From 663cd2f6cc5e3796d001e8c527de0aea8c8bbf68 Mon Sep 17 00:00:00 2001
From: nvalimak <nvalimak@3cdefd35-fc62-479d-8e8d-bae585ffb9ca>
Date: Wed, 4 Mar 2009 19:35:25 +0000
Subject: [PATCH] Added Count

git-svn-id: svn+ssh://idea.nguyen.vg/svn/sxsi/trunk/TextCollection@204 3cdefd35-fc62-479d-8e8d-bae585ffb9ca
---
 CSA.cpp          | 33 +++++++++++++--------------------
 CSA.h            | 11 +++++++++++
 TextCollection.h | 16 +++++++---------
 3 files changed, 31 insertions(+), 29 deletions(-)

diff --git a/CSA.cpp b/CSA.cpp
index 5f69aa8..04ace79 100644
--- a/CSA.cpp
+++ b/CSA.cpp
@@ -40,7 +40,7 @@ const uchar CSA::versionFlag = 2;
 
 ////////////////////////////////////////////////////////////////////////////
 // Class CSA::THuffAlphabetRank
-
+// FIXME Unused code
 CSA::THuffAlphabetRank::THuffAlphabetRank(uchar *s, TextPosition n, TCodeEntry *codetable, unsigned level) {
     left = NULL;
     right = NULL;
@@ -122,26 +122,8 @@ CSA::THuffAlphabetRank::~THuffAlphabetRank() {
         delete bitrank;
 }
 
-/**
- * Saving data fields:
-        BitRank *bitrank;
-        bool leaf;
-        uchar ch;
-        left child;
-        right child;
-*/
-void CSA::THuffAlphabetRank::Save(FILE *file)
-{
-
-}
-
-CSA::THuffAlphabetRank::THuffAlphabetRank(FILE *file)
-{
-
-
-}
 
-/////////////////////////////////////////////f///////////////////////////////
+////////////////////////////////////////////////////////////////////////////
 // Class CSA
 
 /**
@@ -408,6 +390,17 @@ bool CSA::IsLessThan(uchar const*) const
 /******************************************************************
  * Counting queries
  */
+ulong CSA::Count(uchar const * pattern) const
+{
+    TextPosition m = strlen((char *)pattern);
+    if (m == 0)
+        return 0;
+
+    TextPosition sp = 0, ep = 0;
+    unsigned count = (unsigned) Search(pattern, m, &sp, &ep);
+    return count;
+}
+
 unsigned CSA::CountPrefix(uchar const * pattern) const
 {
     TextPosition m = strlen((char *)pattern);
diff --git a/CSA.h b/CSA.h
index ec6e890..df56585 100644
--- a/CSA.h
+++ b/CSA.h
@@ -41,6 +41,7 @@
 #else
 #   define W 32
 #endif
+#undef bitset
 
 
 /**
@@ -73,6 +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;
@@ -81,6 +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;
@@ -128,6 +135,7 @@ public:
     }
 
 private:
+    // FIXME Unused code
     class TCodeEntry {
     public:
         unsigned count;
@@ -137,6 +145,7 @@ private:
     };   
      
 
+    // FIXME Unused code
     class THuffAlphabetRank {
     // using fixed 0...255 alphabet
     private:
@@ -223,6 +232,7 @@ private:
         }
     };
 
+    // FIXME Unused code
     class node {
     private:
         unsigned weight;
@@ -256,6 +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;
diff --git a/TextCollection.h b/TextCollection.h
index ef0eb72..37d5cb5 100644
--- a/TextCollection.h
+++ b/TextCollection.h
@@ -111,7 +111,7 @@ namespace SXSI
          * 
          * Note: Do we need this?
          * Forward iterator would be really in-efficient compared to
-         * getText(k) and getText(k, i, j).
+         * getText(k).
          *
          * TODO Define and implement const_reverse_iterator.
          */
@@ -132,8 +132,12 @@ namespace SXSI
         virtual bool IsLessThan(uchar const *) const = 0;
 
         /**
-         * Counting queries 
-         * 
+         * Counting queries
+         * Result is the number of occurrences.
+         */
+        virtual ulong Count(uchar const *) const = 0;
+        /**
+         * More counting queries
          * Result is the number of documents.
          */
         virtual unsigned CountPrefix(uchar const *) const = 0;
@@ -164,12 +168,6 @@ namespace SXSI
         typedef std::vector<std::pair<DocId, TextPosition> > full_result;
         virtual full_result FullContains(uchar const *) const = 0;
 
-
-        /**
-         *Debug
-         *
-         */
-        virtual TextPosition Lookup(TextPosition) const = 0;
     protected:
         // Protected constructor; call the static function InitTextCollection().
         TextCollection() { };
-- 
2.17.1