X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=TextCollectionBuilder.h;h=9bac6f90e1f0deae57b92dd1b520428aaf59b590;hb=5d74bc076c9d15d607d183c16ff8468ea961a0f9;hp=3c725124cc715ffdcd0e2b6e22c64f17f926c7fe;hpb=40ddf9aca842bdc081b6350a4ebfe36b066c94c9;p=SXSI%2FTextCollection.git diff --git a/TextCollectionBuilder.h b/TextCollectionBuilder.h index 3c72512..9bac6f9 100644 --- a/TextCollectionBuilder.h +++ b/TextCollectionBuilder.h @@ -25,6 +25,17 @@ #include "Tools.h" // Defines ulong and uchar. #include #include // Defines std::pair. +#include // 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 +47,8 @@ namespace SXSI class TextCollectionBuilder { public: - explicit TextCollectionBuilder(unsigned); + explicit TextCollectionBuilder(unsigned samplerate = TEXTCOLLECTION_DEFAULT_SAMPLERATE, + ulong estimatedInputLength = TEXTCOLLECTION_DEFAULT_INPUT_LENGTH); ~TextCollectionBuilder(); /** @@ -60,7 +72,6 @@ namespace SXSI struct TCBuilderRep * p_; // No copy constructor or assignment - TextCollectionBuilder(); TextCollectionBuilder(TextCollectionBuilder const&); TextCollectionBuilder& operator = (TextCollectionBuilder const&); };