X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=CSA.h;h=ec6e890be2bffd93e39452af1bd4d33013ed46e0;hb=a03a69215d4bf9cee1d42ecbc9adcf057b2229dc;hp=bd7b199507476130e1933126e14af006ba74f4ae;hpb=8c8e860bacea7245fb06b7b7e12ca073f45e1a42;p=SXSI%2FTextCollection.git

diff --git a/CSA.h b/CSA.h
index bd7b199..ec6e890 100644
--- a/CSA.h
+++ b/CSA.h
@@ -20,13 +20,28 @@
 
 #ifndef _CSA_H_
 #define _CSA_H_
-
-#include <map>
-#include <vector>
-#include "BitRank.h"
 #include "dynFMI.h"
+#include "BitRank.h"
 #include "TextCollection.h"
 #include "BlockArray.h"
+#include "RLWaveletTree.h"
+#include <set>
+#include <vector>
+
+// Include  from XMLTree/libcds
+#include <basics.h>
+#include <static_bitsequence.h>
+#include <alphabet_mapper.h>
+#include <static_sequence.h>
+
+// Re-define word size to ulong:
+#undef W
+#if __WORDSIZE == 64
+#   define W 64
+#else
+#   define W 32
+#endif
+
 
 /**
  * Implementation of the TextCollection interface
@@ -58,7 +73,7 @@ public:
     void MakeStatic();
     bool EmptyText(DocId) const;
     uchar* GetText(DocId) const;
-    uchar* GetText(DocId, TextPosition, TextPosition) const;
+//    uchar* GetText(DocId, TextPosition, TextPosition) const;
 
     bool IsPrefix(uchar const *) const;
     bool IsSuffix(uchar const *) const;
@@ -86,8 +101,31 @@ public:
     void Load(FILE *, unsigned);
     void Save(FILE *) const;
 
-    // debug
-    TextPosition Lookup(TextPosition) const;
+
+    // Debug FIXME Remove
+    void deleteWT()
+    {
+        delete alphabetrank;
+        alphabetrank = 0;
+        delete [] codetable;
+        codetable = 0;
+    }
+    void deleteSamples()
+    {
+        delete sampled;
+        sampled =0;
+        delete suffixes;
+        suffixes = 0;
+        delete positions;
+        positions = 0;
+        delete suffixDocId;
+        suffixDocId = 0;
+    }
+    void deleteEndmarker()
+    {
+        delete endmarkerDocId;
+        endmarkerDocId = 0;
+    }
 
 private:
     class TCodeEntry {
@@ -110,7 +148,10 @@ private:
         bool leaf;
     public:
         THuffAlphabetRank(uchar *, TextPosition, TCodeEntry *, unsigned);
+        THuffAlphabetRank(FILE *);
         ~THuffAlphabetRank();
+        
+        void Save(FILE *);
         bool Test(uchar *, TextPosition);
         
         inline TextPosition rank(int c, TextPosition i) const { // returns the number of characters c before and including position i
@@ -151,7 +192,7 @@ private:
             } 
             return true;
         }
-        inline int charAtPos(TextPosition i) const {
+        inline uchar access(TextPosition i) const {
             THuffAlphabetRank const * temp=this;
             while (!temp->leaf) {
                 if (temp->bitrank->IsBitSet(i)) {
@@ -163,7 +204,22 @@ private:
                     temp = temp->left;      
             }         
             }
-            return (int)temp->ch;
+            return temp->ch;
+        }
+
+        inline uchar charAtPos(ulong i, TextPosition *rank) const {
+            THuffAlphabetRank const * temp=this;
+            while (!temp->leaf) {
+                if (temp->bitrank->IsBitSet(i)) {
+                    i = temp->bitrank->rank(i)-1;
+                    temp = temp->right;
+                } else {
+                    i = i-temp->bitrank->rank(i);
+                    temp = temp->left;
+                }
+            }
+            (*rank)=i;
+            return temp->ch;
         }
     };
 
@@ -202,23 +258,28 @@ private:
     
     static const unsigned char print = 1;
     static const unsigned char report = 1;
+    static const uchar versionFlag;
     TextPosition n;
     unsigned samplerate;
     unsigned C[256];
     TextPosition bwtEndPos;
-    THuffAlphabetRank *alphabetrank;
-    BitRank *sampled; 
+//    THuffAlphabetRank *alphabetrank;
+    //    RLWaveletTree *alphabetrank;
+    static_sequence * alphabetrank;
+    BSGAP *sampled; 
     BlockArray *suffixes;
+    BlockArray *suffixDocId;
     BlockArray *positions;
     TCodeEntry *codetable;
     DynFMI *dynFMI;
 
     // Total number of texts in the collection
     unsigned numberOfTexts;
+    // Total number of texts including empty texts
+    unsigned numberOfAllTexts;
     // Length of the longest text
     ulong maxTextLength;
-    
-    
+
     // Array of document id's in the order of end-markers in BWT
     // Access by endmarkerDocId[rank_$(L, p) - 1].
     BlockArray *endmarkerDocId;
@@ -226,7 +287,11 @@ private:
     BlockArray *textLength;
     // Array of text starting positions (in the inserted order)
     BlockArray *textStartPos;
- 
+
+    // FIXME Replace with a more succinct data structure
+    std::set<unsigned> emptyTextId;
+    BSGAP *emptyTextRank;
+
     // Private methods
     uchar * BWT(uchar *);
     uchar * LoadFromFile(const char *);
@@ -236,12 +301,27 @@ private:
 
     // Following are not part of the public API
     DocId DocIdAtTextPos(TextPosition) const;
-    unsigned CountEndmarkers(TextPosition, TextPosition) const;
     ulong Search(uchar const *, TextPosition, TextPosition *, TextPosition *) const;
     TextPosition Inverse(TextPosition) const;
     TextPosition LF(uchar c, TextPosition &sp, TextPosition &ep) const;
     TextPosition Psi(TextPosition) const;
     uchar * Substring(TextPosition, TextPosition) const;
+    TextPosition Lookup(TextPosition) const;
+
+    /**
+     * Count end-markers in given interval
+     */
+    inline unsigned CountEndmarkers(TextPosition sp, TextPosition ep) const
+    {
+        if (sp > ep)
+            return 0;
+
+        ulong ranksp = 0;
+        if (sp != 0)
+            ranksp = alphabetrank->rank(0, sp - 1);
+    
+        return alphabetrank->rank(0, ep) - ranksp;
+    }
 };
 
 #endif