GH-4665 Clean up HDT code (#4666)

eclipse-rdf4j · Jul 23, 2023 · 1cf9678 · 1cf9678
2 parents d9d5b19 + 307e5fd
commit 1cf9678
Show file tree

Hide file tree

Showing 19 changed files with 85 additions and 88 deletions.
diff --git a/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/CRC32.java b/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/CRC32.java
@@ -82,6 +82,6 @@ public long getValue() {
 
 	@Override
 	public void reset() {
-		value = 0xFFFFFFFF;
+		value = 0;
 	}
 }
diff --git a/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTArray.java b/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTArray.java
@@ -35,7 +35,7 @@
  * @author Bart Hanssens
  */
 abstract class HDTArray extends HDTPart {
-	protected enum Type {
+	enum Type {
 		LOG64(1),
 		UINT32(2),
 		UINT64(3);
@@ -47,7 +47,7 @@ protected enum Type {
 		 *
 		 * @return value 1,2 or 3
 		 */
-		public int getValue() {
+		int getValue() {
 			return value;
 		}
 
@@ -56,22 +56,22 @@ public int getValue() {
 		}
 	}
 
-	protected int nrbits;
-	protected int entries;
+	int nrbits;
+	int entries;
 
 	/**
 	 * Get the type of the array
 	 *
 	 * @return byte
 	 */
-	protected abstract int getType();
+	abstract int getType();
 
 	/**
 	 * Get number of bits used to encode an entry
 	 *
 	 * @return positive integer value
 	 */
-	protected int getNrBits() {
+	int getNrBits() {
 		return nrbits;
 	}
 
@@ -80,7 +80,7 @@ protected int getNrBits() {
 	 *
 	 * @return positive integer value
 	 */
-	protected int size() {
+	int size() {
 		return entries;
 	}
 
@@ -90,10 +90,10 @@ protected int size() {
 	 * @param i zero-based index
 	 * @return entry
 	 */
-	protected abstract int get(int i);
+	abstract int get(int i);
 
 	@Override
-	protected void parse(InputStream is) throws IOException {
+	void parse(InputStream is) throws IOException {
 		CRC8 crc8 = new CRC8();
 		crc8.update(getType());
 

diff --git a/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTArrayFactory.java b/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTArrayFactory.java
@@ -20,7 +20,7 @@
  */
 class HDTArrayFactory {
 
-	protected static HDTArray parse(InputStream is) throws IOException {
+	static HDTArray parse(InputStream is) throws IOException {
 		int dtype = is.read();
 		if (dtype != HDTArray.Type.LOG64.getValue()) {
 			throw new UnsupportedOperationException("Array section: encoding " + Long.toHexString(dtype) +

diff --git a/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTArrayLog64.java b/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTArrayLog64.java
@@ -34,15 +34,15 @@
  * @author Bart Hanssens
  */
 class HDTArrayLog64 extends HDTArray {
-	private byte buffer[];
+	private byte[] buffer;
 
 	@Override
-	protected int getType() {
+	int getType() {
 		return HDTArray.Type.LOG64.getValue();
 	}
 
 	@Override
-	protected int get(int i) {
+	int get(int i) {
 		// start byte of the value, and start bit in that start byte
 		int bytePos = (i * nrbits) / 8;
 		int bitPos = (i * nrbits) % 8;
@@ -63,14 +63,14 @@ protected int get(int i) {
 	}
 
 	@Override
-	protected void parse(InputStream is) throws IOException {
+	void parse(InputStream is) throws IOException {
 		super.parse(is);
 
 		// don't close CheckedInputStream, as it will close the underlying inputstream
 		try (UncloseableInputStream uis = new UncloseableInputStream(is);
 				CheckedInputStream cis = new CheckedInputStream(uis, new CRC32())) {
 			// read bytes, minimum 1
-			long bytes = (nrbits * entries + 7) / 8;
+			long bytes = ((long) nrbits * entries + 7) / 8;
 			if (bytes > Integer.MAX_VALUE) {
 				throw new UnsupportedOperationException("Maximum number of bytes in array exceeded: " + bytes);
 			}

diff --git a/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTBitmap.java b/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTBitmap.java
@@ -62,7 +62,7 @@ protected void parse(InputStream is) throws IOException {
 						", but only bitmap v1 is supported");
 			}
 
