Merge pull request #179 from ate47/hcat_tree_external

HDTCatTree + HDTGenDisk
rdfhdt · Nov 21, 2022 · 969c049 · 969c049
2 parents 0531f87 + f0969fa
commit 969c049
Show file tree

Hide file tree

Showing 231 changed files with 19,035 additions and 2,108 deletions.
diff --git a/hdt-api/src/main/java/org/rdfhdt/hdt/dictionary/Dictionary.java b/hdt-api/src/main/java/org/rdfhdt/hdt/dictionary/Dictionary.java
@@ -27,13 +27,12 @@
  */
 
 
-import java.io.Closeable;
-import java.util.HashMap;
-import java.util.TreeMap;
-
 import org.rdfhdt.hdt.enums.TripleComponentRole;
 import org.rdfhdt.hdt.header.Header;
 
+import java.io.Closeable;
+import java.util.Map;
+
 
 /**
  * Interface that specifies the basic methods for any Dictionary implementation
@@ -53,7 +52,7 @@ public interface Dictionary extends Closeable {
 	 *            TriplePosition of the id in the dictionary
 	 * @return String
 	 */
-	public CharSequence idToString(long id, TripleComponentRole position);
+	CharSequence idToString(long id, TripleComponentRole position);
 
 	/**
 	 * Returns the id for a given string
@@ -64,11 +63,7 @@ public interface Dictionary extends Closeable {
 	 *            TriplePosition of the string in the dictionary
 	 * @return int
 	 */
-	public long stringToId(CharSequence str, TripleComponentRole position);
-
-	/**
-	 * Returns the number of elements in the dictionary
-	 */
+	long stringToId(CharSequence str, TripleComponentRole position);
 
 	/**
 	 * Returns the data type of a given literal string
@@ -77,54 +72,57 @@ public interface Dictionary extends Closeable {
 	 *            The id to get the data type for
 	 * @return String
 	 */
-	public String dataTypeOfId(long id);
+	CharSequence dataTypeOfId(long id);
 
-	public long getNumberOfElements();
+	/**
+	 * Returns the number of elements in the dictionary
+	 */
+	long getNumberOfElements();
 
 	/**
 	 * Return the combined size of the sections of the dictionary (in bytes)
 	 */
-	public long size();
+	long size();
 
 	/**
 	 * Returns the number of subjects in the dictionary. Note: Includes shared.
 	 */
-	public long getNsubjects();
+	long getNsubjects();
 
 	/**
 	 * Returns the number of predicates in the dictionary.
 	 */
-	public long getNpredicates();
+	long getNpredicates();
 
 	/**
 	 * Returns the number of objects in the dictionary. Note: Includes shared
 	 */
-	public long getNobjects();
+	long getNobjects();
 
 	/**
 	 * Returns the number of subjects/objects in the dictionary.
 	 */
-	public long getNAllObjects();
-	public long getNshared();
+	long getNAllObjects();
+	long getNshared();
 
-	public DictionarySection getSubjects();
+	DictionarySection getSubjects();
 
-	public DictionarySection getPredicates();
+	DictionarySection getPredicates();
 
-	public DictionarySection getObjects();
+	DictionarySection getObjects();
 
-	public TreeMap<String,DictionarySection> getAllObjects();
+	Map<? extends CharSequence, DictionarySection> getAllObjects();
 
-	public DictionarySection getShared();
+	DictionarySection getShared();
 
 	/**
 	 * Fills the header with information from the dictionary
 	 */
-	public void populateHeader(Header header, String rootNode);
+	void populateHeader(Header header, String rootNode);
 
 	/**
 	 * Returns the type of the dictionary (the way it is written onto file/held in memory)
-	 * @return
+	 * @return type
 	 */
-	public String getType();
-}
+	String getType();
+}
diff --git a/hdt-api/src/main/java/org/rdfhdt/hdt/dictionary/DictionarySection.java b/hdt-api/src/main/java/org/rdfhdt/hdt/dictionary/DictionarySection.java
@@ -46,7 +46,7 @@ public interface DictionarySection extends Closeable {
 	 * @return
 	 * 			the corresponding ID in the dictionary
 	 */
-	public long locate(CharSequence s);
+	long locate(CharSequence s);
 
 	/**
 	 * Find the String associated to a given ID
@@ -55,7 +55,7 @@ public interface DictionarySection extends Closeable {
 	 * @return
 	 * 			the corresponding string
 	 */
-	public CharSequence extract(long pos);
+	CharSequence extract(long pos);
 
