diff options
Diffstat (limited to 'src/org/semanticweb/simpleETL')
| -rw-r--r-- | src/org/semanticweb/simpleETL/RDFHandlerWriter.java | 45 | ||||
| -rw-r--r-- | src/org/semanticweb/simpleETL/SimpleETL.java | 71 |
2 files changed, 116 insertions, 0 deletions
diff --git a/src/org/semanticweb/simpleETL/RDFHandlerWriter.java b/src/org/semanticweb/simpleETL/RDFHandlerWriter.java new file mode 100644 index 0000000..e5e2e2a --- /dev/null +++ b/src/org/semanticweb/simpleETL/RDFHandlerWriter.java | |||
| @@ -0,0 +1,45 @@ | |||
| 1 | package org.semanticweb.simpleETL; | ||
| 2 | import org.openrdf.model.Statement; | ||
| 3 | import org.openrdf.rio.RDFHandler; | ||
| 4 | import org.openrdf.rio.RDFHandlerException; | ||
| 5 | import org.openrdf.rio.RDFWriter; | ||
| 6 | |||
| 7 | |||
| 8 | public class RDFHandlerWriter implements RDFHandler { | ||
| 9 | protected RDFWriter m_writer; | ||
| 10 | protected boolean m_started; | ||
| 11 | |||
| 12 | public RDFHandlerWriter(RDFWriter writer){ | ||
| 13 | m_writer = writer; | ||
| 14 | m_started = false; | ||
| 15 | } | ||
| 16 | |||
| 17 | @Override | ||
| 18 | public void endRDF() throws RDFHandlerException { | ||
| 19 | // Do not end | ||
| 20 | } | ||
| 21 | |||
| 22 | @Override | ||
| 23 | public void handleComment(String arg0) throws RDFHandlerException { | ||
| 24 | m_writer.handleComment(arg0); | ||
| 25 | |||
| 26 | } | ||
| 27 | |||
| 28 | @Override | ||
| 29 | public void handleNamespace(String arg0, String arg1) throws RDFHandlerException { | ||
| 30 | m_writer.handleNamespace(arg0, arg1); | ||
| 31 | } | ||
| 32 | |||
| 33 | @Override | ||
| 34 | public void handleStatement(Statement arg0) throws RDFHandlerException { | ||
| 35 | m_writer.handleStatement(arg0); | ||
| 36 | } | ||
| 37 | |||
| 38 | @Override | ||
| 39 | public void startRDF() throws RDFHandlerException { | ||
| 40 | if(!m_started) { | ||
| 41 | m_started = true; | ||
| 42 | m_writer.startRDF(); | ||
| 43 | } | ||
| 44 | } | ||
| 45 | } | ||
diff --git a/src/org/semanticweb/simpleETL/SimpleETL.java b/src/org/semanticweb/simpleETL/SimpleETL.java new file mode 100644 index 0000000..4d4a193 --- /dev/null +++ b/src/org/semanticweb/simpleETL/SimpleETL.java | |||
| @@ -0,0 +1,71 @@ | |||
| 1 | package org.semanticweb.simpleETL; | ||
| 2 | import java.io.File; | ||
| 3 | import java.io.FileInputStream; | ||
| 4 | import java.io.FileOutputStream; | ||
| 5 | import java.util.regex.Pattern; | ||
| 6 | |||
| 7 | import org.openrdf.rio.RDFParser; | ||
| 8 | import org.openrdf.rio.RDFWriter; | ||
| 9 | import org.openrdf.rio.rdfxml.RDFXMLParser; | ||
| 10 | import org.openrdf.rio.turtle.TurtleWriter; | ||
| 11 | |||
| 12 | import uk.ac.ox.cs.pagoda.util.Utility; | ||
| 13 | |||
| 14 | public class SimpleETL { | ||
| 15 | |||
| 16 | protected final static String m_prefix_LUBM = "http://www.lehigh.edu/~zhp2/2004/0401/univ-bench.owl#"; | ||
| 17 | protected final static String m_prefix_UOBM = "http://semantics.crl.ibm.com/univ-bench-dl.owl#"; | ||
| 18 | protected final static String m_prefix_FLY= "http://www.virtualflybrain.org/ontologies/individual_neurons/FC_neuron_GF_an.owl#"; | ||
| 19 | |||
| 20 | String m_prefix; | ||
| 21 | String m_fileToImport; | ||
| 22 | String m_fileToExport; | ||
| 23 | |||
| 24 | public SimpleETL(String prefix, String fileToImport) { | ||
| 25 | m_prefix = prefix; | ||
| 26 | m_fileToImport = fileToImport; | ||
| 27 | m_fileToExport = m_fileToImport.replace(".owl", ".ttl"); | ||
| 28 | } | ||
| 29 | |||
| 30 | public SimpleETL(String prefix, String fileToImport, String outPath) { | ||
| 31 | m_prefix = prefix; | ||
| 32 | m_fileToImport = fileToImport; | ||
| 33 | File file = new File(outPath); | ||
| 34 | if (file.exists() && file.isDirectory()) | ||
| 35 | m_fileToExport = outPath + Utility.FILE_SEPARATOR + "data.ttl"; | ||
| 36 | else | ||
| 37 | m_fileToExport = outPath; | ||
| 38 | // + Utility.FILE_SEPARATOR + m_fileToImport.substring(m_fileToImport.lastIndexOf(Utility.FILE_SEPARATOR), m_fileToImport.lastIndexOf(".")) + ".ttl"; | ||
| 39 | } | ||
| 40 | |||
| 41 | public void rewrite() throws Exception { | ||
| 42 | // RDFParser parser = new TurtleParser(); | ||
| 43 | RDFParser parser = new RDFXMLParser(); | ||
| 44 | |||
| 45 | RDFWriter writer = new TurtleWriter(new FileOutputStream(m_fileToExport)); | ||
| 46 | |||
| 47 | // String m_fileToExport = m_fileToImport.replace(".owl", ".ntriple"); | ||
| 48 | // RDFWriter writer = new NTriplesWriter(new FileOutputStream(m_fileToExport)); | ||
| 49 | |||
| 50 | RDFHandlerWriter multiHandler = new RDFHandlerWriter(writer); | ||
| 51 | parser.setRDFHandler(multiHandler); | ||
| 52 | File fileToImport = new File(m_fileToImport); | ||
| 53 | if(fileToImport.isDirectory()) { | ||
| 54 | for(File file : fileToImport.listFiles()) { | ||
| 55 | if(file.isFile() && (Pattern.matches(".*.owl", file.getName()) || Pattern.matches(".*.rdf", file.getName()))) { | ||
| 56 | Utility.logDebug("Parsing " + file.getName()); | ||
| 57 | parser.parse(new FileInputStream(file), m_prefix); | ||
| 58 | } | ||
| 59 | } | ||
| 60 | } | ||
| 61 | else | ||
| 62 | parser.parse(new FileInputStream(fileToImport), m_prefix); | ||
| 63 | writer.endRDF(); | ||
| 64 | Utility.logInfo("SimpleETL rewriting DONE", | ||
| 65 | "additional ontology data is saved in " + m_fileToExport + "."); | ||
| 66 | } | ||
| 67 | |||
| 68 | public String getExportedFile() { | ||
| 69 | return m_fileToExport; | ||
| 70 | } | ||
| 71 | } | ||
