From 65e613f046f3316769cc66eb27bc891d90ec3dc6 Mon Sep 17 00:00:00 2001
From: nvalimak <nvalimak@3cdefd35-fc62-479d-8e8d-bae585ffb9ca>
Date: Thu, 16 Apr 2009 11:45:02 +0000
Subject: [PATCH] Fix for mem usage of Doc

git-svn-id: svn+ssh://idea.nguyen.vg/svn/sxsi/trunk/TextCollection@320 3cdefd35-fc62-479d-8e8d-bae585ffb9ca
---
 TCImplementation.cpp | 38 ++++++++++++++++++--------------------
 1 file changed, 18 insertions(+), 20 deletions(-)

diff --git a/TCImplementation.cpp b/TCImplementation.cpp
index 5a754c7..8112b45 100644
--- a/TCImplementation.cpp
+++ b/TCImplementation.cpp
@@ -891,7 +891,7 @@ ulong TCImplementation::SearchLessThan(uchar const * pattern, TextPosition m, Te
     {
 //         printf("i = %lu, c = %c, sp = %lu, ep = %lu\n", i, pattern[i], sp, ep);
         c = (int)pattern[--i];
-        uint result = alphabetrank->rank(c,ep);
+        uint result = alphabetrank->rankLessThan(c,ep);
         if (result == ~0u)
             ep = 0;
         else
@@ -939,19 +939,18 @@ void TCImplementation::makewavelet(uchar *bwt)
 //    delete [] bwt;
     //alphabetrank = new RLWaveletTree(bwt, n); // Deletes bwt!
 //  std::cerr << "heap usage: " << HeapProfiler::GetHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
-//    std::cerr << "max heap usage before WT: " << HeapProfiler::GetMaxHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
 
+//    std::cerr << "max heap usage before WT: " << HeapProfiler::GetMaxHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
 //    HeapProfiler::ResetMaxHeapConsumption(); // FIXME remove
 
     alphabet_mapper * am = new alphabet_mapper_none();
     static_bitsequence_builder * bmb = new static_bitsequence_builder_rrr02(8); // FIXME samplerate?
-//    static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(16); // FIXME samplerate?
     wt_coder * wtc = new wt_coder_binary(bwt,n,am);
     alphabetrank = new static_sequence_wvtree(bwt,n,wtc,bmb,am);
     delete bmb;
     bwt = 0; // already deleted
    
-//    std::cerr << "heap usage: " << HeapProfiler::GetHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
+//    std::cerr << "heap usage after WT: " << HeapProfiler::GetHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
 //    std::cerr << "max heap usage after WT: " << HeapProfiler::GetMaxHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
 }
 
@@ -973,9 +972,7 @@ void TCImplementation::maketables()
 
     // Build up arrays for text length and starting positions
     // FIXME Temp, remove
-    //BlockArray* textLength = new BlockArray(numberOfTexts, Tools::CeilLog2(maxTextLength));
     BlockArray* textStartPos = new BlockArray(numberOfTexts, Tools::CeilLog2(this->n));
-    //(*textLength)[0] = l;
     (*textStartPos)[0] = 0; 
 
     // Construct samples
@@ -1000,9 +997,8 @@ void TCImplementation::maketables()
     ulongmax--;
     uint alphabetrank_i_tmp =0;
 
-    //positions:
-    for (ulong i=n-1;i<ulongmax;i--) { // TODO bad solution with ulongmax?
-      // i substitutes SA->GetPos(i)
+    for (ulong i=n-1;i<ulongmax;i--) {
+        // i substitutes SA->GetPos(i)
         x=(i==n-1)?0:i+1;
 
         if (x % samplerate == 0 && posOfSuccEndmarker - x > samplerate) {
@@ -1024,7 +1020,6 @@ void TCImplementation::maketables()
             // Store text length and text start position:
             if (textId < (DocId)numberOfTexts - 1)
             {
-                //(*textLength)[textId + 1] = posOfSuccEndmarker - x;
                 (*textStartPos)[textId + 1] = x;  // x is the position of end-marker.
                 posOfSuccEndmarker = x;
             }
@@ -1032,7 +1027,7 @@ void TCImplementation::maketables()
             // LF-mapping from '\0' does not work with this (pseudo) BWT (see details from Wolfgang's thesis).
             p = textId; // Correct LF-mapping to the last char of the previous text.
         }
-        else
+        else // Now c != '\0', do LF-mapping:
             p = C[c]+alphabetrank_i_tmp-1;
     }
     assert(textId == 0);
@@ -1060,31 +1055,34 @@ void TCImplementation::maketables()
     // FIXME Temp, remove
     delete tmpSuffix;
     delete positions;
-//    delete textLength;
     delete textStartPos;
 
+    //  std::cerr << "heap usage before Doc: " << HeapProfiler::GetHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
+
 
     uint *tmp = new uint[numberOfTexts]; // FIXME Silly...
-//    cout << "Doc: ";
     for (unsigned i = 0; i < numberOfTexts; ++i)
     {
         tmp[i] = ((*endmarkerDocId)[i] + 1) % numberOfTexts;
-        //      cout << tmp[i] << ", ";
+//        cout << tmp[i] << ",";
     }
-//    cout << endl;
+    cout << endl;
     delete endmarkerDocId;
+
     alphabet_mapper * am = new alphabet_mapper_none();
     static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(16); // FIXME samplerate?
-    wt_coder * wtc = new wt_coder_binary(tmp, numberOfTexts, am);
-    Doc = new static_sequence_wvtree(tmp, numberOfTexts, wtc, bmb, am);
+    Doc = new static_sequence_wvtree_noptrs(tmp, numberOfTexts, bmb, am);
     delete bmb;
     delete [] tmp;
 
-    /*    document_result res = Doc->access(1, 2, 0, 1);
-    cout << "result: ";
+//    std::cerr << "heap usage after Doc: " << HeapProfiler::GetHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
+
+    /*document_result res = Doc->access(3, 3, 0, 3);
+    cout << "Access result: ";
     for (document_result::iterator it = res.begin(); it != res.end(); ++it)
         cout << *it << ", ";
-        cout << endl;*/
+        cout << endl;
+        exit(0);*/
 }
 
 
-- 
2.17.1