X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=testTextCollection.cpp;h=166c019279ad90e0f949affddfb3ecc11b830543;hb=b0678a6ba87a1817502e3e3f733bb8860390a445;hp=dd37ff11ca02e3485cc1426c7aa503284a99c048;hpb=eb04a05de55af8957aea1353d21b5e5b98bd1006;p=SXSI%2FTextCollection.git

diff --git a/testTextCollection.cpp b/testTextCollection.cpp
index dd37ff1..166c019 100644
--- a/testTextCollection.cpp
+++ b/testTextCollection.cpp
@@ -1,11 +1,15 @@
 // Test driver for text collection
 #include <iostream>
 #include <cstdlib>
-#include "HeapProfiler.h"
+#include <string>
 using std::cout;
 using std::endl;
+using std::cin;
+using std::string;
 
-#include "TextCollection.h"
+#include "TextCollectionBuilder.h"
+#include "HeapProfiler.h"
+using SXSI::TextCollectionBuilder;
 using SXSI::TextCollection;
 
 void printDocumentResult(TextCollection::document_result dr)
@@ -31,22 +35,74 @@ void printFullResult(TextCollection::full_result fr)
 int main()
 {
   std::string kbd;
-  uchar * str = (uchar*) "";
+  string str;
+  uchar* data;
+  int i = 0 ,j = 0;
+  int heap_base = HeapProfiler::GetHeapConsumption();
+  std::cerr << "Initial heap usage : " << heap_base << "\n";
+  TextCollectionBuilder *tcb = new TextCollectionBuilder(32);
+  heap_base = HeapProfiler::GetHeapConsumption ();
+  std::cerr << "Heap usage after InitTextCollection : " << heap_base << "\n";
+  Tools::StartTimer();
+
+  while (not(cin.eof())){
+      getline(cin,str); // Read line by line.
+//    cin >> str; // Read word by word.
+    data = (uchar *) str.c_str();
+    if (str.size() == 0)
+        continue;
+       
+    tcb->InsertText(data);    
+    i++;
+    j+= str.size();
+    str.clear();
+    if ( i % 100000 == 0)  {
+      std::cerr << "Inserted : " << i << " strings\n";
+      std::cerr << "Number of bytes inserted : " << j << "b \n";
+      std::cerr << "Heap usage used for strings: " << HeapProfiler::GetHeapConsumption() - heap_base
+		<< "bytes\n";
+      std::cerr << "Ratio is : " << (float) (HeapProfiler::GetHeapConsumption() - heap_base) / ((float) j) <<"\n";
+      
+    };
+    
+    };
+/**/
+  //the whole file as 20 strings:
+  /* uchar *temp = Tools::GetFileContents("data/english.100MB", 0);
+  ulong n = strlen((char *)temp);
+  std::cout << "n = " << n << std::endl;
+  ulong offset = n/40;
+  uchar *it = temp;
+  for (i = 0; i < 5; ++i)
+  {
+      it[offset] = '\0';
+      tcb->InsertText(it);
+      std::cout << "inserted " << strlen((char *)it) << " bytes." << std::endl;
+      it += offset +1;
+  }
+  it -= offset+1;
 
-  TextCollection *csa = TextCollection::InitTextCollection(64);
-  for(int i = 0; i<1000000; i++){
-    csa->InsertText(str);    
-  };
+  if (it > temp + n)
+      std::cout << "over bounds" << std::endl;
+  delete [] temp;*/
+  
+  HeapProfiler::ResetMaxHeapConsumption();
+
+  std::cerr << "Creating new text collection with " << i << " strings (total " << j/1024 << " kb)\n"; 
   std::cerr << "Before MakeStatic() [press enter]\n";
-  std::cin >> kbd;
-  std::cerr << "heap usage: " << HeapProfiler::GetHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
-  csa->MakeStatic();
+  //std::cin >> kbd;
+  // This will print the maximum mem usage during construction time:
+  std::cerr << "max heap usage: " << HeapProfiler::GetMaxHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
+  TextCollection* tc = tcb->InitTextCollection();
+  delete tcb; tcb = 0;
   std::cerr << "After MakeStatic() [press enter]\n";
-  std::cin >> kbd;
+  // This will print the maximum mem usage during MakeStatic():
+  std::cerr << "max heap usage: " << HeapProfiler::GetMaxHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
+  //std::cin >> kbd;
   std::cerr << "heap usage: " << HeapProfiler::GetHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
-  delete csa;
+  delete tc;
   std::cerr << "After Delete [press enter]\n";
-  std::cerr << "heap usage: " << HeapProfiler::GetHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
-  std::cin >> kbd;
+  std::cerr << "heap usage: " << HeapProfiler::GetHeapConsumption() << " bytes" << std::endl;
+  //std::cin >> kbd;
   return 0;
 }