X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=testTextCollection.cpp;h=a19d2b20fc2dbddc539d1d1883f37b176469fb81;hb=refs%2Fheads%2Fmaster;hp=d005b8f489e607b26db23ebdd8351f3688ae1d14;hpb=7d27a4450ed429e3b63e9d3ba7217a28cbbf9a31;p=SXSI%2FTextCollection.git

diff --git a/testTextCollection.cpp b/testTextCollection.cpp
index d005b8f..a19d2b2 100644
--- a/testTextCollection.cpp
+++ b/testTextCollection.cpp
@@ -1,9 +1,15 @@
 // Test driver for text collection
 #include <iostream>
+#include <cstdlib>
+#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)
@@ -28,42 +34,75 @@ void printFullResult(TextCollection::full_result fr)
 
 int main()
 {
-    uchar *text = (uchar*) "acabab";
-    TextCollection *csa = TextCollection::InitTextCollection(1);
-    csa->InsertText(text);
-    text = (uchar*) "abaca";
-    csa->InsertText(text);
-    text = (uchar*) "abacb";
-    csa->InsertText(text);
+  std::string kbd;
+  string str;
+  uchar* data;
+  int i = 0 ,j = 0;
+  int heap_base = HeapProfiler::GetHeapConsumption();
+  std::cerr << "Initial heap usage : " << heap_base << "\n";
+  TextCollectionBuilder *tcb = TextCollectionBuilder::create(5);
+  heap_base = HeapProfiler::GetHeapConsumption ();
+  std::cerr << "Heap usage after InitTextCollection : " << heap_base << "\n";
+  Tools::StartTimer();
 
-    csa->MakeStatic();
+  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";
+      
+    };
     
-    text = csa->GetText(0);
-    cout << "Text 0: \"" << text << "\"" << endl;
-    delete [] text;
-    text = csa->GetText(1);
-    cout << "Text 1: \"" << text << "\"" << endl;
-    delete [] text;
-    text = csa->GetText(2);
-    cout << "Text 2: \"" << text << "\"" << endl;
-    delete [] text;
-    
-    text = csa->GetText(2, 2, 4);
-    cout << "Substring of Text 3: \"" << text << "\"" << endl;
-    delete [] text;
-    
-    printf("n:o contains: %u\n", csa->CountContains((uchar *)"ac"));
-    printf("n:o suffix: %u\n", csa->CountSuffix((uchar *)"b"));
-    printf("n:o equal: %u\n", csa->CountEqual((uchar *)"acabab"));
-    printf("is equal: %u\n", csa->IsEqual((uchar *)"abacb"));
-    
-    TextCollection::document_result dr;
-    dr = csa->Contains((uchar*)"ab");
-    printDocumentResult(dr);
+    };
+/**/
+  //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::full_result fr;
-    fr = csa->FullContains((uchar *)"ab");
-    printFullResult(fr);
+  if (it > temp + n)
+      std::cout << "over bounds" << std::endl;
+  delete [] temp;*/
+  
 
-    delete csa;
+  std::cerr << "Creating new text collection with " << i << " strings (total " << j/1024 << " kb)\n"; 
+  std::cerr << "max heap usage: " << HeapProfiler::GetMaxHeapConsumption()/(1024*1024) << " Mbytes" << std::endl;
+  HeapProfiler::ResetMaxHeapConsumption();
+  std::cerr << "Before MakeStatic() [press enter]\n";
+  //std::cin >> kbd;
+  // This will print the maximum mem usage during construction time:
+  TextCollection* tc = tcb->InitTextCollection();//SXSI::TextStorage::TYPE_LZ_INDEX);
+  delete tcb; tcb = 0;
+  std::cerr << "After MakeStatic() [press enter]\n";
+  // 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 tc;
+  std::cerr << "After Delete [press enter]\n";
+  std::cerr << "heap usage: " << HeapProfiler::GetHeapConsumption() << " bytes" << std::endl;
+  //std::cin >> kbd;
+  return 0;
 }