Added some performance measures

src/main/java/data/MultipleHMKeys.java

@@ -5,49 +5,16 @@ import java.util.Objects;
 /*
 Created for when words are sorted by multiple keys, i.e. not just lemmas but lemmas and msd simultaneously.
 */
-public final class MultipleHMKeys {
-    private final String key, lemma, wordType, msd;
-    private MultipleHMKeys actual_obj;
-    public MultipleHMKeys(String key) {
-        this.key = key;
-        this.lemma = "";
-        this.wordType = "";
-        this.msd = "";
-    }
-
-    public MultipleHMKeys(String key, String lemma, String wordType, String msd) {
-        this.key = key;
-        this.lemma = lemma;
-        this.wordType = wordType;
-        this.msd = msd;
-    }
-
-    public String getKey() {
-        return key;
-    }
-
-    public String getLemma() {
-        return lemma;
-    }
-
-    public String getWordType() {
-        return wordType;
-    }
-
-    public String getMsd() {
-        return msd;
-    }
+public interface MultipleHMKeys {
+    String getK1();
+    String getK2();
+    String getK3();
+    String getK4();
+    String getK5();
 
     @Override
-    public int hashCode() {
-        return Objects.hash(key, lemma, wordType, msd);
-    }
+    int hashCode();
 
     @Override
-    public boolean equals(Object obj) {
-        return (obj instanceof MultipleHMKeys) && ((MultipleHMKeys) obj).key.equals(key)
-                && ((MultipleHMKeys) obj).lemma.equals(lemma)
-                && ((MultipleHMKeys) obj).wordType.equals(wordType)
-                && ((MultipleHMKeys) obj).msd.equals(msd);
-    }
+    boolean equals(Object obj);
 }

src/main/java/data/MultipleHMKeys1.java

@@ -0,0 +1,44 @@
+package data;
+
+import java.util.Objects;
+
+/*
+Created for when words are sorted by multiple keys, i.e. not just lemmas but lemmas and msd simultaneously.
+*/
+public final class MultipleHMKeys1 implements MultipleHMKeys {
+    private final String k1;
+
+    public MultipleHMKeys1(String k1) {
+        this.k1 = k1;
+    }
+
+    public String getK1() {
+        return k1;
+    }
+
+    public String getK2() {
+        return null;
+    }
+
+    public String getK3() {
+        return null;
+    }
+
+    public String getK4() {
+        return null;
+    }
+
+    public String getK5() {
+        return null;
+    }
+
+    @Override
+    public int hashCode() {
+        return k1.hashCode();
+    }
+
+    @Override
+    public boolean equals(Object obj) {
+        return (obj instanceof MultipleHMKeys1) && ((MultipleHMKeys1) obj).k1.equals(k1);
+    }
+}

src/main/java/data/MultipleHMKeys2.java

@@ -0,0 +1,49 @@
+package data;
+
+import java.util.Objects;
+
+/*
+Created for when words are sorted by multiple keys, i.e. not just lemmas but lemmas and msd simultaneously.
+*/
+public final class MultipleHMKeys2 implements MultipleHMKeys {
+    private final String k1, k2;
+
+    public MultipleHMKeys2(String k1, String k2) {
+        this.k1 = k1;
+        this.k2 = k2;
+    }
+
+    public String getK1() {
+        return k1;
+    }
+
+    public String getK2() {
+        return k2;
+    }
+
+    public String getK3() {
+        return null;
+    }
+
+    public String getK4() {
+        return null;
+    }
+
+    public String getK5() {
+        return null;
+    }
+
+    @Override
+    public int hashCode() {
+        return Objects.hash(k1, k2);
+//        return key.hashCode();
+    }
+
+    @Override
+    public boolean equals(Object obj) {
+        return (obj instanceof MultipleHMKeys2) && ((MultipleHMKeys2) obj).k1.equals(k1)
+                && ((MultipleHMKeys2) obj).k2.equals(k2);
+
+//        return (obj instanceof MultipleHMKeys) && ((MultipleHMKeys) obj).key.equals(key);
+    }
+}

src/main/java/data/MultipleHMKeys3.java

@@ -0,0 +1,48 @@
+package data;
+
+import java.util.Objects;
+
+/*
+Created for when words are sorted by multiple keys, i.e. not just lemmas but lemmas and msd simultaneously.
+*/
+public final class MultipleHMKeys3 implements MultipleHMKeys {
+    private final String k1, k2, k3;
+
+    public MultipleHMKeys3(String k1, String k2, String k3) {
+        this.k1 = k1;
+        this.k2 = k2;
+        this.k3 = k3;
+    }
+
+    public String getK1() {
+        return k1;
+    }
+
+    public String getK2() {
+        return k2;
+    }
+
+    public String getK3() {
+        return k3;
+    }
+
+    public String getK4() {
+        return null;
+    }
+
+    public String getK5() {
+        return null;
+    }
+
+    @Override
+    public int hashCode() {
+        return Objects.hash(k1, k2, k3);
+    }
+
+    @Override
+    public boolean equals(Object obj) {
+        return (obj instanceof MultipleHMKeys3) && ((MultipleHMKeys3) obj).k1.equals(k1)
+                && ((MultipleHMKeys3) obj).k2.equals(k2)
+                && ((MultipleHMKeys3) obj).k3.equals(k3);
+    }
+}

