Adding support for logical operators (overloading) plus a potential overflow fix.

Author: lemire

headers/ewah.h

@@ -189,6 +189,9 @@ template <class uword = uint32_t> class EWAHBoolArray {
     logicaland(a, answer);
     return answer;
   }
+  EWAHBoolArray operator&(const EWAHBoolArray &a) const {
+    return logicaland(a);
+  }
 
   /**
    * computes the logical and with another compressed bitmap
@@ -199,6 +202,9 @@ template <class uword = uint32_t> class EWAHBoolArray {
    */
   void logicalandnot(const EWAHBoolArray &a, EWAHBoolArray &container) const;
 
+  EWAHBoolArray operator-(const EWAHBoolArray &a) const {
+    return logicalandnot(a);
+  }
 
   /**
    * computes the logical and not with another compressed bitmap
@@ -211,6 +217,8 @@ template <class uword = uint32_t> class EWAHBoolArray {
     logicalandnot(a, answer);
     return answer;
   }
+
+
   /**
    * tests whether the bitmaps "intersect" (have at least one 1-bit at the same
    * position). This function does not modify the existing bitmaps.
@@ -229,6 +237,7 @@ template <class uword = uint32_t> class EWAHBoolArray {
   void logicalor(const EWAHBoolArray &a, EWAHBoolArray &container) const;
 
 
+
   /**
    * computes the size (in number of set bits) of the logical or with another compressed bitmap
    * Running time complexity is proportional to the sum of the compressed
@@ -275,6 +284,10 @@ template <class uword = uint32_t> class EWAHBoolArray {
     return answer;
   }
 
+  EWAHBoolArray operator|(const EWAHBoolArray &a) const {
+    return logicalor(a);
+  }
+
   /**
    * computes the logical xor with another compressed bitmap
    * answer goes into container
@@ -294,6 +307,10 @@ template <class uword = uint32_t> class EWAHBoolArray {
     logicalxor(a, answer);
     return answer;
   }
+
+  EWAHBoolArray operator^(const EWAHBoolArray &a) const {
+    return logicalxor(a);
+  }
   /**
    * clear the content of the bitmap. It does not
    * release the memory.
@@ -1502,7 +1519,7 @@ size_t EWAHBoolArray<uword>::addStreamOfDirtyWords(const uword *v,
   buffer.push_back(0);
   lastRLW = buffer.size() - 1;
   ++wordadded;
-  wordadded += addStreamOfDirtyWords(v + howmanywecanadd, NumberOfLiteralWords - howmanywecanadd);
+  wordadded += addStreamOfDirtyWords(v + howmanywecanadd, number - howmanywecanadd);
   return wordadded;
 }
 
@@ -1528,7 +1545,7 @@ void EWAHBoolArray<uword>::fastaddStreamOfDirtyWords(const uword *v,
   //buffer.insert(buffer.end(), v, v+howmanywecanadd);// seems slower than push_back?
   buffer.push_back(0);
   lastRLW = buffer.size() - 1;
-  fastaddStreamOfDirtyWords(v + howmanywecanadd, NumberOfLiteralWords - howmanywecanadd);
+  fastaddStreamOfDirtyWords(v + howmanywecanadd, number - howmanywecanadd);
 }
 
 
@@ -1558,7 +1575,7 @@ size_t EWAHBoolArray<uword>::addStreamOfNegatedDirtyWords(const uword *v,
   buffer.push_back(0);
   lastRLW = buffer.size() - 1;
   ++wordadded;
-  wordadded += addStreamOfDirtyWords(v + howmanywecanadd, NumberOfLiteralWords - howmanywecanadd);
+  wordadded += addStreamOfDirtyWords(v + howmanywecanadd, number - howmanywecanadd);
   return wordadded;
 }
 

src/unit.cpp

@@ -183,6 +183,55 @@ template <class uword> bool testCardinalityEWAHBoolArray() {
   return true;
 }
 
+template <class uword> bool testLargeDirty() {
+  cout << "[testing LargeDirty] sizeof(uword)=" << sizeof(uword)
+       << endl;
+  size_t N = 80000000;
+  vector<uint32_t> bigarray1(N);
+  vector<uint32_t> bigarray2(N);
+  for(size_t k = 0; k < N; k++) {
+    bigarray1[k] = (uint32_t)(8 * k);
+    bigarray2[k] = (uint32_t)(32 * k);
+  }
+  EWAHBoolArray<uword> b1 = EWAHBoolArray<uword>();
+  EWAHBoolArray<uword> b2 = EWAHBoolArray<uword>();
+  for(size_t k = 0; k < N; k++) {
+    b1.set(8 * k);
+    b2.set(64 * k);
+  }
+
+  EWAHBoolArray<uword> b3 = EWAHBoolArray<uword>::bitmapOf(3,1,10,1000);
+  EWAHBoolArray<uword> b4 = EWAHBoolArray<uword>::bitmapOf(3,1,10,1000);
+
+  if (b1.numberOfOnes() != N) {
+    return false;
+  }
+  if (b2.numberOfOnes() != N) {
+    return false;
+  }
+  if (b3.numberOfOnes() != 3) {
+    return false;
+  }
+  b3 = b3 | b1;
+  b4 = b4 | b2;
+  b3 = b3 | b2;
+  b4 = b4 | b1;
+  if( b3 != b4 ) {
+    return false;
+  }
+  b4 = b4 - b1;
+  if (b4.numberOfOnes() != 3) {
+    return false;
+  }
+  b3 = b3 ^ b4;
+  if( b3 != b1 ) {
+    return false;
+  }
+
+  return true;
+}
+
+
 template <class uword> bool testAndNotEWAHBoolArray() {
   cout << "[testing AndNotEWAHBoolArray] sizeof(uword)=" << sizeof(uword) << endl;
   EWAHBoolArray<uword> b1 = EWAHBoolArray<uword>::bitmapOf(1,1);
@@ -1423,6 +1472,13 @@ int main(void) {
   if (!funnytest()) {
     ++failures;
   }
+  if (!testLargeDirty<uint64_t>())
+    ++failures;
+  if (!testLargeDirty<uint16_t>())
+    ++failures;
+  if (!testLargeDirty<uint32_t>())
+    ++failures;
+
   if (!countstest4<uint64_t>())
     ++failures;
   if (!countstest4<uint16_t>())