VIVO-907 Create the TripleStoreQuirks interface

Use it to handle the odd behavior that a particular triple store might have.
In this case, the odd behavior is that TDB mangles XMLSchema#nonNegativeInteger to XMLSchema#integer in literal types.

webapp/src/edu/cornell/mannlib/vitro/webapp/modules/tripleSource/TripleSource.java

@@ -24,4 +24,7 @@ public interface TripleSource extends Application.Module{
 
 	OntModelCache getShortTermOntModels(RDFService shortTermRdfService,
 			OntModelCache longTermOntModelCache);
+
+	/** Ways in which this TripleSource behaves oddly. */
+	TripleStoreQuirks getQuirks();
 }

webapp/src/edu/cornell/mannlib/vitro/webapp/modules/tripleSource/TripleStoreQuirks.java

@@ -0,0 +1,18 @@
+/* $This file is distributed under the terms of the license in /doc/license.txt$ */
+
+package edu.cornell.mannlib.vitro.webapp.modules.tripleSource;
+
+import com.hp.hpl.jena.rdf.model.Model;
+
+/**
+ * TODO
+ */
+public interface TripleStoreQuirks {
+
+	/**
+	 * Test to see whether the FileGraph must be updated to reflect the current
+	 * state of the file.
+	 */
+	boolean hasFileGraphChanged(Model fromFile, Model previous, String graphURI);
+
+}

webapp/src/edu/cornell/mannlib/vitro/webapp/servlet/setup/FileGraphSetup.java

@@ -218,12 +218,15 @@ public class FileGraphSetup implements ServletContextListener {
 				path.getFileName(), dbModel.size(), fileModel.size()));
         
 
-        boolean isIsomorphic = dbModel.isIsomorphicWith(fileModel);
+		// Isomorphism has some quirky issues with TDB, and perhaps also with Virtuoso.
+		boolean isChanged = ApplicationUtils.instance()
+				.getContentTripleSource().getQuirks()
+				.hasFileGraphChanged(fileModel, dbModel, graphURI);
         
         if (dbModel.isEmpty()  && !fileModel.isEmpty()) {
             dbModel.add(fileModel);
             modelChanged = true;
-        } else if (!isIsomorphic) {
+        } else if (isChanged) {
             log.info("Updating " + path + " because graphs are not isomorphic");
             log.info("dbModel: " + dbModel.size() + " ; fileModel: " + fileModel.size());
             dbModel.removeAll();

webapp/src/edu/cornell/mannlib/vitro/webapp/triplesource/impl/DefaultTripleStoreQuirks.java

@@ -0,0 +1,19 @@
+/* $This file is distributed under the terms of the license in /doc/license.txt$ */
+
+package edu.cornell.mannlib.vitro.webapp.triplesource.impl;
+
+import com.hp.hpl.jena.rdf.model.Model;
+
+import edu.cornell.mannlib.vitro.webapp.modules.tripleSource.TripleStoreQuirks;
+
+/**
+ * The behavior for non-quirky TripleSource implementations.
+ */
+public class DefaultTripleStoreQuirks implements TripleStoreQuirks {
+
+	@Override
+	public boolean hasFileGraphChanged(Model fromFile, Model previous, String graphURI) {
+		return !fromFile.isIsomorphicWith(previous);
+	}
+
+}

webapp/src/edu/cornell/mannlib/vitro/webapp/triplesource/impl/sdb/ContentTripleSourceSDB.java

@@ -38,11 +38,13 @@ import edu.cornell.mannlib.vitro.webapp.modelaccess.ontmodels.UnionModelsOntMode
 import edu.cornell.mannlib.vitro.webapp.modules.Application;
 import edu.cornell.mannlib.vitro.webapp.modules.ComponentStartupStatus;
 import edu.cornell.mannlib.vitro.webapp.modules.tripleSource.ContentTripleSource;
+import edu.cornell.mannlib.vitro.webapp.modules.tripleSource.TripleStoreQuirks;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.RDFService;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.RDFServiceFactory;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.impl.jena.sdb.RDFServiceFactorySDB;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.impl.logging.LoggingRDFServiceFactory;
 import edu.cornell.mannlib.vitro.webapp.servlet.setup.JenaDataSourceSetupBase;
+import edu.cornell.mannlib.vitro.webapp.triplesource.impl.DefaultTripleStoreQuirks;
 import edu.cornell.mannlib.vitro.webapp.utils.logging.ToString;
 
