X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=CSA.h;h=df56585064171bbc4336b81e30aa34c4439a17bc;hb=663cd2f6cc5e3796d001e8c527de0aea8c8bbf68;hp=e5852b9dc2b30b3fd10c9cfd42f211e61bb94c10;hpb=2e59e4488d8d8ab43d6a28d40f81c7df6f811d56;p=SXSI%2FTextCollection.git

diff --git a/CSA.h b/CSA.h
index e5852b9..df56585 100644
--- a/CSA.h
+++ b/CSA.h
@@ -20,25 +20,29 @@
 
 #ifndef _CSA_H_
 #define _CSA_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>
 
-//clash between TextCollection/Tools.h and libcds/includes/basics.h
-// Number of bits in ulong vs uint.
+// Re-define word size to ulong:
 #undef W
-#undef WW
-#undef Wminusone
+#if __WORDSIZE == 64
+#   define W 64
+#else
+#   define W 32
+#endif
+#undef bitset
 
-#include "dynFMI.h"
-#include "BitRank.h"
-#include "TextCollection.h"
-#include "BlockArray.h"
-#include "RLWaveletTree.h"
-#include <set>
-#include <vector>
 
 /**
  * Implementation of the TextCollection interface
@@ -70,6 +74,11 @@ public:
     void MakeStatic();
     bool EmptyText(DocId) const;
     uchar* GetText(DocId) const;
+    /**
+     * Next method is not supported:
+     * Supporting GetText for some substring [i,j]
+     * would require more space.
+     */
 //    uchar* GetText(DocId, TextPosition, TextPosition) const;
 
     bool IsPrefix(uchar const *) const;
@@ -78,6 +87,7 @@ public:
     bool IsContains(uchar const *) const;
     bool IsLessThan(uchar const *) const;
 
+    ulong Count(uchar const *) const;
     unsigned CountPrefix(uchar const *) const;
     unsigned CountSuffix(uchar const *) const;
     unsigned CountEqual(uchar const *) const;
@@ -125,6 +135,7 @@ public:
     }
 
 private:
+    // FIXME Unused code
     class TCodeEntry {
     public:
         unsigned count;
@@ -134,6 +145,7 @@ private:
     };   
      
 
+    // FIXME Unused code
     class THuffAlphabetRank {
     // using fixed 0...255 alphabet
     private:
@@ -220,6 +232,7 @@ private:
         }
     };
 
+    // FIXME Unused code
     class node {
     private:
         unsigned weight;
@@ -253,6 +266,7 @@ private:
         static TCodeEntry *makecodetable(uchar *, TextPosition);
     };
     
+    // FIXME Unused code
     static const unsigned char print = 1;
     static const unsigned char report = 1;
     static const uchar versionFlag;