Add exact counting to count_unique

Author: fbreitwieser

VERSION

@@ -1 +1 @@
-0.4.9
+0.4.10

src/count_unique.cpp

@@ -43,7 +43,8 @@ int usage(int exit_code) {
 "  -s             Use sparse representation for smaller cardinalities\n"
 "  -t             Test mode - print cardinalities regularily\n"
 "  -y             Show relative error along with cardinality estimates\n"
-"  -e             Use improved cardinality estimator by Otmar Ertl, too\n";
+"  -e             Use improved cardinality estimator by Otmar Ertl, too\n"
+"  -E             Use exact cardinality counting (implemented w/ unordered_set, not working w/ test mode)\n ";
     return exit_code;
   }
 
@@ -119,6 +120,7 @@ int main(int argc, char **argv) {
   bool test_mode = false;
   bool heule_too = true;
   bool ertl_too = false;
+  bool exact_counting = false;
   bool flajolet_too = false;
   bool show_rel_error = false;
   bool use_stdin = true;
@@ -127,11 +129,12 @@ int main(int argc, char **argv) {
 
   int c;
 
-  while ((c = getopt (argc, argv, "shtep:r:yx:f")) != -1)
+  while ((c = getopt (argc, argv, "shtep:r:yx:fE")) != -1)
     switch (c) {
       case 's': sparse = true; break;
       case 't': test_mode = true; break;
       case 'e': ertl_too = true; break;
+      case 'E': exact_counting = true; break;
       case 'f': flajolet_too = true; break;
       case 'y': show_rel_error = true; break;
       case 'p': p = stoi(optarg); break;
@@ -157,7 +160,7 @@ int main(int argc, char **argv) {
   HyperLogLogPlusMinus<uint64_t> hll(p, sparse); // unique k-mer count per taxon
   //HyperLogLogPlusMinus<uint64_t> hll(p, sparse, wang_mixer); // unique k-mer count per taxon
 
-  if (test_mode) {
+  if (test_mode && ! exact_counting) {
     cout << "observed\testimate_heule";
     if (flajolet_too) {
       cout << "\testimate_flajolet";
@@ -178,14 +181,24 @@ int main(int argc, char **argv) {
     cout << '\n';
   } 
   uint64_t ctr = 0;
-  
+  unordered_set<uint64_t> exact_counter;
+
   if (use_stdin) {
     uint64_t nr;
     while (cin >> nr) {
-      add_to_hll(hll, nr, ctr, test_mode, show_rel_error, heule_too, flajolet_too, ertl_too);
+      if (exact_counting) {
+	exact_counter.insert(nr);
+      } else {
+        add_to_hll(hll, nr, ctr, test_mode, show_rel_error, heule_too, flajolet_too, ertl_too);
+      }
+    }
+    if (!test_mode) {
+      if (exact_counting) {
+        cout << exact_counter.size() << "\n";
+      } else {
+        print_card(hll, ctr, show_rel_error, heule_too, flajolet_too, ertl_too);
+      }
     }
-    if (!test_mode) 
-      print_card(hll, ctr, show_rel_error, heule_too, flajolet_too, ertl_too);
   } else {
     // get random seed from random_device RNG
     std::random_device rd;
@@ -197,11 +210,21 @@ int main(int argc, char **argv) {
     for (size_t j = 0; j < n_redo; ++j) {
 
       for(size_t i = 0; i < n_rand; i++) {
-        add_to_hll(hll, distr(rng), ctr, test_mode, show_rel_error, heule_too, flajolet_too, ertl_too);
+        if (exact_counting) {
+	  exact_counter.insert(distr(rng));
+        } else {
+          add_to_hll(hll, distr(rng), ctr, test_mode, show_rel_error, heule_too, flajolet_too, ertl_too);
+	}
+      }
+      if (!test_mode) {
+        if (exact_counting) {
+          cout << exact_counter.size() << "\n";
+        } else {
+          print_card(hll, ctr, show_rel_error, heule_too, flajolet_too, ertl_too);
+        }
       }
-      if (!test_mode) 
-        print_card(hll, ctr, show_rel_error, heule_too, flajolet_too, ertl_too);
       hll.reset();
+      exact_counter.clear();
       ctr = 0;
     }
   }