 /**
@@ -77,6 +79,8 @@ public class ContentTripleSourceSDB extends ContentTripleSource {
 	static final boolean DEFAULT_TESTONBORROW = true;
 	static final boolean DEFAULT_TESTONRETURN = true;
 	
+	private final TripleStoreQuirks quirks = new DefaultTripleStoreQuirks();
+
 	private ServletContext ctx;
 	private ComboPooledDataSource ds;
 	private RDFServiceFactory rdfServiceFactory;
@@ -205,6 +209,11 @@ public class ContentTripleSourceSDB extends ContentTripleSource {
 		return new UnionModelsOntModelsCache(combinedCache, CONTENT_UNIONS);
 	}
 
+	@Override
+	public TripleStoreQuirks getQuirks() {
+		return quirks;
+	}
+
 	@Override
 	public String toString() {
 		return "ContentTripleSourceSDB[" + ToString.hashHex(this) + "]";

webapp/src/edu/cornell/mannlib/vitro/webapp/triplesource/impl/sparql/ContentTripleSourceSPARQL.java

@@ -13,11 +13,13 @@ import edu.cornell.mannlib.vitro.webapp.modelaccess.ontmodels.OntModelCache;
 import edu.cornell.mannlib.vitro.webapp.modules.Application;
 import edu.cornell.mannlib.vitro.webapp.modules.ComponentStartupStatus;
 import edu.cornell.mannlib.vitro.webapp.modules.tripleSource.ContentTripleSource;
+import edu.cornell.mannlib.vitro.webapp.modules.tripleSource.TripleStoreQuirks;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.RDFService;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.RDFServiceFactory;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.impl.RDFServiceFactorySingle;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.impl.logging.LoggingRDFServiceFactory;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.impl.sparql.RDFServiceSparql;
+import edu.cornell.mannlib.vitro.webapp.triplesource.impl.DefaultTripleStoreQuirks;
 import edu.cornell.mannlib.vitro.webapp.utils.configuration.Property;
 import edu.cornell.mannlib.vitro.webapp.utils.configuration.Validation;
 import edu.cornell.mannlib.vitro.webapp.utils.logging.ToString;
@@ -36,6 +38,8 @@ public class ContentTripleSourceSPARQL extends ContentTripleSource {
 	private String endpointURI;
 	private String updateEndpointURI; // Optional
 	
+	private final TripleStoreQuirks quirks = new DefaultTripleStoreQuirks();
+
 	private RDFService rdfService;
 	private RDFServiceFactory rdfServiceFactory;
 	private Dataset dataset;
@@ -132,6 +136,11 @@ public class ContentTripleSourceSPARQL extends ContentTripleSource {
 		return longTermOntModelCache;
 	}
 
+	@Override
+	public TripleStoreQuirks getQuirks() {
+		return quirks;
+	}
+
 	@Override
 	public String toString() {
 		return "ContentTripleSourceSPARQL[" + ToString.hashHex(this)

webapp/src/edu/cornell/mannlib/vitro/webapp/triplesource/impl/tdb/ConfigurationTripleSourceTDB.java

@@ -18,6 +18,7 @@ import edu.cornell.mannlib.vitro.webapp.modelaccess.ontmodels.OntModelCache;
 import edu.cornell.mannlib.vitro.webapp.modules.Application;
 import edu.cornell.mannlib.vitro.webapp.modules.ComponentStartupStatus;
 import edu.cornell.mannlib.vitro.webapp.modules.tripleSource.ConfigurationTripleSource;
+import edu.cornell.mannlib.vitro.webapp.modules.tripleSource.TripleStoreQuirks;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.RDFService;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.RDFServiceFactory;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.impl.RDFServiceFactorySingle;
@@ -39,9 +40,10 @@ import edu.cornell.mannlib.vitro.webapp.utils.logging.ToString;
  * Memory-map all of the configuration models, and add the standard decorators.
  */
 public class ConfigurationTripleSourceTDB extends ConfigurationTripleSource {
-
 	private static final String DIRECTORY_TDB = "tdbModels";
 
+	private final TripleStoreQuirks quirks = new TDBTripleStoreQuirks();
+
 	private volatile RDFService rdfService;
 	private RDFServiceFactory rdfServiceFactory;
 	private RDFService unclosableRdfService;
@@ -112,6 +114,11 @@ public class ConfigurationTripleSourceTDB extends ConfigurationTripleSource {
 		return longTermOntModelCache;
 	}
 