-			long b = (int) VByte.decode(cis);
+			long b = VByte.decode(cis);
 			if (b > Integer.MAX_VALUE) {
 				throw new UnsupportedOperationException("Maximum number of entries in bitmap exceeded: " + b);
 			}

diff --git a/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTDictionary.java b/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTDictionary.java
@@ -36,13 +36,13 @@
  * @author Bart Hanssens
  */
 class HDTDictionary extends HDTPart {
-	protected final static byte[] DICT_FORMAT = "<http://purl.org/HDT/hdt#dictionaryFour>"
+	final static byte[] DICT_FORMAT = "<http://purl.org/HDT/hdt#dictionaryFour>"
 			.getBytes(StandardCharsets.US_ASCII);
-	protected final static String DICT_MAPPING = "mapping";
-	protected final static String DICT_ELEMENTS = "elements";
+	final static String DICT_MAPPING = "mapping";
+	final static String DICT_ELEMENTS = "elements";
 
 	@Override
-	protected void parse(InputStream is) throws IOException {
+	void parse(InputStream is) throws IOException {
 		// don't close CheckedInputStream, as it will close the underlying inputstream
 		try (UncloseableInputStream uis = new UncloseableInputStream(is);
 				CheckedInputStream cis = new CheckedInputStream(uis, new CRC16())) {

diff --git a/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTDictionarySection.java b/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTDictionarySection.java
@@ -18,7 +18,7 @@
  * @author Bart Hanssens
  */
 abstract class HDTDictionarySection extends HDTPart {
-	protected enum Type {
+	enum Type {
 		PLAIN(1),
 		FRONT(2),
 		HTFC(3),
@@ -28,7 +28,7 @@ protected enum Type {
 
 		private final int value;
 
-		protected int getValue() {
+		int getValue() {
 			return value;
 		}
 
@@ -42,23 +42,23 @@ protected int getValue() {
 	 *
 	 * @return
 	 */
-	protected abstract int size();
+	abstract int size();
 
 	/**
 	 * Get the entry
 	 *
 	 * @param i zero-based index
 	 * @return
 	 */
-	protected abstract byte[] get(int i) throws IOException;
+	abstract byte[] get(int i) throws IOException;
 
 	/**
 	 * Constructor
 	 *
 	 * @param pos  position
 	 * @param name name
 	 */
-	protected HDTDictionarySection(String name, long pos) {
+	HDTDictionarySection(String name, long pos) {
 		super(name, pos);
 	}
 }
diff --git a/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTDictionarySectionFactory.java b/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTDictionarySectionFactory.java
@@ -28,7 +28,7 @@ class HDTDictionarySectionFactory {
 	 * @return dictionary section
 	 * @throws IOException
 	 */
-	protected static HDTDictionarySection parse(InputStream is, String name, long pos) throws IOException {
+	static HDTDictionarySection parse(InputStream is, String name, long pos) throws IOException {
 		int dtype = is.read();
 		if (dtype != HDTDictionarySection.Type.FRONT.getValue()) {
 			throw new UnsupportedOperationException("Dictionary " + name + ": encoding "

diff --git a/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTDictionarySectionPFC.java b/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTDictionarySectionPFC.java
@@ -74,17 +74,17 @@ protected boolean removeEldestEntry(Map.Entry eldest) {
 	 * @param name
 	 * @param pos
 	 */
-	protected HDTDictionarySectionPFC(String name, long pos) {
+	HDTDictionarySectionPFC(String name, long pos) {
 		super(name, pos);
 	}
 
 	@Override
-	protected int size() {
+	int size() {
 		return totalStrings;
 	}
 
 	@Override
-	protected byte[] get(int i) throws IOException {
+	byte[] get(int i) throws IOException {
 		// HDT index start counting from 1
 		int idx = i - 1;
 
@@ -101,7 +101,7 @@ protected byte[] get(int i) throws IOException {
 	}
 
 	@Override
-	protected void parse(InputStream is) throws IOException {
+	void parse(InputStream is) throws IOException {
 		CRC8 crc8 = new CRC8();
 		crc8.update((byte) HDTDictionarySection.Type.FRONT.getValue());
 