src/main/java/data/MultipleHMKeys4.java

@@ -0,0 +1,50 @@
+package data;
+
+import java.util.Objects;
+
+/*
+Created for when words are sorted by multiple keys, i.e. not just lemmas but lemmas and msd simultaneously.
+*/
+public final class MultipleHMKeys4 implements MultipleHMKeys {
+    private final String k1, k2, k3, k4;
+
+    public MultipleHMKeys4(String k1, String k2, String k3, String k4) {
+        this.k1 = k1;
+        this.k2 = k2;
+        this.k3 = k3;
+        this.k4 = k4;
+    }
+
+    public String getK1() {
+        return k1;
+    }
+
+    public String getK2() {
+        return k2;
+    }
+
+    public String getK3() {
+        return k3;
+    }
+
+    public String getK4() {
+        return k4;
+    }
+
+    public String getK5() {
+        return null;
+    }
+
+    @Override
+    public int hashCode() {
+        return Objects.hash(k1, k2, k3, k4);
+    }
+
+    @Override
+    public boolean equals(Object obj) {
+        return (obj instanceof MultipleHMKeys4) && ((MultipleHMKeys4) obj).k1.equals(k1)
+                && ((MultipleHMKeys4) obj).k2.equals(k2)
+                && ((MultipleHMKeys4) obj).k3.equals(k3)
+                && ((MultipleHMKeys4) obj).k4.equals(k4);
+    }
+}

src/main/java/data/MultipleHMKeys5.java

@@ -0,0 +1,52 @@
+package data;
+
+import java.util.Objects;
+
+/*
+Created for when words are sorted by multiple keys, i.e. not just lemmas but lemmas and msd simultaneously.
+*/
+public final class MultipleHMKeys5 implements MultipleHMKeys {
+    private final String k1, k2, k3, k4, k5;
+
+    public MultipleHMKeys5(String k1, String k2, String k3, String k4, String k5) {
+        this.k1 = k1;
+        this.k2 = k2;
+        this.k3 = k3;
+        this.k4 = k4;
+        this.k5 = k5;
+    }
+
+    public String getK1() {
+        return k1;
+    }
+
+    public String getK2() {
+        return k2;
+    }
+
+    public String getK3() {
+        return k3;
+    }
+
+    public String getK4() {
+        return k4;
+    }
+
+    public String getK5() {
+        return k5;
+    }
+
+    @Override
+    public int hashCode() {
+        return Objects.hash(k1, k2, k3, k4, k5);
+    }
+
+    @Override
+    public boolean equals(Object obj) {
+        return (obj instanceof MultipleHMKeys5) && ((MultipleHMKeys5) obj).k1.equals(k1)
+                && ((MultipleHMKeys5) obj).k2.equals(k2)
+                && ((MultipleHMKeys5) obj).k3.equals(k3)
+                && ((MultipleHMKeys5) obj).k4.equals(k4)
+                && ((MultipleHMKeys5) obj).k5.equals(k5);
+    }
+}

src/main/java/data/Sentence.java

@@ -7,30 +7,30 @@ public class Sentence {
 
 
 	private List<Word> words;
-	private String taksonomija;
+	private List<String> taxonomy;
 
 	// GOS
 	private String type;
 	private Map<String, String> properties;
 
-	public Sentence(List<Word> words, String taksonomija) {
+	public Sentence(List<Word> words, List<String> taxonomy) {
 		this.words = words;
-		this.taksonomija = taksonomija;
+		this.taxonomy = taxonomy;
 	}
 
-	public Sentence(List<Word> words) {
-		this.words = words;
-	}
+//	public Sentence(List<Word> words) {
+//		this.words = words;
+//	}
 
-	public Sentence(List<Word> words, String taksonomija, Map<String, String> properties) {
+	public Sentence(List<Word> words, List<String> taxonomy, Map<String, String> properties) {
 		this.words = words;
-		this.taksonomija = taksonomija;
+		this.taxonomy = taxonomy;
 		this.properties = properties;
 	}
 
-	public Sentence(List<Word> words, String taksonomija, String type) {
+	public Sentence(List<Word> words, List<String> taxonomy, String type) {
 		this.words = words;
-		this.taksonomija = taksonomija;
+		this.taxonomy = taxonomy;
 		this.type = type;
 	}
 
@@ -38,8 +38,8 @@ public class Sentence {
 		return words;
 	}
 
-	public String getTaxonomy() {
-		return taksonomija;
+	public List<String> getTaxonomy() {
+		return taxonomy;
 	}
 
 	public List<Word> getSublist(int indexFrom, int indexTo) {

src/main/java/data/StatisticsNew.java

@@ -213,7 +213,7 @@ public class StatisticsNew {
 		removeMinimalOccurrences(taxonomyResult.get("Total"), filter.getMinimalOccurrences());
 		removeMinimalTaxonomy(taxonomyResult, filter.getMinimalTaxonomy());
 		stats.add(ImmutablePair.of(resultTitle, getSortedResult(taxonomyResult.get("Total"), Util.getValidInt(limit))));
-		Export.SetToCSV(stats, corpus.getChosenResultsLocation(), headerInfoBlock(), taxonomyResult);
+		Export.SetToCSV(stats, corpus.getChosenResultsLocation(), headerInfoBlock(), taxonomyResult, filter);
 		return true;
 	}
 
