Add capacity() getter to ApproxMostFrequentStreamSummary (facebookinc…

…ubator#12212)

Summary:
Pull Request resolved: facebookincubator#12212

# Context

This diff adds a simple getter to retrieve the capacity parameter (after we hit setCapacity()).

We use this internally to perform some sanity checking when we have potentially two different SteamSummaries and we want to ensure capacities are in a sane spot.

This seems reasonably safe to add, the capacity is set by the user and having something to read back out the set value doesn't seem too dangerous. It's also good for sanity checks and unit tests - e.g. as deserializing a stream summary gives back 0 capacity.

Reviewed By: Yuhta

Differential Revision: D68870501

fbshipit-source-id: d7b67eb9f7f375b3a046a94dd3cde00ed82006e8

velox/functions/lib/ApproxMostFrequentStreamSummary.h

@@ -68,6 +68,9 @@ struct ApproxMostFrequentStreamSummary {
   /// Return the number of distinct values currently being tracked.
   int size() const;
 
+  // Return the maximum number of distinct values that can be tracked.
+  int capacity() const;
+
   /// Return the pointer to values data.  The number of values equals to size().
   const T* values() const {
     return queue_.values();
@@ -107,6 +110,11 @@ int ApproxMostFrequentStreamSummary<T, A>::size() const {
   return queue_.size();
 }
 
+template <typename T, typename A>
+int ApproxMostFrequentStreamSummary<T, A>::capacity() const {
+  return capacity_;
+}
+
 template <typename T, typename A>
 void ApproxMostFrequentStreamSummary<T, A>::insert(T value, int64_t count) {
   auto index = queue_.getValueIndex(value);

velox/functions/lib/tests/ApproxMostFrequentStreamSummaryTest.cpp

@@ -184,5 +184,17 @@ TEST(ApproxMostFrequentStreamSummaryTest, mergeSerialized) {
   }
 }
 
+TEST(ApproxMostFrequentStreamSummaryTest, capacity) {
+  constexpr int kCapacity = 30;
+  ApproxMostFrequentStreamSummary<int> summary;
+  summary.setCapacity(kCapacity);
+  ASSERT_EQ(summary.capacity(), 30);
+}
+
+TEST(ApproxMostFrequentStreamSummaryTest, unsetCapacity) {
+  ApproxMostFrequentStreamSummary<int> summary;
+  ASSERT_EQ(summary.capacity(), 0);
+}
+
 } // namespace
 } // namespace facebook::velox::functions