X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=TextCollectionBuilder.h;h=9bac6f90e1f0deae57b92dd1b520428aaf59b590;hb=4bae924993abf1aa9107fe916dd9eb2ee2fea956;hp=bcb3672515b613179f7cb58b580837720ff0b0cc;hpb=ee8a3e526fe7f39cdc075263824faf6c17389297;p=SXSI%2FTextCollection.git diff --git a/TextCollectionBuilder.h b/TextCollectionBuilder.h index bcb3672..9bac6f9 100644 --- a/TextCollectionBuilder.h +++ b/TextCollectionBuilder.h @@ -27,6 +27,16 @@ #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 { struct TCBuilderRep; // Pimpl @@ -37,7 +47,8 @@ namespace SXSI class TextCollectionBuilder { public: - explicit TextCollectionBuilder(unsigned); + explicit TextCollectionBuilder(unsigned samplerate = TEXTCOLLECTION_DEFAULT_SAMPLERATE, + ulong estimatedInputLength = TEXTCOLLECTION_DEFAULT_INPUT_LENGTH); ~TextCollectionBuilder(); /** @@ -61,7 +72,6 @@ namespace SXSI struct TCBuilderRep * p_; // No copy constructor or assignment - TextCollectionBuilder(); TextCollectionBuilder(TextCollectionBuilder const&); TextCollectionBuilder& operator = (TextCollectionBuilder const&); };