X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=TextCollectionBuilder.h;h=1eb0275550fb05145972dfd0ad5399cf3400cdd0;hb=d660b6ec5cd55019d17188810b783a2e3a94fa49;hp=13734d22f74ec09f8c18cb2c04b39992dd131f89;hpb=f58cd5c3504095d721eb4e21bfb0a70a0a4928a2;p=SXSI%2FTextCollection.git

diff --git a/TextCollectionBuilder.h b/TextCollectionBuilder.h
index 13734d2..1eb0275 100644
--- a/TextCollectionBuilder.h
+++ b/TextCollectionBuilder.h
@@ -22,6 +22,7 @@
 #define _SXSI_TextCollectionBuilder_h_
 
 #include "TextCollection.h"
+#include "TextStorage.h"
 #include "Tools.h" // Defines ulong and uchar.
 #include <vector>
 #include <utility> // Defines std::pair.
@@ -33,6 +34,8 @@
 // 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
@@ -45,7 +48,8 @@ namespace SXSI
     class TextCollectionBuilder
     {
     public:
-        explicit TextCollectionBuilder(unsigned samplerate = TEXTCOLLECTION_DEFAULT_SAMPLERATE);
+        explicit TextCollectionBuilder(unsigned samplerate = TEXTCOLLECTION_DEFAULT_SAMPLERATE, 
+                                       ulong estimatedInputLength =  TEXTCOLLECTION_DEFAULT_INPUT_LENGTH);
         ~TextCollectionBuilder();
         
         /** 
@@ -60,10 +64,13 @@ namespace SXSI
         /**
          * 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:
         struct TCBuilderRep * p_;