Added missing #include <cstring> for std::strlen and std::memset

[SXSI/TextCollection.git] / timeTextCollection.cpp

diff --git a/timeTextCollection.cpp b/timeTextCollection.cpp
index dbe1664..47d1976 100644 (file)
--- a/timeTextCollection.cpp
+++ b/timeTextCollection.cpp
@@ -9,9 +9,18 @@ using std::string;
 #include <sys/time.h>
 #include <time.h>
 
+static struct timeval t1;
+static struct timeval t2;
 
-#include "TextCollection.h"
+#define STARTTIMER() (gettimeofday(&t1,NULL))
+#define STOPTIMER()  (gettimeofday(&t2,NULL))
+#define GETTIME()    (((t2.tv_sec  - t1.tv_sec) * 1000000.0 + (t2.tv_usec  - t1.tv_usec))/1000.0)
+                      
+
+
+#include "TextCollectionBuilder.h"
 using SXSI::TextCollection;
+using SXSI::TextCollectionBuilder;
 
 int main(int argc, char**argv)
 {
@@ -19,8 +28,6 @@ int main(int argc, char**argv)
   string buffer;
   unsigned int max_str = 0;
   unsigned int num_str = 0;
-  struct timeval t1;
-  struct timeval t2;
   double time;
 
   string words[] =  { "abcd", "abc", "mirrors", "attires", "mature",
@@ -29,23 +36,23 @@ int main(int argc, char**argv)
   
   
   
-  TextCollection *csa = TextCollection::InitTextCollection(64);
+  TextCollectionBuilder *tcb = new TextCollectionBuilder(64);
 
 
-  gettimeofday(&t1,NULL);
+  STARTTIMER();
   std::cerr << "Filling collection\n";
+  // read only 100000 strings 
   while (not(cin.eof()) && num_str < 100000 ){
       getline(cin,str); // Read line by line.
       if (str.compare("----------") == 0){
-       csa->InsertText((unsigned char*) buffer.c_str());
+       tcb->InsertText((unsigned char*) buffer.c_str());
 
        if (num_str % 10000 == 0){
-               gettimeofday(&t2,NULL);
-               time = ((t2.tv_sec  - t1.tv_sec) * 1000000.0 
-                       + (t2.tv_usec  - t1.tv_usec))/1000.0;
-               std::cerr << "Added " << num_str << " strings in "
-                         << time  << " ms\n";
-               gettimeofday(&t1,NULL);
+         STOPTIMER();
+         time = GETTIME();
+         std::cerr << "Added " << num_str << " strings in "
+                   << time  << " ms\n";
+         STARTTIMER();
        };
 
        num_str++;
@@ -58,47 +65,46 @@ int main(int argc, char**argv)
        buffer.append(str);
   };
   std::cerr << "Calling MakeStatic()\n";
-  csa->MakeStatic();
+  
+  TextCollection *tc = tcb->InitTextCollection();
+
   std::cerr << "Statistics: " << num_str << " strings, " << max_str << " = max length\n";
   int count;
   bool is;
   TextCollection::document_result res;
-  for (int i = 0; i < 14; i++){
-    gettimeofday(&t1,NULL);
-    is = csa->IsContains((unsigned char*) words[i].c_str());
-    gettimeofday(&t2,NULL);
-    time = ((t2.tv_sec  - t1.tv_sec) * 1000000.0 
-           + (t2.tv_usec  - t1.tv_usec))/1000.0;
+  for (unsigned int i = 0; i < (sizeof(words)/sizeof(char*)) ; i++){
+  
+    STARTTIMER();
+    is = tc->IsContains((unsigned char*) words[i].c_str());
+    STOPTIMER();
+    time = GETTIME();
 
     std::cerr << is << ", " << time << ", ";
 
 
-    gettimeofday(&t1,NULL);
-    count = csa->Count((unsigned char*) words[i].c_str());
-    gettimeofday(&t2,NULL);
-    time = ((t2.tv_sec  - t1.tv_sec) * 1000000.0 
-           + (t2.tv_usec  - t1.tv_usec))/1000.0;
+    STARTTIMER();
+    count = tc->Count((unsigned char*) words[i].c_str());
+    STOPTIMER();
+    time = GETTIME();
 
     std::cerr << count << ", " << time << ", ";
 
-    gettimeofday(&t1,NULL);
-    count = csa->CountContains((unsigned char*) words[i].c_str());
-    gettimeofday(&t2,NULL);
-    time = ((t2.tv_sec  - t1.tv_sec) * 1000000.0 
-           + (t2.tv_usec  - t1.tv_usec))/1000.0;
+
+    STARTTIMER();
+    count = tc->CountContains((unsigned char*) words[i].c_str());
+    STOPTIMER();
+    time = GETTIME();
 
     std::cerr << count << ", " << time << ", ";
     
 
-    gettimeofday(&t1,NULL);
-    res = csa->Contains((unsigned char*) words[i].c_str());
-    gettimeofday(&t2,NULL);
-    time = ((t2.tv_sec  - t1.tv_sec) * 1000000.0 
-           + (t2.tv_usec  - t1.tv_usec))/1000.0;
+    STARTTIMER();
+    res = tc->Contains((unsigned char*) words[i].c_str());
+    STOPTIMER();
+    time = GETTIME();
     
     std::cerr << time << "\n";
     
-
     
   };