Added TextStorage class

[SXSI/TextCollection.git] / TCImplementation.cpp

diff --git a/TCImplementation.cpp b/TCImplementation.cpp
index a8303d8..5959e6a 100644 (file)
--- a/TCImplementation.cpp
+++ b/TCImplementation.cpp
@@ -19,6 +19,7 @@
  *****************************************************************************/
 #include "TCImplementation.h"
 
+//#define DEBUG_MEMUSAGE
 #ifdef DEBUG_MEMUSAGE
 #include "HeapProfiler.h" // FIXME remove
 #endif
@@ -40,7 +41,7 @@ namespace SXSI
 {
 
 // Save file version info
-const uchar TCImplementation::versionFlag = 4;
+const uchar TCImplementation::versionFlag = 6;
 
 /**
  * Constructor inits an empty dynamic FM-index.
@@ -63,10 +64,12 @@ bool TCImplementation::EmptyText(DocId k) const
     return false; // Empty texts are not indexed
 }
 
-uchar* TCImplementation::GetText(DocId k) const
+uchar * TCImplementation::GetText(DocId k) const
 {
     assert(k < (DocId)numberOfTexts);
-    TextPosition i = k;
+
+    return textStorage->GetText(k);
+/*    TextPosition i = k;
 
     string result;
     // Reserve average string length to avoid reallocs
@@ -87,7 +90,7 @@ uchar* TCImplementation::GetText(DocId k) const
     res[i] = '\0';   
     for (ulong j = 0; j < i; ++j)
         res[i-j-1] = result[j];
-    return res;
+        return res;*/
 }
 /*
  * Not supported
@@ -769,6 +772,7 @@ void TCImplementation::Save(FILE *file) const
         throw std::runtime_error("TCImplementation::Save(): file write error (maxTextLength).");
 
     Doc->save(file);
+    textStorage->Save(file);
     fflush(file);
 }
 
@@ -815,6 +819,7 @@ TCImplementation::TCImplementation(FILE *file, unsigned samplerate_)
         throw std::runtime_error("TCImplementation::Load(): file read error (maxTextLength).");
 
     Doc = static_sequence::load(file);
+    textStorage = new TextStorage(file);
 
     // FIXME Construct data structures with new samplerate
     //maketables(); 
@@ -906,6 +911,7 @@ TCImplementation::~TCImplementation() {
     delete suffixes;
     delete suffixDocId;
     delete Doc;
+    delete textStorage;
 }
 
 void TCImplementation::makewavelet(uchar *bwt)
@@ -1036,6 +1042,8 @@ void TCImplementation::maketables()
     p=bwtEndPos;
     textId = numberOfTexts;
 
+    TextStorageBuilder tsbuilder(n);
+
     /**
      * Second pass: populate tables suffixes and suffixDocId.
      */
@@ -1052,6 +1060,8 @@ void TCImplementation::maketables()
         }
 
         uchar c = alphabetrank->access(p, alphabetrank_i_tmp);
+        tsbuilder[i] = c;
+
         if (c == '\0')
         {
             --textId;
@@ -1062,9 +1072,10 @@ void TCImplementation::maketables()
             p = C[c]+alphabetrank_i_tmp-1;
     }
     assert(textId == 0);
-
     delete textStartPos;
 
+    textStorage = tsbuilder.InitTextStorage();
+
 #ifdef DEBUG_MEMUSAGE
     std::cerr << "max heap usage before Doc: " << HeapProfiler::GetMaxHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
     HeapProfiler::ResetMaxHeapConsumption();