check if union gadget needs kxq rebuilding before getting estimates/bounds

Author: jmalkin

hll/include/HllArray-internal.hpp

@@ -594,6 +594,42 @@ bool HllArray<A>::isRebuildKxqCurminFlag() const {
   return rebuild_kxq_curmin_;
 }
 
+template<typename A>
+void HllArray<A>::check_rebuild_kxq_cur_min() {
+  if (!rebuild_kxq_curmin_) { return; }
+
+  uint8_t cur_min = 64;
+  uint32_t num_at_cur_min = 0;
+  double kxq0 = 1 << this->lgConfigK_;
+  double kxq1 = 0;
+
+  auto it = this->begin(true); // want all points to adjust cur_min
+  const auto end = this->end();
+  while (it != end) {
+    uint8_t v = HllUtil<A>::getValue(*it);
+    if (v > 0) {
+      if (v < 32) { kxq0 += INVERSE_POWERS_OF_2[v] - 1.0; }
+      else        { kxq1 += INVERSE_POWERS_OF_2[v] - 1.0; }
+    }
+    if (v > cur_min) { ++it; continue; }
+    if (v < cur_min) {
+      cur_min = v;
+      num_at_cur_min = 1;
+    } else {
+      ++num_at_cur_min;
+    }    
+    ++it;
+  }
+
+  kxq0_ = kxq0;
+  kxq1_ = kxq1;
+  curMin_ = cur_min;
+  numAtCurMin_ = num_at_cur_min;
+  rebuild_kxq_curmin_ = false;
+  // HipAccum is not affected
+
+}
+
 template<typename A>
 typename HllArray<A>::const_iterator HllArray<A>::begin(bool all) const {
   return const_iterator(hllByteArr_.data(), 1 << this->lgConfigK_, 0, this->tgtHllType_, nullptr, 0, all);

hll/include/HllArray.hpp

@@ -87,8 +87,9 @@ class HllArray : public HllSketchImpl<A> {
 
     virtual AuxHashMap<A>* getAuxHashMap() const;
 
-    virtual void setRebuildKxqCurminFlag(bool rebuild);
-    virtual bool isRebuildKxqCurminFlag() const;
+    void setRebuildKxqCurminFlag(bool rebuild);
+    bool isRebuildKxqCurminFlag() const;
+    void check_rebuild_kxq_cur_min();
 
     class const_iterator;
     virtual const_iterator begin(bool all = false) const;

hll/include/HllUnion-internal.hpp

@@ -131,21 +131,29 @@ void hll_union_alloc<A>::coupon_update(uint32_t coupon) {
 
 template<typename A>
 double hll_union_alloc<A>::get_estimate() const {
+  if (gadget_.sketch_impl->getCurMode() == hll_mode::HLL)
+    static_cast<HllArray<A>*>(gadget_.sketch_impl)->check_rebuild_kxq_cur_min();
   return gadget_.get_estimate();
 }
 
 template<typename A>
 double hll_union_alloc<A>::get_composite_estimate() const {
+  if (gadget_.sketch_impl->getCurMode() == hll_mode::HLL)
+    static_cast<HllArray<A>*>(gadget_.sketch_impl)->check_rebuild_kxq_cur_min();
   return gadget_.get_composite_estimate();
 }
 
 template<typename A>
 double hll_union_alloc<A>::get_lower_bound(uint8_t num_std_dev) const {
+  if (gadget_.sketch_impl->getCurMode() == hll_mode::HLL)
+    static_cast<HllArray<A>*>(gadget_.sketch_impl)->check_rebuild_kxq_cur_min();
   return gadget_.get_lower_bound(num_std_dev);
 }
 
 template<typename A>
 double hll_union_alloc<A>::get_upper_bound(uint8_t num_std_dev) const {
+  if (gadget_.sketch_impl->getCurMode() == hll_mode::HLL)
+    static_cast<HllArray<A>*>(gadget_.sketch_impl)->check_rebuild_kxq_cur_min();
   return gadget_.get_upper_bound(num_std_dev);
 }