Author: rwesten
Date: Sun Mar 4 14:08:34 2012
New Revision: 1296810
URL: http://svn.apache.org/viewvc?rev=1296810&view=rev
Log:
STANBOL-512: Now the DC terms namespace is used for all Dublin Core properties
Modified:
incubator/stanbol/trunk/enhancer/engines/tika/src/main/java/org/apache/stanbol/enhancer/engines/tika/metadata/OntologyMappings.java
incubator/stanbol/trunk/enhancer/engines/tika/src/test/java/org/apache/stanbol/enhancer/engines/tika/TikaEngineTest.java
Modified:
incubator/stanbol/trunk/enhancer/engines/tika/src/main/java/org/apache/stanbol/enhancer/engines/tika/metadata/OntologyMappings.java
URL:
http://svn.apache.org/viewvc/incubator/stanbol/trunk/enhancer/engines/tika/src/main/java/org/apache/stanbol/enhancer/engines/tika/metadata/OntologyMappings.java?rev=1296810&r1=1296809&r2=1296810&view=diff
==============================================================================
---
incubator/stanbol/trunk/enhancer/engines/tika/src/main/java/org/apache/stanbol/enhancer/engines/tika/metadata/OntologyMappings.java
(original)
+++
incubator/stanbol/trunk/enhancer/engines/tika/src/main/java/org/apache/stanbol/enhancer/engines/tika/metadata/OntologyMappings.java
Sun Mar 4 14:08:34 2012
@@ -33,7 +33,6 @@ import org.apache.clerezza.rdf.core.MGra
import org.apache.clerezza.rdf.core.Resource;
import org.apache.clerezza.rdf.core.TypedLiteral;
import org.apache.clerezza.rdf.core.UriRef;
-import org.apache.clerezza.rdf.ontologies.DC;
import org.apache.clerezza.rdf.ontologies.OWL;
import org.apache.clerezza.rdf.ontologies.RDFS;
import org.apache.clerezza.rdf.ontologies.SKOS;
@@ -163,46 +162,46 @@ public class OntologyMappings implements
* @param mappings The ontology mappings to add the DC mappings
*/
public static void addDcMappings(OntologyMappings mappings) {
-
+ String dc = NamespaceEnum.dc.getNamespace();
mappings.addMapping(
- new PropertyMapping(DC.contributor,
+ new PropertyMapping(dc+"contributor",
DublinCore.CONTRIBUTOR,MSOffice.LAST_AUTHOR));
mappings.addMapping(
- new PropertyMapping(DC.coverage,DublinCore.COVERAGE));
+ new PropertyMapping(dc+"coverage",DublinCore.COVERAGE));
mappings.addMappings(
- new PropertyMapping(DC.creator,
+ new PropertyMapping(dc+"creator",
DublinCore.CREATOR,MSOffice.AUTHOR,"initial-creator"));
mappings.addMappings(
- new PropertyMapping(DC.description,DublinCore.DESCRIPTION));
+ new PropertyMapping(dc+"description",DublinCore.DESCRIPTION));
mappings.addMappings(
- new PropertyMapping(DC.format,
+ new PropertyMapping(dc+"format",
DublinCore.FORMAT,HttpHeaders.CONTENT_TYPE));
mappings.addMappings(
- new PropertyMapping(DC.identifier,DublinCore.IDENTIFIER));
+ new PropertyMapping(dc+"identifier",DublinCore.IDENTIFIER));
mappings.addMappings(
- new PropertyMapping(DC.language,
+ new PropertyMapping(dc+"language",
DublinCore.LANGUAGE,HttpHeaders.CONTENT_LANGUAGE));
mappings.addMappings(
- new PropertyMapping(NamespaceEnum.dc+"modified",XSD.dateTime,
+ new PropertyMapping(dc+"modified",XSD.dateTime,
DublinCore.MODIFIED,"Last-Modified"));
mappings.addMappings(
- new PropertyMapping(DC.publisher,
+ new PropertyMapping(dc+"publisher",
DublinCore.PUBLISHER,MSOffice.COMPANY));
mappings.addMappings(
- new PropertyMapping(DC.relation,DublinCore.RELATION));
+ new PropertyMapping(dc+"relation",DublinCore.RELATION));
mappings.addMappings(
- new PropertyMapping(DC.rights,DublinCore.RIGHTS));
+ new PropertyMapping(dc+"rights",DublinCore.RIGHTS));
mappings.addMappings(
- new PropertyMapping(DC.source,DublinCore.SOURCE));
+ new PropertyMapping(dc+"source",DublinCore.SOURCE));
mappings.addMappings(
- new PropertyMapping(DC.subject,
+ new PropertyMapping(dc+"subject",
DublinCore.SUBJECT,MSOffice.KEYWORDS));
mappings.addMappings(
- new PropertyMapping(DC.title,DublinCore.TITLE));
+ new PropertyMapping(dc+"title",DublinCore.TITLE));
mappings.addMappings(
- new PropertyMapping(DC.type,DublinCore.TYPE));
+ new PropertyMapping(dc+"type",DublinCore.TYPE));
mappings.addMappings(
- new
PropertyMapping(DC.date,XSD.dateTime,DublinCore.DATE.getName()));
