GH-5148 improved soft fail on corruption for values.id and values.has…

…h files.

core/sail/nativerdf/src/main/java/org/eclipse/rdf4j/sail/nativerdf/NativeStatementIterator.java

@@ -10,6 +10,8 @@
  *******************************************************************************/
 package org.eclipse.rdf4j.sail.nativerdf;
 
+import static org.eclipse.rdf4j.sail.nativerdf.NativeStore.SOFT_FAIL_ON_CORRUPT_DATA;
+
 import java.io.IOException;
 
 import org.eclipse.rdf4j.common.io.ByteArrayUtil;
@@ -20,6 +22,9 @@
 import org.eclipse.rdf4j.model.Value;
 import org.eclipse.rdf4j.sail.SailException;
 import org.eclipse.rdf4j.sail.nativerdf.btree.RecordIterator;
+import org.eclipse.rdf4j.sail.nativerdf.model.CorruptIRI;
+import org.eclipse.rdf4j.sail.nativerdf.model.CorruptIRIOrBNode;
+import org.eclipse.rdf4j.sail.nativerdf.model.CorruptUnknownValue;
 
 /**
  * A statement iterator that wraps a RecordIterator containing statement records and translates these records to
@@ -74,6 +79,17 @@ public Statement getNextElement() throws SailException {
 			if (contextID != 0) {
 				context = valueStore.getResource(contextID);
 			}
+			if (SOFT_FAIL_ON_CORRUPT_DATA) {
+				if (subj == null) {
+					subj = new CorruptIRIOrBNode(valueStore.getRevision(), subjID, null);
+				}
+				if (pred == null) {
+					pred = new CorruptIRI(valueStore.getRevision(), predID, null, null);
+				}
+				if (obj == null) {
+					obj = new CorruptUnknownValue(valueStore.getRevision(), objID, null);
+				}
+			}
 
 			return valueStore.createStatement(subj, pred, obj, context);
 		} catch (IOException e) {

core/sail/nativerdf/src/main/java/org/eclipse/rdf4j/sail/nativerdf/NativeStore.java

@@ -24,6 +24,7 @@
 import org.apache.commons.io.FileUtils;
 import org.eclipse.rdf4j.collection.factory.api.CollectionFactory;
 import org.eclipse.rdf4j.collection.factory.mapdb.MapDb3CollectionFactory;
+import org.eclipse.rdf4j.common.annotation.InternalUseOnly;
 import org.eclipse.rdf4j.common.concurrent.locks.Lock;
 import org.eclipse.rdf4j.common.concurrent.locks.LockManager;
 import org.eclipse.rdf4j.common.io.MavenUtil;
@@ -62,6 +63,15 @@ public class NativeStore extends AbstractNotifyingSail implements FederatedServi
 
 	private static final String VERSION = MavenUtil.loadVersion("org.eclipse.rdf4j", "rdf4j-sail-nativerdf", "devel");
 
+	/**
+	 * Do not throw an exception when corrupt data is detected. Instead, try to return as much data as possible.
+	 *
+	 * Variable can be set through the system property org.eclipse.rdf4j.sail.nativerdf.softFailOnCorruptData.
+	 */
+	@InternalUseOnly
+	public static boolean SOFT_FAIL_ON_CORRUPT_DATA = "true"
+			.equalsIgnoreCase(System.getProperty("org.eclipse.rdf4j.sail.nativerdf.softFailOnCorruptData"));;
+
 	private static final Cleaner REMOVE_STORES_USED_FOR_MEMORY_OVERFLOW = Cleaner.create();
 
 	/**

core/sail/nativerdf/src/main/java/org/eclipse/rdf4j/sail/nativerdf/ValueStore.java

@@ -10,6 +10,8 @@
  *******************************************************************************/
 package org.eclipse.rdf4j.sail.nativerdf;
 
+import static org.eclipse.rdf4j.sail.nativerdf.NativeStore.SOFT_FAIL_ON_CORRUPT_DATA;
+
 import java.io.File;
 import java.io.IOException;
 import java.io.UnsupportedEncodingException;
@@ -128,12 +130,6 @@ public class ValueStore extends SimpleValueFactory {
 	 */
 	private final ConcurrentCache<String, Integer> namespaceIDCache;
 
-	/**
-	 * Do not throw an exception in case a value cannot be loaded, e.g. due to a corrupt value store.
-	 */
-	public static boolean SOFT_FAIL_ON_CORRUPT_DATA = "true"
-			.equalsIgnoreCase(System.getProperty("org.eclipse.rdf4j.sail.nativerdf.softFailOnCorruptData"));;
-
 	/*--------------*
 	 * Constructors *
 	 *--------------*/

core/sail/nativerdf/src/main/java/org/eclipse/rdf4j/sail/nativerdf/datastore/DataFile.java

