first input
dwinter
parents:
diff changeset
115
1384a0d382fa first input
dwinter
parents:
diff changeset
116 static public void main(String args[]) throws IOException, SQLException, ClassNotFoundException{
1384a0d382fa first input
dwinter
parents:
diff changeset
117 Class.forName("virtuoso.jdbc4.Driver");
1384a0d382fa first input
dwinter
parents:
diff changeset
118 if (args.length <2){
1384a0d382fa first input
dwinter
parents:
diff changeset
119 System.out.println("Usage: storeRDF username password");
1384a0d382fa first input
dwinter
parents:
diff changeset
120 }
1384a0d382fa first input
dwinter
parents:
diff changeset
121 StoreRDFToVirtuoso st = new StoreRDFToVirtuoso(args[0],args[1]);
1384a0d382fa first input
dwinter
parents:
diff changeset
122
1384a0d382fa first input
dwinter
parents:
diff changeset
123 st.run("/Users/dwinter/Documents/Projekte/ECHO-eSciDoc-MPDL/escidocMPIWG/NamedIdentityManager/data/PNDrdf_1.rdf.txt");
1384a0d382fa first input
dwinter
parents:
diff changeset
124
1384a0d382fa first input
dwinter
parents:
diff changeset
125 st.close();
1384a0d382fa first input
dwinter
parents:
diff changeset
126 }
1384a0d382fa first input
dwinter
parents:
diff changeset
127
1384a0d382fa first input
dwinter
parents:
diff changeset
128 private void close() throws SQLException {
1384a0d382fa first input
dwinter
parents:
diff changeset
129 con.close();
1384a0d382fa first input
dwinter
parents:
diff changeset
130
1384a0d382fa first input
dwinter
parents:
diff changeset
131 }
1384a0d382fa first input
dwinter
parents:
diff changeset
132 }