@@ -376,7 +376,7 @@ public class StatisticsNew {
 	}
 
 	public void updateResultsNestedSuffix(String key, String stringValue) {
-		MultipleHMKeys mkStringValue = new MultipleHMKeys(stringValue);
+		MultipleHMKeys mkStringValue = new MultipleHMKeys1(stringValue);
 
 		if (resultNestedSuffix.containsKey(key)) {
 			// if not in map
@@ -397,7 +397,7 @@ public class StatisticsNew {
 	}
 
 	public void updateResultsNestedPrefix(String key, String stringValue) {
-		MultipleHMKeys mkStringValue = new MultipleHMKeys(stringValue);
+		MultipleHMKeys mkStringValue = new MultipleHMKeys1(stringValue);
 
 		if (resultNestedPrefix.containsKey(key)) {
 			// if not in map

src/main/java/data/Word.java

@@ -16,8 +16,7 @@ public class Word implements Serializable {
 	private String word;
 	private String lemma;
 	private String msd;
-//	private String msd;
-	private List<String> taxonomy;
+	private String normalizedWord;
 	private final HashSet<Character> VOWELS = new HashSet<>(Arrays.asList('a', 'e', 'i', 'o', 'u'));
 
 	/**
@@ -41,7 +40,8 @@ public class Word implements Serializable {
 	//private char besedna_vrsta;
 	public Word(String word, String lemma, String msd) {
 		this.lemma = lemma;
-		this.msd = normalizeMsd(msd);
+		this.msd = msd; //normalizeMsd(msd);
+		this.normalizedWord = "";
 
 		// veliko zacetnico ohranimo samo za lastna imena
 		if (!ValidationUtil.isEmpty(this.msd) && !(this.msd.charAt(0) == 'S'
@@ -53,12 +53,11 @@ public class Word implements Serializable {
 		}
 	}
 
-	//private char besedna_vrsta;
-	public Word(String word, String lemma, String msd, List<String> taxonomy) {
+	public Word(String word, String lemma, String msd, String normalizedWord) {
 		this.lemma = lemma;
 //		this.msd = normalizeMsd(msd);
 		this.msd = msd;
-		this.taxonomy = taxonomy;
+		this.normalizedWord = normalizedWord;
 
 		// veliko zacetnico ohranimo samo za lastna imena
 		if (!ValidationUtil.isEmpty(this.msd) && !(this.msd.charAt(0) == 'S'
@@ -73,21 +72,21 @@ public class Word implements Serializable {
 	public Word() {
 	}
 
-	/**
-	 * Appends a number of '-' to msds which are not properly sized.
-	 * E.g. nouns should have 5 attributes, yet the last one isn't always defined (Somei vs. Sometd)
-	 *
-	 * @param msdInput
-	 *
-	 * @return
-	 */
-	private String normalizeMsd(String msdInput) {
-		if (ValidationUtil.isEmpty(msdInput)) {
-			return "";
-		} else {
-			return StringUtils.rightPad(msdInput, Msd.getMsdLengthForType(msdInput), PAD_CHARACTER);
-		}
-	}
+//	/**
+//	 * Appends a number of '-' to msds which are not properly sized.
+//	 * E.g. nouns should have 5 attributes, yet the last one isn't always defined (Somei vs. Sometd)
+//	 *
+//	 * @param msdInput
+//	 *
+//	 * @return
+//	 */
+//	private String normalizeMsd(String msdInput) {
+//		if (ValidationUtil.isEmpty(msdInput)) {
+//			return "";
+//		} else {
+//			return StringUtils.rightPad(msdInput, Msd.getMsdLengthForType(msdInput), PAD_CHARACTER);
+//		}
+//	}
 
 	public Word(String word) {
 		this.word = word;
@@ -119,10 +118,6 @@ public class Word implements Serializable {
 		this.word = word;
 	}
 
-	public List<String> getTaxonomy() {
-		return taxonomy;
-	}
-
 	public String getLemma() {
 		return lemma;
 	}
@@ -139,6 +134,14 @@ public class Word implements Serializable {
 		this.msd = msd;
 	}
 
+	public String getNormalizedWord() {
+		return normalizedWord;
+	}
+
+	public void setNormalizedWord(String normalizedWord) {
+		this.normalizedWord = normalizedWord;
+	}
+
 	public String toString() {
 		StringBuilder sb = new StringBuilder();
 
@@ -150,6 +153,8 @@ public class Word implements Serializable {
 				.append("\n")
 				.append("msd:\t")
 				.append(getMsd())
+				.append("normalized word:\t")
+				.append(getNormalizedWord())
 				.append("\n");
 
 		return sb.toString();