 	/**
 	 * Size in bytes of the strings held in the dictionary section.
@@ -67,7 +67,7 @@ public interface DictionarySection extends Closeable {
 	 * Number of entries in the dictionary section.
 	 * @return long
 	 */
-	public long getNumberOfElements();
+	long getNumberOfElements();
 
 	/**
 	 * Iterator over all entries in the dictionary, sorted lexicographically.

diff --git a/hdt-api/src/main/java/org/rdfhdt/hdt/enums/CompressionType.java b/hdt-api/src/main/java/org/rdfhdt/hdt/enums/CompressionType.java
@@ -0,0 +1,52 @@
+package org.rdfhdt.hdt.enums;
+
+/**
+ * A compression type
+ * @author Antoine Willerval
+ */
+public enum CompressionType {
+
+	/**
+	 * gzip compression (.gz .tgz)
+	 */
+	GZIP("gz", "tgz"),
+	/**
+	 * bzip compression (.bz2 .bz)
+	 */
+	BZIP("bz2", "bz"),
+	/**
+	 * bzip compression (.xz)
+	 */
+	XZ("xz"),
+	/**
+	 * no compression
+	 */
+	NONE;
+
+	/**
+	 * try to guess a compression of a file with its name
+	 * @param fileName the file name to guess
+	 * @return the compression type or none if it can't be guessed
+	 */
+	public static CompressionType guess(String fileName) {
+		String str = fileName.toLowerCase();
+
+		int idx = str.lastIndexOf('.');
+		if(idx!=-1) {
+			String ext = str.substring(idx + 1);
+			for (CompressionType type: values()) {
+				for (String typeExt : type.ext) {
+					if (typeExt.equals(ext)) {
+						return type;
+					}
+				}
+			}
+		}
+		return NONE;
+	}
+
+	private final String[] ext;
+	CompressionType(String... ext) {
+		this.ext = ext;
+	}
+}
diff --git a/hdt-api/src/main/java/org/rdfhdt/hdt/enums/RDFNotation.java b/hdt-api/src/main/java/org/rdfhdt/hdt/enums/RDFNotation.java
@@ -110,26 +110,34 @@ public static RDFNotation parse(String str) {
 			return NTRIPLES;
 		}
 		str = str.toLowerCase();
-		if(str.equals("ntriples")||str.equals("nt")) {
-			return NTRIPLES;
-		} else if(str.equals("n3")) {
-			return N3;
-		} else if(str.equals("nq")||str.equals("nquad")) {
-			return NQUAD;
-		} else if(str.equals("rdfxml")||str.equals("rdf-xml") || str.equals("owl")) {
-			return RDFXML;
-		} else if(str.equals("turtle")) {
-			return TURTLE;
-		} else if(str.equals("rar")) {
-			return RAR;
-		} else if(str.equals("tar")||str.equals("tgz")||str.equals("tbz")||str.equals("tbz2")) {
-			return TAR;
-		} else if(str.equals("zip")) {
-			return ZIP;
-		} else if(str.equals("list")) {
-			return LIST;
-		} else if(str.equals("hdt")) {
-			return HDT;
+		switch (str) {
+			case "ntriples":
+			case "nt":
+				return NTRIPLES;
+			case "n3":
+				return N3;
+			case "nq":
+			case "nquad":
+				return NQUAD;
+			case "rdfxml":
+			case "rdf-xml":
+			case "owl":
+				return RDFXML;
+			case "turtle":
+				return TURTLE;
+			case "rar":
+				return RAR;
+			case "tar":
+			case "tgz":
+			case "tbz":
+			case "tbz2":
+				return TAR;
+			case "zip":
+				return ZIP;
+			case "list":
+				return LIST;
+			case "hdt":
+				return HDT;
 		}
 		throw new IllegalArgumentException();
 	}
@@ -147,7 +155,7 @@ public static RDFNotation guess(String fileName) throws IllegalArgumentException
 
 		int idx = str.lastIndexOf('.');		
 		if(idx!=-1) {
-			String ext = str.substring(idx+1, str.length());
+			String ext = str.substring(idx+1);
 			if(ext.equals("gz") || ext.equals("bz") || ext.equals("bz2")|| ext.equals("xz")) {
 				str = str.substring(0,idx);
 			}