@@ -112,7 +112,7 @@ protected void parse(InputStream is) throws IOException {
 				CheckedInputStream cis = new CheckedInputStream(uis, crc8)) {
 
 			long val = VByte.decode(cis);
-			if (totalStrings > Integer.MAX_VALUE) {
+			if (val > Integer.MAX_VALUE) {
 				throw new UnsupportedOperationException(getDebugPartStr() + " max number of strings exceeded: " + val);
 			}
 			totalStrings = (int) val;

diff --git a/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTGlobal.java b/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTGlobal.java
@@ -38,12 +38,12 @@
  * @author Bart Hanssens
  */
 class HDTGlobal extends HDTPart {
-	protected final static byte[] GLOBAL_FORMAT = "<http://purl.org/HDT/hdt#HDTv1>".getBytes(StandardCharsets.US_ASCII);
-	protected final static String GLOBAL_BASEURI = "BaseUri";
-	protected final static String GLOBAL_SOFTWARE = "Software";
+	final static byte[] GLOBAL_FORMAT = "<http://purl.org/HDT/hdt#HDTv1>".getBytes(StandardCharsets.US_ASCII);
+	final static String GLOBAL_BASEURI = "BaseUri";
+	final static String GLOBAL_SOFTWARE = "Software";
 
 	@Override
-	protected void parse(InputStream is) throws IOException {
+	void parse(InputStream is) throws IOException {
 		// don't close CheckedInputStream, as it will close the underlying inputstream
 		try (UncloseableInputStream uis = new UncloseableInputStream(is);
 				CheckedInputStream cis = new CheckedInputStream(uis, new CRC16())) {

diff --git a/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTHeader.java b/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTHeader.java
@@ -37,13 +37,13 @@
  */
 class HDTHeader extends HDTPart {
 
-	protected final static byte[] HEADER_FORMAT = "ntriples".getBytes(StandardCharsets.US_ASCII);
-	protected final static String HEADER_LENGTH = "length";
+	final static byte[] HEADER_FORMAT = "ntriples".getBytes(StandardCharsets.US_ASCII);
+	final static String HEADER_LENGTH = "length";
 
 	private byte[] headerData;
 
 	@Override
-	protected void parse(InputStream is) throws IOException {
+	void parse(InputStream is) throws IOException {
 		// don't close CheckedInputStream, as it will close the underlying inputstream
 		try (UncloseableInputStream uis = new UncloseableInputStream(is);
 				CheckedInputStream cis = new CheckedInputStream(uis, new CRC16())) {
@@ -63,7 +63,7 @@ protected void parse(InputStream is) throws IOException {
 	 *
 	 * @return byte array
 	 */
-	protected byte[] getHeaderData() {
+	byte[] getHeaderData() {
 		return headerData;
 	}
 
@@ -75,7 +75,7 @@ protected byte[] getHeaderData() {
 	 * @throws IOException
 	 */
 	private byte[] parseHeaderData(InputStream is, int len) throws IOException {
-		byte b[] = new byte[len];
+		byte[] b = new byte[len];
 		is.read(b);
 		return b;
 	}

diff --git a/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTParser.java b/core/rio/hdt/src/main/java/org/eclipse/rdf4j/rio/hdt/HDTParser.java
@@ -10,7 +10,6 @@
  *******************************************************************************/
 package org.eclipse.rdf4j.rio.hdt;
 
-import java.io.FileInputStream;
 import java.io.IOException;
 import java.io.InputStream;
 import java.io.Reader;
@@ -97,9 +96,9 @@ public synchronized void parse(InputStream in, String baseURI)
 			throw new IllegalArgumentException("Input stream must not be 'null'");
 		}
 
-		if (in instanceof FileInputStream) {
-			// "TODO: use more optimized way to parse the file, eg. filechannel / membuffer"
-		}
+//		if (in instanceof FileInputStream) {
+//			 TODO: use more optimized way to parse the file, eg. filechannel / membuffer
+//		}
 
 		HDTDictionarySection shared = null;
 		HDTDictionarySection subjects = null;
@@ -163,9 +162,10 @@ public synchronized void parse(InputStream in, String baseURI)
 			rdfHandler.startRDF();
 		}
 
-		int cnt = 0;
+		assert shared != null;
 		int size = shared.size();
 
+		assert section != null;
 		while (section.hasNext()) {
 			int[] t = section.next();
 			byte[] s = getSO(t[0], size, shared, subjects);
@@ -188,7 +188,7 @@ public synchronized void parse(InputStream in, String baseURI)
 	 */
 	@Override
 	public synchronized void parse(Reader reader, String baseURI)
-			throws IOException, RDFParseException, RDFHandlerException {
+			throws RDFParseException, RDFHandlerException {
 		throw new UnsupportedOperationException("HDT is binary, text readers not supported.");
 	}