+	@Override
+	public TripleStoreQuirks getQuirks() {
+		return quirks;
+	}
+
 	@Override
 	public String toString() {
 		return "ConfigurationTripleSourceTDB[" + ToString.hashHex(this) + "]";

webapp/src/edu/cornell/mannlib/vitro/webapp/triplesource/impl/tdb/ContentTripleSourceTDB.java

@@ -16,6 +16,7 @@ import edu.cornell.mannlib.vitro.webapp.modelaccess.ontmodels.OntModelCache;
 import edu.cornell.mannlib.vitro.webapp.modules.Application;
 import edu.cornell.mannlib.vitro.webapp.modules.ComponentStartupStatus;
 import edu.cornell.mannlib.vitro.webapp.modules.tripleSource.ContentTripleSource;
+import edu.cornell.mannlib.vitro.webapp.modules.tripleSource.TripleStoreQuirks;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.RDFService;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.RDFServiceFactory;
 import edu.cornell.mannlib.vitro.webapp.rdfservice.impl.RDFServiceFactorySingle;
@@ -39,6 +40,8 @@ import edu.cornell.mannlib.vitro.webapp.utils.logging.ToString;
  * Memory-map the small content models, and add the standard decorators.
  */
 public class ContentTripleSourceTDB extends ContentTripleSource {
+	private final TripleStoreQuirks quirks = new TDBTripleStoreQuirks();
+
 	private String tdbPath;
 
 	private volatile RDFService rdfService;
@@ -124,6 +127,11 @@ public class ContentTripleSourceTDB extends ContentTripleSource {
 		return longTermOntModelCache;
 	}
 
+	@Override
+	public TripleStoreQuirks getQuirks() {
+		return quirks;
+	}
+
 	@Override
 	public String toString() {
 		return "ContentTripleSourceTDB[" + ToString.hashHex(this) + "]";

webapp/src/edu/cornell/mannlib/vitro/webapp/triplesource/impl/tdb/TDBTripleStoreQuirks.java

@@ -0,0 +1,58 @@
+/* $This file is distributed under the terms of the license in /doc/license.txt$ */
+
+package edu.cornell.mannlib.vitro.webapp.triplesource.impl.tdb;
+
+import static com.hp.hpl.jena.datatypes.xsd.XSDDatatype.XSDinteger;
+import static com.hp.hpl.jena.datatypes.xsd.XSDDatatype.XSDnonNegativeInteger;
+
+import java.io.ByteArrayInputStream;
+import java.io.ByteArrayOutputStream;
+import java.io.InputStream;
+
+import org.apache.commons.logging.Log;
+import org.apache.commons.logging.LogFactory;
+
+import com.hp.hpl.jena.rdf.model.Model;
+import com.hp.hpl.jena.rdf.model.ModelFactory;
+
+import edu.cornell.mannlib.vitro.webapp.triplesource.impl.DefaultTripleStoreQuirks;
+
+/**
+ * TDB has some odd behaviors to deal with.
+ */
+public class TDBTripleStoreQuirks extends DefaultTripleStoreQuirks {
+	private static final Log log = LogFactory
+			.getLog(TDBTripleStoreQuirks.class);
+
+	/**
+	 * When the file graph was previously written to the TDB store, TDB mangled
+	 * some of the literal types: any type of XMLSchema#nonNegativeInteger was
+	 * changed to XMLSchema#integer.
+	 * 
+	 * We need to mangle our new model in the same way before comparing to the
+	 * previous one.
+	 */
+	@Override
+	public boolean hasFileGraphChanged(Model fromFile, Model previous,
+			String graphURI) {
+		try {
+			ByteArrayOutputStream buffer = new ByteArrayOutputStream();
+			fromFile.write(buffer, "N-TRIPLE");
+			String fromString = buffer.toString("UTF-8");
+
+			String mangleString = fromString.replace(
+					XSDnonNegativeInteger.getURI(), XSDinteger.getURI());
+			InputStream mangleStream = new ByteArrayInputStream(
+					mangleString.getBytes("UTF-8"));
+			Model mangled = ModelFactory.createDefaultModel();
+			mangled.read(mangleStream, null, "N-TRIPLE");
+
+			return !mangled.isIsomorphicWith(previous);
+		} catch (Exception e) {
+			log.warn("Failed to test for changes in filegraph. "
+					+ "Change assumed.", e);
+			return true;
+		}
+	}
+
+}