X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=TextCollectionBuilder.cpp;h=467bf949677e7ae05d6ee279828b2fb0853fbdd4;hb=d660b6ec5cd55019d17188810b783a2e3a94fa49;hp=1d49f397ec4e52b74d6c32b9ae304daddd711cc2;hpb=6e35318fa5b3d5630aa8e5c8ac019d62a47b8948;p=SXSI%2FTextCollection.git diff --git a/TextCollectionBuilder.cpp b/TextCollectionBuilder.cpp index 1d49f39..467bf94 100644 --- a/TextCollectionBuilder.cpp +++ b/TextCollectionBuilder.cpp @@ -43,6 +43,9 @@ TextCollectionBuilder::TextCollectionBuilder(unsigned samplerate, ulong estimate p_->numberOfSamples = 0; // Current params: 8 bytes, no samples, buffer size n/10 bytes. + // Buffer size is always at least 15MB: + if (estimatedInputLength < TEXTCOLLECTION_DEFAULT_INPUT_LENGTH) + estimatedInputLength = TEXTCOLLECTION_DEFAULT_INPUT_LENGTH; p_->sa = new CSA::RLCSABuilder(8, 0, estimatedInputLength/10); assert(p_->sa->isOk()); @@ -92,7 +95,7 @@ void TextCollectionBuilder::InsertText(uchar const * text) } -TextCollection * TextCollectionBuilder::InitTextCollection() +TextCollection * TextCollectionBuilder::InitTextCollection(char type) { uchar * bwt = 0; CSA::usint length = 0; @@ -148,7 +151,7 @@ TextCollection * TextCollectionBuilder::InitTextCollection() } TextCollection *result = new TCImplementation(bwt, (ulong)length, - p_->samplerate, p_->numberOfTexts, p_->maxTextLength, p_->numberOfSamples); + p_->samplerate, p_->numberOfTexts, p_->maxTextLength, p_->numberOfSamples, type); return result; }