@@ -10,6 +10,8 @@
  *******************************************************************************/
 package org.eclipse.rdf4j.sail.nativerdf.datastore;
 
+import static org.eclipse.rdf4j.sail.nativerdf.NativeStore.SOFT_FAIL_ON_CORRUPT_DATA;
+
 import java.io.Closeable;
 import java.io.File;
 import java.io.IOException;
@@ -18,7 +20,6 @@
 import java.util.NoSuchElementException;
 
 import org.eclipse.rdf4j.common.io.NioFile;
-import org.eclipse.rdf4j.sail.nativerdf.ValueStore;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
@@ -203,38 +204,49 @@ public byte[] getData(long offset) throws IOException {
 				(data[3]) & 0x000000ff;
 
 		// If the data length is larger than 750MB, we are likely reading the wrong data. Probably data corruption.
-		if (dataLength > 750 * 1024 * 1024) {
-			if (ValueStore.SOFT_FAIL_ON_CORRUPT_DATA) {
+		if (dataLength > 128 * 1024 * 1024) {
+			if (SOFT_FAIL_ON_CORRUPT_DATA) {
 				logger.error(
 						"Data length is {}MB which is larger than 750MB. This is likely data corruption. Truncating length to 32 MB.",
 						dataLength / ((1024 * 1024)));
 				dataLength = 32 * 1024 * 1024;
 			}
 		}
 
-		// We have either managed to read enough data and can return the required subset of the data, or we have read
-		// too little so we need to execute another read to get the correct data.
-		if (dataLength <= data.length - 4) {
+		try {
 
-			// adjust the approximate average with 1 part actual length and 99 parts previous average up to a sensible
-			// max of 200
-			dataLengthApproximateAverage = (int) (Math.min(200,
-					((dataLengthApproximateAverage / 100.0) * 99) + (dataLength / 100.0)));
+			// We have either managed to read enough data and can return the required subset of the data, or we have
+			// read
+			// too little so we need to execute another read to get the correct data.
+			if (dataLength <= data.length - 4) {
 
-			return Arrays.copyOfRange(data, 4, dataLength + 4);
+				// adjust the approximate average with 1 part actual length and 99 parts previous average up to a
+				// sensible
+				// max of 200
+				dataLengthApproximateAverage = (int) (Math.min(200,
+						((dataLengthApproximateAverage / 100.0) * 99) + (dataLength / 100.0)));
 
-		} else {
+				return Arrays.copyOfRange(data, 4, dataLength + 4);
 
-			// adjust the approximate average, but favour the actual dataLength since dataLength predictions misses are
-			// costly
-			dataLengthApproximateAverage = Math.min(200, (dataLengthApproximateAverage + dataLength) / 2);
+			} else {
 
-			// we didn't read enough data so we need to execute a new read
-			data = new byte[dataLength];
-			buf = ByteBuffer.wrap(data);
-			nioFile.read(buf, offset + 4L);
+				// adjust the approximate average, but favour the actual dataLength since dataLength predictions misses
+				// are costly
+				dataLengthApproximateAverage = Math.min(200, (dataLengthApproximateAverage + dataLength) / 2);
 
-			return data;
+				// we didn't read enough data so we need to execute a new read
+				data = new byte[dataLength];
+				buf = ByteBuffer.wrap(data);
+				nioFile.read(buf, offset + 4L);
+
+				return data;
+			}
+		} catch (OutOfMemoryError e) {
+			if (dataLength > 128 * 1024 * 1024) {
+				logger.error(
+						"Trying to read large amounts of data may be a sign of data corruption. Consider setting the system property org.eclipse.rdf4j.sail.nativerdf.softFailOnCorruptData to true");
+			}
+			throw e;
 		}
 
 	}

core/sail/nativerdf/src/test/java/org/eclipse/rdf4j/sail/nativerdf/NativeSailStoreCorruptionTest.java

@@ -251,7 +251,7 @@ public void testCorruptValuesHashFile() throws IOException {
 	private List<Statement> getStatements() {
 		List<Statement> list = new ArrayList<>();
 
-		ValueStore.SOFT_FAIL_ON_CORRUPT_DATA = true;
+		NativeStore.SOFT_FAIL_ON_CORRUPT_DATA = true;
 
 		try (RepositoryConnection conn = repo.getConnection()) {
 			StringWriter stringWriter = new StringWriter();
@@ -267,15 +267,12 @@ private List<Statement> getStatements() {
 			}
 			return list;
 		} finally {
-			ValueStore.SOFT_FAIL_ON_CORRUPT_DATA = false;
+			NativeStore.SOFT_FAIL_ON_CORRUPT_DATA = false;
 		}
 	}
 
 	@AfterEach
 	public void after() throws IOException {
 		repo.shutDown();
-		restoreFile(dataDir, "values.hash");
-		restoreFile(dataDir, "values.id");
-		restoreFile(dataDir, "values.dat");
 	}
 }