Removed shift and alphabet size from SA

chinaeva · Apr 1, 2019 · 5442b56 · 5442b56
1 parent e0f960e
commit 5442b56
Show file tree

Hide file tree

Showing 5 changed files with 115 additions and 116 deletions.
diff --git a/com/williamfiset/datastructures/suffixarray/SuffixArray.java b/com/williamfiset/datastructures/suffixarray/SuffixArray.java
@@ -1,52 +1,72 @@
+/**
+ * Abstract class that captures the behavior of a suffix array.
+ *
+ * @author William Fiset, william.alexandre.fiset@gmail.com
+ */
+
 package com.williamfiset.datastructures.suffixarray;
 
 public abstract class SuffixArray {
 
-  protected static final int DEFAULT_ALPHABET_SHIFT = 0;
-  protected static final int DEFAULT_ALPHABET_SIZE = 256;
-
   // Length of the suffix array
-  public final int N;
-
-  protected int shift = DEFAULT_ALPHABET_SHIFT;
-
-  protected int alphabetSize = DEFAULT_ALPHABET_SIZE;
+  protected final int N;
 
   // T is the text
-  public int[] T;
+  protected int[] T;
 
   // The sorted suffix array values.
-  public int[] sa;
+  protected int[] sa;
 
   // Longest Common Prefix array
-  public int [] lcp;
+  protected int [] lcp;
+
+  private boolean constructedSa = false;
+  private boolean constructedLcpArray = false;
 
-  // Designated constructor
-  public SuffixArray(int[] text, int shift, int alphabetSize) {
-
-    if (text == null || alphabetSize <= 0) 
-      throw new IllegalArgumentException();
-
+  public SuffixArray(int[] text) {
+    if (text == null) 
+      throw new IllegalArgumentException("Text cannot be null.");
     this.T = text;
     this.N = text.length;
-
-    this.shift = shift;
-    this.alphabetSize = alphabetSize;
-
-    // Build suffix array
+  }
+
+  public int getTextLength() {
+    return T.length;
+  }
+
+  // Returns the suffix array.
+  public int[] getSa() {
+    buildSuffixArray();
+    return sa;
+  }
+
+  // Returns the LCP array.
+  public int[] getLcpArray() {
+    buildLcpArray();
+    return lcp;
+  }
+
+  // Builds the suffix array by calling the construct() method.
+  private void buildSuffixArray() {
+    if (constructedSa) return;
     construct();
-
-    // Build LCP array
+    constructedSa = true;
+  }
+
+  // Builds the LCP array by first creating the SA and then running the kasai algorithm.
+  private void buildLcpArray() {
+    if (constructedLcpArray) return;
+    buildSuffixArray();
     kasai();
-
+    constructedLcpArray = true;
   }
 
   protected static int[] toIntArray(String s) {
     if (s == null) return null;
-    int[] text = new int[s.length()];
+    int[] t = new int[s.length()];
     for(int i = 0; i < s.length(); i++)
-      text[i] = s.charAt(i);
-    return text;
+      t[i] = s.charAt(i);
+    return t;
   }
 
   // The suffix array construction algorithm is left undefined 
@@ -69,23 +89,21 @@ private void kasai() {
     }
   }
 
-  @Override public String toString() {
-
+  @Override 
+  public String toString() {
     StringBuilder sb = new StringBuilder();
     sb.append("-----i-----SA-----LCP---Suffix\n");
 
     for(int i = 0; i < N; i++) {
       int suffixLen = N - sa[i];
       char[] suffixArray = new char[suffixLen];
       for (int j = sa[i], k = 0; j < N; j++, k++)
-        suffixArray[k] = (char)(T[j] - shift);
+        suffixArray[k] = (char) T[j];
       String suffix = new String(suffixArray);
       String formattedStr = String.format("% 7d % 7d % 7d %s\n", i, sa[i], lcp[i], suffix);
       sb.append(formattedStr);
     }
-
     return sb.toString();
-
   }
 
 }
diff --git a/com/williamfiset/datastructures/suffixarray/SuffixArrayFast.java b/com/williamfiset/datastructures/suffixarray/SuffixArrayFast.java
@@ -9,32 +9,27 @@
 
 public class SuffixArrayFast extends SuffixArray {
 
+  private static final int DEFAULT_ALPHABET_SIZE = 256;
+
+  int alphabetSize;
   int[] sa2, rank, tmp, c;
 
   public SuffixArrayFast(String text) {
-    super(toIntArray(text), DEFAULT_ALPHABET_SHIFT, DEFAULT_ALPHABET_SIZE);
+    this(toIntArray(text), DEFAULT_ALPHABET_SIZE);
   }
 
   public SuffixArrayFast(int[] text) {
-    super(text, DEFAULT_ALPHABET_SHIFT, DEFAULT_ALPHABET_SIZE);
-  }
-
-  // TODO(williamfiset): Get rid of these constructors in favor of
-  // automatically detecting the alphabet size shift required
-  public SuffixArrayFast(String text, int shift) {
-    super(toIntArray(text), shift, DEFAULT_ALPHABET_SHIFT);
-  }
-  public SuffixArrayFast(int[] text, int shift) {
-    super(text, shift, DEFAULT_ALPHABET_SIZE);
+    this(text, DEFAULT_ALPHABET_SIZE);
   }
 
   // Designated constructor
-  public SuffixArrayFast(int[] text, int shift, int alphabetSize) {
-    super(text, shift, alphabetSize);
+  public SuffixArrayFast(int[] text, int alphabetSize) {
+    super(text);
+    this.alphabetSize = alphabetSize;
   }
 