+ new
PropertyMapping(dc+"date",XSD.dateTime,DublinCore.DATE.getName()));
//MS Office -> DC
mappings.addMappings(
new PropertyMapping(NamespaceEnum.dc+"created",XSD.dateTime,
Modified:
incubator/stanbol/trunk/enhancer/engines/tika/src/test/java/org/apache/stanbol/enhancer/engines/tika/TikaEngineTest.java
URL:
http://svn.apache.org/viewvc/incubator/stanbol/trunk/enhancer/engines/tika/src/test/java/org/apache/stanbol/enhancer/engines/tika/TikaEngineTest.java?rev=1296810&r1=1296809&r2=1296810&view=diff
==============================================================================
---
incubator/stanbol/trunk/enhancer/engines/tika/src/test/java/org/apache/stanbol/enhancer/engines/tika/TikaEngineTest.java
(original)
+++
incubator/stanbol/trunk/enhancer/engines/tika/src/test/java/org/apache/stanbol/enhancer/engines/tika/TikaEngineTest.java
Sun Mar 4 14:08:34 2012
@@ -55,9 +55,6 @@ import org.apache.clerezza.rdf.core.Reso
import org.apache.clerezza.rdf.core.Triple;
import org.apache.clerezza.rdf.core.TypedLiteral;
import org.apache.clerezza.rdf.core.UriRef;
-import org.apache.clerezza.rdf.core.impl.TripleImpl;
-import org.apache.clerezza.rdf.core.impl.util.W3CDateFormat;
-import org.apache.clerezza.rdf.ontologies.DC;
import org.apache.clerezza.rdf.ontologies.RDF;
import org.apache.clerezza.rdf.ontologies.XSD;
import org.apache.commons.io.IOUtils;
@@ -312,10 +309,10 @@ public class TikaEngineTest {
"URL: https://issues.apache.org/jira/browse/TIKA-461");
//no check the extracted metadata!
//DC
- verifyValue(ci, DC.date, XSD.dateTime,"2010-09-06T09:25:34Z");
- verifyValue(ci, DC.format, null,"message/rfc822");
- verifyValue(ci, DC.subject, null,"[jira] Commented: (TIKA-461) RFC822
messages not parsed");
- verifyValue(ci, DC.creator, null,"Julien Nioche (JIRA)
<[email protected]>");
+ verifyValue(ci, new UriRef(NamespaceEnum.dc+"date"),
XSD.dateTime,"2010-09-06T09:25:34Z");
+ verifyValue(ci, new UriRef(NamespaceEnum.dc+"format"),
null,"message/rfc822");
+ verifyValue(ci, new UriRef(NamespaceEnum.dc+"subject"), null,"[jira]
Commented: (TIKA-461) RFC822 messages not parsed");
+ verifyValue(ci, new UriRef(NamespaceEnum.dc+"creator"), null,"Julien
Nioche (JIRA) <[email protected]>");
verifyValue(ci, new UriRef(NamespaceEnum.dc+"created"),
XSD.dateTime,"2010-09-06T09:25:34Z");
//Media Ontology
@@ -403,9 +400,9 @@ public class TikaEngineTest {
ContentItem ci = createContentItem("testMP3id3v24.mp3", "audio/mpeg");
assertFalse(engine.canEnhance(ci) == CANNOT_ENHANCE);
engine.computeEnhancements(ci);
- verifyValue(ci,DC.creator,null,"Test Artist");
- verifyValue(ci, DC.title,null,"Test Album");
- verifyValue(ci, DC.format,null,"audio/mpeg");
+ verifyValue(ci,new UriRef(NamespaceEnum.dc+"creator"),null,"Test
Artist");
+ verifyValue(ci, new UriRef(NamespaceEnum.dc+"title"),null,"Test
Album");
+ verifyValue(ci, new
UriRef(NamespaceEnum.dc+"format"),null,"audio/mpeg");
verifyValue(ci, new
UriRef(NamespaceEnum.media+"hasFormat"),null,"audio/mpeg");
verifyValue(ci, new
UriRef(NamespaceEnum.media+"mainOriginalTitle"),null,"Test Album");
verifyValue(ci, new
UriRef(NamespaceEnum.media+"hasContributor"),null,"Test Artist");
@@ -442,10 +439,10 @@ public class TikaEngineTest {
verifyValue(ci, new
UriRef(NamespaceEnum.media+"creationDate"),XSD.dateTime,"2009-08-11T09:09:45");
verifyValues(ci, new
UriRef(NamespaceEnum.media+"hasKeyword"),null,"serbor","moscow-birds","canon-55-250");
//and finally the mapped DC properties
- verifyValue(ci, DC.format,null,"image/jpeg");
- verifyValue(ci, DC.date,XSD.dateTime,"2009-08-11T09:09:45");
+ verifyValue(ci, new
UriRef(NamespaceEnum.dc+"format"),null,"image/jpeg");
+ verifyValue(ci, new
UriRef(NamespaceEnum.dc+"date"),XSD.dateTime,"2009-08-11T09:09:45");
verifyValue(ci, new
UriRef(NamespaceEnum.dc+"modified"),XSD.dateTime,"2009-10-02T23:02:49");
- verifyValues(ci, DC.subject, null,
"serbor","moscow-birds","canon-55-250");
+ verifyValues(ci, new UriRef(NamespaceEnum.dc+"subject"), null,
"serbor","moscow-birds","canon-55-250");
}
@Test