X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=TextCollectionBuilder.h;h=6b3819a2ead0ff2aba0fa7ac9f3f350bd9ac556c;hb=2dba1ef94915ba5c1a06c3756d6dd0333605a5f7;hp=3c725124cc715ffdcd0e2b6e22c64f17f926c7fe;hpb=40ddf9aca842bdc081b6350a4ebfe36b066c94c9;p=SXSI%2FTextCollection.git

diff --git a/TextCollectionBuilder.h b/TextCollectionBuilder.h
index 3c72512..6b3819a 100644
--- a/TextCollectionBuilder.h
+++ b/TextCollectionBuilder.h
@@ -22,9 +22,23 @@
 #define _SXSI_TextCollectionBuilder_h_
 
 #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
+
+// Default samplerate for suffix array samples
+#define TEXTCOLLECTION_DEFAULT_SAMPLERATE 64
+
+// Default input length, used to calculate the buffer size.
+#define TEXTCOLLECTION_DEFAULT_INPUT_LENGTH (150 * 1024 * 1024)
+
 
 namespace SXSI
 {
@@ -36,7 +50,8 @@ namespace SXSI
     class TextCollectionBuilder
     {
     public:
-        explicit TextCollectionBuilder(unsigned);
+        explicit TextCollectionBuilder(unsigned samplerate = TEXTCOLLECTION_DEFAULT_SAMPLERATE, 
+                                       ulong estimatedInputLength =  TEXTCOLLECTION_DEFAULT_INPUT_LENGTH);
         ~TextCollectionBuilder();
         
         /** 
@@ -46,21 +61,28 @@ namespace SXSI
          * Can not be called after makeStatic().
          * The i'th text insertion gets an identifier value i-1.
          * In other words, document identifiers start from 0.
+         *
+         * Second parameter tells if the text will be added to the
+         * index also. If false, text is added only to the TextCollection
+         * and can not be searched for.
          */
-        void InsertText(uchar const *);
+        void InsertText(uchar const *, bool index = true);
         /**
          * Make static
          *
-         * Convert to a static collection; reduces space and time complexities.
+         * Convert to a static collection.
          * New texts can not be inserted after this operation.
+         *
+         * TextStorage type defaults to TYPE_PLAIN_TEXT, another
+         * possible type is TYPE_LZ_INDEX.
          */
-        TextCollection * InitTextCollection();
+        TextCollection * InitTextCollection(char type = TextStorage::TYPE_PLAIN_TEXT);
         
     private:
+        // Using Pimpl idiom to hide RLCSA implementation.
         struct TCBuilderRep * p_;
 
         // No copy constructor or assignment
-        TextCollectionBuilder();
         TextCollectionBuilder(TextCollectionBuilder const&);
         TextCollectionBuilder& operator = (TextCollectionBuilder const&);
     };