Added SWCSA

[SXSI/TextCollection.git] / TextCollectionBuilder.h

diff --git a/TextCollectionBuilder.h b/TextCollectionBuilder.h
index 6b3819a..43d4204 100644 (file)
--- a/TextCollectionBuilder.h
+++ b/TextCollectionBuilder.h
@@ -23,18 +23,10 @@
 
 #include "TextCollection.h"
 #include "TextStorage.h"
-#include "Tools.h" // Defines ulong and uchar.
-
-#include <string>
-#include <vector>
-#include <utility> // Defines std::pair.
-#include <cstring> // Defines std::strlen, added by Kim
-
-// Un-comment to compare BWT against a BWT generated from class dynFMI:
-//#define TCB_TEST_BWT
+#include "Tools.h"
 
 // Default samplerate for suffix array samples
-#define TEXTCOLLECTION_DEFAULT_SAMPLERATE 64
+#define TEXTCOLLECTION_DEFAULT_SAMPLERATE 32
 
 // Default input length, used to calculate the buffer size.
 #define TEXTCOLLECTION_DEFAULT_INPUT_LENGTH (150 * 1024 * 1024)
@@ -42,17 +34,23 @@
 
 namespace SXSI
 {
-    struct TCBuilderRep; // Pimpl
-    
     /**
-     * Build an instance of the TextCollection class.
+     * Builder for an instance of the TextCollection class.
      */
     class TextCollectionBuilder
     {
     public:
-        explicit TextCollectionBuilder(unsigned samplerate = TEXTCOLLECTION_DEFAULT_SAMPLERATE, 
-                                       ulong estimatedInputLength =  TEXTCOLLECTION_DEFAULT_INPUT_LENGTH);
-        ~TextCollectionBuilder();
+        // Index type defaults to FM-index.
+        // SWCSA can be used for natural language inputs.
+        // NB: Current SWCSA uses a lot of memory during construction!
+        enum index_type_t { index_type_default, index_type_swcsa }; 
+
+        static TextCollectionBuilder* create(unsigned samplerate = TEXTCOLLECTION_DEFAULT_SAMPLERATE, 
+                                             index_type_t type = index_type_default,
+                                             ulong estimatedInputLength =  TEXTCOLLECTION_DEFAULT_INPUT_LENGTH);
+        
+
+        virtual ~TextCollectionBuilder() { };
         
         /** 
          * Insert text
@@ -66,7 +64,7 @@ namespace SXSI
          * index also. If false, text is added only to the TextCollection
          * and can not be searched for.
          */
-        void InsertText(uchar const *, bool index = true);
+        virtual void InsertText(uchar const *, bool index = true) = 0;
         /**
          * Make static
          *
@@ -76,15 +74,18 @@ namespace SXSI
          * TextStorage type defaults to TYPE_PLAIN_TEXT, another
          * possible type is TYPE_LZ_INDEX.
          */
-        TextCollection * InitTextCollection(char type = TextStorage::TYPE_PLAIN_TEXT);
+        virtual TextCollection * InitTextCollection(char type = TextStorage::TYPE_PLAIN_TEXT) = 0;
         
-    private:
-        // Using Pimpl idiom to hide RLCSA implementation.
-        struct TCBuilderRep * p_;
+    protected:
+        // Protected constructor; use the static method TextCollectionBuilder::create()
+        TextCollectionBuilder() { };
 
+    private:
         // No copy constructor or assignment
         TextCollectionBuilder(TextCollectionBuilder const&);
         TextCollectionBuilder& operator = (TextCollectionBuilder const&);
     };
+
 }
+
 #endif