Added RLCSA index option

[SXSI/TextCollection.git] / SWCSAWrapper.h

diff --git a/SWCSAWrapper.h b/SWCSAWrapper.h
index 731fc83..8081ec7 100644 (file)
--- a/SWCSAWrapper.h
+++ b/SWCSAWrapper.h
@@ -1,7 +1,7 @@
 /******************************************************************************
  *   Copyright (C) 2010 by Niko Välimäki                                      *
  *                                                                            *
- *   FMIndex implementation for the TextCollection interface                  *
+ *   SWCSA implementation for the TextCollection interface                  *
  *                                                                            *
  *   This program is free software; you can redistribute it and/or modify     *
  *   it under the terms of the GNU Lesser General Public License as published *
@@ -57,7 +57,7 @@ class SWCSAWrapper : public SXSI::TextCollection {
 public:
     SWCSAWrapper(uchar * text, ulong length, unsigned samplerate,
                  unsigned numberOfTexts_)
-        : index(0), offsets(0), n(length), seSize(0), numberOfTexts(numberOfTexts_)
+        : index(0), offsets(0), offit(0), n(length), seSize(0), numberOfTexts(numberOfTexts_)
     {
        // Inicializes the arrays used to detect if a char is valid or not.
        StartValid();
@@ -125,7 +125,9 @@ public:
         fprintf(stderr,"\n-----------------------\nnumber of words = %lu\n------------------------\n", seSize);
         encoder.setBit(seSize);
         offsets = new CSA::DeltaVector(encoder, seSize+1);
-        std::cerr << "Number of texts: " << numberOfTexts << " vs " << offsets->rank(seSize - 1) << std::endl;
+        offit = new CSA::DeltaVector::Iterator(*offsets);
+
+        std::cerr << "Number of texts: " << numberOfTexts << " vs " << offit->rank(seSize - 1) << std::endl;
 
         char opt[100];
         snprintf(opt, 99, "sA=%u;sAinv=%u;sPsi=%u", samplerate, samplerate, samplerate);
@@ -147,6 +149,7 @@ public:
             std::exit(r);
         }
         index = 0;
+        delete offit; offit = 0;
         delete offsets; offsets = 0;
     }
 
@@ -182,8 +185,8 @@ public:
     { 
         ulong from, to, l;
         uchar *text;
-        from = offsets->select(i);
-        to = offsets->select(i+1); // ADD one 1-bit in to end!!!
+        from = offit->select(i);
+        to = offit->select(i+1); // ADD one 1-bit in to end!!!
         
         int r = extractWords(index, from, to, &text, &l);
         if (r)
@@ -249,7 +252,7 @@ public:
         document_result dr;
         dr.reserve(numocc+1);
         for (ulong i = 0; i < numocc; ++i)
-            dr.push_back(offsets->rank(occ[i])-1);
+            dr.push_back(offit->rank(occ[i])-1);
 
         free(occ);
         return dr;
@@ -272,7 +275,7 @@ public:
 
     // Index from/to disk
     SWCSAWrapper(FILE *file, char const *filename)
-        : index(0), offsets(0), n(0), seSize(0), numberOfTexts(0)
+        : index(0), offsets(0), offit(0), n(0), seSize(0), numberOfTexts(0)
     {
         uchar verFlag = 0;
         if (std::fread(&verFlag, 1, 1, file) != 1)
@@ -287,7 +290,8 @@ public:
         if (std::fread(&(this->numberOfTexts), sizeof(unsigned), 1, file) != 1)
             throw std::runtime_error("FMIndex::Load(): file read error (numberOfTexts).");
         
-        offsets = new CSA::DeltaVector(file);   
+        offsets = new CSA::DeltaVector(file);
+        offit = new CSA::DeltaVector::Iterator(*offsets);
 
         // FIXME Const correctness is broken!
         int r = load_index((char *)filename, &index);
@@ -331,6 +335,7 @@ public:
 private:
     void *index;
     CSA::DeltaVector *offsets;
+    CSA::DeltaVector::Iterator *offit;
     
     TextPosition n;
     ulong seSize;