-  @Override protected void construct() {
-    
+  @Override
+  protected void construct() {
     sa = new int[N];
     sa2 = new int[N];
     rank = new int[N];

diff --git a/com/williamfiset/datastructures/suffixarray/SuffixArrayMed.java b/com/williamfiset/datastructures/suffixarray/SuffixArrayMed.java
@@ -10,49 +10,36 @@
 
 public class SuffixArrayMed extends SuffixArray {
 
-  // Helper class which sorts suffix ranks
-  static class SuffixRankTuple implements Comparable <SuffixRankTuple> {
+  // Wrapper class to help sort suffix ranks
+  static class SuffixRankTuple implements Comparable<SuffixRankTuple> {
 
     int firstHalf, secondHalf, originalIndex;
 
     // Sort Suffix ranks first on the first half then the second half
-    @Override public int compareTo(SuffixRankTuple other) {
+    @Override 
+    public int compareTo(SuffixRankTuple other) {
       int cmp = Integer.compare(firstHalf, other.firstHalf);
       if (cmp == 0) return Integer.compare(secondHalf, other.secondHalf);
       return cmp;
     }
 
-    @Override public String toString() {
+    @Override
+    public String toString() {
       return originalIndex + " -> (" + firstHalf + ", " + secondHalf + ")";
     }
-
   }
 
   public SuffixArrayMed(String text) {
-    super(toIntArray(text), DEFAULT_ALPHABET_SHIFT, DEFAULT_ALPHABET_SIZE);
+    super(toIntArray(text));
   }
 
   public SuffixArrayMed(int[] text) {
-    super(text, DEFAULT_ALPHABET_SHIFT, DEFAULT_ALPHABET_SIZE);
-  }
-
-  // TODO(williamfiset): Get rid of these constructors in favor of
-  // automatically detecting the alphabet size shift required
-  public SuffixArrayMed(String text, int shift) {
-    super(toIntArray(text), shift, DEFAULT_ALPHABET_SHIFT);
-  }
-  public SuffixArrayMed(int[] text, int shift) {
-    super(text, shift, DEFAULT_ALPHABET_SIZE);
-  }
-
-  // Designated constructor
-  public SuffixArrayMed(int[] text, int shift, int alphabetSize) {
-    super(text, shift, alphabetSize);
+    super(text);
   }
 
   // Construct a suffix array in O(nlog^2(n))
-  @Override protected void construct() {
-
+  @Override 
+  protected void construct() {
     sa = new int[N];
 
     // Maintain suffix ranks in both a matrix with two rows containing the
@@ -114,7 +101,6 @@ public SuffixArrayMed(int[] text, int shift, int alphabetSize) {
     suffixRanks[0] = suffixRanks[1] = null;
     suffixRanks = null;
     ranks = null;
-
   }
 
   public static void main(String[] args) {

diff --git a/com/williamfiset/datastructures/suffixarray/SuffixArraySlow.java b/com/williamfiset/datastructures/suffixarray/SuffixArraySlow.java
@@ -9,8 +9,7 @@
 
 public class SuffixArraySlow extends SuffixArray {
 
-  static class Suffix implements Comparable <Suffix> {
-
+  private static class Suffix implements Comparable<Suffix> {
     // Starting position of suffix in text
     final int index, len;
     final int[] text;
@@ -22,7 +21,8 @@ public Suffix(int[] text, int index) {
     }
 
     // Compare the two suffixes inspired by Robert Sedgewick and Kevin Wayne
-    @Override public int compareTo(Suffix other) {
+    @Override
+    public int compareTo(Suffix other) {
       if (this == other) return 0;
       int min_len = Math.min(len, other.len);
       for (int i = 0; i < min_len; i++) {
@@ -32,42 +32,27 @@ public Suffix(int[] text, int index) {
       return len - other.len;
     }
 
-    @Override public String toString() {
+    @Override
+    public String toString() {
       return new String(text, index, len);
     }
-
   }
 
   // Contains all the suffixes of the SuffixArray
   Suffix[] suffixes;
 
   public SuffixArraySlow(String text) {
-    super(toIntArray(text), DEFAULT_ALPHABET_SHIFT, DEFAULT_ALPHABET_SIZE);
+    super(toIntArray(text));
   }
 
   public SuffixArraySlow(int[] text) {
-    super(text, DEFAULT_ALPHABET_SHIFT, DEFAULT_ALPHABET_SIZE);
-  }
-
-  // TODO(williamfiset): Get rid of these constructors in favor of
-  // automatically detecting the alphabet size shift required
-  public SuffixArraySlow(String text, int shift) {
-    super(toIntArray(text), shift, DEFAULT_ALPHABET_SHIFT);
-  }
-  public SuffixArraySlow(int[] text, int shift) {
-    super(text, shift, DEFAULT_ALPHABET_SIZE);
+    super(text);
   }
 
-  // Designated constructor
-  public SuffixArraySlow(int[] text, int shift, int alphabetSize) {
-    super(text, shift, alphabetSize);
-  }
-
-  // Suffix array construction. This acutally takes O(n^2log(n))
-  // time since sorting takes on average O(nlog(n)) and each String
-  // comparision takes O(n)
-  @Override protected void construct() {
-
+  // Suffix array construction. This actually takes O(n^2log(n)) time since sorting takes on
+  // average O(nlog(n)) and each String comparison takes O(n).
+  @Override
+  protected void construct() {
     sa = new int[N];
     suffixes = new Suffix[N];
 
@@ -83,12 +68,10 @@ public SuffixArraySlow(int[] text, int shift, int alphabetSize) {
     }
 
     suffixes = null;
-
   }
 
   public static void main(String[] args) {
     SuffixArraySlow sa = new SuffixArraySlow("ABBABAABAA");
     System.out.println(sa);
   }
-
 }